ft-ops.cc 235 KB
Newer Older
1 2
/* -*- mode: C++; c-basic-offset: 4; indent-tabs-mode: nil -*- */
// vim: ft=cpp:expandtab:ts=8:sw=4:softtabstop=4:
Bradley C. Kuszmaul's avatar
Bradley C. Kuszmaul committed
3
#ident "$Id$"
4
#ident "Copyright (c) 2007-2012 Tokutek Inc.  All rights reserved."
5
#ident "The technology is licensed by the Massachusetts Institute of Technology, Rutgers State University of New Jersey, and the Research Foundation of State University of New York at Stony Brook under United States of America Serial No. 11/760379 and to the patents and/or patent applications resulting from it."
6

7 8
/*

9
Managing the tree shape:  How insertion, deletion, and querying work
10

11
When we insert a message into the FT_HANDLE, here's what happens.
12

13
to insert a message at the root
14

15 16 17 18
    - find the root node
    - capture the next msn of the root node and assign it to the message
    - split the root if it needs to be split
    - insert the message into the root buffer
John Esmet's avatar
John Esmet committed
19
    - if the root is too full, then toku_ft_flush_some_child() of the root on a flusher thread
20

21
flusher functions use an advice struct with provides some functions to
22
call that tell it what to do based on the context of the flush. see ft-flusher.h
23 24 25 26 27

to flush some child, given a parent and some advice
    - pick the child using advice->pick_child()
    - remove that childs buffer from the parent
    - flush the buffer to the child
28
    - if the child has stable reactivity and
29
      advice->should_recursively_flush() is true, then
John Esmet's avatar
John Esmet committed
30
      toku_ft_flush_some_child() of the child
31 32 33
    - otherwise split the child if it needs to be split
    - otherwise maybe merge the child if it needs to be merged

34
flusher threads:
35

36 37 38 39 40 41 42 43 44 45 46 47 48 49
    flusher threads are created on demand as the result of internal nodes
    becoming gorged by insertions. this allows flushing to be done somewhere
    other than the client thread. these work items are enqueued onto
    the cachetable kibbutz and are done in a first in first out order.

cleaner threads:

    the cleaner thread wakes up every so often (say, 1 second) and chooses
    a small number (say, 5) of nodes as candidates for a flush. the one
    with the largest cache pressure is chosen to be flushed. cache pressure
    is a function of the size of the node in the cachetable plus the work done.
    the cleaner thread need not actually do a flush when awoken, so only
    nodes that have sufficient cache pressure are flushed.

50
checkpointing:
51 52 53 54 55 56 57 58 59

    the checkpoint thread wakes up every minute to checkpoint dirty nodes
    to disk. at the time of this writing, nodes during checkpoint are
    locked and cannot be queried or flushed to. a design in which nodes
    are copied before checkpoint is being considered as a way to reduce
    the performance variability caused by a checkpoint locking too
    many nodes and preventing other threads from traversing down the tree,
    for a query or otherwise.

60
To shrink a file: Let X be the size of the reachable data.
61 62 63
    We define an acceptable bloat constant of C.  For example we set C=2 if we are willing to allow the file to be as much as 2X in size.
    The goal is to find the smallest amount of stuff we can move to get the file down to size CX.
    That seems like a difficult problem, so we use the following heuristics:
64 65 66 67 68 69 70 71
       If we can relocate the last block to an lower location, then do so immediately.        (The file gets smaller right away, so even though the new location
         may even not be in the first CX bytes, we are making the file smaller.)
       Otherwise all of the earlier blocks are smaller than the last block (of size L).         So find the smallest region that has L free bytes in it.
         (This can be computed in one pass)
         Move the first allocated block in that region to some location not in the interior of the region.
               (Outside of the region is OK, and reallocating the block at the edge of the region is OK).
            This has the effect of creating a smaller region with at least L free bytes in it.
         Go back to the top (because by now some other block may have been allocated or freed).
72 73 74 75 76 77 78 79 80
    Claim: if there are no other allocations going on concurrently, then this algorithm will shrink the file reasonably efficiently.  By this I mean that
       each block of shrinkage does the smallest amount of work possible.  That doesn't mean that the work overall is minimized.
    Note: If there are other allocations and deallocations going on concurrently, we might never get enough space to move the last block.  But it takes a lot
      of allocations and deallocations to make that happen, and it's probably reasonable for the file not to shrink in this case.

To split or merge a child of a node:
Split_or_merge (node, childnum) {
  If the child needs to be split (it's a leaf with too much stuff or a nonleaf with too much fanout)
    fetch the node and the child into main memory.
81
    split the child, producing two nodes A and B, and also a pivot.   Don't worry if the resulting child is still too big or too small.         Fix it on the next pass.
82
    fixup node to point at the two new children.  Don't worry about the node getting too much fanout.
83
    return;
84 85 86 87
  If the child needs to be merged (it's a leaf with too little stuff (less than 1/4 full) or a nonleaf with too little fanout (less than 1/4)
    fetch node, the child  and a sibling of the child into main memory.
    move all messages from the node to the two children (so that the FIFOs are empty)
    If the two siblings together fit into one node then
88
      merge the two siblings.
89 90 91
      fixup the node to point at one child
    Otherwise
      load balance the content of the two nodes
92
    Don't worry about the resulting children having too many messages or otherwise being too big or too small.        Fix it on the next pass.
93 94
  }
}
95

96 97
Here's how querying works:

98 99
lookups:
    - As of Dr. No, we don't do any tree shaping on lookup.
100
    - We don't promote eagerly or use aggressive promotion or passive-aggressive
101
    promotion.        We just push messages down according to the traditional FT_HANDLE
102 103
    algorithm on insertions.
    - when a node is brought into memory, we apply ancestor messages above it.
104 105 106 107

basement nodes, bulk fetch,  and partial fetch:
    - leaf nodes are comprised of N basement nodes, each of nominal size. when
    a query hits a leaf node. it may require one or more basement nodes to be in memory.
108
    - for point queries, we do not read the entire node into memory. instead,
109 110
      we only read in the required basement node
    - for range queries, cursors may return cursor continue in their callback
111 112 113 114
      to take a the shortcut path until the end of the basement node.
    - for range queries, cursors may prelock a range of keys (with or without a txn).
      the fractal tree will prefetch nodes aggressively until the end of the range.
    - without a prelocked range, range queries behave like successive point queries.
115 116

*/
117

118
#include "checkpoint.h"
119 120 121 122 123 124 125 126 127
#include "ft.h"
#include "ft-cachetable-wrappers.h"
#include "ft-flusher.h"
#include "ft-internal.h"
#include "ft_layout_version.h"
#include "key.h"
#include "log-internal.h"
#include "sub_block.h"
#include "txn_manager.h"
John Esmet's avatar
John Esmet committed
128
#include "leafentry.h"
129
#include "xids.h"
130

131
#include <toku_race_tools.h>
John Esmet's avatar
John Esmet committed
132

133
#include <portability/toku_atomic.h>
John Esmet's avatar
John Esmet committed
134

135 136 137 138 139 140
#include <util/mempool.h>
#include <util/partitioned_counter.h>
#include <util/rwlock.h>
#include <util/sort.h>

#include <stdint.h>
141

142
static const uint32_t this_version = FT_LAYOUT_VERSION;
143

144 145 146
/* Status is intended for display to humans to help understand system behavior.
 * It does not need to be perfectly thread-safe.
 */
147
static FT_STATUS_S ft_status;
148

149 150 151 152 153 154 155 156
#define STATUS_INIT(k,t,l) do {                                         \
        ft_status.status[k].keyname = #k;                               \
        ft_status.status[k].type    = t;                                \
        ft_status.status[k].legend  = "brt: " l;                        \
        if (t == PARCOUNT) {                                            \
            ft_status.status[k].value.parcount = create_partitioned_counter(); \
        }                                                               \
    } while (0)
157

158 159
static toku_mutex_t ft_open_close_lock;

160 161 162 163 164
static void
status_init(void)
{
    // Note, this function initializes the keyname, type, and legend fields.
    // Value fields are initialized to zero by compiler.
165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180
    STATUS_INIT(FT_UPDATES,                                PARCOUNT, "dictionary updates");
    STATUS_INIT(FT_UPDATES_BROADCAST,                      PARCOUNT, "dictionary broadcast updates");
    STATUS_INIT(FT_DESCRIPTOR_SET,                         PARCOUNT, "descriptor set");
    STATUS_INIT(FT_MSN_DISCARDS,                           PARCOUNT, "messages ignored by leaf due to msn");
    STATUS_INIT(FT_TOTAL_RETRIES,                          PARCOUNT, "total search retries due to TRY_AGAIN");
    STATUS_INIT(FT_SEARCH_TRIES_GT_HEIGHT,                 PARCOUNT, "searches requiring more tries than the height of the tree");
    STATUS_INIT(FT_SEARCH_TRIES_GT_HEIGHTPLUS3,            PARCOUNT, "searches requiring more tries than the height of the tree plus three");
    STATUS_INIT(FT_CREATE_LEAF,                            PARCOUNT, "leaf nodes created");
    STATUS_INIT(FT_CREATE_NONLEAF,                         PARCOUNT, "nonleaf nodes created");
    STATUS_INIT(FT_DESTROY_LEAF,                           PARCOUNT, "leaf nodes destroyed");
    STATUS_INIT(FT_DESTROY_NONLEAF,                        PARCOUNT, "nonleaf nodes destroyed");
    STATUS_INIT(FT_MSG_BYTES_IN,                           PARCOUNT, "bytes of messages injected at root (all trees)");
    STATUS_INIT(FT_MSG_BYTES_OUT,                          PARCOUNT, "bytes of messages flushed from h1 nodes to leaves");
    STATUS_INIT(FT_MSG_BYTES_CURR,                         PARCOUNT, "bytes of messages currently in trees (estimate)");
    STATUS_INIT(FT_MSG_NUM,                                PARCOUNT, "messages injected at root");
    STATUS_INIT(FT_MSG_NUM_BROADCAST,                      PARCOUNT, "broadcast messages injected at root");
181

182 183 184 185 186 187 188 189
    STATUS_INIT(FT_NUM_BASEMENTS_DECOMPRESSED_NORMAL,      PARCOUNT, "basements decompressed as a target of a query");
    STATUS_INIT(FT_NUM_BASEMENTS_DECOMPRESSED_AGGRESSIVE,  PARCOUNT, "basements decompressed for prelocked range");
    STATUS_INIT(FT_NUM_BASEMENTS_DECOMPRESSED_PREFETCH,    PARCOUNT, "basements decompressed for prefetch");
    STATUS_INIT(FT_NUM_BASEMENTS_DECOMPRESSED_WRITE,       PARCOUNT, "basements decompressed for write");
    STATUS_INIT(FT_NUM_MSG_BUFFER_DECOMPRESSED_NORMAL,     PARCOUNT, "buffers decompressed as a target of a query");
    STATUS_INIT(FT_NUM_MSG_BUFFER_DECOMPRESSED_AGGRESSIVE, PARCOUNT, "buffers decompressed for prelocked range");
    STATUS_INIT(FT_NUM_MSG_BUFFER_DECOMPRESSED_PREFETCH,   PARCOUNT, "buffers decompressed for prefetch");
    STATUS_INIT(FT_NUM_MSG_BUFFER_DECOMPRESSED_WRITE,      PARCOUNT, "buffers decompressed for write");
190

191 192 193 194 195 196 197 198 199 200
    // Eviction statistics:
    STATUS_INIT(FT_FULL_EVICTIONS_LEAF,                    PARCOUNT, "leaf node full evictions");
    STATUS_INIT(FT_FULL_EVICTIONS_LEAF_BYTES,              PARCOUNT, "leaf node full evictions (bytes)");
    STATUS_INIT(FT_FULL_EVICTIONS_NONLEAF,                 PARCOUNT, "nonleaf node full evictions");
    STATUS_INIT(FT_FULL_EVICTIONS_NONLEAF_BYTES,           PARCOUNT, "nonleaf node full evictions (bytes)");
    STATUS_INIT(FT_PARTIAL_EVICTIONS_LEAF,                 PARCOUNT, "leaf node partial evictions");
    STATUS_INIT(FT_PARTIAL_EVICTIONS_LEAF_BYTES,           PARCOUNT, "leaf node partial evictions (bytes)");
    STATUS_INIT(FT_PARTIAL_EVICTIONS_NONLEAF,              PARCOUNT, "nonleaf node partial evictions");
    STATUS_INIT(FT_PARTIAL_EVICTIONS_NONLEAF_BYTES,        PARCOUNT, "nonleaf node partial evictions (bytes)");

201 202 203
    // Disk read statistics: 
    //
    // Pivots: For queries, prefetching, or writing.
204
    STATUS_INIT(FT_NUM_PIVOTS_FETCHED_QUERY,               PARCOUNT, "pivots fetched for query");
205
    STATUS_INIT(FT_BYTES_PIVOTS_FETCHED_QUERY,             PARCOUNT, "pivots fetched for query (bytes)");
206
    STATUS_INIT(FT_TOKUTIME_PIVOTS_FETCHED_QUERY,          TOKUTIME, "pivots fetched for query (seconds)");
207
    STATUS_INIT(FT_NUM_PIVOTS_FETCHED_PREFETCH,            PARCOUNT, "pivots fetched for prefetch");
208
    STATUS_INIT(FT_BYTES_PIVOTS_FETCHED_PREFETCH,          PARCOUNT, "pivots fetched for prefetch (bytes)");
209
    STATUS_INIT(FT_TOKUTIME_PIVOTS_FETCHED_PREFETCH,       TOKUTIME, "pivots fetched for prefetch (seconds)");
210
    STATUS_INIT(FT_NUM_PIVOTS_FETCHED_WRITE,               PARCOUNT, "pivots fetched for write");
211
    STATUS_INIT(FT_BYTES_PIVOTS_FETCHED_WRITE,             PARCOUNT, "pivots fetched for write (bytes)");
212 213
    STATUS_INIT(FT_TOKUTIME_PIVOTS_FETCHED_WRITE,          TOKUTIME, "pivots fetched for write (seconds)");
    // Basements: For queries, aggressive fetching in prelocked range, prefetching, or writing.
214
    STATUS_INIT(FT_NUM_BASEMENTS_FETCHED_NORMAL,           PARCOUNT, "basements fetched as a target of a query");
215
    STATUS_INIT(FT_BYTES_BASEMENTS_FETCHED_NORMAL,         PARCOUNT, "basements fetched as a target of a query (bytes)");
216
    STATUS_INIT(FT_TOKUTIME_BASEMENTS_FETCHED_NORMAL,      TOKUTIME, "basements fetched as a target of a query (seconds)");
217
    STATUS_INIT(FT_NUM_BASEMENTS_FETCHED_AGGRESSIVE,       PARCOUNT, "basements fetched for prelocked range");
218
    STATUS_INIT(FT_BYTES_BASEMENTS_FETCHED_AGGRESSIVE,     PARCOUNT, "basements fetched for prelocked range (bytes)");
219
    STATUS_INIT(FT_TOKUTIME_BASEMENTS_FETCHED_AGGRESSIVE,  TOKUTIME, "basements fetched for prelocked range (seconds)");
220
    STATUS_INIT(FT_NUM_BASEMENTS_FETCHED_PREFETCH,         PARCOUNT, "basements fetched for prefetch");
221
    STATUS_INIT(FT_BYTES_BASEMENTS_FETCHED_PREFETCH,       PARCOUNT, "basements fetched for prefetch (bytes)");
222
    STATUS_INIT(FT_TOKUTIME_BASEMENTS_FETCHED_PREFETCH,    TOKUTIME, "basements fetched for prefetch (seconds)");
223
    STATUS_INIT(FT_NUM_BASEMENTS_FETCHED_WRITE,            PARCOUNT, "basements fetched for write");
224
    STATUS_INIT(FT_BYTES_BASEMENTS_FETCHED_WRITE,          PARCOUNT, "basements fetched for write (bytes)");
225 226
    STATUS_INIT(FT_TOKUTIME_BASEMENTS_FETCHED_WRITE,       TOKUTIME, "basements fetched for write (seconds)");
    // Buffers: For queries, aggressive fetching in prelocked range, prefetching, or writing.
227
    STATUS_INIT(FT_NUM_MSG_BUFFER_FETCHED_NORMAL,          PARCOUNT, "buffers fetched as a target of a query");
228
    STATUS_INIT(FT_BYTES_MSG_BUFFER_FETCHED_NORMAL,        PARCOUNT, "buffers fetched as a target of a query (bytes)");
229
    STATUS_INIT(FT_TOKUTIME_MSG_BUFFER_FETCHED_NORMAL,     TOKUTIME, "buffers fetched as a target of a query (seconds)");
230
    STATUS_INIT(FT_NUM_MSG_BUFFER_FETCHED_AGGRESSIVE,      PARCOUNT, "buffers fetched for prelocked range");
231
    STATUS_INIT(FT_BYTES_MSG_BUFFER_FETCHED_AGGRESSIVE,    PARCOUNT, "buffers fetched for prelocked range (bytes)");
232
    STATUS_INIT(FT_TOKUTIME_MSG_BUFFER_FETCHED_AGGRESSIVE, TOKUTIME, "buffers fetched for prelocked range (seconds)");
233
    STATUS_INIT(FT_NUM_MSG_BUFFER_FETCHED_PREFETCH,        PARCOUNT, "buffers fetched for prefetch");
234
    STATUS_INIT(FT_BYTES_MSG_BUFFER_FETCHED_PREFETCH,      PARCOUNT, "buffers fetched for prefetch (bytes)");
235
    STATUS_INIT(FT_TOKUTIME_MSG_BUFFER_FETCHED_PREFETCH,   TOKUTIME, "buffers fetched for prefetch (seconds)");
236
    STATUS_INIT(FT_NUM_MSG_BUFFER_FETCHED_WRITE,           PARCOUNT, "buffers fetched for write");
237
    STATUS_INIT(FT_BYTES_MSG_BUFFER_FETCHED_WRITE,         PARCOUNT, "buffers fetched for write (bytes)");
238
    STATUS_INIT(FT_TOKUTIME_MSG_BUFFER_FETCHED_WRITE,      TOKUTIME, "buffers fetched for write (seconds)");
239 240
    
    // Disk write statistics.
241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259
    //
    // Leaf/Nonleaf: Not for checkpoint
    STATUS_INIT(FT_DISK_FLUSH_LEAF,                                         PARCOUNT, "leaf nodes flushed to disk (not for checkpoint)");
    STATUS_INIT(FT_DISK_FLUSH_LEAF_BYTES,                                   PARCOUNT, "leaf nodes flushed to disk (not for checkpoint) (bytes)");
    STATUS_INIT(FT_DISK_FLUSH_LEAF_UNCOMPRESSED_BYTES,                      PARCOUNT, "leaf nodes flushed to disk (not for checkpoint) (uncompressed bytes)");
    STATUS_INIT(FT_DISK_FLUSH_LEAF_TOKUTIME,                                TOKUTIME, "leaf nodes flushed to disk (not for checkpoint) (seconds)");
    STATUS_INIT(FT_DISK_FLUSH_NONLEAF,                                      PARCOUNT, "nonleaf nodes flushed to disk (not for checkpoint)");
    STATUS_INIT(FT_DISK_FLUSH_NONLEAF_BYTES,                                PARCOUNT, "nonleaf nodes flushed to disk (not for checkpoint) (bytes)");
    STATUS_INIT(FT_DISK_FLUSH_NONLEAF_UNCOMPRESSED_BYTES,                   PARCOUNT, "nonleaf nodes flushed to disk (not for checkpoint) (uncompressed bytes)");
    STATUS_INIT(FT_DISK_FLUSH_NONLEAF_TOKUTIME,                             TOKUTIME, "nonleaf nodes flushed to disk (not for checkpoint) (seconds)");
    // Leaf/Nonleaf: For checkpoint
    STATUS_INIT(FT_DISK_FLUSH_LEAF_FOR_CHECKPOINT,                          PARCOUNT, "leaf nodes flushed to disk (for checkpoint)");
    STATUS_INIT(FT_DISK_FLUSH_LEAF_BYTES_FOR_CHECKPOINT,                    PARCOUNT, "leaf nodes flushed to disk (for checkpoint) (bytes)");
    STATUS_INIT(FT_DISK_FLUSH_LEAF_UNCOMPRESSED_BYTES_FOR_CHECKPOINT,       PARCOUNT, "leaf nodes flushed to disk (for checkpoint) (uncompressed bytes)");
    STATUS_INIT(FT_DISK_FLUSH_LEAF_TOKUTIME_FOR_CHECKPOINT,                 TOKUTIME, "leaf nodes flushed to disk (for checkpoint) (seconds)");
    STATUS_INIT(FT_DISK_FLUSH_NONLEAF_FOR_CHECKPOINT,                       PARCOUNT, "nonleaf nodes flushed to disk (for checkpoint)");
    STATUS_INIT(FT_DISK_FLUSH_NONLEAF_BYTES_FOR_CHECKPOINT,                 PARCOUNT, "nonleaf nodes flushed to disk (for checkpoint) (bytes)");
    STATUS_INIT(FT_DISK_FLUSH_NONLEAF_UNCOMPRESSED_BYTES_FOR_CHECKPOINT,    PARCOUNT, "nonleaf nodes flushed to disk (for checkpoint) (uncompressed bytes)");
    STATUS_INIT(FT_DISK_FLUSH_NONLEAF_TOKUTIME_FOR_CHECKPOINT,              TOKUTIME, "nonleaf nodes flushed to disk (for checkpoint) (seconds)");
260

261 262 263 264 265 266
    // CPU time statistics for [de]serialization and [de]compression.
    STATUS_INIT(FT_NODE_COMPRESS_TOKUTIME,                             TOKUTIME, "node compression to memory (seconds)");
    STATUS_INIT(FT_NODE_SERIALIZE_TOKUTIME,                            TOKUTIME, "node serialization to memory (seconds)");
    STATUS_INIT(FT_NODE_DECOMPRESS_TOKUTIME,                           TOKUTIME, "node decompression to memory (seconds)");
    STATUS_INIT(FT_NODE_DESERIALIZE_TOKUTIME,                          TOKUTIME, "node deserialization to memory (seconds)");

267
    // Promotion statistics.
268 269 270 271 272 273 274 275 276 277 278 279 280 281
    STATUS_INIT(FT_PRO_NUM_ROOT_SPLIT,                     PARCOUNT, "promotion: roots split");
    STATUS_INIT(FT_PRO_NUM_ROOT_H0_INJECT,                 PARCOUNT, "promotion: leaf roots injected into");
    STATUS_INIT(FT_PRO_NUM_ROOT_H1_INJECT,                 PARCOUNT, "promotion: h1 roots injected into");
    STATUS_INIT(FT_PRO_NUM_INJECT_DEPTH_0,                 PARCOUNT, "promotion: injections at depth 0");
    STATUS_INIT(FT_PRO_NUM_INJECT_DEPTH_1,                 PARCOUNT, "promotion: injections at depth 1");
    STATUS_INIT(FT_PRO_NUM_INJECT_DEPTH_2,                 PARCOUNT, "promotion: injections at depth 2");
    STATUS_INIT(FT_PRO_NUM_INJECT_DEPTH_3,                 PARCOUNT, "promotion: injections at depth 3");
    STATUS_INIT(FT_PRO_NUM_INJECT_DEPTH_GT3,               PARCOUNT, "promotion: injections lower than depth 3");
    STATUS_INIT(FT_PRO_NUM_STOP_NONEMPTY_BUF,              PARCOUNT, "promotion: stopped because of a nonempty buffer");
    STATUS_INIT(FT_PRO_NUM_STOP_H1,                        PARCOUNT, "promotion: stopped at height 1");
    STATUS_INIT(FT_PRO_NUM_STOP_LOCK_CHILD,                PARCOUNT, "promotion: stopped because the child was locked or not at all in memory");
    STATUS_INIT(FT_PRO_NUM_STOP_CHILD_INMEM,               PARCOUNT, "promotion: stopped because the child was not fully in memory");
    STATUS_INIT(FT_PRO_NUM_DIDNT_WANT_PROMOTE,             PARCOUNT, "promotion: stopped anyway, after locking the child");

282
    ft_status.initialized = true;
283
}
284 285 286 287 288 289 290
static void status_destroy(void) {
    for (int i = 0; i < FT_STATUS_NUM_ROWS; ++i) {
        if (ft_status.status[i].type == PARCOUNT) {
            destroy_partitioned_counter(ft_status.status[i].value.parcount);
        }
    }
}
291 292 293
#undef STATUS_INIT

void
294 295
toku_ft_get_status(FT_STATUS s) {
    *s = ft_status;
296 297
}

298 299 300 301 302 303 304
#define STATUS_INC(x, d)                                                            \
    do {                                                                            \
        if (ft_status.status[x].type == PARCOUNT) {                                 \
            increment_partitioned_counter(ft_status.status[x].value.parcount, d);   \
        } else {                                                                    \
            toku_sync_fetch_and_add(&ft_status.status[x].value.num, d);             \
        }                                                                           \
305
    } while (0)
306

307
bool is_entire_node_in_memory(FTNODE node) {
308 309 310 311 312 313 314
    for (int i = 0; i < node->n_children; i++) {
        if(BP_STATE(node,i) != PT_AVAIL) {
            return false;
        }
    }
    return true;
}
315

316
void
317 318
toku_assert_entire_node_in_memory(FTNODE UU() node) {
    paranoid_invariant(is_entire_node_in_memory(node));
319 320
}

Yoni Fogel's avatar
Yoni Fogel committed
321
static uint32_t
322
get_leaf_num_entries(FTNODE node) {
Yoni Fogel's avatar
Yoni Fogel committed
323
    uint32_t result = 0;
324
    int i;
325
    toku_assert_entire_node_in_memory(node);
326
    for ( i = 0; i < node->n_children; i++) {
327
        result += toku_omt_size(BLB_BUFFER(node, i));
328 329 330 331
    }
    return result;
}

332
static enum reactivity
Zardosht Kasheff's avatar
Zardosht Kasheff committed
333
get_leaf_reactivity (FTNODE node, uint32_t nodesize) {
334
    enum reactivity re = RE_STABLE;
Leif Walsh's avatar
Leif Walsh committed
335
    toku_assert_entire_node_in_memory(node);
336
    paranoid_invariant(node->height==0);
337
    unsigned int size = toku_serialize_ftnode_size(node);
Zardosht Kasheff's avatar
Zardosht Kasheff committed
338
    if (size > nodesize && get_leaf_num_entries(node) > 1) {
Leif Walsh's avatar
Leif Walsh committed
339 340
        re = RE_FISSIBLE;
    }
Zardosht Kasheff's avatar
Zardosht Kasheff committed
341
    else if ((size*4) < nodesize && !BLB_SEQINSERT(node, node->n_children-1)) {
Leif Walsh's avatar
Leif Walsh committed
342
        re = RE_FUSIBLE;
343 344
    }
    return re;
345 346
}

347
enum reactivity
348
get_nonleaf_reactivity (FTNODE node) {
349
    paranoid_invariant(node->height>0);
350
    int n_children = node->n_children;
351 352 353 354 355
    if (n_children > TREE_FANOUT) return RE_FISSIBLE;
    if (n_children*4 < TREE_FANOUT) return RE_FUSIBLE;
    return RE_STABLE;
}

356
enum reactivity
Zardosht Kasheff's avatar
Zardosht Kasheff committed
357
get_node_reactivity (FTNODE node, uint32_t nodesize) {
358
    toku_assert_entire_node_in_memory(node);
359
    if (node->height==0)
Zardosht Kasheff's avatar
Zardosht Kasheff committed
360
        return get_leaf_reactivity(node, nodesize);
361
    else
362
        return get_nonleaf_reactivity(node);
363 364
}

365 366 367
unsigned int
toku_bnc_nbytesinbuf(NONLEAF_CHILDINFO bnc)
{
368
    return toku_fifo_buffer_size_in_use(bnc->buffer);
369 370
}

Yoni Fogel's avatar
Yoni Fogel committed
371
// return true if the size of the buffers plus the amount of work done is large enough.   (But return false if there is nothing to be flushed (the buffers empty)).
372
bool
Zardosht Kasheff's avatar
Zardosht Kasheff committed
373
toku_ft_nonleaf_is_gorged (FTNODE node, uint32_t nodesize) {
Yoni Fogel's avatar
Yoni Fogel committed
374
    uint64_t size = toku_serialize_ftnode_size(node);
375

Yoni Fogel's avatar
Yoni Fogel committed
376
    bool buffers_are_empty = true;
377
    toku_assert_entire_node_in_memory(node);
Zardosht Kasheff's avatar
Zardosht Kasheff committed
378 379 380 381
    //
    // the nonleaf node is gorged if the following holds true:
    //  - the buffers are non-empty
    //  - the total workdone by the buffers PLUS the size of the buffers
Zardosht Kasheff's avatar
Zardosht Kasheff committed
382
    //     is greater than nodesize (which as of Maxwell should be
Zardosht Kasheff's avatar
Zardosht Kasheff committed
383 384
    //     4MB)
    //
385
    paranoid_invariant(node->height > 0);
Zardosht Kasheff's avatar
Zardosht Kasheff committed
386 387 388
    for (int child = 0; child < node->n_children; ++child) {
        size += BP_WORKDONE(node, child);
    }
389
    for (int child = 0; child < node->n_children; ++child) {
390
        if (toku_bnc_nbytesinbuf(BNC(node, child)) > 0) {
Yoni Fogel's avatar
Yoni Fogel committed
391
            buffers_are_empty = false;
392 393 394
            break;
        }
    }
Zardosht Kasheff's avatar
Zardosht Kasheff committed
395
    return ((size > nodesize)
396 397
            &&
            (!buffers_are_empty));
398 399
}

400 401
static void ft_verify_flags(FT UU(ft), FTNODE UU(node)) {
    paranoid_invariant(ft->h->flags == node->flags);
402
}
403

404
int toku_ft_debug_mode = 0;
405

Yoni Fogel's avatar
Yoni Fogel committed
406
uint32_t compute_child_fullhash (CACHEFILE cf, FTNODE node, int childnum) {
407 408
    paranoid_invariant(node->height>0);
    paranoid_invariant(childnum<node->n_children);
409
    return toku_cachetable_hash(cf, BP_BLOCKNUM(node, childnum));
410 411
}

412 413 414 415 416 417
int
toku_bnc_n_entries(NONLEAF_CHILDINFO bnc)
{
    return toku_fifo_n_entries(bnc->buffer);
}

418
static const DBT *prepivotkey (FTNODE node, int childnum, const DBT * const lower_bound_exclusive) {
419
    if (childnum==0)
420
        return lower_bound_exclusive;
421
    else {
422
        return &node->childkeys[childnum-1];
423 424 425
    }
}

426
static const DBT *postpivotkey (FTNODE node, int childnum, const DBT * const upper_bound_inclusive) {
427
    if (childnum+1 == node->n_children)
428
        return upper_bound_inclusive;
429
    else {
430
        return &node->childkeys[childnum];
431 432
    }
}
433
static struct pivot_bounds next_pivot_keys (FTNODE node, int childnum, struct pivot_bounds const * const old_pb) {
434
    struct pivot_bounds pb = {.lower_bound_exclusive = prepivotkey(node, childnum, old_pb->lower_bound_exclusive),
435
                              .upper_bound_inclusive = postpivotkey(node, childnum, old_pb->upper_bound_inclusive)};
436 437
    return pb;
}
438

439
// how much memory does this child buffer consume?
440
long
441
toku_bnc_memory_size(NONLEAF_CHILDINFO bnc)
442 443 444
{
    return (sizeof(*bnc) +
            toku_fifo_memory_footprint(bnc->buffer) +
445 446 447
            bnc->fresh_message_tree.memory_size() +
            bnc->stale_message_tree.memory_size() +
            bnc->broadcast_list.memory_size());
448 449 450 451 452 453
}

// how much memory in this child buffer holds useful data?
// originally created solely for use by test program(s).
long
toku_bnc_memory_used(NONLEAF_CHILDINFO bnc)
454 455
{
    return (sizeof(*bnc) +
456
            toku_fifo_memory_size_in_use(bnc->buffer) +
457 458 459
            bnc->fresh_message_tree.memory_size() +
            bnc->stale_message_tree.memory_size() +
            bnc->broadcast_list.memory_size());
460 461
}

Zardosht Kasheff's avatar
Zardosht Kasheff committed
462
static long
463
get_avail_internal_node_partition_size(FTNODE node, int i)
Zardosht Kasheff's avatar
Zardosht Kasheff committed
464
{
465
    paranoid_invariant(node->height > 0);
466
    return toku_bnc_memory_size(BNC(node, i));
Zardosht Kasheff's avatar
Zardosht Kasheff committed
467 468
}

Zardosht Kasheff's avatar
Zardosht Kasheff committed
469

470
static long
471
ftnode_cachepressure_size(FTNODE node)
Zardosht Kasheff's avatar
Zardosht Kasheff committed
472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490
{
    long retval = 0;
    bool totally_empty = true;
    if (node->height == 0) {
        goto exit;
    }
    else {
        for (int i = 0; i < node->n_children; i++) {
            if (BP_STATE(node,i) == PT_INVALID || BP_STATE(node,i) == PT_ON_DISK) {
                continue;
            }
            else if (BP_STATE(node,i) == PT_COMPRESSED) {
                SUB_BLOCK sb = BSB(node, i);
                totally_empty = false;
                retval += sb->compressed_size;
            }
            else if (BP_STATE(node,i) == PT_AVAIL) {
                totally_empty = totally_empty && (toku_bnc_n_entries(BNC(node, i)) == 0);
                retval += get_avail_internal_node_partition_size(node, i);
491
                retval += BP_WORKDONE(node, i);
Zardosht Kasheff's avatar
Zardosht Kasheff committed
492 493
            }
            else {
494
                abort();
Zardosht Kasheff's avatar
Zardosht Kasheff committed
495 496 497 498 499 500 501 502 503 504
            }
        }
    }
exit:
    if (totally_empty) {
        return 0;
    }
    return retval;
}

505
static long
506
ftnode_memory_size (FTNODE node)
507
// Effect: Estimate how much main memory a node requires.
508
{
509 510 511
    long retval = 0;
    int n_children = node->n_children;
    retval += sizeof(*node);
512
    retval += (n_children)*(sizeof(node->bp[0]));
513
    retval += node->totalchildkeylens;
514 515 516 517 518 519 520

    // now calculate the sizes of the partitions
    for (int i = 0; i < n_children; i++) {
        if (BP_STATE(node,i) == PT_INVALID || BP_STATE(node,i) == PT_ON_DISK) {
            continue;
        }
        else if (BP_STATE(node,i) == PT_COMPRESSED) {
521
            SUB_BLOCK sb = BSB(node, i);
522 523 524 525 526
            retval += sizeof(*sb);
            retval += sb->compressed_size;
        }
        else if (BP_STATE(node,i) == PT_AVAIL) {
            if (node->height > 0) {
Zardosht Kasheff's avatar
Zardosht Kasheff committed
527
                retval += get_avail_internal_node_partition_size(node, i);
528 529
            }
            else {
530
                BASEMENTNODE bn = BLB(node, i);
531
                retval += sizeof(*bn);
532 533 534 535 536 537 538
                {
                    // include fragmentation overhead but do not include space in the
                    // mempool that has not yet been allocated for leaf entries
                    size_t poolsize = toku_mempool_footprint(&bn->buffer_mempool);
                    invariant (poolsize >= BLB_NBYTESINBUF(node,i));
                    retval += poolsize;
                }
539 540 541 542 543
                OMT curr_omt = BLB_BUFFER(node, i);
                retval += (toku_omt_memory_size(curr_omt));
            }
        }
        else {
544
            abort();
545
        }
546
    }
547
    return retval;
548
}
Bradley C. Kuszmaul's avatar
Rename  
Bradley C. Kuszmaul committed
549

550 551 552
PAIR_ATTR make_ftnode_pair_attr(FTNODE node) {
    long size = ftnode_memory_size(node);
    long cachepressure_size = ftnode_cachepressure_size(node);
Zardosht Kasheff's avatar
Zardosht Kasheff committed
553
    PAIR_ATTR result={
554 555 556 557 558
        .size = size,
        .nonleaf_size = (node->height > 0) ? size : 0,
        .leaf_size = (node->height > 0) ? 0 : size,
        .rollback_size = 0,
        .cache_pressure_size = cachepressure_size,
Yoni Fogel's avatar
Yoni Fogel committed
559
        .is_valid = true
560 561
    };
    return result;
Zardosht Kasheff's avatar
Zardosht Kasheff committed
562 563
}

564 565
PAIR_ATTR make_invalid_pair_attr(void) {
    PAIR_ATTR result={
566 567 568 569 570
        .size = 0,
        .nonleaf_size = 0,
        .leaf_size = 0,
        .rollback_size = 0,
        .cache_pressure_size = 0,
Yoni Fogel's avatar
Yoni Fogel committed
571
        .is_valid = false
572 573
    };
    return result;
574
}
Zardosht Kasheff's avatar
Zardosht Kasheff committed
575 576


577
// assign unique dictionary id
578
static uint64_t dict_id_serial = 1;
579 580
static DICTIONARY_ID
next_dict_id(void) {
581
    uint64_t i = toku_sync_fetch_and_add(&dict_id_serial, 1);
582
    assert(i);        // guarantee unique dictionary id by asserting 64-bit counter never wraps
583 584 585 586
    DICTIONARY_ID d = {.dictid = i};
    return d;
}

587
//
588
// Given a bfe and a childnum, returns whether the query that constructed the bfe
589 590 591
// wants the child available.
// Requires: bfe->child_to_read to have been set
//
592
bool
593
toku_bfe_wants_child_available (struct ftnode_fetch_extra* bfe, int childnum)
594
{
595 596
    if (bfe->type == ftnode_fetch_all ||
        (bfe->type == ftnode_fetch_subset && bfe->child_to_read == childnum))
597
    {
598
        return true;
599 600
    }
    else {
601
        return false;
602 603 604
    }
}

605
int
606
toku_bfe_leftmost_child_wanted(struct ftnode_fetch_extra *bfe, FTNODE node)
607
{
608
    paranoid_invariant(bfe->type == ftnode_fetch_subset || bfe->type == ftnode_fetch_prefetch);
609 610
    if (bfe->left_is_neg_infty) {
        return 0;
611
    } else if (bfe->range_lock_left_key.data == nullptr) {
612 613
        return -1;
    } else {
614
        return toku_ftnode_which_child(node, &bfe->range_lock_left_key, &bfe->h->cmp_descriptor, bfe->h->compare_fun);
615 616 617 618
    }
}

int
619
toku_bfe_rightmost_child_wanted(struct ftnode_fetch_extra *bfe, FTNODE node)
620
{
621
    paranoid_invariant(bfe->type == ftnode_fetch_subset || bfe->type == ftnode_fetch_prefetch);
622 623
    if (bfe->right_is_pos_infty) {
        return node->n_children - 1;
624
    } else if (bfe->range_lock_right_key.data == nullptr) {
625 626
        return -1;
    } else {
627
        return toku_ftnode_which_child(node, &bfe->range_lock_right_key, &bfe->h->cmp_descriptor, bfe->h->compare_fun);
628 629
    }
}
630

631
static int
632
ft_cursor_rightmost_child_wanted(FT_CURSOR cursor, FT_HANDLE brt, FTNODE node)
633 634 635
{
    if (cursor->right_is_pos_infty) {
        return node->n_children - 1;
636
    } else if (cursor->range_lock_right_key.data == nullptr) {
637 638
        return -1;
    } else {
639
        return toku_ftnode_which_child(node, &cursor->range_lock_right_key, &brt->ft->cmp_descriptor, brt->ft->compare_fun);
640 641
    }
}
642

643
STAT64INFO_S
644
toku_get_and_clear_basement_stats(FTNODE leafnode) {
645 646 647
    invariant(leafnode->height == 0);
    STAT64INFO_S deltas = ZEROSTATS;
    for (int i = 0; i < leafnode->n_children; i++) {
Zardosht Kasheff's avatar
Zardosht Kasheff committed
648 649 650 651 652
        BASEMENTNODE bn = BLB(leafnode, i);
        invariant(BP_STATE(leafnode,i) == PT_AVAIL);
        deltas.numrows  += bn->stat64_delta.numrows;
        deltas.numbytes += bn->stat64_delta.numbytes;
        bn->stat64_delta = ZEROSTATS;
653 654 655 656
    }
    return deltas;
}

657 658 659
void toku_ft_status_update_flush_reason(FTNODE node, 
        uint64_t uncompressed_bytes_flushed, uint64_t bytes_written,
        tokutime_t write_time, bool for_checkpoint) {
Zardosht Kasheff's avatar
Zardosht Kasheff committed
660 661
    if (node->height == 0) {
        if (for_checkpoint) {
662
            STATUS_INC(FT_DISK_FLUSH_LEAF_FOR_CHECKPOINT, 1);
663 664 665
            STATUS_INC(FT_DISK_FLUSH_LEAF_BYTES_FOR_CHECKPOINT, bytes_written);
            STATUS_INC(FT_DISK_FLUSH_LEAF_UNCOMPRESSED_BYTES_FOR_CHECKPOINT, uncompressed_bytes_flushed);
            STATUS_INC(FT_DISK_FLUSH_LEAF_TOKUTIME_FOR_CHECKPOINT, write_time);
Zardosht Kasheff's avatar
Zardosht Kasheff committed
666 667
        }
        else {
668
            STATUS_INC(FT_DISK_FLUSH_LEAF, 1);
669 670 671
            STATUS_INC(FT_DISK_FLUSH_LEAF_BYTES, bytes_written);
            STATUS_INC(FT_DISK_FLUSH_LEAF_UNCOMPRESSED_BYTES, uncompressed_bytes_flushed);
            STATUS_INC(FT_DISK_FLUSH_LEAF_TOKUTIME, write_time);
Zardosht Kasheff's avatar
Zardosht Kasheff committed
672 673 674 675
        }
    }
    else {
        if (for_checkpoint) {
676
            STATUS_INC(FT_DISK_FLUSH_NONLEAF_FOR_CHECKPOINT, 1);
677 678 679
            STATUS_INC(FT_DISK_FLUSH_NONLEAF_BYTES_FOR_CHECKPOINT, bytes_written);
            STATUS_INC(FT_DISK_FLUSH_NONLEAF_UNCOMPRESSED_BYTES_FOR_CHECKPOINT, uncompressed_bytes_flushed);
            STATUS_INC(FT_DISK_FLUSH_NONLEAF_TOKUTIME_FOR_CHECKPOINT, write_time);
Zardosht Kasheff's avatar
Zardosht Kasheff committed
680 681
        }
        else {
682
            STATUS_INC(FT_DISK_FLUSH_NONLEAF, 1);
683 684 685
            STATUS_INC(FT_DISK_FLUSH_NONLEAF_BYTES, bytes_written);
            STATUS_INC(FT_DISK_FLUSH_NONLEAF_UNCOMPRESSED_BYTES, uncompressed_bytes_flushed);
            STATUS_INC(FT_DISK_FLUSH_NONLEAF_TOKUTIME, write_time);
Zardosht Kasheff's avatar
Zardosht Kasheff committed
686 687 688 689
        }
    }
}

690 691
static void ftnode_update_disk_stats(
    FTNODE ftnode,
692
    FT ft,
Yoni Fogel's avatar
Yoni Fogel committed
693
    bool for_checkpoint
694
    )
Zardosht Kasheff's avatar
Zardosht Kasheff committed
695 696 697
{
    STAT64INFO_S deltas = ZEROSTATS;
    // capture deltas before rebalancing basements for serialization
698
    deltas = toku_get_and_clear_basement_stats(ftnode);
699 700 701 702 703 704 705 706 707 708
    // locking not necessary here with respect to checkpointing
    // in Clayface (because of the pending lock and cachetable lock
    // in toku_cachetable_begin_checkpoint)
    // essentially, if we are dealing with a for_checkpoint 
    // parameter in a function that is called by the flush_callback,
    // then the cachetable needs to ensure that this is called in a safe
    // manner that does not interfere with the beginning
    // of a checkpoint, which it does with the cachetable lock
    // and pending lock
    toku_ft_update_stats(&ft->h->on_disk_stats, deltas);
Zardosht Kasheff's avatar
Zardosht Kasheff committed
709
    if (for_checkpoint) {
710
        toku_ft_update_stats(&ft->checkpoint_header->on_disk_stats, deltas);
Zardosht Kasheff's avatar
Zardosht Kasheff committed
711 712 713
    }
}

714
static void ftnode_clone_partitions(FTNODE node, FTNODE cloned_node) {
Zardosht Kasheff's avatar
Zardosht Kasheff committed
715 716
    for (int i = 0; i < node->n_children; i++) {
        BP_BLOCKNUM(cloned_node,i) = BP_BLOCKNUM(node,i);
717
        paranoid_invariant(BP_STATE(node,i) == PT_AVAIL);
Zardosht Kasheff's avatar
Zardosht Kasheff committed
718 719 720
        BP_STATE(cloned_node,i) = PT_AVAIL;
        BP_WORKDONE(cloned_node, i) = BP_WORKDONE(node, i);
        if (node->height == 0) {
721
            set_BLB(cloned_node, i, toku_clone_bn(BLB(node,i)));
Zardosht Kasheff's avatar
Zardosht Kasheff committed
722 723 724 725 726 727 728
        }
        else {
            set_BNC(cloned_node, i, toku_clone_nl(BNC(node,i)));
        }
    }
}

729 730 731 732 733 734 735 736 737 738 739 740 741
void toku_ftnode_checkpoint_complete_callback(void *value_data) {
    FTNODE node = static_cast<FTNODE>(value_data);
    if (node->height > 0) {
        for (int i = 0; i < node->n_children; ++i) {
            if (BP_STATE(node, i) == PT_AVAIL) {
                NONLEAF_CHILDINFO bnc = BNC(node, i);
                bnc->flow[1] = bnc->flow[0];
                bnc->flow[0] = 0;
            }
        }
    }
}

742
void toku_ftnode_clone_callback(
743 744 745
    void* value_data,
    void** cloned_value_data,
    PAIR_ATTR* new_attr,
Yoni Fogel's avatar
Yoni Fogel committed
746
    bool for_checkpoint,
Zardosht Kasheff's avatar
Zardosht Kasheff committed
747 748 749
    void* write_extraargs
    )
{
750
    FTNODE node = static_cast<FTNODE>(value_data);
Zardosht Kasheff's avatar
Zardosht Kasheff committed
751
    toku_assert_entire_node_in_memory(node);
752 753
    FT ft = static_cast<FT>(write_extraargs);
    FTNODE XCALLOC(cloned_node);
Zardosht Kasheff's avatar
Zardosht Kasheff committed
754
    if (node->height == 0) {
755
        // set header stats, must be done before rebalancing
756
        ftnode_update_disk_stats(node, ft, for_checkpoint);
757
        // rebalance the leaf node
758
        rebalance_ftnode_leaf(node, ft->h->basementnodesize);
Zardosht Kasheff's avatar
Zardosht Kasheff committed
759 760
    }

John Esmet's avatar
John Esmet committed
761
    cloned_node->oldest_referenced_xid_known = node->oldest_referenced_xid_known;
Zardosht Kasheff's avatar
Zardosht Kasheff committed
762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778
    cloned_node->max_msn_applied_to_node_on_disk = node->max_msn_applied_to_node_on_disk;
    cloned_node->flags = node->flags;
    cloned_node->thisnodename = node->thisnodename;
    cloned_node->layout_version = node->layout_version;
    cloned_node->layout_version_original = node->layout_version_original;
    cloned_node->layout_version_read_from_disk = node->layout_version_read_from_disk;
    cloned_node->build_id = node->build_id;
    cloned_node->height = node->height;
    cloned_node->dirty = node->dirty;
    cloned_node->fullhash = node->fullhash;
    cloned_node->n_children = node->n_children;
    cloned_node->totalchildkeylens = node->totalchildkeylens;

    XMALLOC_N(node->n_children-1, cloned_node->childkeys);
    XMALLOC_N(node->n_children, cloned_node->bp);
    // clone pivots
    for (int i = 0; i < node->n_children-1; i++) {
779
        toku_clone_dbt(&cloned_node->childkeys[i], node->childkeys[i]);
Zardosht Kasheff's avatar
Zardosht Kasheff committed
780 781
    }
    // clone partition
782
    ftnode_clone_partitions(node, cloned_node);
Zardosht Kasheff's avatar
Zardosht Kasheff committed
783 784 785 786

    // clear dirty bit
    node->dirty = 0;
    cloned_node->dirty = 0;
787
    node->layout_version_read_from_disk = FT_LAYOUT_VERSION;
Zardosht Kasheff's avatar
Zardosht Kasheff committed
788 789
    // set new pair attr if necessary
    if (node->height == 0) {
790
        *new_attr = make_ftnode_pair_attr(node);
Zardosht Kasheff's avatar
Zardosht Kasheff committed
791 792
    }
    else {
Yoni Fogel's avatar
Yoni Fogel committed
793
        new_attr->is_valid = false;
Zardosht Kasheff's avatar
Zardosht Kasheff committed
794 795 796 797
    }
    *cloned_value_data = cloned_node;
}

798
static void ft_leaf_run_gc(FTNODE node, FT ft);
Zardosht Kasheff's avatar
Zardosht Kasheff committed
799

800
void toku_ftnode_flush_callback(
801
    CACHEFILE UU(cachefile),
802 803
    int fd,
    BLOCKNUM nodename,
804
    void *ftnode_v,
805 806 807 808
    void** disk_data,
    void *extraargs,
    PAIR_ATTR size __attribute__((unused)),
    PAIR_ATTR* new_size,
Yoni Fogel's avatar
Yoni Fogel committed
809 810 811
    bool write_me,
    bool keep_me,
    bool for_checkpoint,
Zardosht Kasheff's avatar
Zardosht Kasheff committed
812
    bool is_clone
813
    )
Zardosht Kasheff's avatar
Zardosht Kasheff committed
814
{
815 816
    FT h = (FT) extraargs;
    FTNODE ftnode = (FTNODE) ftnode_v;
817 818 819
    FTNODE_DISK_DATA* ndd = (FTNODE_DISK_DATA*)disk_data;
    assert(ftnode->thisnodename.b==nodename.b);
    int height = ftnode->height;
820
    if (write_me) {
821
        toku_assert_entire_node_in_memory(ftnode);
822 823 824
        if (height == 0) {
            ft_leaf_run_gc(ftnode, h);
        }
Zardosht Kasheff's avatar
Zardosht Kasheff committed
825 826 827
        if (height == 0 && !is_clone) {
            ftnode_update_disk_stats(ftnode, h, for_checkpoint);
        }
828 829 830
        int r = toku_serialize_ftnode_to(fd, ftnode->thisnodename, ftnode, ndd, !is_clone, h, for_checkpoint);
        assert_zero(r);
        ftnode->layout_version_read_from_disk = FT_LAYOUT_VERSION;
831 832
    }
    if (!keep_me) {
833
        if (!is_clone) {
834 835 836 837 838 839 840 841
            long node_size = ftnode_memory_size(ftnode);
            if (ftnode->height == 0) {
                STATUS_INC(FT_FULL_EVICTIONS_LEAF, 1);
                STATUS_INC(FT_FULL_EVICTIONS_LEAF_BYTES, node_size);
            } else {
                STATUS_INC(FT_FULL_EVICTIONS_NONLEAF, 1);
                STATUS_INC(FT_FULL_EVICTIONS_NONLEAF_BYTES, node_size);
            }
842 843 844
            toku_free(*disk_data);
        }
        else {
845 846 847 848 849
            if (ftnode->height == 0) {
                for (int i = 0; i < ftnode->n_children; i++) {
                    if (BP_STATE(ftnode,i) == PT_AVAIL) {
                        BASEMENTNODE bn = BLB(ftnode, i);
                        toku_ft_decrease_stats(&h->in_memory_stats, bn->stat64_delta);
850 851 852 853
                    }
                }
            }
        }
854
        toku_ftnode_free(&ftnode);
855
    }
Zardosht Kasheff's avatar
Zardosht Kasheff committed
856
    else {
857
        *new_size = make_ftnode_pair_attr(ftnode);
Zardosht Kasheff's avatar
Zardosht Kasheff committed
858
    }
859 860
}

861
void
862
toku_ft_status_update_pivot_fetch_reason(struct ftnode_fetch_extra *bfe)
863
{
864
    if (bfe->type == ftnode_fetch_prefetch) {
865
        STATUS_INC(FT_NUM_PIVOTS_FETCHED_PREFETCH, 1);
866
        STATUS_INC(FT_BYTES_PIVOTS_FETCHED_PREFETCH, bfe->bytes_read);
867
        STATUS_INC(FT_TOKUTIME_PIVOTS_FETCHED_PREFETCH, bfe->io_time);
868
    } else if (bfe->type == ftnode_fetch_all) {
869
        STATUS_INC(FT_NUM_PIVOTS_FETCHED_WRITE, 1);
870
        STATUS_INC(FT_BYTES_PIVOTS_FETCHED_WRITE, bfe->bytes_read);
871
        STATUS_INC(FT_TOKUTIME_PIVOTS_FETCHED_WRITE, bfe->io_time);
872
    } else if (bfe->type == ftnode_fetch_subset) {
873
        STATUS_INC(FT_NUM_PIVOTS_FETCHED_QUERY, 1);
874
        STATUS_INC(FT_BYTES_PIVOTS_FETCHED_QUERY, bfe->bytes_read);
875
        STATUS_INC(FT_TOKUTIME_PIVOTS_FETCHED_QUERY, bfe->io_time);
876 877
    }
}
878

Zardosht Kasheff's avatar
Zardosht Kasheff committed
879
int toku_ftnode_fetch_callback (CACHEFILE UU(cachefile), PAIR p, int fd, BLOCKNUM nodename, uint32_t fullhash,
880
                                 void **ftnode_pv,  void** disk_data, PAIR_ATTR *sizep, int *dirtyp, void *extraargs) {
881
    assert(extraargs);
882 883 884 885
    assert(*ftnode_pv == NULL);
    FTNODE_DISK_DATA* ndd = (FTNODE_DISK_DATA*)disk_data;
    struct ftnode_fetch_extra *bfe = (struct ftnode_fetch_extra *)extraargs;
    FTNODE *node=(FTNODE*)ftnode_pv;
886 887 888
    // deserialize the node, must pass the bfe in because we cannot
    // evaluate what piece of the the node is necessary until we get it at
    // least partially into memory
889 890 891
    int r = toku_deserialize_ftnode_from(fd, nodename, fullhash, node, ndd, bfe);
    if (r != 0) {
        if (r == TOKUDB_BAD_CHECKSUM) {
892
            fprintf(stderr,
893 894
                    "Checksum failure while reading node in file %s.\n",
                    toku_cachefile_fname_in_env(cachefile));
895
        } else {
896
            fprintf(stderr, "Error deserializing node, errno = %d", r);
897 898
        }
        // make absolutely sure we crash before doing anything else.
899
        abort();
900
    }
901

902
    if (r == 0) {
903
        *sizep = make_ftnode_pair_attr(*node);
Zardosht Kasheff's avatar
Zardosht Kasheff committed
904
        (*node)->ct_pair = p;
905
        *dirtyp = (*node)->dirty;  // deserialize could mark the node as dirty (presumably for upgrade)
906
    }
907 908 909
    return r;
}

910 911
void toku_ftnode_pe_est_callback(
    void* ftnode_pv,
Zardosht Kasheff's avatar
Zardosht Kasheff committed
912
    void* disk_data,
913 914
    long* bytes_freed_estimate,
    enum partial_eviction_cost *cost,
Zardosht Kasheff's avatar
Zardosht Kasheff committed
915 916 917
    void* UU(write_extraargs)
    )
{
918
    paranoid_invariant(ftnode_pv != NULL);
Zardosht Kasheff's avatar
Zardosht Kasheff committed
919
    long bytes_to_free = 0;
920
    FTNODE node = static_cast<FTNODE>(ftnode_pv);
921
    if (node->dirty || node->height == 0 ||
922
        node->layout_version_read_from_disk < FT_FIRST_LAYOUT_VERSION_WITH_BASEMENT_NODES) {
Zardosht Kasheff's avatar
Zardosht Kasheff committed
923 924 925 926 927 928 929 930 931 932 933 934 935
        *bytes_freed_estimate = 0;
        *cost = PE_CHEAP;
        goto exit;
    }

    //
    // we are dealing with a clean internal node
    //
    *cost = PE_EXPENSIVE;
    // now lets get an estimate for how much data we can free up
    // we estimate the compressed size of data to be how large
    // the compressed data is on disk
    for (int i = 0; i < node->n_children; i++) {
John Esmet's avatar
John Esmet committed
936
        if (BP_STATE(node,i) == PT_AVAIL && BP_SHOULD_EVICT(node,i)) {
Zardosht Kasheff's avatar
Zardosht Kasheff committed
937 938 939 940
            // calculate how much data would be freed if
            // we compress this node and add it to
            // bytes_to_free

941
            // first get an estimate for how much space will be taken
Zardosht Kasheff's avatar
Zardosht Kasheff committed
942 943
            // after compression, it is simply the size of compressed
            // data on disk plus the size of the struct that holds it
944
            FTNODE_DISK_DATA ndd = (FTNODE_DISK_DATA) disk_data;
Yoni Fogel's avatar
Yoni Fogel committed
945
            uint32_t compressed_data_size = BP_SIZE(ndd, i);
Zardosht Kasheff's avatar
Zardosht Kasheff committed
946 947 948
            compressed_data_size += sizeof(struct sub_block);

            // now get the space taken now
Yoni Fogel's avatar
Yoni Fogel committed
949
            uint32_t decompressed_data_size = get_avail_internal_node_partition_size(node,i);
Zardosht Kasheff's avatar
Zardosht Kasheff committed
950 951 952 953 954 955 956 957 958
            bytes_to_free += (decompressed_data_size - compressed_data_size);
        }
    }

    *bytes_freed_estimate = bytes_to_free;
exit:
    return;
}

959
static void
960
compress_internal_node_partition(FTNODE node, int i, enum toku_compression_method compression_method)
961 962 963 964 965
{
    // if we should evict, compress the
    // message buffer into a sub_block
    assert(BP_STATE(node, i) == PT_AVAIL);
    assert(node->height > 0);
966
    SUB_BLOCK XMALLOC(sb);
967
    sub_block_init(sb);
968
    toku_create_compressed_partition_from_available(node, i, compression_method, sb);
969

970 971 972 973 974 975
    // now free the old partition and replace it with this
    destroy_nonleaf_childinfo(BNC(node,i));
    set_BSB(node, i, sb);
    BP_STATE(node,i) = PT_COMPRESSED;
}

976
void toku_evict_bn_from_memory(FTNODE node, int childnum, FT h) {
977 978 979
    // free the basement node
    assert(!node->dirty);
    BASEMENTNODE bn = BLB(node, childnum);
980
    toku_ft_decrease_stats(&h->in_memory_stats, bn->stat64_delta);
981 982 983 984 985 986 987
    struct mempool * mp = &bn->buffer_mempool;
    toku_mempool_destroy(mp);
    destroy_basement_node(bn);
    set_BNULL(node, childnum);
    BP_STATE(node, childnum) = PT_ON_DISK;
}

988
// callback for partially evicting a node
989 990
int toku_ftnode_pe_callback (void *ftnode_pv, PAIR_ATTR UU(old_attr), PAIR_ATTR* new_attr, void* extraargs) {
    FTNODE node = (FTNODE)ftnode_pv;
991
    FT ft = (FT) extraargs;
992
    // Don't partially evict dirty nodes
Zardosht Kasheff's avatar
Zardosht Kasheff committed
993 994 995
    if (node->dirty) {
        goto exit;
    }
996 997
    // Don't partially evict nodes whose partitions can't be read back
    // from disk individually
998
    if (node->layout_version_read_from_disk < FT_FIRST_LAYOUT_VERSION_WITH_BASEMENT_NODES) {
999 1000
        goto exit;
    }
Zardosht Kasheff's avatar
Zardosht Kasheff committed
1001
    //
1002
    // partial eviction for nonleaf nodes
Zardosht Kasheff's avatar
Zardosht Kasheff committed
1003 1004 1005 1006 1007
    //
    if (node->height > 0) {
        for (int i = 0; i < node->n_children; i++) {
            if (BP_STATE(node,i) == PT_AVAIL) {
                if (BP_SHOULD_EVICT(node,i)) {
1008 1009 1010
                    long size_before = ftnode_memory_size(node);
                    compress_internal_node_partition(node, i, ft->h->compression_method);
                    long delta = size_before - ftnode_memory_size(node);
1011
                    STATUS_INC(FT_PARTIAL_EVICTIONS_NONLEAF, 1);
1012
                    STATUS_INC(FT_PARTIAL_EVICTIONS_NONLEAF_BYTES, delta);
Zardosht Kasheff's avatar
Zardosht Kasheff committed
1013 1014 1015 1016 1017 1018 1019 1020 1021
                }
                else {
                    BP_SWEEP_CLOCK(node,i);
                }
            }
            else {
                continue;
            }
        }
Zardosht Kasheff's avatar
Zardosht Kasheff committed
1022 1023 1024 1025 1026 1027 1028 1029 1030 1031
    }
    //
    // partial eviction strategy for basement nodes:
    //  if the bn is compressed, evict it
    //  else: check if it requires eviction, if it does, evict it, if not, sweep the clock count
    //
    else {
        for (int i = 0; i < node->n_children; i++) {
            // Get rid of compressed stuff no matter what.
            if (BP_STATE(node,i) == PT_COMPRESSED) {
1032
                long size_before = ftnode_memory_size(node);
1033
                SUB_BLOCK sb = BSB(node, i);
Zardosht Kasheff's avatar
Zardosht Kasheff committed
1034
                toku_free(sb->compressed_ptr);
1035
                toku_free(sb);
Zardosht Kasheff's avatar
Zardosht Kasheff committed
1036
                set_BNULL(node, i);
Zardosht Kasheff's avatar
Zardosht Kasheff committed
1037
                BP_STATE(node,i) = PT_ON_DISK;
1038 1039 1040
                long delta = size_before - ftnode_memory_size(node);
                STATUS_INC(FT_PARTIAL_EVICTIONS_LEAF, 1);
                STATUS_INC(FT_PARTIAL_EVICTIONS_LEAF_BYTES, delta);
Zardosht Kasheff's avatar
Zardosht Kasheff committed
1041 1042 1043
            }
            else if (BP_STATE(node,i) == PT_AVAIL) {
                if (BP_SHOULD_EVICT(node,i)) {
1044
                    long size_before = ftnode_memory_size(node);
1045
                    toku_evict_bn_from_memory(node, i, ft);
1046 1047 1048
                    long delta = size_before - ftnode_memory_size(node);
                    STATUS_INC(FT_PARTIAL_EVICTIONS_LEAF, 1);
                    STATUS_INC(FT_PARTIAL_EVICTIONS_LEAF_BYTES, delta);
Zardosht Kasheff's avatar
Zardosht Kasheff committed
1049 1050 1051 1052 1053 1054 1055 1056 1057
                }
                else {
                    BP_SWEEP_CLOCK(node,i);
                }
            }
            else if (BP_STATE(node,i) == PT_ON_DISK) {
                continue;
            }
            else {
1058
                abort();
Zardosht Kasheff's avatar
Zardosht Kasheff committed
1059 1060 1061
            }
        }
    }
Zardosht Kasheff's avatar
Zardosht Kasheff committed
1062 1063

exit:
1064
    *new_attr = make_ftnode_pair_attr(node);
Zardosht Kasheff's avatar
Zardosht Kasheff committed
1065 1066 1067
    return 0;
}

1068 1069 1070 1071 1072
// We touch the clock while holding a read lock.
// DRD reports a race but we want to ignore it.
// Using a valgrind suppressions file is better than the DRD_IGNORE_VAR macro because it's more targeted.
// We need a function to have something a drd suppression can reference
// see src/tests/drd.suppressions (unsafe_touch_clock)
1073
static void unsafe_touch_clock(FTNODE node, int i) {
1074 1075
    BP_TOUCH_CLOCK(node, i);
}
1076

Zardosht Kasheff's avatar
Zardosht Kasheff committed
1077
// Callback that states if a partial fetch of the node is necessary
1078 1079 1080 1081
// Currently, this function is responsible for the following things:
//  - reporting to the cachetable whether a partial fetch is required (as required by the contract of the callback)
//  - A couple of things that are NOT required by the callback, but we do for efficiency and simplicity reasons:
//   - for queries, set the value of bfe->child_to_read so that the query that called this can proceed with the query
1082
//      as opposed to having to evaluate toku_ft_search_which_child again. This is done to make the in-memory query faster
1083 1084 1085
//   - touch the necessary partition's clock. The reason we do it here is so that there is one central place it is done, and not done
//      by all the various callers
//
Yoni Fogel's avatar
Yoni Fogel committed
1086
bool toku_ftnode_pf_req_callback(void* ftnode_pv, void* read_extraargs) {
1087
    // placeholder for now
Yoni Fogel's avatar
Yoni Fogel committed
1088
    bool retval = false;
1089 1090
    FTNODE node = (FTNODE) ftnode_pv;
    struct ftnode_fetch_extra *bfe = (struct ftnode_fetch_extra *) read_extraargs;
Zardosht Kasheff's avatar
Zardosht Kasheff committed
1091 1092
    //
    // The three types of fetches that the brt layer may request are:
1093 1094 1095
    //  - ftnode_fetch_none: no partitions are necessary (example use: stat64)
    //  - ftnode_fetch_subset: some subset is necessary (example use: toku_ft_search)
    //  - ftnode_fetch_all: entire node is necessary (example use: flush, split, merge)
1096
    // The code below checks if the necessary partitions are already in memory,
Yoni Fogel's avatar
Yoni Fogel committed
1097
    // and if they are, return false, and if not, return true
Zardosht Kasheff's avatar
Zardosht Kasheff committed
1098
    //
1099
    if (bfe->type == ftnode_fetch_none) {
Yoni Fogel's avatar
Yoni Fogel committed
1100
        retval = false;
1101
    }
1102
    else if (bfe->type == ftnode_fetch_all) {
Yoni Fogel's avatar
Yoni Fogel committed
1103
        retval = false;
1104
        for (int i = 0; i < node->n_children; i++) {
1105
            unsafe_touch_clock(node,i);
Zardosht Kasheff's avatar
Zardosht Kasheff committed
1106 1107 1108
            // if we find a partition that is not available,
            // then a partial fetch is required because
            // the entire node must be made available
1109
            if (BP_STATE(node,i) != PT_AVAIL) {
Yoni Fogel's avatar
Yoni Fogel committed
1110
                retval = true;
1111 1112 1113
            }
        }
    }
1114
    else if (bfe->type == ftnode_fetch_subset) {
1115 1116 1117 1118 1119
        // we do not take into account prefetching yet
        // as of now, if we need a subset, the only thing
        // we can possibly require is a single basement node
        // we find out what basement node the query cares about
        // and check if it is available
1120 1121
        paranoid_invariant(bfe->h->compare_fun);
        paranoid_invariant(bfe->search);
1122
        bfe->child_to_read = toku_ft_search_which_child(
Zardosht Kasheff's avatar
Zardosht Kasheff committed
1123
            &bfe->h->cmp_descriptor,
Zardosht Kasheff's avatar
Zardosht Kasheff committed
1124
            bfe->h->compare_fun,
1125 1126 1127
            node,
            bfe->search
            );
1128
        unsafe_touch_clock(node,bfe->child_to_read);
Yoni Fogel's avatar
Yoni Fogel committed
1129
        // child we want to read is not available, must set retval to true
1130 1131
        retval = (BP_STATE(node, bfe->child_to_read) != PT_AVAIL);
    }
1132
    else if (bfe->type == ftnode_fetch_prefetch) {
1133 1134
        // makes no sense to have prefetching disabled
        // and still call this function
1135
        paranoid_invariant(!bfe->disable_prefetching);
1136 1137 1138 1139
        int lc = toku_bfe_leftmost_child_wanted(bfe, node);
        int rc = toku_bfe_rightmost_child_wanted(bfe, node);
        for (int i = lc; i <= rc; ++i) {
            if (BP_STATE(node, i) != PT_AVAIL) {
Yoni Fogel's avatar
Yoni Fogel committed
1140
                retval = true;
1141 1142
            }
        }
1143 1144 1145
    }
    else {
        // we have a bug. The type should be known
1146
        abort();
1147 1148
    }
    return retval;
1149 1150
}

1151
static void
1152
ft_status_update_partial_fetch_reason(
1153 1154 1155 1156
    struct ftnode_fetch_extra* bfe,
    int childnum,
    enum pt_state state,
    bool is_leaf
1157 1158 1159 1160
    )
{
    invariant(state == PT_COMPRESSED || state == PT_ON_DISK);
    if (is_leaf) {
1161
        if (bfe->type == ftnode_fetch_prefetch) {
1162
            if (state == PT_COMPRESSED) {
1163
                STATUS_INC(FT_NUM_BASEMENTS_DECOMPRESSED_PREFETCH, 1);
1164
            } else {
1165
                STATUS_INC(FT_NUM_BASEMENTS_FETCHED_PREFETCH, 1);
1166
                STATUS_INC(FT_BYTES_BASEMENTS_FETCHED_PREFETCH, bfe->bytes_read);
1167
                STATUS_INC(FT_TOKUTIME_BASEMENTS_FETCHED_PREFETCH, bfe->io_time);
1168
            }
1169
        } else if (bfe->type == ftnode_fetch_all) {
1170
            if (state == PT_COMPRESSED) {
1171
                STATUS_INC(FT_NUM_BASEMENTS_DECOMPRESSED_WRITE, 1);
1172
            } else {
1173
                STATUS_INC(FT_NUM_BASEMENTS_FETCHED_WRITE, 1);
1174
                STATUS_INC(FT_BYTES_BASEMENTS_FETCHED_WRITE, bfe->bytes_read);
1175
                STATUS_INC(FT_TOKUTIME_BASEMENTS_FETCHED_WRITE, bfe->io_time);
1176
            }
1177
        } else if (childnum == bfe->child_to_read) {
1178
            if (state == PT_COMPRESSED) {
1179
                STATUS_INC(FT_NUM_BASEMENTS_DECOMPRESSED_NORMAL, 1);
1180
            } else {
1181
                STATUS_INC(FT_NUM_BASEMENTS_FETCHED_NORMAL, 1);
1182
                STATUS_INC(FT_BYTES_BASEMENTS_FETCHED_NORMAL, bfe->bytes_read);
1183
                STATUS_INC(FT_TOKUTIME_BASEMENTS_FETCHED_NORMAL, bfe->io_time);
1184 1185 1186
            }
        } else {
            if (state == PT_COMPRESSED) {
1187
                STATUS_INC(FT_NUM_BASEMENTS_DECOMPRESSED_AGGRESSIVE, 1);
1188
            } else {
1189
                STATUS_INC(FT_NUM_BASEMENTS_FETCHED_AGGRESSIVE, 1);
1190
                STATUS_INC(FT_BYTES_BASEMENTS_FETCHED_AGGRESSIVE, bfe->bytes_read);
1191
                STATUS_INC(FT_TOKUTIME_BASEMENTS_FETCHED_AGGRESSIVE, bfe->io_time);
1192 1193 1194 1195
            }
        }
    }
    else {
1196
        if (bfe->type == ftnode_fetch_prefetch) {
1197
            if (state == PT_COMPRESSED) {
1198
                STATUS_INC(FT_NUM_MSG_BUFFER_DECOMPRESSED_PREFETCH, 1);
1199
            } else {
1200
                STATUS_INC(FT_NUM_MSG_BUFFER_FETCHED_PREFETCH, 1);
1201
                STATUS_INC(FT_BYTES_MSG_BUFFER_FETCHED_PREFETCH, bfe->bytes_read);
1202
                STATUS_INC(FT_TOKUTIME_MSG_BUFFER_FETCHED_PREFETCH, bfe->io_time);
1203
            }
1204
        } else if (bfe->type == ftnode_fetch_all) {
1205
            if (state == PT_COMPRESSED) {
1206
                STATUS_INC(FT_NUM_MSG_BUFFER_DECOMPRESSED_WRITE, 1);
1207
            } else {
1208
                STATUS_INC(FT_NUM_MSG_BUFFER_FETCHED_WRITE, 1);
1209
                STATUS_INC(FT_BYTES_MSG_BUFFER_FETCHED_WRITE, bfe->bytes_read);
1210
                STATUS_INC(FT_TOKUTIME_MSG_BUFFER_FETCHED_WRITE, bfe->io_time);
1211
            }
1212
        } else if (childnum == bfe->child_to_read) {
1213
            if (state == PT_COMPRESSED) {
1214
                STATUS_INC(FT_NUM_MSG_BUFFER_DECOMPRESSED_NORMAL, 1);
1215
            } else {
1216
                STATUS_INC(FT_NUM_MSG_BUFFER_FETCHED_NORMAL, 1);
1217
                STATUS_INC(FT_BYTES_MSG_BUFFER_FETCHED_NORMAL, bfe->bytes_read);
1218
                STATUS_INC(FT_TOKUTIME_MSG_BUFFER_FETCHED_NORMAL, bfe->io_time);
1219 1220 1221
            }
        } else {
            if (state == PT_COMPRESSED) {
1222
                STATUS_INC(FT_NUM_MSG_BUFFER_DECOMPRESSED_AGGRESSIVE, 1);
1223
            } else {
1224
                STATUS_INC(FT_NUM_MSG_BUFFER_FETCHED_AGGRESSIVE, 1);
1225
                STATUS_INC(FT_BYTES_MSG_BUFFER_FETCHED_AGGRESSIVE, bfe->bytes_read);
1226
                STATUS_INC(FT_TOKUTIME_MSG_BUFFER_FETCHED_AGGRESSIVE, bfe->io_time);
1227 1228 1229 1230 1231
            }
        }
    }
}

1232 1233 1234 1235 1236 1237 1238 1239 1240 1241
void toku_ft_status_update_serialize_times(tokutime_t serialize_time, tokutime_t compress_time) {
    STATUS_INC(FT_NODE_SERIALIZE_TOKUTIME, serialize_time);
    STATUS_INC(FT_NODE_COMPRESS_TOKUTIME, compress_time);
}

void toku_ft_status_update_deserialize_times(tokutime_t deserialize_time, tokutime_t decompress_time) {
    STATUS_INC(FT_NODE_DESERIALIZE_TOKUTIME, deserialize_time);
    STATUS_INC(FT_NODE_DECOMPRESS_TOKUTIME, decompress_time);
}

1242
// callback for partially reading a node
1243 1244
// could have just used toku_ftnode_fetch_callback, but wanted to separate the two cases to separate functions
int toku_ftnode_pf_callback(void* ftnode_pv, void* disk_data, void* read_extraargs, int fd, PAIR_ATTR* sizep) {
1245
    int r = 0;
1246 1247 1248
    FTNODE node = (FTNODE) ftnode_pv;
    FTNODE_DISK_DATA ndd = (FTNODE_DISK_DATA) disk_data;
    struct ftnode_fetch_extra *bfe = (struct ftnode_fetch_extra *) read_extraargs;
1249
    // there must be a reason this is being called. If we get a garbage type or the type is ftnode_fetch_none,
1250
    // then something went wrong
1251
    assert((bfe->type == ftnode_fetch_subset) || (bfe->type == ftnode_fetch_all) || (bfe->type == ftnode_fetch_prefetch));
1252 1253
    // determine the range to prefetch
    int lc, rc;
1254
    if (!bfe->disable_prefetching &&
1255
        (bfe->type == ftnode_fetch_subset || bfe->type == ftnode_fetch_prefetch)
1256
        )
1257
    {
1258 1259 1260 1261 1262 1263
        lc = toku_bfe_leftmost_child_wanted(bfe, node);
        rc = toku_bfe_rightmost_child_wanted(bfe, node);
    } else {
        lc = -1;
        rc = -1;
    }
1264 1265 1266 1267
    for (int i = 0; i < node->n_children; i++) {
        if (BP_STATE(node,i) == PT_AVAIL) {
            continue;
        }
1268
        if ((lc <= i && i <= rc) || toku_bfe_wants_child_available(bfe, i)) {
1269 1270
            enum pt_state state = BP_STATE(node, i);
            if (state == PT_COMPRESSED) {
1271
                r = toku_deserialize_bp_from_compressed(node, i, bfe);
1272 1273
            } else {
                invariant(state == PT_ON_DISK);
1274
                r = toku_deserialize_bp_from_disk(node, ndd, i, fd, bfe);
Zardosht Kasheff's avatar
Zardosht Kasheff committed
1275
            }
1276
            ft_status_update_partial_fetch_reason(bfe, i, state, (node->height == 0));
1277
        }
1278

1279 1280
        if (r != 0) {
            if (r == TOKUDB_BAD_CHECKSUM) {
1281 1282 1283
                fprintf(stderr,
                        "Checksum failure while reading node partition in file %s.\n",
                        toku_cachefile_fname_in_env(bfe->h->cf));
1284
            } else {
1285 1286
                fprintf(stderr,
                        "Error while reading node partition %d\n",
1287
                        get_maybe_error_errno());
1288
            }
1289
            abort();
1290
        }
1291
    }
1292

1293
    *sizep = make_ftnode_pair_attr(node);
1294

1295
    return 0;
1296 1297
}

1298
static int
Yoni Fogel's avatar
Yoni Fogel committed
1299
leafval_heaviside_le (uint32_t klen, void *kval,
1300
                      struct cmd_leafval_heaviside_extra *be) {
1301
    DBT dbt;
1302
    DBT const * const key = be->key;
Zardosht Kasheff's avatar
Zardosht Kasheff committed
1303
    FAKE_DB(db, be->desc);
Zardosht Kasheff's avatar
Zardosht Kasheff committed
1304
    return be->compare_fun(&db,
1305 1306
                           toku_fill_dbt(&dbt, kval, klen),
                           key);
1307 1308
}

1309 1310 1311
//TODO: #1125 optimize
int
toku_cmd_leafval_heaviside (OMTVALUE lev, void *extra) {
1312 1313
    LEAFENTRY CAST_FROM_VOIDP(le, lev);
    struct cmd_leafval_heaviside_extra *CAST_FROM_VOIDP(be, extra);
Yoni Fogel's avatar
Yoni Fogel committed
1314
    uint32_t keylen;
1315 1316
    void*     key = le_key_and_len(le, &keylen);
    return leafval_heaviside_le(keylen, key,
1317
                                be);
1318 1319
}

1320
static int
1321
ft_compare_pivot(DESCRIPTOR desc, ft_compare_func cmp, const DBT *key, const DBT *pivot)
1322
{
1323
    int r;
Zardosht Kasheff's avatar
Zardosht Kasheff committed
1324
    FAKE_DB(db, desc);
1325
    r = cmp(&db, key, pivot);
1326
    return r;
1327 1328
}

1329

1330
// destroys the internals of the ftnode, but it does not free the values
1331
// that are stored
1332
// this is common functionality for toku_ftnode_free and rebalance_ftnode_leaf
1333
// MUST NOT do anything besides free the structures that have been allocated
1334
void toku_destroy_ftnode_internals(FTNODE node)
1335
{
1336
    for (int i=0; i<node->n_children-1; i++) {
1337
        toku_destroy_dbt(&node->childkeys[i]);
1338 1339 1340
    }
    toku_free(node->childkeys);
    node->childkeys = NULL;
1341 1342 1343 1344

    for (int i=0; i < node->n_children; i++) {
        if (BP_STATE(node,i) == PT_AVAIL) {
            if (node->height > 0) {
1345
                destroy_nonleaf_childinfo(BNC(node,i));
1346
            } else {
1347
                destroy_basement_node(BLB(node, i));
1348
            }
1349
        } else if (BP_STATE(node,i) == PT_COMPRESSED) {
Zardosht Kasheff's avatar
Zardosht Kasheff committed
1350 1351
            SUB_BLOCK sb = BSB(node,i);
            toku_free(sb->compressed_ptr);
1352
            toku_free(sb);
1353
        } else {
1354
            paranoid_invariant(is_BNULL(node, i));
1355
        }
1356
        set_BNULL(node, i);
1357 1358 1359
    }
    toku_free(node->bp);
    node->bp = NULL;
1360 1361 1362
}

/* Frees a node, including all the stuff in the hash table. */
1363 1364
void toku_ftnode_free(FTNODE *nodep) {
    FTNODE node = *nodep;
1365
    if (node->height == 0) {
1366
        for (int i = 0; i < node->n_children; i++) {
1367
            if (BP_STATE(node,i) == PT_AVAIL) {
1368 1369
                struct mempool * mp = &(BLB_BUFFER_MEMPOOL(node, i));
                toku_mempool_destroy(mp);
1370
            }
1371
        }
1372
        STATUS_INC(FT_DESTROY_LEAF, 1);
1373
    } else {
1374
        STATUS_INC(FT_DESTROY_NONLEAF, 1);
1375
    }
1376
    toku_destroy_ftnode_internals(node);
1377
    toku_free(node);
1378
    *nodep = nullptr;
1379 1380
}

1381
void
Zardosht Kasheff's avatar
Zardosht Kasheff committed
1382
toku_initialize_empty_ftnode (FTNODE n, BLOCKNUM nodename, int height, int num_children, int layout_version, unsigned int flags)
1383
// Effect: Fill in N as an empty ftnode.
1384
{
1385 1386
    paranoid_invariant(layout_version != 0);
    paranoid_invariant(height >= 0);
1387

1388
    if (height == 0) {
1389
        STATUS_INC(FT_CREATE_LEAF, 1);
1390
    } else {
1391
        STATUS_INC(FT_CREATE_NONLEAF, 1);
1392
    }
1393

1394
    n->max_msn_applied_to_node_on_disk = ZERO_MSN;    // correct value for root node, harmless for others
1395
    n->flags = flags;
Bradley C. Kuszmaul's avatar
Rename  
Bradley C. Kuszmaul committed
1396
    n->thisnodename = nodename;
1397
    n->layout_version               = layout_version;
1398 1399 1400
    n->layout_version_original = layout_version;
    n->layout_version_read_from_disk = layout_version;
    n->height = height;
1401
    n->totalchildkeylens = 0;
1402
    n->childkeys = 0;
1403
    n->bp = 0;
1404
    n->n_children = num_children;
1405
    n->oldest_referenced_xid_known = TXNID_NONE;
1406

1407
    if (num_children > 0) {
1408 1409
        XMALLOC_N(num_children-1, n->childkeys);
        XMALLOC_N(num_children, n->bp);
1410
        for (int i = 0; i < num_children; i++) {
1411 1412
            BP_BLOCKNUM(n,i).b=0;
            BP_STATE(n,i) = PT_INVALID;
1413
            BP_WORKDONE(n,i) = 0;
Zardosht Kasheff's avatar
Zardosht Kasheff committed
1414
            BP_INIT_TOUCHED_CLOCK(n, i);
1415
            set_BNULL(n,i);
1416
            if (height > 0) {
1417 1418 1419
                set_BNC(n, i, toku_create_empty_nl());
            } else {
                set_BLB(n, i, toku_create_empty_bn());
1420
            }
1421
        }
Bradley C. Kuszmaul's avatar
Rename  
Bradley C. Kuszmaul committed
1422
    }
1423
    n->dirty = 1;  // special case exception, it's okay to mark as dirty because the basements are empty
Bradley C. Kuszmaul's avatar
Rename  
Bradley C. Kuszmaul committed
1424 1425
}

1426
static void
Zardosht Kasheff's avatar
Zardosht Kasheff committed
1427 1428 1429
ft_init_new_root(FT ft, FTNODE oldroot, FTNODE *newrootp)
// Effect:  Create a new root node whose two children are the split of oldroot.
//  oldroot is unpinned in the process.
1430 1431
//  Leave the new root pinned.
{
Zardosht Kasheff's avatar
Zardosht Kasheff committed
1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452
    FTNODE newroot;

    BLOCKNUM old_blocknum = oldroot->thisnodename;
    uint32_t old_fullhash = oldroot->fullhash;
    PAIR old_pair = oldroot->ct_pair;
    
    int new_height = oldroot->height+1;
    uint32_t new_fullhash;
    BLOCKNUM new_blocknum;
    PAIR new_pair = NULL;

    cachetable_put_empty_node_with_dep_nodes(
        ft,
        1,
        &oldroot,
        &new_blocknum,
        &new_fullhash,
        &newroot
        );
    new_pair = newroot->ct_pair;
    
1453 1454
    assert(newroot);
    assert(new_height > 0);
Zardosht Kasheff's avatar
Zardosht Kasheff committed
1455 1456 1457 1458 1459 1460 1461 1462 1463 1464
    toku_initialize_empty_ftnode (
        newroot, 
        new_blocknum, 
        new_height, 
        1, 
        ft->h->layout_version, 
        ft->h->flags
        );
    MSN msna = oldroot->max_msn_applied_to_node_on_disk;
    newroot->max_msn_applied_to_node_on_disk = msna;
1465
    BP_STATE(newroot,0) = PT_AVAIL;
1466
    newroot->dirty = 1;
Zardosht Kasheff's avatar
Zardosht Kasheff committed
1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483

    // now do the "switcheroo"
    BP_BLOCKNUM(newroot,0) = new_blocknum;
    newroot->thisnodename = old_blocknum;
    newroot->fullhash = old_fullhash;
    newroot->ct_pair = old_pair;

    oldroot->thisnodename = new_blocknum;
    oldroot->fullhash = new_fullhash;
    oldroot->ct_pair = new_pair;

    toku_cachetable_swap_pair_values(old_pair, new_pair);

    toku_ft_split_child(
        ft,
        newroot,
        0, // childnum to split
1484 1485
        oldroot,
        SPLIT_EVENLY
Zardosht Kasheff's avatar
Zardosht Kasheff committed
1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502
        );

    // ft_split_child released locks on newroot
    // and oldroot, so now we repin and
    // return to caller
    struct ftnode_fetch_extra bfe;
    fill_bfe_for_full_read(&bfe, ft);
    toku_pin_ftnode_off_client_thread(
        ft,
        old_blocknum,
        old_fullhash,
        &bfe,
        PL_WRITE_EXPENSIVE, // may_modify_node
        0,
        NULL,
        newrootp
        );
1503
}
Bradley C. Kuszmaul's avatar
Rename  
Bradley C. Kuszmaul committed
1504

1505
static void
1506
init_childinfo(FTNODE node, int childnum, FTNODE child) {
1507 1508
    BP_BLOCKNUM(node,childnum) = child->thisnodename;
    BP_STATE(node,childnum) = PT_AVAIL;
1509 1510
    BP_WORKDONE(node, childnum)   = 0;
    set_BNC(node, childnum, toku_create_empty_nl());
1511 1512 1513
}

static void
1514
init_childkey(FTNODE node, int childnum, const DBT *pivotkey) {
1515
    toku_clone_dbt(&node->childkeys[childnum], *pivotkey);
1516
    node->totalchildkeylens += pivotkey->size;
1517 1518
}

1519
// Used only by test programs: append a child node to a parent node
1520
void
1521
toku_ft_nonleaf_append_child(FTNODE node, FTNODE child, const DBT *pivotkey) {
1522 1523
    int childnum = node->n_children;
    node->n_children++;
1524
    XREALLOC_N(node->n_children, node->bp);
1525
    init_childinfo(node, childnum, child);
1526
    XREALLOC_N(node->n_children-1, node->childkeys);
1527
    if (pivotkey) {
1528 1529
        invariant(childnum > 0);
        init_childkey(node, childnum-1, pivotkey);
1530
    }
1531
    node->dirty = 1;
1532 1533
}

1534
static void
1535
ft_leaf_delete_leafentry (
1536
    BASEMENTNODE bn,
Yoni Fogel's avatar
Yoni Fogel committed
1537
    uint32_t idx,
1538 1539
    LEAFENTRY le
    )
1540 1541 1542 1543 1544 1545 1546
// Effect: Delete leafentry
//   idx is the location where it is
//   le is the leafentry to be deleted
{
    // Figure out if one of the other keys is the same key

    {
1547
        int r = toku_omt_delete_at(bn->buffer, idx);
1548
        assert_zero(r);
1549 1550
    }

1551
    bn->n_bytes_in_buffer -= leafentry_disksize(le);
1552

1553
    toku_mempool_mfree(&bn->buffer_mempool, 0, leafentry_memsize(le)); // Must pass 0, since le is no good any more.
1554 1555
}

1556
void
1557
toku_ft_bn_apply_cmd_once (
Zardosht Kasheff's avatar
Zardosht Kasheff committed
1558
    BASEMENTNODE bn,
1559
    const FT_MSG cmd,
Yoni Fogel's avatar
Yoni Fogel committed
1560
    uint32_t idx,
Zardosht Kasheff's avatar
Zardosht Kasheff committed
1561
    LEAFENTRY le,
Zardosht Kasheff's avatar
Zardosht Kasheff committed
1562
    TXNID oldest_referenced_xid,
1563
    GC_INFO gc_info,
1564 1565
    uint64_t *workdone,
    STAT64INFO stats_to_update
1566 1567
    )
// Effect: Apply cmd to leafentry (msn is ignored)
1568
//         Calculate work done by message on leafentry and add it to caller's workdone counter.
1569 1570 1571
//   idx is the location where it goes
//   le is old leafentry
{
1572
    size_t newsize=0, oldsize=0, workdone_this_le=0;
1573
    LEAFENTRY new_le=0;
1574
    void *maybe_free = 0;
1575 1576
    int64_t numbytes_delta = 0;  // how many bytes of user data (not including overhead) were added or deleted from this row
    int64_t numrows_delta = 0;   // will be +1 or -1 or 0 (if row was added or deleted or not)
1577 1578

    if (le)
1579
        oldsize = leafentry_memsize(le);
1580

John Esmet's avatar
John Esmet committed
1581
    // toku_le_apply_msg() may call mempool_malloc_from_omt() to allocate more space.
1582
    // That means le is guaranteed to not cause a sigsegv but it may point to a mempool that is
1583
    // no longer in use.  We'll have to release the old mempool later.
1584
    toku_le_apply_msg(cmd, le, oldest_referenced_xid, gc_info, &newsize, &new_le, &bn->buffer, &bn->buffer_mempool, &maybe_free, &numbytes_delta);
1585

1586 1587 1588
    if (new_le) {
        paranoid_invariant(newsize == leafentry_disksize(new_le));
    }
1589
    if (le && new_le) {
1590 1591
        bn->n_bytes_in_buffer -= oldsize;
        bn->n_bytes_in_buffer += newsize;
1592

1593 1594 1595 1596
        // This mfree must occur after the mempool_malloc so that when
        // the mempool is compressed everything is accounted for.  But
        // we must compute the size before doing the mempool mfree
        // because otherwise the le pointer is no good.
1597
        toku_mempool_mfree(&bn->buffer_mempool, 0, oldsize); // Must pass 0, since le may be no good any more.
1598

1599 1600 1601 1602 1603 1604
        {
            int r = toku_omt_set_at(bn->buffer, new_le, idx);
            invariant(r==0);
        }

        workdone_this_le = (oldsize > newsize ? oldsize : newsize);  // work done is max of le size before and after message application
1605

1606
    } else {           // we did not just replace a row, so ...
1607 1608
        if (le) {
            //            ... we just deleted a row ...
1609
            // It was there, note that it's gone and remove it from the mempool
1610
            ft_leaf_delete_leafentry (bn, idx, le);
1611 1612 1613 1614 1615 1616 1617
            workdone_this_le = oldsize;
            numrows_delta = -1;
        }
        if (new_le) {
            //            ... or we just added a row
            int r = toku_omt_insert_at(bn->buffer, new_le, idx);
            invariant(r==0);
1618
            bn->n_bytes_in_buffer += newsize;
1619 1620
            workdone_this_le = newsize;
            numrows_delta = 1;
1621
        }
1622
    }
1623
    if (workdone) {  // test programs may call with NULL
1624
        *workdone += workdone_this_le;
1625
    }
1626

1627 1628 1629
    // if we created a new mempool buffer, free the old one
    if (maybe_free) toku_free(maybe_free);

1630 1631 1632
    // now update stat64 statistics
    bn->stat64_delta.numrows  += numrows_delta;
    bn->stat64_delta.numbytes += numbytes_delta;
1633 1634 1635 1636 1637
    // the only reason stats_to_update may be null is for tests
    if (stats_to_update) {
        stats_to_update->numrows += numrows_delta;
        stats_to_update->numbytes += numbytes_delta;
    }
1638

1639 1640
}

1641 1642
static const uint32_t setval_tag = 0xee0ccb99; // this was gotten by doing "cat /dev/random|head -c4|od -x" to get a random number.  We want to make sure that the user actually passes us the setval_extra_s that we passed in.
struct setval_extra_s {
Yoni Fogel's avatar
Yoni Fogel committed
1643 1644
    uint32_t  tag;
    bool did_set_val;
1645
    int         setval_r;    // any error code that setval_fun wants to return goes here.
1646
    // need arguments for toku_ft_bn_apply_cmd_once
1647
    BASEMENTNODE bn;
1648
    MSN msn;              // captured from original message, not currently used
1649 1650
    XIDS xids;
    const DBT *key;
Yoni Fogel's avatar
Yoni Fogel committed
1651
    uint32_t idx;
1652
    LEAFENTRY le;
Zardosht Kasheff's avatar
Zardosht Kasheff committed
1653
    TXNID oldest_referenced_xid;
1654
    GC_INFO gc_info;
1655
    uint64_t * workdone;  // set by toku_ft_bn_apply_cmd_once()
1656
    STAT64INFO stats_to_update;
1657 1658 1659 1660 1661
};

/*
 * If new_val == NULL, we send a delete message instead of an insert.
 * This happens here instead of in do_delete() for consistency.
1662 1663
 * setval_fun() is called from handlerton, passing in svextra_v
 * from setval_extra_s input arg to brt->update_fun().
1664 1665
 */
static void setval_fun (const DBT *new_val, void *svextra_v) {
1666
    struct setval_extra_s *CAST_FROM_VOIDP(svextra, svextra_v);
1667 1668
    paranoid_invariant(svextra->tag==setval_tag);
    paranoid_invariant(!svextra->did_set_val);
Yoni Fogel's avatar
Yoni Fogel committed
1669
    svextra->did_set_val = true;
1670 1671

    {
1672
        // can't leave scope until toku_ft_bn_apply_cmd_once if
Zardosht Kasheff's avatar
Zardosht Kasheff committed
1673 1674
        // this is a delete
        DBT val;
1675
        FT_MSG_S msg = { FT_NONE, svextra->msn, svextra->xids,
1676
                         .u = { .id = {svextra->key, NULL} } };
Zardosht Kasheff's avatar
Zardosht Kasheff committed
1677
        if (new_val) {
1678
            msg.type = FT_INSERT;
Zardosht Kasheff's avatar
Zardosht Kasheff committed
1679 1680
            msg.u.id.val = new_val;
        } else {
1681
            msg.type = FT_DELETE_ANY;
Zardosht Kasheff's avatar
Zardosht Kasheff committed
1682 1683 1684
            toku_init_dbt(&val);
            msg.u.id.val = &val;
        }
1685
        toku_ft_bn_apply_cmd_once(svextra->bn, &msg,
1686
                                  svextra->idx, svextra->le,
1687
                                  svextra->oldest_referenced_xid, svextra->gc_info,
1688
                                  svextra->workdone, svextra->stats_to_update);
Zardosht Kasheff's avatar
Zardosht Kasheff committed
1689
        svextra->setval_r = 0;
1690
    }
1691 1692
}

1693
// We are already past the msn filter (in toku_ft_bn_apply_cmd(), which calls do_update()),
1694
// so capturing the msn in the setval_extra_s is not strictly required.         The alternative
1695 1696
// would be to put a dummy msn in the messages created by setval_fun(), but preserving
// the original msn seems cleaner and it preserves accountability at a lower layer.
1697
static int do_update(ft_update_func update_fun, DESCRIPTOR desc, BASEMENTNODE bn, FT_MSG cmd, uint32_t idx,
1698
                     LEAFENTRY le,
Zardosht Kasheff's avatar
Zardosht Kasheff committed
1699
                     TXNID oldest_referenced_xid,
1700
                     GC_INFO gc_info,
1701 1702
                     uint64_t * workdone,
                     STAT64INFO stats_to_update) {
1703 1704 1705 1706 1707 1708 1709 1710 1711
    LEAFENTRY le_for_update;
    DBT key;
    const DBT *keyp;
    const DBT *update_function_extra;
    DBT vdbt;
    const DBT *vdbtp;

    // the location of data depends whether this is a regular or
    // broadcast update
1712
    if (cmd->type == FT_UPDATE) {
Zardosht Kasheff's avatar
Zardosht Kasheff committed
1713 1714
        // key is passed in with command (should be same as from le)
        // update function extra is passed in with command
1715
        STATUS_INC(FT_UPDATES, 1);
Zardosht Kasheff's avatar
Zardosht Kasheff committed
1716 1717
        keyp = cmd->u.id.key;
        update_function_extra = cmd->u.id.val;
1718
    } else if (cmd->type == FT_UPDATE_BROADCAST_ALL) {
Zardosht Kasheff's avatar
Zardosht Kasheff committed
1719 1720
        // key is not passed in with broadcast, it comes from le
        // update function extra is passed in with command
1721
        paranoid_invariant(le);  // for broadcast updates, we just hit all leafentries
Zardosht Kasheff's avatar
Zardosht Kasheff committed
1722
                     // so this cannot be null
1723
        paranoid_invariant(cmd->u.id.key->size == 0);
1724
        STATUS_INC(FT_UPDATES_BROADCAST, 1);
Zardosht Kasheff's avatar
Zardosht Kasheff committed
1725 1726
        keyp = toku_fill_dbt(&key, le_key(le), le_keylen(le));
        update_function_extra = cmd->u.id.val;
1727
    } else {
1728
        abort();
1729 1730 1731
    }

    if (le && !le_latest_is_del(le)) {
Zardosht Kasheff's avatar
Zardosht Kasheff committed
1732
        // if the latest val exists, use it, and we'll use the leafentry later
Yoni Fogel's avatar
Yoni Fogel committed
1733
        uint32_t vallen;
Zardosht Kasheff's avatar
Zardosht Kasheff committed
1734 1735
        void *valp = le_latest_val_and_len(le, &vallen);
        vdbtp = toku_fill_dbt(&vdbt, valp, vallen);
1736
    } else {
Zardosht Kasheff's avatar
Zardosht Kasheff committed
1737 1738
        // otherwise, the val and leafentry are both going to be null
        vdbtp = NULL;
1739
    }
Zardosht Kasheff's avatar
Zardosht Kasheff committed
1740
    le_for_update = le;
1741

Yoni Fogel's avatar
Yoni Fogel committed
1742
    struct setval_extra_s setval_extra = {setval_tag, false, 0, bn, cmd->msn, cmd->xids,
1743
                                          keyp, idx, le_for_update, oldest_referenced_xid, gc_info, workdone, stats_to_update};
1744
    // call handlerton's brt->update_fun(), which passes setval_extra to setval_fun()
Zardosht Kasheff's avatar
Zardosht Kasheff committed
1745
    FAKE_DB(db, desc);
Zardosht Kasheff's avatar
Zardosht Kasheff committed
1746 1747 1748 1749 1750 1751 1752
    int r = update_fun(
        &db,
        keyp,
        vdbtp,
        update_function_extra,
        setval_fun, &setval_extra
        );
1753 1754 1755 1756 1757

    if (r == 0) { r = setval_extra.setval_r; }
    return r;
}

1758
// Should be renamed as something like "apply_cmd_to_basement()."
1759
void
1760 1761 1762
toku_ft_bn_apply_cmd (
    ft_compare_func compare_fun,
    ft_update_func update_fun,
Zardosht Kasheff's avatar
Zardosht Kasheff committed
1763
    DESCRIPTOR desc,
1764
    BASEMENTNODE bn,
1765
    FT_MSG cmd,
1766
    TXNID oldest_referenced_xid_known,
1767
    GC_INFO gc_info, 
1768 1769
    uint64_t *workdone,
    STAT64INFO stats_to_update
1770
    )
1771
// Effect:
1772 1773
//   Put a cmd into a leaf.
//   Calculate work done by message on leafnode and add it to caller's workdone counter.
1774
// The leaf could end up "too big" or "too small".  The caller must fix that up.
1775 1776 1777 1778
{
    LEAFENTRY storeddata;
    OMTVALUE storeddatav=NULL;

Yoni Fogel's avatar
Yoni Fogel committed
1779
    uint32_t omt_size;
1780
    int r;
Zardosht Kasheff's avatar
Zardosht Kasheff committed
1781
    struct cmd_leafval_heaviside_extra be = {compare_fun, desc, cmd->u.id.key};
1782

1783 1784
    unsigned int doing_seqinsert = bn->seqinsert;
    bn->seqinsert = 0;
1785

1786
    switch (cmd->type) {
1787 1788
    case FT_INSERT_NO_OVERWRITE:
    case FT_INSERT: {
Yoni Fogel's avatar
Yoni Fogel committed
1789
        uint32_t idx;
1790 1791 1792 1793
        if (doing_seqinsert) {
            idx = toku_omt_size(bn->buffer);
            r = toku_omt_fetch(bn->buffer, idx-1, &storeddatav);
            if (r != 0) goto fz;
1794
            CAST_FROM_VOIDP(storeddata, storeddatav);
1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 1805
            int cmp = toku_cmd_leafval_heaviside(storeddata, &be);
            if (cmp >= 0) goto fz;
            r = DB_NOTFOUND;
        } else {
        fz:
            r = toku_omt_find_zero(bn->buffer, toku_cmd_leafval_heaviside, &be,
                                   &storeddatav, &idx);
        }
        if (r==DB_NOTFOUND) {
            storeddata = 0;
        } else {
1806
            assert_zero(r);
1807
            CAST_FROM_VOIDP(storeddata, storeddatav);
1808
        }
1809
        toku_ft_bn_apply_cmd_once(bn, cmd, idx, storeddata, oldest_referenced_xid_known, gc_info, workdone, stats_to_update);
1810 1811 1812 1813 1814

        // if the insertion point is within a window of the right edge of
        // the leaf then it is sequential
        // window = min(32, number of leaf entries/16)
        {
Yoni Fogel's avatar
Yoni Fogel committed
1815 1816
            uint32_t s = toku_omt_size(bn->buffer);
            uint32_t w = s / 16;
1817 1818 1819 1820 1821 1822 1823 1824
            if (w == 0) w = 1;
            if (w > 32) w = 32;

            // within the window?
            if (s - idx <= w)
                bn->seqinsert = doing_seqinsert + 1;
        }
        break;
1825
    }
1826 1827 1828
    case FT_DELETE_ANY:
    case FT_ABORT_ANY:
    case FT_COMMIT_ANY: {
Yoni Fogel's avatar
Yoni Fogel committed
1829
        uint32_t idx;
1830 1831 1832 1833 1834
        // Apply to all the matches

        r = toku_omt_find_zero(bn->buffer, toku_cmd_leafval_heaviside, &be,
                               &storeddatav, &idx);
        if (r == DB_NOTFOUND) break;
1835
        assert_zero(r);
1836
        CAST_FROM_VOIDP(storeddata, storeddatav);
1837
        toku_ft_bn_apply_cmd_once(bn, cmd, idx, storeddata, oldest_referenced_xid_known, gc_info, workdone, stats_to_update);
1838 1839

        break;
1840
    }
1841
    case FT_OPTIMIZE_FOR_UPGRADE:
1842
        // fall through so that optimize_for_upgrade performs rest of the optimize logic
1843 1844
    case FT_COMMIT_BROADCAST_ALL:
    case FT_OPTIMIZE:
1845 1846
        // Apply to all leafentries
        omt_size = toku_omt_size(bn->buffer);
Yoni Fogel's avatar
Yoni Fogel committed
1847
        for (uint32_t idx = 0; idx < omt_size; ) {
1848 1849
            r = toku_omt_fetch(bn->buffer, idx, &storeddatav);
            assert_zero(r);
1850
            CAST_FROM_VOIDP(storeddata, storeddatav);
1851 1852
            int deleted = 0;
            if (!le_is_clean(storeddata)) { //If already clean, nothing to do.
1853
                toku_ft_bn_apply_cmd_once(bn, cmd, idx, storeddata, oldest_referenced_xid_known, gc_info, workdone, stats_to_update);
Yoni Fogel's avatar
Yoni Fogel committed
1854
                uint32_t new_omt_size = toku_omt_size(bn->buffer);
1855
                if (new_omt_size != omt_size) {
1856
                    paranoid_invariant(new_omt_size+1 == omt_size);
1857 1858 1859 1860 1861 1862 1863 1864 1865
                    //Item was deleted.
                    deleted = 1;
                }
            }
            if (deleted)
                omt_size--;
            else
                idx++;
        }
1866
        paranoid_invariant(toku_omt_size(bn->buffer) == omt_size);
1867 1868

        break;
1869 1870
    case FT_COMMIT_BROADCAST_TXN:
    case FT_ABORT_BROADCAST_TXN:
1871 1872
        // Apply to all leafentries if txn is represented
        omt_size = toku_omt_size(bn->buffer);
Yoni Fogel's avatar
Yoni Fogel committed
1873
        for (uint32_t idx = 0; idx < omt_size; ) {
1874 1875
            r = toku_omt_fetch(bn->buffer, idx, &storeddatav);
            assert_zero(r);
1876
            CAST_FROM_VOIDP(storeddata, storeddatav);
1877 1878
            int deleted = 0;
            if (le_has_xids(storeddata, cmd->xids)) {
1879
                toku_ft_bn_apply_cmd_once(bn, cmd, idx, storeddata, oldest_referenced_xid_known, gc_info, workdone, stats_to_update);
Yoni Fogel's avatar
Yoni Fogel committed
1880
                uint32_t new_omt_size = toku_omt_size(bn->buffer);
1881
                if (new_omt_size != omt_size) {
1882
                    paranoid_invariant(new_omt_size+1 == omt_size);
1883 1884 1885 1886 1887 1888 1889 1890 1891
                    //Item was deleted.
                    deleted = 1;
                }
            }
            if (deleted)
                omt_size--;
            else
                idx++;
        }
1892
        paranoid_invariant(toku_omt_size(bn->buffer) == omt_size);
1893 1894

        break;
1895
    case FT_UPDATE: {
Yoni Fogel's avatar
Yoni Fogel committed
1896
        uint32_t idx;
1897 1898 1899
        r = toku_omt_find_zero(bn->buffer, toku_cmd_leafval_heaviside, &be,
                               &storeddatav, &idx);
        if (r==DB_NOTFOUND) {
1900
            r = do_update(update_fun, desc, bn, cmd, idx, NULL, oldest_referenced_xid_known, gc_info, workdone, stats_to_update);
1901
        } else if (r==0) {
1902
            CAST_FROM_VOIDP(storeddata, storeddatav);
1903
            r = do_update(update_fun, desc, bn, cmd, idx, storeddata, oldest_referenced_xid_known, gc_info, workdone, stats_to_update);
1904 1905
        } // otherwise, a worse error, just return it
        break;
1906
    }
1907
    case FT_UPDATE_BROADCAST_ALL: {
1908
        // apply to all leafentries.
Yoni Fogel's avatar
Yoni Fogel committed
1909 1910
        uint32_t idx = 0;
        uint32_t num_leafentries_before;
1911 1912
        while (idx < (num_leafentries_before = toku_omt_size(bn->buffer))) {
            r = toku_omt_fetch(bn->buffer, idx, &storeddatav);
1913
            assert_zero(r);
1914
            CAST_FROM_VOIDP(storeddata, storeddatav);
1915
            r = do_update(update_fun, desc, bn, cmd, idx, storeddata, oldest_referenced_xid_known, gc_info, workdone, stats_to_update);
1916
            assert_zero(r);
1917 1918 1919 1920 1921 1922 1923

            if (num_leafentries_before == toku_omt_size(bn->buffer)) {
                // we didn't delete something, so increment the index.
                idx++;
            }
        }
        break;
1924
    }
1925
    case FT_NONE: break; // don't do anything
1926 1927
    }

1928
    return;
1929 1930
}

1931 1932
static inline int
key_msn_cmp(const DBT *a, const DBT *b, const MSN amsn, const MSN bmsn,
1933
            DESCRIPTOR descriptor, ft_compare_func key_cmp)
1934
{
Zardosht Kasheff's avatar
Zardosht Kasheff committed
1935
    FAKE_DB(db, descriptor);
Zardosht Kasheff's avatar
Zardosht Kasheff committed
1936
    int r = key_cmp(&db, a, b);
1937
    if (r == 0) {
1938 1939 1940 1941 1942 1943 1944
        if (amsn.msn > bmsn.msn) {
            r = +1;
        } else if (amsn.msn < bmsn.msn) {
            r = -1;
        } else {
            r = 0;
        }
1945 1946 1947 1948 1949
    }
    return r;
}

int
1950
toku_fifo_entry_key_msn_heaviside(const int32_t &offset, const struct toku_fifo_entry_key_msn_heaviside_extra &extra)
1951
{
1952
    const struct fifo_entry *query = toku_fifo_get_entry(extra.fifo, offset);
1953
    DBT qdbt;
1954
    const DBT *query_key = fill_dbt_for_fifo_entry(&qdbt, query);
1955 1956 1957
    const DBT *target_key = extra.key;
    return key_msn_cmp(query_key, target_key, query->msn, extra.msn,
                       extra.desc, extra.cmp);
1958 1959 1960
}

int
1961
toku_fifo_entry_key_msn_cmp(const struct toku_fifo_entry_key_msn_cmp_extra &extra, const int32_t &ao, const int32_t &bo)
1962
{
1963 1964
    const struct fifo_entry *a = toku_fifo_get_entry(extra.fifo, ao);
    const struct fifo_entry *b = toku_fifo_get_entry(extra.fifo, bo);
1965 1966 1967 1968
    DBT adbt, bdbt;
    const DBT *akey = fill_dbt_for_fifo_entry(&adbt, a);
    const DBT *bkey = fill_dbt_for_fifo_entry(&bdbt, b);
    return key_msn_cmp(akey, bkey, a->msn, b->msn,
1969
                       extra.desc, extra.cmp);
1970 1971
}

1972
void toku_bnc_insert_msg(NONLEAF_CHILDINFO bnc, const void *key, ITEMLEN keylen, const void *data, ITEMLEN datalen, enum ft_msg_type type, MSN msn, XIDS xids, bool is_fresh, DESCRIPTOR desc, ft_compare_func cmp)
1973 1974 1975
// Effect: Enqueue the message represented by the parameters into the
//   bnc's buffer, and put it in either the fresh or stale message tree,
//   or the broadcast list.
1976 1977
//
// This is only exported for tests.
1978
{
1979
    int32_t offset;
1980 1981
    int r = toku_fifo_enq(bnc->buffer, key, keylen, data, datalen, type, msn, xids, is_fresh, &offset);
    assert_zero(r);
1982
    if (ft_msg_type_applies_once(type)) {
1983 1984
        DBT keydbt;
        struct toku_fifo_entry_key_msn_heaviside_extra extra = { .desc = desc, .cmp = cmp, .fifo = bnc->buffer, .key = toku_fill_dbt(&keydbt, key, keylen), .msn = msn };
1985
        if (is_fresh) {
1986
            r = bnc->fresh_message_tree.insert<struct toku_fifo_entry_key_msn_heaviside_extra, toku_fifo_entry_key_msn_heaviside>(offset, extra, nullptr);
1987
            assert_zero(r);
1988
        } else {
1989
            r = bnc->stale_message_tree.insert<struct toku_fifo_entry_key_msn_heaviside_extra, toku_fifo_entry_key_msn_heaviside>(offset, extra, nullptr);
1990
            assert_zero(r);
1991
        }
1992 1993
    } else {
        invariant(ft_msg_type_applies_all(type) || ft_msg_type_does_nothing(type));
1994 1995
        const uint32_t idx = bnc->broadcast_list.size();
        r = bnc->broadcast_list.insert_at(offset, idx);
1996
        assert_zero(r);
1997
    }
1998 1999 2000 2001
}

// append a cmd to a nonleaf node's child buffer
// should be static, but used by test programs
2002
void toku_ft_append_to_child_buffer(ft_compare_func compare_fun, DESCRIPTOR desc, FTNODE node, int childnum, enum ft_msg_type type, MSN msn, XIDS xids, bool is_fresh, const DBT *key, const DBT *val) {
2003
    paranoid_invariant(BP_STATE(node,childnum) == PT_AVAIL);
2004
    toku_bnc_insert_msg(BNC(node, childnum), key->data, key->size, val->data, val->size, type, msn, xids, is_fresh, desc, compare_fun);
2005
    node->dirty = 1;
2006 2007
}

2008
static void ft_nonleaf_cmd_once_to_child(ft_compare_func compare_fun, DESCRIPTOR desc, FTNODE node, int target_childnum, FT_MSG cmd, bool is_fresh, size_t flow_deltas[])
2009 2010 2011
// Previously we had passive aggressive promotion, but that causes a lot of I/O a the checkpoint.  So now we are just putting it in the buffer here.
// Also we don't worry about the node getting overfull here.  It's the caller's problem.
{
2012 2013 2014
    unsigned int childnum = (target_childnum >= 0
                             ? target_childnum
                             : toku_ftnode_which_child(node, cmd->u.id.key, desc, compare_fun));
2015
    toku_ft_append_to_child_buffer(compare_fun, desc, node, childnum, cmd->type, cmd->msn, cmd->xids, is_fresh, cmd->u.id.key, cmd->u.id.val);
2016 2017 2018
    NONLEAF_CHILDINFO bnc = BNC(node, childnum);
    bnc->flow[0] += flow_deltas[0];
    bnc->flow[1] += flow_deltas[1];
2019
}
2020

2021 2022 2023 2024
/* Find the leftmost child that may contain the key.
 * If the key exists it will be in the child whose number
 * is the return value of this function.
 */
2025 2026
int toku_ftnode_which_child(FTNODE node, const DBT *k,
                            DESCRIPTOR desc, ft_compare_func cmp) {
2027
    // a funny case of no pivots
2028
    if (node->n_children <= 1) return 0;
2029 2030

    // check the last key to optimize seq insertions
2031
    int n = node->n_children-1;
2032
    int c = ft_compare_pivot(desc, cmp, k, &node->childkeys[n-1]);
2033
    if (c > 0) return n;
2034 2035 2036 2037 2038 2039

    // binary search the pivots
    int lo = 0;
    int hi = n-1; // skip the last one, we checked it above
    int mi;
    while (lo < hi) {
2040
        mi = (lo + hi) / 2;
2041
        c = ft_compare_pivot(desc, cmp, k, &node->childkeys[mi]);
2042 2043 2044 2045 2046 2047 2048 2049 2050
        if (c > 0) {
            lo = mi+1;
            continue;
        }
        if (c < 0) {
            hi = mi;
            continue;
        }
        return mi;
2051 2052
    }
    return lo;
2053 2054
}

Leif Walsh's avatar
Leif Walsh committed
2055
// Used for HOT.
2056
int
2057
toku_ftnode_hot_next_child(FTNODE node,
2058 2059 2060
                           const DBT *k,
                           DESCRIPTOR desc,
                           ft_compare_func cmp) {
Leif Walsh's avatar
Leif Walsh committed
2061 2062 2063 2064 2065
    int low = 0;
    int hi = node->n_children - 1;
    int mi;
    while (low < hi) {
        mi = (low + hi) / 2;
2066
        int r = ft_compare_pivot(desc, cmp, k, &node->childkeys[mi]);
Leif Walsh's avatar
Leif Walsh committed
2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 2077 2078 2079 2080
        if (r > 0) {
            low = mi + 1;
        } else if (r < 0) {
            hi = mi;
        } else {
            // if they were exactly equal, then we want the sub-tree under
            // the next pivot.
            return mi + 1;
        }
    }
    invariant(low == hi);
    return low;
}

2081 2082
// TODO Use this function to clean up other places where bits of messages are passed around
//      such as toku_bnc_insert_msg() and the call stack above it.
2083
static uint64_t
2084
ft_msg_size(FT_MSG msg) {
2085
    size_t keyval_size = msg->u.id.key->size + msg->u.id.val->size;
2086
    size_t xids_size = xids_get_serialize_size(msg->xids);
2087
    return keyval_size + KEY_VALUE_OVERHEAD + FT_CMD_OVERHEAD + xids_size;
2088 2089
}

2090
static void
2091
ft_nonleaf_cmd_all (ft_compare_func compare_fun, DESCRIPTOR desc, FTNODE node, FT_MSG cmd, bool is_fresh, size_t flow_deltas[])
2092 2093
// Effect: Put the cmd into a nonleaf node.  We put it into all children, possibly causing the children to become reactive.
//  We don't do the splitting and merging.  That's up to the caller after doing all the puts it wants to do.
2094
//  The re_array[i] gets set to the reactivity of any modified child i.         (And there may be several such children.)
2095
{
2096 2097
    for (int i = 0; i < node->n_children; i++) {
        ft_nonleaf_cmd_once_to_child(compare_fun, desc, node, i, cmd, is_fresh, flow_deltas);
2098 2099 2100
    }
}

Yoni Fogel's avatar
Yoni Fogel committed
2101
static bool
2102
ft_msg_applies_once(FT_MSG cmd)
2103
{
2104
    return ft_msg_type_applies_once(cmd->type);
2105 2106
}

Yoni Fogel's avatar
Yoni Fogel committed
2107
static bool
2108
ft_msg_applies_all(FT_MSG cmd)
2109
{
2110
    return ft_msg_type_applies_all(cmd->type);
2111 2112
}

Yoni Fogel's avatar
Yoni Fogel committed
2113
static bool
2114
ft_msg_does_nothing(FT_MSG cmd)
2115
{
2116
    return ft_msg_type_does_nothing(cmd->type);
2117 2118 2119
}

static void
2120
ft_nonleaf_put_cmd (ft_compare_func compare_fun, DESCRIPTOR desc, FTNODE node, int target_childnum, FT_MSG cmd, bool is_fresh, size_t flow_deltas[])
2121 2122
// Effect: Put the cmd into a nonleaf node.  We may put it into a child, possibly causing the child to become reactive.
//  We don't do the splitting and merging.  That's up to the caller after doing all the puts it wants to do.
2123
//  The re_array[i] gets set to the reactivity of any modified child i.         (And there may be several such children.)
2124 2125
//
{
2126 2127

    //
2128
    // see comments in toku_ft_leaf_apply_cmd
2129 2130
    // to understand why we handle setting
    // node->max_msn_applied_to_node_on_disk here,
2131
    // and don't do it in toku_ft_node_put_cmd
2132
    //
2133
    MSN cmd_msn = cmd->msn;
2134 2135
    invariant(cmd_msn.msn > node->max_msn_applied_to_node_on_disk.msn);
    node->max_msn_applied_to_node_on_disk = cmd_msn;
2136

2137 2138 2139 2140 2141
    if (ft_msg_applies_once(cmd)) {
        ft_nonleaf_cmd_once_to_child(compare_fun, desc, node, target_childnum, cmd, is_fresh, flow_deltas);
    } else if (ft_msg_applies_all(cmd)) {
        ft_nonleaf_cmd_all(compare_fun, desc, node, cmd, is_fresh, flow_deltas);
    } else {
2142
        paranoid_invariant(ft_msg_does_nothing(cmd));
2143 2144 2145
    }
}

2146 2147
// Garbage collect one leaf entry.
static void
2148
ft_basement_node_gc_once(BASEMENTNODE bn,
Yoni Fogel's avatar
Yoni Fogel committed
2149
                          uint32_t index,
2150
                          LEAFENTRY leaf_entry,
2151 2152 2153
                          const xid_omt_t &snapshot_xids,
                          const rx_omt_t &referenced_xids,
                          const xid_omt_t &live_root_txns,
2154
                          TXNID oldest_referenced_xid_known,
2155 2156
                          STAT64INFO_S * delta)
{
2157
    paranoid_invariant(leaf_entry);
2158

John Esmet's avatar
John Esmet committed
2159 2160 2161 2162 2163 2164
    // Don't run garbage collection on non-mvcc leaf entries.
    if (leaf_entry->type != LE_MVCC) {
        goto exit;
    }

    // Don't run garbage collection if this leafentry decides it's not worth it.
2165
    if (!toku_le_worth_running_garbage_collection(leaf_entry, oldest_referenced_xid_known)) {
2166 2167 2168
        goto exit;
    }

2169 2170 2171 2172 2173 2174
    size_t oldsize;
    oldsize = 0;
    size_t newsize;
    newsize = 0;
    LEAFENTRY new_leaf_entry;
    new_leaf_entry = NULL;
2175 2176 2177 2178

    // The mempool doesn't free itself.  When it allocates new memory,
    // this pointer will be set to the older memory that must now be
    // freed.
2179 2180
    void * maybe_free;
    maybe_free = NULL;
2181 2182 2183

    // Cache the size of the leaf entry.
    oldsize = leafentry_memsize(leaf_entry);
2184 2185 2186 2187
    // These will represent the number of bytes and rows changed as
    // part of the garbage collection.
    int64_t numbytes_delta;
    int64_t numrows_delta;
John Esmet's avatar
John Esmet committed
2188 2189 2190 2191 2192 2193 2194 2195 2196
    toku_le_garbage_collect(leaf_entry,
                            &new_leaf_entry,
                            &newsize,
                            &bn->buffer,
                            &bn->buffer_mempool,
                            &maybe_free,
                            snapshot_xids,
                            referenced_xids,
                            live_root_txns,
2197
                            oldest_referenced_xid_known,
2198
                            &numbytes_delta);
2199

2200
    numrows_delta = 0;
2201 2202 2203 2204 2205 2206 2207 2208 2209 2210 2211
    if (new_leaf_entry) {
        // If we have a new leaf entry, we must update the size of the
        // memory object.
        bn->n_bytes_in_buffer -= oldsize;
        bn->n_bytes_in_buffer += newsize;
        toku_mempool_mfree(&bn->buffer_mempool, 0, oldsize);
        toku_omt_set_at(bn->buffer, new_leaf_entry, index);
        numrows_delta = 0;
    } else {
        // Our garbage collection removed the leaf entry so we must
        // remove it from the mempool.
2212
        ft_leaf_delete_leafentry (bn, index, leaf_entry);
2213 2214 2215 2216 2217 2218 2219 2220 2221 2222 2223 2224 2225 2226 2227
        numrows_delta = -1;
    }

    // If we created a new mempool buffer we must free the
    // old/original buffer.
    if (maybe_free) {
        toku_free(maybe_free);
    }

    // Update stats.
    bn->stat64_delta.numrows += numrows_delta;
    bn->stat64_delta.numbytes += numbytes_delta;
    delta->numrows += numrows_delta;
    delta->numbytes += numbytes_delta;

2228
exit:
2229 2230 2231 2232 2233 2234
    return;
}

// Garbage collect all leaf entries for a given basement node.
static void
basement_node_gc_all_les(BASEMENTNODE bn,
2235 2236 2237
                         const xid_omt_t &snapshot_xids,
                         const rx_omt_t &referenced_xids,
                         const xid_omt_t &live_root_txns,
2238
                         TXNID oldest_referenced_xid_known,
2239 2240 2241
                         STAT64INFO_S * delta)
{
    int r = 0;
Yoni Fogel's avatar
Yoni Fogel committed
2242 2243
    uint32_t index = 0;
    uint32_t num_leafentries_before;
2244 2245 2246 2247
    while (index < (num_leafentries_before = toku_omt_size(bn->buffer))) {
        OMTVALUE storedatav = NULL;
        LEAFENTRY leaf_entry;
        r = toku_omt_fetch(bn->buffer, index, &storedatav);
2248
        assert_zero(r);
2249
        CAST_FROM_VOIDP(leaf_entry, storedatav);
2250
        ft_basement_node_gc_once(bn, index, leaf_entry, snapshot_xids, referenced_xids, live_root_txns, oldest_referenced_xid_known, delta);
2251 2252 2253 2254 2255 2256 2257
        // Check if the leaf entry was deleted or not.
        if (num_leafentries_before == toku_omt_size(bn->buffer)) {
            ++index;
        }
    }
}

John Esmet's avatar
John Esmet committed
2258
// Garbage collect all leaf entires in all basement nodes.
2259
static void
2260
ft_leaf_gc_all_les(FTNODE node,
John Esmet's avatar
John Esmet committed
2261 2262 2263 2264
                   FT ft,
                   const xid_omt_t &snapshot_xids,
                   const rx_omt_t &referenced_xids,
                   const xid_omt_t &live_root_txns,
2265
                   TXNID oldest_referenced_xid_known)
2266 2267
{
    toku_assert_entire_node_in_memory(node);
2268
    paranoid_invariant_zero(node->height);
2269 2270 2271 2272 2273 2274 2275
    // Loop through each leaf entry, garbage collecting as we go.
    for (int i = 0; i < node->n_children; ++i) {
        // Perform the garbage collection.
        BASEMENTNODE bn = BLB(node, i);
        STAT64INFO_S delta;
        delta.numrows = 0;
        delta.numbytes = 0;
2276
        basement_node_gc_all_les(bn, snapshot_xids, referenced_xids, live_root_txns, oldest_referenced_xid_known, &delta);
John Esmet's avatar
John Esmet committed
2277
        toku_ft_update_stats(&ft->in_memory_stats, delta);
2278 2279 2280
    }
}

2281 2282 2283 2284 2285 2286 2287 2288 2289
static void
ft_leaf_run_gc(FTNODE node, FT ft) {
    TOKULOGGER logger = toku_cachefile_logger(ft->cf);
    if (logger) {
        xid_omt_t snapshot_txnids;
        rx_omt_t referenced_xids;
        xid_omt_t live_root_txns;
        toku_txn_manager_clone_state_for_gc(
            logger->txn_manager,
2290 2291 2292
            &snapshot_txnids,
            &referenced_xids,
            &live_root_txns
2293 2294 2295 2296 2297 2298 2299 2300 2301
            );
        
        // Perform garbage collection. Provide a full snapshot of the transaction
        // system plus the oldest known referenced xid that could have had messages
        // applied to this leaf.
        //
        // Using the oldest xid in either the referenced_xids or live_root_txns
        // snapshots is not sufficient, because there could be something older that is neither
        // live nor referenced, but instead aborted somewhere above us as a message in the tree.
2302
        ft_leaf_gc_all_les(node, ft, snapshot_txnids, referenced_xids, live_root_txns, node->oldest_referenced_xid_known);
2303 2304 2305 2306
        
        // Free the OMT's we used for garbage collecting.
        snapshot_txnids.destroy();
        referenced_xids.destroy();
2307
        live_root_txns.destroy();
2308 2309 2310
    }
}

2311
void toku_bnc_flush_to_child(
John Esmet's avatar
John Esmet committed
2312
    FT ft,
2313
    NONLEAF_CHILDINFO bnc,
John Esmet's avatar
John Esmet committed
2314
    FTNODE child,
2315
    TXNID oldest_referenced_xid_known
Zardosht Kasheff's avatar
Zardosht Kasheff committed
2316
    )
2317
{
2318 2319
    paranoid_invariant(bnc);
    paranoid_invariant(toku_fifo_n_entries(bnc->buffer)>0);
2320
    STAT64INFO_S stats_delta = {0,0};
2321
    size_t remaining_memsize = toku_fifo_buffer_size_in_use(bnc->buffer);
2322 2323 2324 2325
    FIFO_ITERATE(
        bnc->buffer, key, keylen, val, vallen, type, msn, xids, is_fresh,
        ({
            DBT hk,hv;
2326 2327
            FT_MSG_S ftcmd = { type, msn, xids, .u = { .id = { toku_fill_dbt(&hk, key, keylen),
                                                               toku_fill_dbt(&hv, val, vallen) } } };
2328 2329 2330 2331 2332 2333 2334 2335 2336 2337
            size_t flow_deltas[] = { 0, 0 };
            if (remaining_memsize <= bnc->flow[0]) {
                // this message is in the current checkpoint's worth of
                // the end of the fifo
                flow_deltas[0] = FIFO_CURRENT_ENTRY_MEMSIZE;
            } else if (remaining_memsize <= bnc->flow[0] + bnc->flow[1]) {
                // this message is in the last checkpoint's worth of the
                // end of the fifo
                flow_deltas[1] = FIFO_CURRENT_ENTRY_MEMSIZE;
            }
2338
            toku_ft_node_put_cmd(
John Esmet's avatar
John Esmet committed
2339 2340 2341
                ft->compare_fun,
                ft->update_fun,
                &ft->cmp_descriptor,
2342
                child,
2343
                -1,
2344
                &ftcmd,
2345
                is_fresh,
2346
                make_gc_info(true), // mvcc_needed
2347
                flow_deltas,
2348
                &stats_delta
Zardosht Kasheff's avatar
Zardosht Kasheff committed
2349
                );
2350
            remaining_memsize -= FIFO_CURRENT_ENTRY_MEMSIZE;
2351
        }));
John Esmet's avatar
John Esmet committed
2352
    child->oldest_referenced_xid_known = oldest_referenced_xid_known;
John Esmet's avatar
John Esmet committed
2353

2354
    invariant(remaining_memsize == 0);
2355
    if (stats_delta.numbytes || stats_delta.numrows) {
John Esmet's avatar
John Esmet committed
2356
        toku_ft_update_stats(&ft->in_memory_stats, stats_delta);
2357
    }
2358 2359
    if (child->height == 0) {
        ft_leaf_run_gc(child, ft);
2360 2361 2362 2363
        size_t buffsize = toku_fifo_buffer_size_in_use(bnc->buffer);
        STATUS_INC(FT_MSG_BYTES_OUT, buffsize);
        // may be misleading if there's a broadcast message in there
        STATUS_INC(FT_MSG_BYTES_CURR, -buffsize);
Zardosht Kasheff's avatar
Zardosht Kasheff committed
2364
    }
2365 2366
}

2367 2368 2369 2370 2371 2372
bool toku_bnc_should_promote(FT ft, NONLEAF_CHILDINFO bnc) {
    static const double factor = 0.125;
    const uint64_t flow_threshold = ft->h->nodesize * factor;
    return bnc->flow[0] >= flow_threshold || bnc->flow[1] >= flow_threshold;
}

Leif Walsh's avatar
Leif Walsh committed
2373
void
2374 2375 2376
toku_ft_node_put_cmd (
    ft_compare_func compare_fun,
    ft_update_func update_fun,
Zardosht Kasheff's avatar
Zardosht Kasheff committed
2377
    DESCRIPTOR desc,
2378
    FTNODE node,
2379
    int target_childnum,
2380
    FT_MSG cmd,
2381
    bool is_fresh,
2382
    GC_INFO gc_info,
2383
    size_t flow_deltas[],
2384
    STAT64INFO stats_to_update
Zardosht Kasheff's avatar
Zardosht Kasheff committed
2385
    )
2386 2387
// Effect: Push CMD into the subtree rooted at NODE.
//   If NODE is a leaf, then
2388
//   put CMD into leaf, applying it to the leafentries
2389 2390
//   If NODE is a nonleaf, then push the cmd into the FIFO(s) of the relevent child(ren).
//   The node may become overfull.  That's not our problem.
2391
{
2392
    toku_assert_entire_node_in_memory(node);
2393
    //
2394
    // see comments in toku_ft_leaf_apply_cmd
2395 2396 2397 2398
    // to understand why we don't handle setting
    // node->max_msn_applied_to_node_on_disk here,
    // and instead defer to these functions
    //
2399
    if (node->height==0) {
2400
        toku_ft_leaf_apply_cmd(compare_fun, update_fun, desc, node, target_childnum, cmd, gc_info, nullptr, stats_to_update);
2401
    } else {
2402
        ft_nonleaf_put_cmd(compare_fun, desc, node, target_childnum, cmd, is_fresh, flow_deltas);
2403
    }
2404 2405
}

2406
static const struct pivot_bounds infinite_bounds = {.lower_bound_exclusive=NULL,
2407
                                                    .upper_bound_inclusive=NULL};
2408

Zardosht Kasheff's avatar
Zardosht Kasheff committed
2409

2410
// Effect: applies the cmd to the leaf if the appropriate basement node is in memory.
2411 2412
//           This function is called during message injection and/or flushing, so the entire
//           node MUST be in memory.
2413 2414 2415
void toku_ft_leaf_apply_cmd(
    ft_compare_func compare_fun,
    ft_update_func update_fun,
2416
    DESCRIPTOR desc,
2417
    FTNODE node,
2418
    int target_childnum,  // which child to inject to, or -1 if unknown
2419
    FT_MSG cmd,
2420
    GC_INFO gc_info,
2421 2422
    uint64_t *workdone,
    STAT64INFO stats_to_update
2423
    )
Zardosht Kasheff's avatar
Zardosht Kasheff committed
2424
{
2425
    VERIFY_NODE(t, node);
2426
    toku_assert_entire_node_in_memory(node);
2427

2428
    //
2429
    // Because toku_ft_leaf_apply_cmd is called with the intent of permanently
2430 2431
    // applying a message to a leaf node (meaning the message is permanently applied
    // and will be purged from the system after this call, as opposed to
2432
    // toku_apply_ancestors_messages_to_node, which applies a message
2433 2434
    // for a query, but the message may still reside in the system and
    // be reapplied later), we mark the node as dirty and
2435 2436
    // take the opportunity to update node->max_msn_applied_to_node_on_disk.
    //
2437
    node->dirty = 1;
2438

2439 2440 2441
    //
    // we cannot blindly update node->max_msn_applied_to_node_on_disk,
    // we must check to see if the msn is greater that the one already stored,
2442
    // because the cmd may have already been applied earlier (via
2443
    // toku_apply_ancestors_messages_to_node) to answer a query
2444
    //
2445
    // This is why we handle node->max_msn_applied_to_node_on_disk both here
2446
    // and in ft_nonleaf_put_cmd, as opposed to in one location, toku_ft_node_put_cmd.
2447
    //
2448
    MSN cmd_msn = cmd->msn;
2449 2450 2451
    if (cmd_msn.msn > node->max_msn_applied_to_node_on_disk.msn) {
        node->max_msn_applied_to_node_on_disk = cmd_msn;
    }
2452

John Esmet's avatar
John Esmet committed
2453 2454
    // Pass the oldest possible live xid value to each basementnode
    // when we apply messages to them.
John Esmet's avatar
John Esmet committed
2455
    TXNID oldest_referenced_xid_known = node->oldest_referenced_xid_known;
John Esmet's avatar
John Esmet committed
2456

2457
    if (ft_msg_applies_once(cmd)) {
2458 2459 2460 2461 2462 2463
        unsigned int childnum = (target_childnum >= 0
                                 ? target_childnum
                                 : toku_ftnode_which_child(node, cmd->u.id.key, desc, compare_fun));
        BASEMENTNODE bn = BLB(node, childnum);
        if (cmd->msn.msn > bn->max_msn_applied.msn) {
            bn->max_msn_applied = cmd->msn;
2464
            toku_ft_bn_apply_cmd(compare_fun,
2465 2466 2467 2468
                                 update_fun,
                                 desc,
                                 bn,
                                 cmd,
2469
                                 oldest_referenced_xid_known,
2470
                                 gc_info,
2471 2472
                                 workdone,
                                 stats_to_update);
2473
        } else {
2474
            STATUS_INC(FT_MSN_DISCARDS, 1);
2475 2476
        }
    }
2477
    else if (ft_msg_applies_all(cmd)) {
2478
        for (int childnum=0; childnum<node->n_children; childnum++) {
2479 2480
            if (cmd->msn.msn > BLB(node, childnum)->max_msn_applied.msn) {
                BLB(node, childnum)->max_msn_applied = cmd->msn;
2481
                toku_ft_bn_apply_cmd(compare_fun,
2482 2483 2484 2485
                                      update_fun,
                                      desc,
                                      BLB(node, childnum),
                                      cmd,
2486
                                      oldest_referenced_xid_known,
2487
                                      gc_info,
2488 2489
                                      workdone,
                                      stats_to_update);
2490
            } else {
2491
                STATUS_INC(FT_MSN_DISCARDS, 1);
2492
            }
2493
        }
2494
    }
2495
    else if (!ft_msg_does_nothing(cmd)) {
2496
        abort();
2497 2498 2499 2500
    }
    VERIFY_NODE(t, node);
}

Zardosht Kasheff's avatar
Zardosht Kasheff committed
2501 2502 2503 2504 2505 2506
static void inject_message_in_locked_node(
    FT ft, 
    FTNODE node, 
    int childnum, 
    FT_MSG_S *cmd, 
    size_t flow_deltas[],
2507 2508
    TXNID oldest_referenced_xid,
    GC_INFO gc_info
Zardosht Kasheff's avatar
Zardosht Kasheff committed
2509 2510
    ) 
{
2511 2512 2513 2514 2515
    // No guarantee that we're the writer, but oh well.
    // TODO(leif): Implement "do I have the lock or is it someone else?"
    // check in frwlock.  Should be possible with TOKU_PTHREAD_DEBUG, nop
    // otherwise.
    invariant(toku_ctpair_is_write_locked(node->ct_pair));
2516
    toku_assert_entire_node_in_memory(node);
John Esmet's avatar
John Esmet committed
2517 2518 2519 2520

    // Update the oldest known referenced xid for this node if it is younger
    // than the one currently known. Otherwise, it's better to keep the heurstic
    // we have and ignore this one.
2521 2522
    if (oldest_referenced_xid >= node->oldest_referenced_xid_known) {
        node->oldest_referenced_xid_known = oldest_referenced_xid;
John Esmet's avatar
John Esmet committed
2523 2524
    }

2525 2526 2527 2528 2529
    // Get the MSN from the header.  Now that we have a write lock on the
    // node we're injecting into, we know no other thread will get an MSN
    // after us and get that message into our subtree before us.
    cmd->msn.msn = toku_sync_add_and_fetch(&ft->h->max_msn_in_ft.msn, 1);
    paranoid_invariant(cmd->msn.msn > node->max_msn_applied_to_node_on_disk.msn);
2530
    STAT64INFO_S stats_delta = {0,0};
2531
    toku_ft_node_put_cmd(
2532 2533 2534
        ft->compare_fun,
        ft->update_fun,
        &ft->cmp_descriptor,
2535
        node,
2536
        childnum,
2537
        cmd,
2538
        true,
2539
        gc_info,
2540
        flow_deltas,
2541
        &stats_delta
2542
        );
2543
    if (stats_delta.numbytes || stats_delta.numrows) {
2544
        toku_ft_update_stats(&ft->in_memory_stats, stats_delta);
2545
    }
2546
    //
2547
    // assumption is that toku_ft_node_put_cmd will
2548 2549 2550
    // mark the node as dirty.
    // enforcing invariant here.
    //
2551
    paranoid_invariant(node->dirty != 0);
2552

2553
    // TODO: Why not at height 0?
2554 2555
    // update some status variables
    if (node->height != 0) {
2556
        uint64_t msgsize = ft_msg_size(cmd);
2557 2558 2559
        STATUS_INC(FT_MSG_BYTES_IN, msgsize);
        STATUS_INC(FT_MSG_BYTES_CURR, msgsize);
        STATUS_INC(FT_MSG_NUM, 1);
2560
        if (ft_msg_applies_all(cmd)) {
2561
            STATUS_INC(FT_MSG_NUM_BROADCAST, 1);
2562
        }
2563
    }
2564 2565 2566 2567

    // verify that msn of latest message was captured in root node
    paranoid_invariant(cmd->msn.msn == node->max_msn_applied_to_node_on_disk.msn);

John Esmet's avatar
John Esmet committed
2568
    // if we call toku_ft_flush_some_child, then that function unpins the root
2569 2570
    // otherwise, we unpin ourselves
    if (node->height > 0 && toku_ft_nonleaf_is_gorged(node, ft->h->nodesize)) {
John Esmet's avatar
John Esmet committed
2571
        toku_ft_flush_node_on_background_thread(ft, node);
2572 2573 2574 2575
    }
    else {
        toku_unpin_ftnode(ft, node);
    }
Bradley C. Kuszmaul's avatar
Rename  
Bradley C. Kuszmaul committed
2576 2577
}

2578 2579 2580 2581 2582 2583 2584 2585 2586
// seqinsert_loc is a bitmask.
// The root counts as being both on the "left extreme" and on the "right extreme".
// Therefore, at the root, you're at LEFT_EXTREME | RIGHT_EXTREME.
typedef char seqinsert_loc;
static const seqinsert_loc NEITHER_EXTREME = 0;
static const seqinsert_loc LEFT_EXTREME = 1;
static const seqinsert_loc RIGHT_EXTREME = 2;

static bool process_maybe_reactive_child(FT ft, FTNODE parent, FTNODE child, int childnum, seqinsert_loc loc)
2587
// Effect:
2588 2589 2590 2591 2592 2593 2594
//  If child needs to be split or merged, do that.
//  parent and child will be unlocked if this happens
//  also, the batched pin will have ended if this happens
// Requires: parent and child are read locked
// Returns:
//  true if relocking is needed
//  false otherwise
2595
{
2596 2597 2598 2599 2600 2601 2602 2603 2604 2605 2606 2607 2608 2609 2610 2611 2612 2613 2614 2615 2616 2617 2618 2619 2620 2621 2622 2623 2624 2625 2626 2627 2628 2629 2630 2631 2632 2633 2634 2635 2636 2637 2638 2639 2640 2641 2642 2643 2644 2645 2646 2647 2648 2649 2650 2651 2652 2653 2654 2655 2656 2657 2658 2659 2660 2661 2662 2663 2664 2665 2666 2667 2668 2669 2670 2671 2672 2673 2674 2675 2676
    enum reactivity re = get_node_reactivity(child, ft->h->nodesize);
    enum reactivity newre;
    BLOCKNUM child_blocknum;
    uint32_t child_fullhash;
    switch (re) {
    case RE_STABLE:
        return false;
    case RE_FISSIBLE:
        {
            // We only have a read lock on the parent.  We need to drop both locks, and get write locks.
            BLOCKNUM parent_blocknum = parent->thisnodename;
            uint32_t parent_fullhash = toku_cachetable_hash(ft->cf, parent_blocknum);
            int parent_height = parent->height;
            int parent_n_children = parent->n_children;
            toku_unpin_ftnode_read_only(ft, child);
            toku_unpin_ftnode_read_only(ft, parent);
            struct ftnode_fetch_extra bfe;
            fill_bfe_for_full_read(&bfe, ft);
            FTNODE newparent, newchild;
            toku_pin_ftnode_off_client_thread_batched(ft, parent_blocknum, parent_fullhash, &bfe, PL_WRITE_CHEAP, 0, nullptr, &newparent);
            if (newparent->height != parent_height || newparent->n_children != parent_n_children ||
                childnum >= newparent->n_children || toku_bnc_n_entries(BNC(newparent, childnum))) {
                // If the height changed or childnum is now off the end, something clearly got split or merged out from under us.
                // If something got injected in this node, then it got split or merged and we shouldn't be splitting it.
                // But we already unpinned the child so we need to have the caller re-try the pins.
                toku_unpin_ftnode_read_only(ft, newparent);
                return true;
            }
            // It's ok to reuse the same childnum because if we get something
            // else we need to split, well, that's crazy, but let's go ahead
            // and split it.
            child_blocknum = BP_BLOCKNUM(newparent, childnum);
            child_fullhash = compute_child_fullhash(ft->cf, newparent, childnum);
            toku_pin_ftnode_off_client_thread_batched(ft, child_blocknum, child_fullhash, &bfe, PL_WRITE_CHEAP, 1, &newparent, &newchild);
            newre = get_node_reactivity(newchild, ft->h->nodesize);
            if (newre == RE_FISSIBLE) {
                enum split_mode split_mode;
                if (newparent->height == 1 && (loc & LEFT_EXTREME) && childnum == 0) {
                    split_mode = SPLIT_RIGHT_HEAVY;
                } else if (newparent->height == 1 && (loc & RIGHT_EXTREME) && childnum == newparent->n_children - 1) {
                    split_mode = SPLIT_LEFT_HEAVY;
                } else {
                    split_mode = SPLIT_EVENLY;
                }
                toku_ft_split_child(ft, newparent, childnum, newchild, split_mode);
            } else {
                // some other thread already got it, just unpin and tell the
                // caller to retry
                toku_unpin_ftnode_read_only(ft, newchild);
                toku_unpin_ftnode_read_only(ft, newparent);
            }
            return true;
        }
    case RE_FUSIBLE:
        {
            if (parent->height == 1) {
                // prevent re-merging of recently unevenly-split nodes
                if (((loc & LEFT_EXTREME) && childnum <= 1) ||
                    ((loc & RIGHT_EXTREME) && childnum >= parent->n_children - 2)) {
                    return false;
                }
            }

            int parent_height = parent->height;
            BLOCKNUM parent_blocknum = parent->thisnodename;
            uint32_t parent_fullhash = toku_cachetable_hash(ft->cf, parent_blocknum);
            toku_unpin_ftnode_read_only(ft, child);
            toku_unpin_ftnode_read_only(ft, parent);
            struct ftnode_fetch_extra bfe;
            fill_bfe_for_full_read(&bfe, ft);
            FTNODE newparent, newchild;
            toku_pin_ftnode_off_client_thread_batched(ft, parent_blocknum, parent_fullhash, &bfe, PL_WRITE_CHEAP, 0, nullptr, &newparent);
            if (newparent->height != parent_height || childnum >= newparent->n_children) {
                // looks like this is the root and it got merged, let's just start over (like in the split case above)
                toku_unpin_ftnode_read_only(ft, newparent);
                return true;
            }
            child_blocknum = BP_BLOCKNUM(newparent, childnum);
            child_fullhash = compute_child_fullhash(ft->cf, newparent, childnum);
            toku_pin_ftnode_off_client_thread_batched(ft, child_blocknum, child_fullhash, &bfe, PL_READ, 1, &newparent, &newchild);
            newre = get_node_reactivity(newchild, ft->h->nodesize);
2677
            if (newre == RE_FUSIBLE && newparent->n_children >= 2) {
2678 2679 2680
                toku_unpin_ftnode_read_only(ft, newchild);
                toku_ft_merge_child(ft, newparent, childnum);
            } else {
2681 2682 2683 2684 2685 2686 2687 2688 2689
                // Could be a weird case where newparent has only one
                // child.  In this case, we want to inject here but we've
                // already unpinned the caller's copy of parent so we have
                // to ask them to re-pin, or they could (very rarely)
                // dereferenced memory in a freed node.  TODO: we could
                // give them back the copy of the parent we pinned.
                //
                // Otherwise, some other thread already got it, just unpin
                // and tell the caller to retry
2690 2691 2692 2693 2694
                toku_unpin_ftnode_read_only(ft, newchild);
                toku_unpin_ftnode_read_only(ft, newparent);
            }
            return true;
        }
2695
    }
2696 2697
    abort();
}
2698

2699
static void inject_message_at_this_blocknum(FT ft, CACHEKEY cachekey, uint32_t fullhash, FT_MSG_S *cmd, size_t flow_deltas[], TXNID oldest_referenced_xid, GC_INFO gc_info)
2700 2701 2702 2703
// Effect:
//  Inject cmd into the node at this blocknum (cachekey).
//  Gets a write lock on the node for you.
{
2704
    FTNODE node;
2705 2706 2707 2708 2709 2710
    struct ftnode_fetch_extra bfe;
    fill_bfe_for_full_read(&bfe, ft);
    toku_pin_ftnode_off_client_thread_batched(ft, cachekey, fullhash, &bfe, PL_WRITE_CHEAP, 0, NULL, &node);
    toku_assert_entire_node_in_memory(node);
    paranoid_invariant(node->fullhash==fullhash);
    ft_verify_flags(ft, node);
2711
    inject_message_in_locked_node(ft, node, -1, cmd, flow_deltas, oldest_referenced_xid, gc_info);
2712
}
Zardosht Kasheff's avatar
Zardosht Kasheff committed
2713

2714 2715 2716 2717 2718 2719 2720 2721 2722 2723
__attribute__((const))
static inline bool should_inject_in_node(seqinsert_loc loc, int height, int depth)
// We should inject directly in a node if:
//  - it's a leaf, or
//  - it's a height 1 node not at either extreme, or
//  - it's a depth 2 node not at either extreme
{
    return (height == 0 || (loc == NEITHER_EXTREME && (height <= 1 || depth >= 2)));
}

Zardosht Kasheff's avatar
Zardosht Kasheff committed
2724 2725 2726 2727 2728 2729 2730
static void push_something_in_subtree(
    FT ft, 
    FTNODE subtree_root, 
    int target_childnum, 
    FT_MSG_S *cmd, 
    size_t flow_deltas[], 
    TXNID oldest_referenced_xid, 
2731
    GC_INFO gc_info,
Zardosht Kasheff's avatar
Zardosht Kasheff committed
2732 2733 2734 2735
    int depth, 
    seqinsert_loc loc, 
    bool just_did_split_or_merge
    )
2736 2737 2738 2739 2740 2741 2742 2743 2744 2745 2746 2747 2748 2749 2750 2751 2752 2753 2754 2755 2756 2757 2758 2759 2760 2761 2762 2763 2764 2765 2766 2767 2768 2769 2770 2771
// Effects:
//  Assign cmd an MSN from ft->h.
//  Put cmd in the subtree rooted at node.  Due to promotion the message may not be injected directly in this node.
//  Unlock node or schedule it to be unlocked (after a background flush).
//   Either way, the caller is not responsible for unlocking node.
// Requires:
//  subtree_root is read locked and fully in memory.
// Notes:
//  In Ming, the basic rules of promotion are as follows:
//   Don't promote broadcast messages.
//   Don't promote past non-empty buffers.
//   Otherwise, promote at most to height 1 or depth 2 (whichever is highest), as far as the birdie asks you to promote.
//    We don't promote to leaves because injecting into leaves is expensive, mostly because of #5605 and some of #5552.
//    We don't promote past depth 2 because we found that gives us enough parallelism without costing us too much pinning work.
//
//    This is true with the following caveats:
//     We always promote all the way to the leaves on the rightmost and leftmost edges of the tree, for sequential insertions.
//      (That means we can promote past depth 2 near the edges of the tree.)
//
//   When the birdie is still saying we should promote, we use get_and_pin so that we wait to get the node.
//   If the birdie doesn't say to promote, we try maybe_get_and_pin.  If we get the node cheaply, and it's dirty, we promote anyway.
{
    toku_assert_entire_node_in_memory(subtree_root);
    if (should_inject_in_node(loc, subtree_root->height, depth)) {
        switch (depth) {
        case 0:
            STATUS_INC(FT_PRO_NUM_INJECT_DEPTH_0, 1); break;
        case 1:
            STATUS_INC(FT_PRO_NUM_INJECT_DEPTH_1, 1); break;
        case 2:
            STATUS_INC(FT_PRO_NUM_INJECT_DEPTH_2, 1); break;
        case 3:
            STATUS_INC(FT_PRO_NUM_INJECT_DEPTH_3, 1); break;
        default:
            STATUS_INC(FT_PRO_NUM_INJECT_DEPTH_GT3, 1); break;
        }
2772
        inject_message_in_locked_node(ft, subtree_root, target_childnum, cmd, flow_deltas, oldest_referenced_xid, gc_info);
2773 2774 2775 2776 2777 2778 2779 2780 2781 2782 2783 2784 2785 2786 2787 2788 2789 2790 2791 2792 2793 2794 2795 2796 2797 2798 2799 2800 2801 2802 2803 2804 2805 2806 2807 2808 2809 2810 2811 2812 2813 2814 2815 2816 2817 2818 2819 2820 2821 2822 2823 2824 2825 2826 2827 2828 2829 2830 2831 2832 2833 2834 2835 2836 2837 2838 2839 2840 2841 2842 2843 2844 2845 2846 2847 2848 2849 2850 2851 2852 2853 2854 2855 2856 2857 2858 2859 2860
    } else {
        int r;
        int childnum;
        NONLEAF_CHILDINFO bnc;

        // toku_ft_root_put_cmd should not have called us otherwise.
        paranoid_invariant(ft_msg_applies_once(cmd));

        childnum = (target_childnum >= 0 ? target_childnum
                    : toku_ftnode_which_child(subtree_root, cmd->u.id.key, &ft->cmp_descriptor, ft->compare_fun));
        bnc = BNC(subtree_root, childnum);

        if (toku_bnc_n_entries(bnc) > 0) {
            // The buffer is non-empty, give up on promoting.
            STATUS_INC(FT_PRO_NUM_STOP_NONEMPTY_BUF, 1);
            goto relock_and_push_here;
        }

        seqinsert_loc next_loc;
        if ((loc & LEFT_EXTREME) && childnum == 0) {
            next_loc = LEFT_EXTREME;
        } else if ((loc & RIGHT_EXTREME) && childnum == subtree_root->n_children - 1) {
            next_loc = RIGHT_EXTREME;
        } else {
            next_loc = NEITHER_EXTREME;
        }

        if (next_loc == NEITHER_EXTREME && subtree_root->height <= 1) {
            // Never promote to leaf nodes except on the edges
            STATUS_INC(FT_PRO_NUM_STOP_H1, 1);
            goto relock_and_push_here;
        }

        {
            const BLOCKNUM child_blocknum = BP_BLOCKNUM(subtree_root, childnum);
            toku_verify_blocknum_allocated(ft->blocktable, child_blocknum);
            const uint32_t child_fullhash = toku_cachetable_hash(ft->cf, child_blocknum);

            FTNODE child;
            {
                const int child_height = subtree_root->height - 1;
                const int child_depth = depth + 1;
                // If we're locking a leaf, or a height 1 node or depth 2
                // node in the middle, we know we won't promote further
                // than that, so just get a write lock now.
                const pair_lock_type lock_type = (should_inject_in_node(next_loc, child_height, child_depth)
                                                  ? PL_WRITE_CHEAP
                                                  : PL_READ);
                if (next_loc != NEITHER_EXTREME || (toku_bnc_should_promote(ft, bnc) && depth <= 1)) {
                    // If we're on either extreme, or the birdie wants to
                    // promote and we're in the top two levels of the
                    // tree, don't stop just because someone else has the
                    // node locked.
                    struct ftnode_fetch_extra bfe;
                    fill_bfe_for_full_read(&bfe, ft);
                    toku_pin_ftnode_off_client_thread_batched(ft, child_blocknum, child_fullhash, &bfe, lock_type, 0, nullptr, &child);
                } else {
                    r = toku_maybe_pin_ftnode_clean(ft, child_blocknum, child_fullhash, lock_type, &child);
                    if (r != 0) {
                        // We couldn't get the child cheaply, so give up on promoting.
                        STATUS_INC(FT_PRO_NUM_STOP_LOCK_CHILD, 1);
                        goto relock_and_push_here;
                    }
                    if (is_entire_node_in_memory(child)) {
                        // toku_pin_ftnode... touches the clock but toku_maybe_pin_ftnode... doesn't.
                        // This prevents partial eviction.
                        for (int i = 0; i < child->n_children; ++i) {
                            BP_TOUCH_CLOCK(child, i);
                        }
                    } else {
                        // We got the child, but it's not fully in memory.  Give up on promoting.
                        STATUS_INC(FT_PRO_NUM_STOP_CHILD_INMEM, 1);
                        goto unlock_child_and_push_here;
                    }
                }
            }
            paranoid_invariant_notnull(child);

            if (!just_did_split_or_merge) {
                BLOCKNUM subtree_root_blocknum = subtree_root->thisnodename;
                uint32_t subtree_root_fullhash = toku_cachetable_hash(ft->cf, subtree_root_blocknum);
                const bool did_split_or_merge = process_maybe_reactive_child(ft, subtree_root, child, childnum, loc);
                if (did_split_or_merge) {
                    // Need to re-pin this node and try at this level again.
                    FTNODE newparent;
                    struct ftnode_fetch_extra bfe;
                    fill_bfe_for_full_read(&bfe, ft); // should be fully in memory, we just split it
                    toku_pin_ftnode_off_client_thread_batched(ft, subtree_root_blocknum, subtree_root_fullhash, &bfe, PL_READ, 0, nullptr, &newparent);
2861
                    push_something_in_subtree(ft, newparent, -1, cmd, flow_deltas, oldest_referenced_xid, gc_info, depth, loc, true);
2862 2863 2864
                    return;
                }
            }
2865

2866
            if (next_loc != NEITHER_EXTREME || child->dirty || toku_bnc_should_promote(ft, bnc)) {
2867
                push_something_in_subtree(ft, child, -1, cmd, flow_deltas, oldest_referenced_xid, gc_info, depth + 1, next_loc, false);
2868 2869 2870 2871 2872 2873
                toku_sync_fetch_and_add(&bnc->flow[0], flow_deltas[0]);
                // The recursive call unpinned the child, but
                // we're responsible for unpinning subtree_root.
                toku_unpin_ftnode_read_only(ft, subtree_root);
                return;
            }
2874

2875 2876 2877 2878 2879 2880 2881 2882 2883 2884 2885 2886 2887 2888 2889 2890 2891 2892 2893 2894 2895 2896 2897 2898 2899 2900 2901 2902
            STATUS_INC(FT_PRO_NUM_DIDNT_WANT_PROMOTE, 1);
        unlock_child_and_push_here:
            // We locked the child, but we decided not to promote.
            // Unlock the child, and fall through to the next case.
            toku_unpin_ftnode_read_only(ft, child);
        }
    relock_and_push_here:
        // Give up on promoting.
        // We have subtree_root read-locked and we don't have a child locked.
        // Drop the read lock, grab a write lock, and inject here.
        {
            // Right now we have a read lock on subtree_root, but we want
            // to inject into it so we get a write lock instead.
            BLOCKNUM subtree_root_blocknum = subtree_root->thisnodename;
            uint32_t subtree_root_fullhash = toku_cachetable_hash(ft->cf, subtree_root_blocknum);
            toku_unpin_ftnode_read_only(ft, subtree_root);
            switch (depth) {
            case 0:
                STATUS_INC(FT_PRO_NUM_INJECT_DEPTH_0, 1); break;
            case 1:
                STATUS_INC(FT_PRO_NUM_INJECT_DEPTH_1, 1); break;
            case 2:
                STATUS_INC(FT_PRO_NUM_INJECT_DEPTH_2, 1); break;
            case 3:
                STATUS_INC(FT_PRO_NUM_INJECT_DEPTH_3, 1); break;
            default:
                STATUS_INC(FT_PRO_NUM_INJECT_DEPTH_GT3, 1); break;
            }
2903
            inject_message_at_this_blocknum(ft, subtree_root_blocknum, subtree_root_fullhash, cmd, flow_deltas, oldest_referenced_xid, gc_info);
2904 2905 2906 2907
        }
    }
}

2908 2909 2910 2911 2912 2913
void toku_ft_root_put_cmd(
    FT ft, 
    FT_MSG_S *cmd, 
    TXNID oldest_referenced_xid,
    GC_INFO gc_info
    )
2914 2915 2916
// Effect:
//  - assign msn to cmd and update msn in the header
//  - push the cmd into the ft
2917

2918 2919 2920 2921 2922 2923 2924 2925 2926 2927 2928 2929 2930 2931 2932 2933 2934 2935 2936
// As of Clayface, the root blocknum is a constant, so preventing a
// race between message injection and the split of a root is the job
// of the cachetable's locking rules.
//
// We also hold the MO lock for a number of reasons, but an important
// one is to make sure that a begin_checkpoint may not start while
// this code is executing. A begin_checkpoint does (at least) two things
// that can interfere with the operations here:
//  - Copies the header to a checkpoint header. Because we may change
//    the max_msn_in_ft below, we don't want the header to be copied in
//    the middle of these operations.
//  - Takes note of the log's LSN. Because this put operation has
//    already been logged, this message injection must be included
//    in any checkpoint that contains this put's logentry.
//    Holding the mo lock throughout this function ensures that fact.
{
    // blackhole fractal trees drop all messages, so do nothing.
    if (ft->blackhole) {
        return;
Zardosht Kasheff's avatar
Zardosht Kasheff committed
2937
    }
Zardosht Kasheff's avatar
Zardosht Kasheff committed
2938

2939 2940 2941 2942 2943 2944 2945 2946 2947 2948 2949 2950 2951 2952 2953 2954 2955 2956 2957 2958 2959 2960 2961 2962 2963 2964 2965 2966 2967 2968 2969 2970 2971 2972 2973 2974 2975 2976 2977 2978 2979 2980 2981 2982 2983 2984 2985 2986 2987 2988 2989 2990 2991 2992 2993 2994 2995 2996 2997 2998
    FTNODE node;

    uint32_t fullhash;
    CACHEKEY root_key;
    toku_calculate_root_offset_pointer(ft, &root_key, &fullhash);
    struct ftnode_fetch_extra bfe;
    fill_bfe_for_full_read(&bfe, ft);

    size_t flow_deltas[] = { toku_ft_msg_memsize_in_fifo(cmd), 0 };

    pair_lock_type lock_type;
    lock_type = PL_READ; // try first for a read lock
    // If we need to split the root, we'll have to change from a read lock
    // to a write lock and check again.  We change the variable lock_type
    // and jump back to here.
 change_lock_type:
    // get the root node
    toku_pin_ftnode_off_client_thread_batched(ft, root_key, fullhash, &bfe, lock_type, 0, NULL, &node);
    toku_assert_entire_node_in_memory(node);
    paranoid_invariant(node->fullhash==fullhash);
    ft_verify_flags(ft, node);

    // First handle a reactive root.
    // This relocking for split algorithm will cause every message
    // injection thread to change lock type back and forth, when only one
    // of them needs to in order to handle the split.  That's not great,
    // but root splits are incredibly rare.
    enum reactivity re = get_node_reactivity(node, ft->h->nodesize);
    switch (re) {
    case RE_STABLE:
    case RE_FUSIBLE: // cannot merge anything at the root
        if (lock_type != PL_READ) {
            // We thought we needed to split, but someone else got to
            // it before us.  Downgrade to a read lock.
            toku_unpin_ftnode_read_only(ft, node);
            lock_type = PL_READ;
            goto change_lock_type;
        }
        break;
    case RE_FISSIBLE:
        if (lock_type == PL_READ) {
            // Here, we only have a read lock on the root.  In order
            // to split it, we need a write lock, but in the course of
            // gaining the write lock, someone else may have gotten in
            // before us and split it.  So we upgrade to a write lock
            // and check again.
            toku_unpin_ftnode_read_only(ft, node);
            lock_type = PL_WRITE_CHEAP;
            goto change_lock_type;
        } else {
            // We have a write lock, now we can split.
            ft_init_new_root(ft, node, &node);
            // Then downgrade back to a read lock, and we can finally
            // do the injection.
            toku_unpin_ftnode_off_client_thread(ft, node);
            lock_type = PL_READ;
            STATUS_INC(FT_PRO_NUM_ROOT_SPLIT, 1);
            goto change_lock_type;
        }
        break;
Zardosht Kasheff's avatar
Zardosht Kasheff committed
2999
    }
3000 3001 3002 3003 3004 3005 3006 3007 3008 3009 3010 3011 3012
    // If we get to here, we have a read lock and the root doesn't
    // need to be split.  It's safe to inject the message.
    paranoid_invariant(lock_type == PL_READ);
    // We cannot assert that we have the read lock because frwlock asserts
    // that its mutex is locked when we check if there are any readers.
    // That wouldn't give us a strong guarantee that we have the read lock
    // anyway.

    // Now, either inject here or promote.  We decide based on a heuristic:
    if (node->height == 0 || !ft_msg_applies_once(cmd)) {
        // If the root's a leaf or we're injecting a broadcast, drop the read lock and inject here.
        toku_unpin_ftnode_read_only(ft, node);
        STATUS_INC(FT_PRO_NUM_ROOT_H0_INJECT, 1);
3013
        inject_message_at_this_blocknum(ft, root_key, fullhash, cmd, flow_deltas, oldest_referenced_xid, gc_info);
3014 3015
    } else if (node->height > 1) {
        // If the root's above height 1, we are definitely eligible for promotion.
3016
        push_something_in_subtree(ft, node, -1, cmd, flow_deltas, oldest_referenced_xid, gc_info, 0, LEFT_EXTREME | RIGHT_EXTREME, false);
3017 3018 3019 3020 3021 3022
    } else {
        // The root's height 1.  We may be eligible for promotion here.
        // On the extremes, we want to promote, in the middle, we don't.
        int childnum = toku_ftnode_which_child(node, cmd->u.id.key, &ft->cmp_descriptor, ft->compare_fun);
        if (childnum == 0 || childnum == node->n_children - 1) {
            // On the extremes, promote.  We know which childnum we're going to, so pass that down too.
3023
            push_something_in_subtree(ft, node, childnum, cmd, flow_deltas, oldest_referenced_xid, gc_info, 0, LEFT_EXTREME | RIGHT_EXTREME, false);
3024 3025 3026 3027
        } else {
            // At height 1 in the middle, don't promote, drop the read lock and inject here.
            toku_unpin_ftnode_read_only(ft, node);
            STATUS_INC(FT_PRO_NUM_ROOT_H1_INJECT, 1);
3028
            inject_message_at_this_blocknum(ft, root_key, fullhash, cmd, flow_deltas, oldest_referenced_xid, gc_info);
3029
        }
3030
    }
3031 3032
}

3033
// Effect: Insert the key-val pair into brt.
3034 3035
void toku_ft_insert (FT_HANDLE brt, DBT *key, DBT *val, TOKUTXN txn) {
    toku_ft_maybe_insert(brt, key, val, txn, false, ZERO_LSN, true, FT_INSERT);
3036 3037
}

3038
void toku_ft_load_recovery(TOKUTXN txn, FILENUM old_filenum, char const * new_iname, int do_fsync, int do_log, LSN *load_lsn) {
3039
    paranoid_invariant(txn);
Yoni Fogel's avatar
Yoni Fogel committed
3040
    toku_txn_force_fsync_on_commit(txn);  //If the txn commits, the commit MUST be in the log
3041
                                          //before the (old) file is actually unlinked
Yoni Fogel's avatar
Yoni Fogel committed
3042 3043
    TOKULOGGER logger = toku_txn_logger(txn);

3044
    BYTESTRING new_iname_bs = {.len=(uint32_t) strlen(new_iname), .data=(char*)new_iname};
3045 3046
    toku_logger_save_rollback_load(txn, old_filenum, &new_iname_bs);
    if (do_log && logger) {
3047
        TXNID_PAIR xid = toku_txn_get_txnid(txn);
3048
        toku_log_load(logger, load_lsn, do_fsync, txn, xid, old_filenum, new_iname_bs);
Yoni Fogel's avatar
Yoni Fogel committed
3049 3050 3051
    }
}

3052 3053 3054 3055
// 2954
// this function handles the tasks needed to be recoverable
//  - write to rollback log
//  - write to recovery log
3056
void toku_ft_hot_index_recovery(TOKUTXN txn, FILENUMS filenums, int do_fsync, int do_log, LSN *hot_index_lsn)
3057
{
3058
    paranoid_invariant(txn);
3059 3060 3061
    TOKULOGGER logger = toku_txn_logger(txn);

    // write to the rollback log
3062 3063
    toku_logger_save_rollback_hot_index(txn, &filenums);
    if (do_log && logger) {
3064
        TXNID_PAIR xid = toku_txn_get_txnid(txn);
3065
        // write to the recovery log
3066
        toku_log_hot_index(logger, hot_index_lsn, do_fsync, txn, xid, filenums);
3067 3068
    }
}
3069

3070 3071
// Effect: Optimize the ft.
void toku_ft_optimize (FT_HANDLE brt) {
3072
    TOKULOGGER logger = toku_cachefile_logger(brt->ft->cf);
3073
    if (logger) {
3074
        TXNID oldest = toku_txn_manager_get_oldest_living_xid(logger->txn_manager);
3075

3076 3077 3078 3079 3080 3081
        XIDS root_xids = xids_get_root_xids();
        XIDS message_xids;
        if (oldest == TXNID_NONE_LIVING) {
            message_xids = root_xids;
        }
        else {
3082 3083
            int r = xids_create_child(root_xids, &message_xids, oldest);
            invariant(r == 0);
3084
        }
3085

3086 3087 3088 3089
        DBT key;
        DBT val;
        toku_init_dbt(&key);
        toku_init_dbt(&val);
3090
        FT_MSG_S ftcmd = { FT_OPTIMIZE, ZERO_MSN, message_xids, .u = { .id = {&key,&val} } };
3091
        toku_ft_root_put_cmd(brt->ft, &ftcmd, TXNID_NONE, make_gc_info(true));
3092 3093
        xids_destroy(&message_xids);
    }
3094 3095
}

3096
void toku_ft_load(FT_HANDLE brt, TOKUTXN txn, char const * new_iname, int do_fsync, LSN *load_lsn) {
3097
    FILENUM old_filenum = toku_cachefile_filenum(brt->ft->cf);
Yoni Fogel's avatar
Yoni Fogel committed
3098
    int do_log = 1;
3099
    toku_ft_load_recovery(txn, old_filenum, new_iname, do_fsync, do_log, load_lsn);
Yoni Fogel's avatar
Yoni Fogel committed
3100 3101
}

3102 3103
// ft actions for logging hot index filenums
void toku_ft_hot_index(FT_HANDLE brt __attribute__ ((unused)), TOKUTXN txn, FILENUMS filenums, int do_fsync, LSN *lsn) {
3104
    int do_log = 1;
3105
    toku_ft_hot_index_recovery(txn, filenums, do_fsync, do_log, lsn);
3106 3107
}

3108
void
3109
toku_ft_log_put (TOKUTXN txn, FT_HANDLE brt, const DBT *key, const DBT *val) {
3110
    TOKULOGGER logger = toku_txn_logger(txn);
3111
    if (logger) {
3112 3113
        BYTESTRING keybs = {.len=key->size, .data=(char *) key->data};
        BYTESTRING valbs = {.len=val->size, .data=(char *) val->data};
3114
        TXNID_PAIR xid = toku_txn_get_txnid(txn);
3115
        toku_log_enq_insert(logger, (LSN*)0, 0, txn, toku_cachefile_filenum(brt->ft->cf), xid, keybs, valbs);
3116 3117 3118
    }
}

3119
void
3120
toku_ft_log_put_multiple (TOKUTXN txn, FT_HANDLE src_ft, FT_HANDLE *brts, uint32_t num_fts, const DBT *key, const DBT *val) {
3121
    assert(txn);
3122
    assert(num_fts > 0);
3123 3124
    TOKULOGGER logger = toku_txn_logger(txn);
    if (logger) {
3125
        FILENUM         fnums[num_fts];
3126
        uint32_t i;
3127
        for (i = 0; i < num_fts; i++) {
3128
            fnums[i] = toku_cachefile_filenum(brts[i]->ft->cf);
3129
        }
3130 3131 3132
        FILENUMS filenums = {.num = num_fts, .filenums = fnums};
        BYTESTRING keybs = {.len=key->size, .data=(char *) key->data};
        BYTESTRING valbs = {.len=val->size, .data=(char *) val->data};
3133
        TXNID_PAIR xid = toku_txn_get_txnid(txn);
3134 3135
        FILENUM src_filenum = src_ft ? toku_cachefile_filenum(src_ft->ft->cf) : FILENUM_NONE;
        toku_log_enq_insert_multiple(logger, (LSN*)0, 0, txn, src_filenum, filenums, xid, keybs, valbs);
3136 3137 3138
    }
}

3139
void toku_ft_maybe_insert (FT_HANDLE ft_h, DBT *key, DBT *val, TOKUTXN txn, bool oplsn_valid, LSN oplsn, bool do_logging, enum ft_msg_type type) {
3140
    paranoid_invariant(type==FT_INSERT || type==FT_INSERT_NO_OVERWRITE);
3141
    XIDS message_xids = xids_get_root_xids(); //By default use committed messages
3142
    TXNID_PAIR xid = toku_txn_get_txnid(txn);
3143
    if (txn) {
3144 3145 3146 3147
        BYTESTRING keybs = {key->size, (char *) key->data};
        toku_logger_save_rollback_cmdinsert(txn, toku_cachefile_filenum(ft_h->ft->cf), &keybs);
        toku_txn_maybe_note_ft(txn, ft_h->ft);
        message_xids = toku_txn_get_xids(txn);
Bradley C. Kuszmaul's avatar
Rename  
Bradley C. Kuszmaul committed
3148
    }
3149
    TOKULOGGER logger = toku_txn_logger(txn);
3150
    if (do_logging && logger) {
3151 3152
        BYTESTRING keybs = {.len=key->size, .data=(char *) key->data};
        BYTESTRING valbs = {.len=val->size, .data=(char *) val->data};
3153
        if (type == FT_INSERT) {
3154
            toku_log_enq_insert(logger, (LSN*)0, 0, txn, toku_cachefile_filenum(ft_h->ft->cf), xid, keybs, valbs);
3155 3156
        }
        else {
3157
            toku_log_enq_insert_no_overwrite(logger, (LSN*)0, 0, txn, toku_cachefile_filenum(ft_h->ft->cf), xid, keybs, valbs);
3158
        }
3159 3160
    }

3161
    LSN treelsn;
3162
    if (oplsn_valid && oplsn.lsn <= (treelsn = toku_ft_checkpoint_lsn(ft_h->ft)).lsn) {
3163
        // do nothing
3164
    } else {
3165
        TXNID oldest_referenced_xid = (txn) ? txn->oldest_referenced_xid : TXNID_NONE;
3166
        toku_ft_send_insert(ft_h, key, val, message_xids, type, oldest_referenced_xid, make_gc_info(txn ? !txn->for_recovery : false));
3167
    }
3168
}
3169

3170
static void
3171
ft_send_update_msg(FT_HANDLE brt, FT_MSG_S *msg, TOKUTXN txn) {
3172
    msg->xids = (txn
3173 3174
                 ? toku_txn_get_xids(txn)
                 : xids_get_root_xids());
3175 3176
    
    TXNID oldest_referenced_xid = (txn) ? txn->oldest_referenced_xid : TXNID_NONE;
3177
    toku_ft_root_put_cmd(brt->ft, msg, oldest_referenced_xid, make_gc_info(txn ? !txn->for_recovery : false));
3178 3179
}

3180
void toku_ft_maybe_update(FT_HANDLE ft_h, const DBT *key, const DBT *update_function_extra,
Yoni Fogel's avatar
Yoni Fogel committed
3181 3182
                      TOKUTXN txn, bool oplsn_valid, LSN oplsn,
                      bool do_logging) {
3183
    TXNID_PAIR xid = toku_txn_get_txnid(txn);
3184
    if (txn) {
3185
        BYTESTRING keybs = { key->size, (char *) key->data };
3186
        toku_logger_save_rollback_cmdupdate(
3187 3188
            txn, toku_cachefile_filenum(ft_h->ft->cf), &keybs);
        toku_txn_maybe_note_ft(txn, ft_h->ft);
3189 3190
    }

3191 3192
    TOKULOGGER logger;
    logger = toku_txn_logger(txn);
3193
    if (do_logging && logger) {
3194
        BYTESTRING keybs = {.len=key->size, .data=(char *) key->data};
3195
        BYTESTRING extrabs = {.len=update_function_extra->size,
3196
                              .data = (char *) update_function_extra->data};
3197
        toku_log_enq_update(logger, NULL, 0, txn,
3198
                                toku_cachefile_filenum(ft_h->ft->cf),
3199
                                xid, keybs, extrabs);
3200 3201 3202
    }

    LSN treelsn;
3203 3204
    if (oplsn_valid && oplsn.lsn <= (treelsn = toku_ft_checkpoint_lsn(ft_h->ft)).lsn) {
        // do nothing
3205
    } else {
3206
        FT_MSG_S msg = { FT_UPDATE, ZERO_MSN, NULL,
3207
                         .u = { .id = { key, update_function_extra } } };
3208
        ft_send_update_msg(ft_h, &msg, txn);
3209 3210 3211
    }
}

3212
void toku_ft_maybe_update_broadcast(FT_HANDLE ft_h, const DBT *update_function_extra,
Yoni Fogel's avatar
Yoni Fogel committed
3213 3214
                                TOKUTXN txn, bool oplsn_valid, LSN oplsn,
                                bool do_logging, bool is_resetting_op) {
3215
    TXNID_PAIR xid = toku_txn_get_txnid(txn);
Yoni Fogel's avatar
Yoni Fogel committed
3216
    uint8_t  resetting = is_resetting_op ? 1 : 0;
3217
    if (txn) {
3218
        toku_logger_save_rollback_cmdupdatebroadcast(txn, toku_cachefile_filenum(ft_h->ft->cf), resetting);
3219
        toku_txn_maybe_note_ft(txn, ft_h->ft);
3220 3221
    }

3222 3223
    TOKULOGGER logger;
    logger = toku_txn_logger(txn);
3224
    if (do_logging && logger) {
3225
        BYTESTRING extrabs = {.len=update_function_extra->size,
3226
                              .data = (char *) update_function_extra->data};
3227
        toku_log_enq_updatebroadcast(logger, NULL, 0, txn,
3228
                                         toku_cachefile_filenum(ft_h->ft->cf),
3229
                                         xid, extrabs, resetting);
3230 3231
    }

3232
    //TODO(yoni): remove treelsn here and similar calls (no longer being used)
3233 3234
    LSN treelsn;
    if (oplsn_valid &&
3235
        oplsn.lsn <= (treelsn = toku_ft_checkpoint_lsn(ft_h->ft)).lsn) {
3236

3237
    } else {
3238 3239
        DBT nullkey;
        const DBT *nullkeyp = toku_init_dbt(&nullkey);
3240
        FT_MSG_S msg = { FT_UPDATE_BROADCAST_ALL, ZERO_MSN, NULL,
3241
                         .u = { .id = { nullkeyp, update_function_extra } } };
3242
        ft_send_update_msg(ft_h, &msg, txn);
3243 3244 3245
    }
}

3246
void toku_ft_send_insert(FT_HANDLE brt, DBT *key, DBT *val, XIDS xids, enum ft_msg_type type, TXNID oldest_referenced_xid, GC_INFO gc_info) {
3247
    FT_MSG_S ftcmd = { type, ZERO_MSN, xids, .u = { .id = { key, val } } };
3248
    toku_ft_root_put_cmd(brt->ft, &ftcmd, oldest_referenced_xid, gc_info);
3249 3250
}

3251
void toku_ft_send_commit_any(FT_HANDLE brt, DBT *key, XIDS xids, TXNID oldest_referenced_xid, GC_INFO gc_info) {
3252
    DBT val;
3253
    FT_MSG_S ftcmd = { FT_COMMIT_ANY, ZERO_MSN, xids, .u = { .id = { key, toku_init_dbt(&val) } } };
3254
    toku_ft_root_put_cmd(brt->ft, &ftcmd, oldest_referenced_xid, gc_info);
3255 3256
}

3257 3258
void toku_ft_delete(FT_HANDLE brt, DBT *key, TOKUTXN txn) {
    toku_ft_maybe_delete(brt, key, txn, false, ZERO_LSN, true);
3259 3260
}

3261
void
3262
toku_ft_log_del(TOKUTXN txn, FT_HANDLE brt, const DBT *key) {
3263
    TOKULOGGER logger = toku_txn_logger(txn);
3264
    if (logger) {
3265
        BYTESTRING keybs = {.len=key->size, .data=(char *) key->data};
3266
        TXNID_PAIR xid = toku_txn_get_txnid(txn);
3267
        toku_log_enq_delete_any(logger, (LSN*)0, 0, txn, toku_cachefile_filenum(brt->ft->cf), xid, keybs);
3268 3269 3270
    }
}

3271
void
3272
toku_ft_log_del_multiple (TOKUTXN txn, FT_HANDLE src_ft, FT_HANDLE *brts, uint32_t num_fts, const DBT *key, const DBT *val) {
3273
    assert(txn);
3274
    assert(num_fts > 0);
3275 3276
    TOKULOGGER logger = toku_txn_logger(txn);
    if (logger) {
3277
        FILENUM         fnums[num_fts];
3278
        uint32_t i;
3279
        for (i = 0; i < num_fts; i++) {
3280
            fnums[i] = toku_cachefile_filenum(brts[i]->ft->cf);
3281
        }
3282 3283 3284
        FILENUMS filenums = {.num = num_fts, .filenums = fnums};
        BYTESTRING keybs = {.len=key->size, .data=(char *) key->data};
        BYTESTRING valbs = {.len=val->size, .data=(char *) val->data};
3285
        TXNID_PAIR xid = toku_txn_get_txnid(txn);
3286 3287
        FILENUM src_filenum = src_ft ? toku_cachefile_filenum(src_ft->ft->cf) : FILENUM_NONE;
        toku_log_enq_delete_multiple(logger, (LSN*)0, 0, txn, src_filenum, filenums, xid, keybs, valbs);
3288 3289 3290
    }
}

3291
void toku_ft_maybe_delete(FT_HANDLE ft_h, DBT *key, TOKUTXN txn, bool oplsn_valid, LSN oplsn, bool do_logging) {
3292
    XIDS message_xids = xids_get_root_xids(); //By default use committed messages
3293
    TXNID_PAIR xid = toku_txn_get_txnid(txn);
3294
    if (txn) {
3295 3296 3297 3298
        BYTESTRING keybs = {key->size, (char *) key->data};
        toku_logger_save_rollback_cmddelete(txn, toku_cachefile_filenum(ft_h->ft->cf), &keybs);
        toku_txn_maybe_note_ft(txn, ft_h->ft);
        message_xids = toku_txn_get_xids(txn);
3299
    }
3300
    TOKULOGGER logger = toku_txn_logger(txn);
3301
    if (do_logging && logger) {
3302
        BYTESTRING keybs = {.len=key->size, .data=(char *) key->data};
3303
        toku_log_enq_delete_any(logger, (LSN*)0, 0, txn, toku_cachefile_filenum(ft_h->ft->cf), xid, keybs);
3304
    }
3305

3306
    LSN treelsn;
3307
    if (oplsn_valid && oplsn.lsn <= (treelsn = toku_ft_checkpoint_lsn(ft_h->ft)).lsn) {
3308
        // do nothing
3309
    } else {
3310
        TXNID oldest_referenced_xid = (txn) ? txn->oldest_referenced_xid : TXNID_NONE;
3311
        toku_ft_send_delete(ft_h, key, message_xids, oldest_referenced_xid, make_gc_info(txn ? !txn->for_recovery : false));
3312
    }
3313 3314
}

3315
void toku_ft_send_delete(FT_HANDLE brt, DBT *key, XIDS xids, TXNID oldest_referenced_xid, GC_INFO gc_info) {
3316
    DBT val; toku_init_dbt(&val);
3317
    FT_MSG_S ftcmd = { FT_DELETE_ANY, ZERO_MSN, xids, .u = { .id = { key, &val } } };
3318
    toku_ft_root_put_cmd(brt->ft, &ftcmd, oldest_referenced_xid, gc_info);
3319 3320
}

3321 3322 3323 3324
/* mempool support */

struct omt_compressor_state {
    struct mempool *new_kvspace;
3325
    OMTVALUE *newvals;
3326 3327
};

Yoni Fogel's avatar
Yoni Fogel committed
3328
static int move_it (OMTVALUE lev, uint32_t idx, void *v) {
3329 3330
    LEAFENTRY CAST_FROM_VOIDP(le, lev);
    struct omt_compressor_state *CAST_FROM_VOIDP(oc, v);
Yoni Fogel's avatar
Yoni Fogel committed
3331
    uint32_t size = leafentry_memsize(le);
3332
    LEAFENTRY CAST_FROM_VOIDP(newdata, toku_mempool_malloc(oc->new_kvspace, size, 1));
3333
    paranoid_invariant_notnull(newdata); // we do this on a fresh mempool, so nothing bad should happen
3334
    memcpy(newdata, le, size);
3335
    oc->newvals[idx] = newdata;
3336 3337 3338 3339
    return 0;
}

// Compress things, and grow the mempool if needed.
3340
static void omt_compress_kvspace (OMT *omtp, struct mempool *memp, size_t added_size, void **maybe_free) {
Yoni Fogel's avatar
Yoni Fogel committed
3341
    uint32_t total_size_needed = memp->free_offset-memp->frag_size + added_size;
3342 3343
    if (total_size_needed+total_size_needed >= memp->size) {
        memp->size = total_size_needed+total_size_needed;
3344 3345 3346 3347
    }
    void *newmem = toku_xmalloc(memp->size);
    struct mempool new_kvspace;
    toku_mempool_init(&new_kvspace, newmem, memp->size);
3348 3349 3350 3351 3352 3353
    uint32_t numvals = toku_omt_size(*omtp);
    OMTVALUE *XMALLOC_N(numvals, newvals);
    struct omt_compressor_state oc = { &new_kvspace, newvals };
    toku_omt_iterate(*omtp, move_it, &oc);
    toku_omt_destroy(omtp);
    toku_omt_create_steal_sorted_array(omtp, &newvals, numvals, numvals);
3354 3355 3356 3357 3358 3359 3360 3361 3362 3363

    if (maybe_free) {
        *maybe_free = memp->base;
    } else {
        toku_free(memp->base);
    }
    *memp = new_kvspace;
}

void *
3364
mempool_malloc_from_omt(OMT *omtp, struct mempool *mp, size_t size, void **maybe_free) {
3365
    void *v = toku_mempool_malloc(mp, size, 1);
3366
    if (v == NULL) {
3367
        omt_compress_kvspace(omtp, mp, size, maybe_free);
3368
        v = toku_mempool_malloc(mp, size, 1);
3369
        paranoid_invariant_notnull(v);
3370 3371 3372 3373
    }
    return v;
}

3374 3375
/* ******************** open,close and create  ********************** */

3376
// Test only function (not used in running system). This one has no env
3377
int toku_open_ft_handle (const char *fname, int is_create, FT_HANDLE *ft_handle_p, int nodesize,
3378 3379 3380 3381
                   int basementnodesize,
                   enum toku_compression_method compression_method,
                   CACHETABLE cachetable, TOKUTXN txn,
                   int (*compare_fun)(DB *, const DBT*,const DBT*)) {
3382
    FT_HANDLE brt;
3383 3384
    const int only_create = 0;

3385
    toku_ft_handle_create(&brt);
3386 3387 3388
    toku_ft_handle_set_nodesize(brt, nodesize);
    toku_ft_handle_set_basementnodesize(brt, basementnodesize);
    toku_ft_handle_set_compression_method(brt, compression_method);
3389
    toku_ft_set_bt_compare(brt, compare_fun);
3390

3391
    int r = toku_ft_handle_open(brt, fname, is_create, only_create, cachetable, txn);
3392
    if (r != 0) {
3393
        return r;
3394 3395
    }

3396
    *ft_handle_p = brt;
3397 3398 3399
    return r;
}

3400 3401 3402 3403 3404 3405 3406 3407 3408 3409 3410 3411 3412 3413
static bool use_direct_io = true;

void toku_ft_set_direct_io (bool direct_io_on) {
    use_direct_io = direct_io_on;
}

static inline int ft_open_maybe_direct(const char *filename, int oflag, int mode) {
    if (use_direct_io) {
        return toku_os_open_direct(filename, oflag, mode);
    } else {
        return toku_os_open(filename, oflag, mode);
    }
}

3414 3415
// open a file for use by the brt
// Requires:  File does not exist.
3416
static int ft_create_file(FT_HANDLE UU(brt), const char *fname, int *fdp) {
3417
    mode_t mode = S_IRWXU|S_IRWXG|S_IRWXO;
3418
    int r;
3419
    int fd;
3420
    int er;
3421
    fd = ft_open_maybe_direct(fname, O_RDWR | O_BINARY, mode);
3422
    assert(fd==-1);
3423 3424
    if ((er = get_maybe_error_errno()) != ENOENT) {
        return er;
3425
    }
3426
    fd = ft_open_maybe_direct(fname, O_RDWR | O_CREAT | O_BINARY, mode);
3427
    if (fd==-1) {
3428
        r = get_error_errno();
3429
        return r;
3430
    }
3431 3432

    r = toku_fsync_directory(fname);
3433
    resource_assert_zero(r);
3434

3435 3436 3437 3438 3439
    *fdp = fd;
    return 0;
}

// open a file for use by the brt.  if the file does not exist, error
3440
static int ft_open_file(const char *fname, int *fdp) {
3441 3442
    mode_t mode = S_IRWXU|S_IRWXG|S_IRWXO;
    int fd;
3443
    fd = ft_open_maybe_direct(fname, O_RDWR | O_BINARY, mode);
3444
    if (fd==-1) {
3445
        return get_error_errno();
3446 3447 3448 3449 3450
    }
    *fdp = fd;
    return 0;
}

3451 3452
void
toku_ft_handle_set_compression_method(FT_HANDLE t, enum toku_compression_method method)
3453
{
3454 3455 3456 3457 3458 3459
    if (t->ft) {
        toku_ft_set_compression_method(t->ft, method);
    }
    else {
        t->options.compression_method = method;
    }
3460 3461
}

3462 3463
void
toku_ft_handle_get_compression_method(FT_HANDLE t, enum toku_compression_method *methodp)
3464
{
3465 3466 3467 3468 3469 3470
    if (t->ft) {
        toku_ft_get_compression_method(t->ft, methodp);
    }
    else {
        *methodp = t->options.compression_method;
    }
3471 3472
}

3473
static int
Yoni Fogel's avatar
Yoni Fogel committed
3474
verify_builtin_comparisons_consistent(FT_HANDLE t, uint32_t flags) {
3475
    if ((flags & TOKU_DB_KEYCMP_BUILTIN) && (t->options.compare_fun != toku_builtin_compare_fun))
3476
        return EINVAL;
3477 3478 3479
    return 0;
}

3480 3481 3482 3483
//
// See comments in toku_db_change_descriptor to understand invariants 
// in the system when this function is called
//
3484
void toku_ft_change_descriptor(
3485
    FT_HANDLE ft_h,
3486 3487
    const DBT* old_descriptor,
    const DBT* new_descriptor,
Yoni Fogel's avatar
Yoni Fogel committed
3488
    bool do_log,
3489
    TOKUTXN txn,
Yoni Fogel's avatar
Yoni Fogel committed
3490
    bool update_cmp_descriptor
3491
    )
3492 3493 3494
{
    DESCRIPTOR_S new_d;

3495 3496 3497 3498 3499
    // if running with txns, save to rollback + write to recovery log
    if (txn) {
        // put information into rollback file
        BYTESTRING old_desc_bs = { old_descriptor->size, (char *) old_descriptor->data };
        BYTESTRING new_desc_bs = { new_descriptor->size, (char *) new_descriptor->data };
3500
        toku_logger_save_rollback_change_fdescriptor(
3501
            txn,
3502
            toku_cachefile_filenum(ft_h->ft->cf),
3503
            &old_desc_bs
3504
            );
3505 3506 3507 3508
        toku_txn_maybe_note_ft(txn, ft_h->ft);

        if (do_log) {
            TOKULOGGER logger = toku_txn_logger(txn);
3509
            TXNID_PAIR xid = toku_txn_get_txnid(txn);
3510
            toku_log_change_fdescriptor(
3511 3512 3513 3514 3515 3516 3517 3518 3519
                logger, NULL, 0,
                txn,
                toku_cachefile_filenum(ft_h->ft->cf),
                xid,
                old_desc_bs,
                new_desc_bs,
                update_cmp_descriptor
                );
        }
3520
    }
3521 3522 3523

    // write new_descriptor to header
    new_d.dbt = *new_descriptor;
3524
    toku_ft_update_descriptor(ft_h->ft, &new_d);
3525
    // very infrequent operation, worth precise threadsafe count
3526
    STATUS_INC(FT_DESCRIPTOR_SET, 1);
3527

3528
    if (update_cmp_descriptor) {
3529
        toku_ft_update_cmp_descriptor(ft_h->ft);
3530
    }
3531 3532
}

3533 3534 3535
static void
toku_ft_handle_inherit_options(FT_HANDLE t, FT ft) {
    struct ft_options options = {
3536 3537 3538 3539
        .nodesize = ft->h->nodesize,
        .basementnodesize = ft->h->basementnodesize,
        .compression_method = ft->h->compression_method,
        .flags = ft->h->flags,
3540 3541 3542 3543
        .compare_fun = ft->compare_fun,
        .update_fun = ft->update_fun
    };
    t->options = options;
Yoni Fogel's avatar
Yoni Fogel committed
3544
    t->did_set_flags = true;
3545 3546
}

3547
// This is the actual open, used for various purposes, such as normal use, recovery, and redirect.
3548 3549
// fname_in_env is the iname, relative to the env_dir  (data_dir is already in iname as prefix).
// The checkpointed version (checkpoint_lsn) of the dictionary must be no later than max_acceptable_lsn .
3550
// Requires: The multi-operation client lock must be held to prevent a checkpoint from occuring.
3551
static int
3552
ft_handle_open(FT_HANDLE ft_h, const char *fname_in_env, int is_create, int only_create, CACHETABLE cachetable, TOKUTXN txn, FILENUM use_filenum, DICTIONARY_ID use_dictionary_id, LSN max_acceptable_lsn) {
3553
    int r;
Yoni Fogel's avatar
Yoni Fogel committed
3554
    bool txn_created = false;
3555 3556
    char *fname_in_cwd = NULL;
    CACHEFILE cf = NULL;
3557
    FT ft = NULL;
Yoni Fogel's avatar
Yoni Fogel committed
3558
    bool did_create = false;
3559
    toku_ft_open_close_lock();
3560

3561 3562
    if (ft_h->did_set_flags) {
        r = verify_builtin_comparisons_consistent(ft_h, ft_h->options.flags);
3563 3564
        if (r!=0) { goto exit; }
    }
3565

3566
    assert(is_create || !only_create);
3567 3568
    FILENUM reserved_filenum;
    reserved_filenum = use_filenum;
3569
    fname_in_cwd = toku_cachetable_get_fname_in_cwd(cachetable, fname_in_env);
3570
    {
3571
        int fd = -1;
3572
        r = ft_open_file(fname_in_cwd, &fd);
John Esmet's avatar
John Esmet committed
3573 3574 3575
        if (reserved_filenum.fileid == FILENUM_NONE.fileid) {
            reserved_filenum = toku_cachetable_reserve_filenum(cachetable);
        }
3576
        if (r==ENOENT && is_create) {
Yoni Fogel's avatar
Yoni Fogel committed
3577
            did_create = true;
3578 3579
            mode_t mode = S_IRWXU|S_IRWXG|S_IRWXO;
            if (txn) {
3580
                BYTESTRING bs = { .len=(uint32_t) strlen(fname_in_env), .data = (char*)fname_in_env };
3581
                toku_logger_save_rollback_fcreate(txn, reserved_filenum, &bs); // bs is a copy of the fname relative to the environment
3582
            }
Yoni Fogel's avatar
Yoni Fogel committed
3583
            txn_created = (bool)(txn!=NULL);
3584
            toku_logger_log_fcreate(txn, fname_in_env, reserved_filenum, mode, ft_h->options.flags, ft_h->options.nodesize, ft_h->options.basementnodesize, ft_h->options.compression_method);
3585
            r = ft_create_file(ft_h, fname_in_cwd, &fd);
3586 3587 3588
            assert_zero(r);
        }
        if (r) { goto exit; }
John Esmet's avatar
John Esmet committed
3589
        r=toku_cachetable_openfd_with_filenum(&cf, cachetable, fd, fname_in_env, reserved_filenum);
3590
        if (r) { goto exit; }
3591
    }
3592
    assert(ft_h->options.nodesize>0);
Yoni Fogel's avatar
Yoni Fogel committed
3593
    bool was_already_open;
3594
    if (is_create) {
3595
        r = toku_read_ft_and_store_in_cachefile(ft_h, cf, max_acceptable_lsn, &ft, &was_already_open);
3596
        if (r==TOKUDB_DICTIONARY_NO_HEADER) {
3597
            toku_ft_create(&ft, &ft_h->options, cf, txn);
3598 3599 3600 3601 3602 3603 3604 3605 3606 3607
        }
        else if (r!=0) {
            goto exit;
        }
        else if (only_create) {
            assert_zero(r);
            r = EEXIST;
            goto exit;
        }
        // if we get here, then is_create was true but only_create was false,
3608 3609
        // so it is ok for toku_read_ft_and_store_in_cachefile to have read
        // the header via toku_read_ft_and_store_in_cachefile
3610
    } else {
3611
        r = toku_read_ft_and_store_in_cachefile(ft_h, cf, max_acceptable_lsn, &ft, &was_already_open);
3612 3613
        if (r) { goto exit; }
    }
3614 3615
    if (!ft_h->did_set_flags) {
        r = verify_builtin_comparisons_consistent(ft_h, ft_h->options.flags);
3616
        if (r) { goto exit; }
3617
    } else if (ft_h->options.flags != ft->h->flags) {                  /* if flags have been set then flags must match */
3618 3619
        r = EINVAL;
        goto exit;
3620
    }
3621
    toku_ft_handle_inherit_options(ft_h, ft);
3622 3623

    if (!was_already_open) {
3624
        if (!did_create) { //Only log the fopen that OPENs the file.  If it was already open, don't log.
3625
            toku_logger_log_fopen(txn, fname_in_env, toku_cachefile_filenum(cf), ft_h->options.flags);
3626
        }
3627
    }
3628 3629
    int use_reserved_dict_id;
    use_reserved_dict_id = use_dictionary_id.dictid != DICTIONARY_ID_NONE.dictid;
3630
    if (!was_already_open) {
3631 3632 3633 3634 3635 3636 3637
        DICTIONARY_ID dict_id;
        if (use_reserved_dict_id) {
            dict_id = use_dictionary_id;
        }
        else {
            dict_id = next_dict_id();
        }
3638
        ft->dict_id = dict_id;
3639 3640
    }
    else {
3641 3642
        // dict_id is already in header
        if (use_reserved_dict_id) {
3643
            assert(ft->dict_id.dictid == use_dictionary_id.dictid);
3644
        }
3645
    }
3646 3647 3648
    assert(ft);
    assert(ft->dict_id.dictid != DICTIONARY_ID_NONE.dictid);
    assert(ft->dict_id.dictid < dict_id_serial);
3649

3650 3651 3652
    // important note here,
    // after this point, where we associate the header
    // with the brt, the function is not allowed to fail
3653 3654
    // Code that handles failure (located below "exit"),
    // depends on this
3655
    toku_ft_note_ft_handle_open(ft, ft_h);
3656
    if (txn_created) {
3657
        assert(txn);
3658
        toku_txn_maybe_note_ft(txn, ft);
3659
    }
3660

3661
    //Opening a brt may restore to previous checkpoint.         Truncate if necessary.
3662
    {
3663
        int fd = toku_cachefile_get_fd (ft->cf);
3664
        toku_maybe_truncate_file_on_open(ft->blocktable, fd);
3665
    }
3666 3667 3668 3669 3670 3671

    r = 0;
exit:
    if (fname_in_cwd) {
        toku_free(fname_in_cwd);
    }
John Esmet's avatar
John Esmet committed
3672
    if (r != 0 && cf) {
3673
        if (ft) {
John Esmet's avatar
John Esmet committed
3674 3675 3676 3677 3678 3679
            // we only call toku_ft_note_ft_handle_open
            // when the function succeeds, so if we are here,
            // then that means we have a reference to the header
            // but we have not linked it to this brt. So,
            // we can simply try to remove the header.
            // We don't need to unlink this brt from the header
3680
            toku_ft_grab_reflock(ft);
Yoni Fogel's avatar
Yoni Fogel committed
3681
            bool needed = toku_ft_needed_unlocked(ft);
3682 3683
            toku_ft_release_reflock(ft);
            if (!needed) {
3684 3685
                // close immediately.
                toku_ft_evict_from_memory(ft, false, ZERO_LSN);
3686 3687
            }
        }
John Esmet's avatar
John Esmet committed
3688
        else {
3689
            toku_cachefile_close(&cf, false, ZERO_LSN);
John Esmet's avatar
John Esmet committed
3690
        }
3691
    }
3692
    toku_ft_open_close_unlock();
3693
    return r;
3694 3695
}

3696
// Open a brt for the purpose of recovery, which requires that the brt be open to a pre-determined FILENUM
3697
// and may require a specific checkpointed version of the file.
3698
// (dict_id is assigned by the ft_handle_open() function.)
3699
int
3700
toku_ft_handle_open_recovery(FT_HANDLE t, const char *fname_in_env, int is_create, int only_create, CACHETABLE cachetable, TOKUTXN txn, FILENUM use_filenum, LSN max_acceptable_lsn) {
3701
    int r;
3702
    assert(use_filenum.fileid != FILENUM_NONE.fileid);
3703
    r = ft_handle_open(t, fname_in_env, is_create, only_create, cachetable,
3704
                 txn, use_filenum, DICTIONARY_ID_NONE, max_acceptable_lsn);
3705 3706 3707
    return r;
}

3708
// Open a brt in normal use.  The FILENUM and dict_id are assigned by the ft_handle_open() function.
3709
// Requires: The multi-operation client lock must be held to prevent a checkpoint from occuring.
3710
int
3711
toku_ft_handle_open(FT_HANDLE t, const char *fname_in_env, int is_create, int only_create, CACHETABLE cachetable, TOKUTXN txn) {
3712
    int r;
3713
    r = ft_handle_open(t, fname_in_env, is_create, only_create, cachetable, txn, FILENUM_NONE, DICTIONARY_ID_NONE, MAX_LSN);
3714 3715 3716
    return r;
}

3717 3718 3719 3720
// clone an ft handle. the cloned handle has a new dict_id but refers to the same fractal tree
int 
toku_ft_handle_clone(FT_HANDLE *cloned_ft_handle, FT_HANDLE ft_handle, TOKUTXN txn) {
    FT_HANDLE result_ft_handle; 
3721
    toku_ft_handle_create(&result_ft_handle);
3722 3723 3724 3725 3726 3727 3728 3729 3730 3731 3732 3733

    // we're cloning, so the handle better have an open ft and open cf
    invariant(ft_handle->ft);
    invariant(ft_handle->ft->cf);

    // inherit the options of the ft whose handle is being cloned.
    toku_ft_handle_inherit_options(result_ft_handle, ft_handle->ft);

    // we can clone the handle by creating a new handle with the same fname
    CACHEFILE cf = ft_handle->ft->cf;
    CACHETABLE ct = toku_cachefile_get_cachetable(cf);
    const char *fname_in_env = toku_cachefile_fname_in_env(cf);
3734
    int r = toku_ft_handle_open(result_ft_handle, fname_in_env, false, false, ct, txn); 
3735 3736 3737 3738 3739 3740 3741 3742
    if (r != 0) {
        toku_ft_handle_close(result_ft_handle);
        result_ft_handle = NULL;
    }
    *cloned_ft_handle = result_ft_handle;
    return r;
}

3743
// Open a brt in normal use.  The FILENUM and dict_id are assigned by the ft_handle_open() function.
3744
int
3745 3746
toku_ft_handle_open_with_dict_id(
    FT_HANDLE t,
3747 3748 3749 3750 3751
    const char *fname_in_env,
    int is_create,
    int only_create,
    CACHETABLE cachetable,
    TOKUTXN txn,
3752
    DICTIONARY_ID use_dictionary_id
3753
    )
3754
{
3755
    int r;
3756
    r = ft_handle_open(
3757 3758 3759 3760 3761 3762 3763 3764
        t,
        fname_in_env,
        is_create,
        only_create,
        cachetable,
        txn,
        FILENUM_NONE,
        use_dictionary_id,
3765 3766
        MAX_LSN
        );
3767 3768 3769 3770
    return r;
}

DICTIONARY_ID
3771
toku_ft_get_dictionary_id(FT_HANDLE brt) {
3772
    FT h = brt->ft;
3773 3774 3775 3776
    DICTIONARY_ID dict_id = h->dict_id;
    return dict_id;
}

3777 3778 3779
void toku_ft_set_flags(FT_HANDLE ft_handle, unsigned int flags) {
    ft_handle->did_set_flags = true;
    ft_handle->options.flags = flags;
3780 3781
}

3782 3783
void toku_ft_get_flags(FT_HANDLE ft_handle, unsigned int *flags) {
    *flags = ft_handle->options.flags;
3784 3785
}

3786
void toku_ft_get_maximum_advised_key_value_lengths (unsigned int *max_key_len, unsigned int *max_val_len)
3787 3788 3789 3790 3791 3792
// return the maximum advisable key value lengths.  The brt doesn't enforce these.
{
    *max_key_len = 32*1024;
    *max_val_len = 32*1024*1024;
}

3793 3794 3795 3796 3797 3798 3799 3800

void toku_ft_handle_set_nodesize(FT_HANDLE ft_handle, unsigned int nodesize) {
    if (ft_handle->ft) {
        toku_ft_set_nodesize(ft_handle->ft, nodesize);
    }
    else {
        ft_handle->options.nodesize = nodesize;
    }
3801 3802
}

3803 3804 3805 3806 3807 3808 3809 3810 3811 3812 3813 3814 3815 3816 3817 3818 3819 3820 3821 3822 3823 3824 3825 3826 3827
void toku_ft_handle_get_nodesize(FT_HANDLE ft_handle, unsigned int *nodesize) {
    if (ft_handle->ft) {
        toku_ft_get_nodesize(ft_handle->ft, nodesize);
    }
    else {
        *nodesize = ft_handle->options.nodesize;
    }
}

void toku_ft_handle_set_basementnodesize(FT_HANDLE ft_handle, unsigned int basementnodesize) {
    if (ft_handle->ft) {
        toku_ft_set_basementnodesize(ft_handle->ft, basementnodesize);
    }
    else {
        ft_handle->options.basementnodesize = basementnodesize;
    }
}

void toku_ft_handle_get_basementnodesize(FT_HANDLE ft_handle, unsigned int *basementnodesize) {
    if (ft_handle->ft) {
        toku_ft_get_basementnodesize(ft_handle->ft, basementnodesize);
    }
    else {
        *basementnodesize = ft_handle->options.basementnodesize;
    }
3828 3829
}

3830
void toku_ft_set_bt_compare(FT_HANDLE brt, int (*bt_compare)(DB*, const DBT*, const DBT*)) {
3831
    brt->options.compare_fun = bt_compare;
3832 3833
}

3834
void toku_ft_set_redirect_callback(FT_HANDLE brt, on_redirect_callback redir_cb, void* extra) {
3835 3836 3837 3838
    brt->redirect_callback = redir_cb;
    brt->redirect_callback_extra = extra;
}

3839
void toku_ft_set_update(FT_HANDLE brt, ft_update_func update_fun) {
3840
    brt->options.update_fun = update_fun;
3841 3842
}

3843
ft_compare_func toku_ft_get_bt_compare (FT_HANDLE brt) {
3844
    return brt->options.compare_fun;
3845 3846
}

3847 3848
static void
ft_remove_handle_ref_callback(FT UU(ft), void *extra) {
3849
    FT_HANDLE CAST_FROM_VOIDP(handle, extra);
3850 3851 3852
    toku_list_remove(&handle->live_ft_handle_link);
}

3853 3854 3855 3856 3857 3858
// close an ft handle during normal operation. the underlying ft may or may not close,
// depending if there are still references. an lsn for this close will come from the logger.
void
toku_ft_handle_close(FT_HANDLE ft_handle) {
    // There are error paths in the ft_handle_open that end with ft_handle->ft==NULL.
    FT ft = ft_handle->ft;
3859
    if (ft) {
3860 3861
        const bool oplsn_valid = false;
        toku_ft_remove_reference(ft, oplsn_valid, ZERO_LSN, ft_remove_handle_ref_callback, ft_handle);
3862
    }
3863
    toku_free(ft_handle);
3864 3865
}

3866 3867 3868 3869 3870 3871 3872 3873 3874 3875 3876 3877 3878 3879 3880 3881 3882
// close an ft handle during recovery. the underlying ft must close, and will use the given lsn.
void 
toku_ft_handle_close_recovery(FT_HANDLE ft_handle, LSN oplsn) {
    FT ft = ft_handle->ft;
    // the ft must exist if closing during recovery. error paths during 
    // open for recovery should close handles using toku_ft_handle_close()
    assert(ft);
    const bool oplsn_valid = true;
    toku_ft_remove_reference(ft, oplsn_valid, oplsn, ft_remove_handle_ref_callback, ft_handle);
    toku_free(ft_handle);
}

// TODO: remove this, callers should instead just use toku_ft_handle_close()
int 
toku_close_ft_handle_nolsn (FT_HANDLE ft_handle, char** UU(error_string)) {
    toku_ft_handle_close(ft_handle);
    return 0;
3883 3884
}

3885 3886
void toku_ft_handle_create(FT_HANDLE *ft_handle_ptr) {
    FT_HANDLE XMALLOC(brt);
3887
    memset(brt, 0, sizeof *brt);
3888
    toku_list_init(&brt->live_ft_handle_link);
3889
    brt->options.flags = 0;
Yoni Fogel's avatar
Yoni Fogel committed
3890
    brt->did_set_flags = false;
3891 3892 3893 3894 3895
    brt->options.nodesize = FT_DEFAULT_NODE_SIZE;
    brt->options.basementnodesize = FT_DEFAULT_BASEMENT_NODE_SIZE;
    brt->options.compression_method = TOKU_DEFAULT_COMPRESSION_METHOD;
    brt->options.compare_fun = toku_builtin_compare_fun;
    brt->options.update_fun = NULL;
3896
    *ft_handle_ptr = brt;
3897
}
3898

3899 3900
/* ************* CURSORS ********************* */

3901
static inline void
3902
ft_cursor_cleanup_dbts(FT_CURSOR c) {
3903 3904
    toku_destroy_dbt(&c->key);
    toku_destroy_dbt(&c->val);
3905 3906
}

3907 3908 3909 3910 3911 3912 3913 3914 3915 3916 3917
//
// This function is used by the leafentry iterators.
// returns TOKUDB_ACCEPT if live transaction context is allowed to read a value
// that is written by transaction with LSN of id
// live transaction context may read value if either id is the root ancestor of context, or if
// id was committed before context's snapshot was taken.
// For id to be committed before context's snapshot was taken, the following must be true:
//  - id < context->snapshot_txnid64 AND id is not in context's live root transaction list
// For the above to NOT be true:
//  - id > context->snapshot_txnid64 OR id is in context's live root transaction list
//
3918
static int
3919
does_txn_read_entry(TXNID id, TOKUTXN context) {
3920 3921
    int rval;
    TXNID oldest_live_in_snapshot = toku_get_oldest_in_live_root_txn_list(context);
3922 3923 3924 3925
    if (oldest_live_in_snapshot == TXNID_NONE && id < context->snapshot_txnid64) {
        rval = TOKUDB_ACCEPT;
    }
    else if (id < oldest_live_in_snapshot || id == context->txnid.parent_id64) {
3926
        rval = TOKUDB_ACCEPT;
3927
    }
3928
    else if (id > context->snapshot_txnid64 || toku_is_txn_in_live_root_txn_list(*context->live_root_txn_list, id)) {
3929
        rval = 0;
3930 3931
    }
    else {
3932
        rval = TOKUDB_ACCEPT;
3933 3934 3935 3936
    }
    return rval;
}

3937
static inline void
3938 3939
ft_cursor_extract_key_and_val(LEAFENTRY le,
                               FT_CURSOR cursor,
Yoni Fogel's avatar
Yoni Fogel committed
3940
                               uint32_t *keylen,
3941
                               void            **key,
Yoni Fogel's avatar
Yoni Fogel committed
3942
                               uint32_t *vallen,
3943
                               void            **val) {
3944
    if (toku_ft_cursor_is_leaf_mode(cursor)) {
3945 3946 3947
        *key = le_key_and_len(le, keylen);
        *val = le;
        *vallen = leafentry_memsize(le);
3948
    } else if (cursor->is_snapshot_read) {
3949 3950 3951 3952 3953 3954 3955 3956
        le_iterate_val(
            le,
            does_txn_read_entry,
            val,
            vallen,
            cursor->ttxn
            );
        *key = le_key_and_len(le, keylen);
3957
    } else {
3958 3959
        *key = le_key_and_len(le, keylen);
        *val = le_latest_val_and_len(le, vallen);
3960
    }
3961 3962
}

3963 3964 3965
int toku_ft_cursor (
    FT_HANDLE brt,
    FT_CURSOR *cursorptr,
3966
    TOKUTXN ttxn,
Yoni Fogel's avatar
Yoni Fogel committed
3967 3968
    bool is_snapshot_read,
    bool disable_prefetching
3969
    )
3970 3971
{
    if (is_snapshot_read) {
3972
        invariant(ttxn != NULL);
3973
        int accepted = does_txn_read_entry(brt->ft->h->root_xid_that_created, ttxn);
3974 3975 3976 3977
        if (accepted!=TOKUDB_ACCEPT) {
            invariant(accepted==0);
            return TOKUDB_MVCC_DICTIONARY_TOO_NEW;
        }
3978
    }
3979
    FT_CURSOR XCALLOC(cursor);
3980
    cursor->ft_handle = brt;
Yoni Fogel's avatar
Yoni Fogel committed
3981
    cursor->prefetching = false;
3982 3983
    toku_init_dbt(&cursor->range_lock_left_key);
    toku_init_dbt(&cursor->range_lock_right_key);
Yoni Fogel's avatar
Yoni Fogel committed
3984 3985
    cursor->left_is_neg_infty = false;
    cursor->right_is_pos_infty = false;
3986
    cursor->is_snapshot_read = is_snapshot_read;
Yoni Fogel's avatar
Yoni Fogel committed
3987
    cursor->is_leaf_mode = false;
3988
    cursor->ttxn = ttxn;
3989
    cursor->disable_prefetching = disable_prefetching;
Yoni Fogel's avatar
Yoni Fogel committed
3990
    cursor->is_temporary = false;
3991 3992 3993 3994
    *cursorptr = cursor;
    return 0;
}

3995
void
3996
toku_ft_cursor_set_temporary(FT_CURSOR ftcursor) {
Yoni Fogel's avatar
Yoni Fogel committed
3997
    ftcursor->is_temporary = true;
3998 3999
}

4000
void
4001
toku_ft_cursor_set_leaf_mode(FT_CURSOR ftcursor) {
Yoni Fogel's avatar
Yoni Fogel committed
4002
    ftcursor->is_leaf_mode = true;
4003 4004 4005
}

int
4006 4007
toku_ft_cursor_is_leaf_mode(FT_CURSOR ftcursor) {
    return ftcursor->is_leaf_mode;
4008 4009
}

4010
void
4011
toku_ft_cursor_set_range_lock(FT_CURSOR cursor, const DBT *left, const DBT *right,
Yoni Fogel's avatar
Yoni Fogel committed
4012
                               bool left_is_neg_infty, bool right_is_pos_infty)
4013
{
4014 4015
    // Destroy any existing keys and then clone the given left, right keys
    toku_destroy_dbt(&cursor->range_lock_left_key);
4016
    if (left_is_neg_infty) {
Yoni Fogel's avatar
Yoni Fogel committed
4017
        cursor->left_is_neg_infty = true;
4018
    } else {
4019
        toku_clone_dbt(&cursor->range_lock_left_key, *left);
4020
    }
4021 4022

    toku_destroy_dbt(&cursor->range_lock_right_key);
4023
    if (right_is_pos_infty) {
Yoni Fogel's avatar
Yoni Fogel committed
4024
        cursor->right_is_pos_infty = true;
4025
    } else {
4026
        toku_clone_dbt(&cursor->range_lock_right_key, *right);
4027 4028 4029
    }
}

4030
void toku_ft_cursor_close(FT_CURSOR cursor) {
4031
    ft_cursor_cleanup_dbts(cursor);
4032 4033
    toku_destroy_dbt(&cursor->range_lock_left_key);
    toku_destroy_dbt(&cursor->range_lock_right_key);
4034
    toku_free(cursor);
4035
}
4036

4037
static inline void ft_cursor_set_prefetching(FT_CURSOR cursor) {
Yoni Fogel's avatar
Yoni Fogel committed
4038
    cursor->prefetching = true;
4039 4040
}

Yoni Fogel's avatar
Yoni Fogel committed
4041
static inline bool ft_cursor_prefetching(FT_CURSOR cursor) {
4042 4043 4044
    return cursor->prefetching;
}

Yoni Fogel's avatar
Yoni Fogel committed
4045 4046
//Return true if cursor is uninitialized.  false otherwise.
static bool
4047
ft_cursor_not_set(FT_CURSOR cursor) {
4048
    assert((cursor->key.data==NULL) == (cursor->val.data==NULL));
Yoni Fogel's avatar
Yoni Fogel committed
4049
    return (bool)(cursor->key.data == NULL);
4050 4051 4052
}

static int
Yoni Fogel's avatar
Yoni Fogel committed
4053
pair_leafval_heaviside_le (uint32_t klen, void *kval,
4054
                           ft_search_t *search) {
4055
    DBT x;
4056
    int cmp = search->compare(search,
4057
                              search->k ? toku_fill_dbt(&x, kval, klen) : 0);
4058 4059
    // The search->compare function returns only 0 or 1
    switch (search->direction) {
4060 4061
    case FT_SEARCH_LEFT:   return cmp==0 ? -1 : +1;
    case FT_SEARCH_RIGHT:  return cmp==0 ? +1 : -1; // Because the comparison runs backwards for right searches.
4062
    }
4063
    abort(); return 0;
4064 4065 4066 4067
}


static int
4068
heaviside_from_search_t (OMTVALUE lev, void *extra) {
4069 4070
    LEAFENTRY CAST_FROM_VOIDP(le, lev);
    ft_search_t *CAST_FROM_VOIDP(search, extra);
Yoni Fogel's avatar
Yoni Fogel committed
4071
    uint32_t keylen;
4072
    void* key = le_key_and_len(le, &keylen);
4073

4074
    return pair_leafval_heaviside_le (keylen, key,
4075
                                      search);
4076 4077 4078
}


4079 4080 4081
//
// Returns true if the value that is to be read is empty.
//
4082
static inline int
4083
is_le_val_del(LEAFENTRY le, FT_CURSOR ftcursor) {
4084
    int rval;
4085
    if (ftcursor->is_snapshot_read) {
Yoni Fogel's avatar
Yoni Fogel committed
4086
        bool is_del;
4087 4088 4089 4090
        le_iterate_is_del(
            le,
            does_txn_read_entry,
            &is_del,
4091
            ftcursor->ttxn
4092 4093
            );
        rval = is_del;
4094 4095
    }
    else {
4096
        rval = le_latest_is_del(le);
4097
    }
4098
    return rval;
4099 4100
}

4101
struct store_fifo_offset_extra {
4102
    int32_t *offsets;
4103 4104 4105
    int i;
};

4106 4107
int store_fifo_offset(const int32_t &offset, const uint32_t UU(idx), struct store_fifo_offset_extra *const extra) __attribute__((nonnull(3)));
int store_fifo_offset(const int32_t &offset, const uint32_t UU(idx), struct store_fifo_offset_extra *const extra)
4108
{
4109 4110
    extra->offsets[extra->i] = offset;
    extra->i++;
4111 4112 4113
    return 0;
}

4114 4115 4116 4117 4118
/**
 * Given pointers to offsets within a FIFO where we can find messages,
 * figure out the MSN of each message, and compare those MSNs.  Returns 1,
 * 0, or -1 if a is larger than, equal to, or smaller than b.
 */
4119 4120
int fifo_offset_msn_cmp(FIFO &fifo, const int32_t &ao, const int32_t &bo);
int fifo_offset_msn_cmp(FIFO &fifo, const int32_t &ao, const int32_t &bo)
4121
{
4122 4123
    const struct fifo_entry *a = toku_fifo_get_entry(fifo, ao);
    const struct fifo_entry *b = toku_fifo_get_entry(fifo, bo);
4124 4125 4126 4127 4128 4129 4130
    if (a->msn.msn > b->msn.msn) {
        return +1;
    }
    if (a->msn.msn < b->msn.msn) {
        return -1;
    }
    return 0;
4131 4132
}

4133 4134
/**
 * Given a fifo_entry, either decompose it into its parameters and call
4135
 * toku_ft_bn_apply_cmd, or discard it, based on its MSN and the MSN of the
4136 4137
 * basement node.
 */
4138
static void
4139
do_bn_apply_cmd(FT_HANDLE t, BASEMENTNODE bn, struct fifo_entry *entry, TXNID oldest_referenced_xid, uint64_t *workdone, STAT64INFO stats_to_update)
4140
{
4141 4142 4143 4144
    // The messages are being iterated over in (key,msn) order or just in
    // msn order, so all the messages for one key, from one buffer, are in
    // ascending msn order.  So it's ok that we don't update the basement
    // node's msn until the end.
4145 4146 4147
    if (entry->msn.msn > bn->max_msn_applied.msn) {
        ITEMLEN keylen = entry->keylen;
        ITEMLEN vallen = entry->vallen;
4148
        enum ft_msg_type type = fifo_entry_get_msg_type(entry);
4149 4150 4151
        MSN msn = entry->msn;
        const XIDS xids = (XIDS) &entry->xids_s;
        bytevec key = xids_get_end_of_array(xids);
Yoni Fogel's avatar
Yoni Fogel committed
4152
        bytevec val = (uint8_t*)key + entry->keylen;
4153 4154 4155 4156

        DBT hk;
        toku_fill_dbt(&hk, key, keylen);
        DBT hv;
4157
        FT_MSG_S ftcmd = { type, msn, xids, .u = { .id = { &hk, toku_fill_dbt(&hv, val, vallen) } } };
4158
        toku_ft_bn_apply_cmd(
4159 4160 4161
            t->ft->compare_fun,
            t->ft->update_fun,
            &t->ft->cmp_descriptor,
4162
            bn,
4163
            &ftcmd,
4164 4165
            oldest_referenced_xid,
            make_gc_info(true), //mvcc is needed
4166
            workdone,
4167 4168
            stats_to_update
            );
4169
    } else {
4170
        STATUS_INC(FT_MSN_DISCARDS, 1);
4171
    }
4172 4173 4174 4175 4176
    // We must always mark entry as stale since it has been marked
    // (using omt::iterate_and_mark_range)
    // It is possible to call do_bn_apply_cmd even when it won't apply the message because
    // the node containing it could have been evicted and brought back in.
    entry->is_fresh = false;
4177 4178
}

4179
struct iterate_do_bn_apply_cmd_extra {
4180
    FT_HANDLE t;
4181
    BASEMENTNODE bn;
4182
    NONLEAF_CHILDINFO bnc;
4183
    TXNID oldest_referenced_xid;
4184
    uint64_t *workdone;
4185
    STAT64INFO stats_to_update;
4186 4187
};

4188 4189
int iterate_do_bn_apply_cmd(const int32_t &offset, const uint32_t UU(idx), struct iterate_do_bn_apply_cmd_extra *const e) __attribute__((nonnull(3)));
int iterate_do_bn_apply_cmd(const int32_t &offset, const uint32_t UU(idx), struct iterate_do_bn_apply_cmd_extra *const e)
4190
{
4191
    struct fifo_entry *entry = toku_fifo_get_entry(e->bnc->buffer, offset);
4192
    do_bn_apply_cmd(e->t, e->bn, entry, e->oldest_referenced_xid, e->workdone, e->stats_to_update);
4193 4194 4195
    return 0;
}

4196 4197 4198 4199 4200 4201 4202 4203 4204 4205 4206 4207 4208 4209 4210
/**
 * Given the bounds of the basement node to which we will apply messages,
 * find the indexes within message_tree which contain the range of
 * relevant messages.
 *
 * The message tree contains offsets into the buffer, where messages are
 * found.  The pivot_bounds are the lower bound exclusive and upper bound
 * inclusive, because they come from pivot keys in the tree.  We want OMT
 * indices, which must have the lower bound be inclusive and the upper
 * bound exclusive.  We will get these by telling toku_omt_find to look
 * for something strictly bigger than each of our pivot bounds.
 *
 * Outputs the OMT indices in lbi (lower bound inclusive) and ube (upper
 * bound exclusive).
 */
4211
template<typename find_bounds_omt_t>
4212
static void
4213 4214
find_bounds_within_message_tree(
    DESCRIPTOR desc,       /// used for cmp
4215
    ft_compare_func cmp,  /// used to compare keys
4216
    const find_bounds_omt_t &message_tree,      /// tree holding FIFO offsets, in which we want to look for indices
4217 4218
    FIFO buffer,           /// buffer in which messages are found
    struct pivot_bounds const * const bounds,  /// key bounds within the basement node we're applying messages to
4219 4220
    uint32_t *lbi,        /// (output) "lower bound inclusive" (index into message_tree)
    uint32_t *ube         /// (output) "upper bound exclusive" (index into message_tree)
4221 4222 4223
    )
{
    int r = 0;
4224 4225

    if (bounds->lower_bound_exclusive) {
4226 4227 4228 4229 4230
        // By setting msn to MAX_MSN and by using direction of +1, we will
        // get the first message greater than (in (key, msn) order) any
        // message (with any msn) with the key lower_bound_exclusive.
        // This will be a message we want to try applying, so it is the
        // "lower bound inclusive" within the message_tree.
4231 4232 4233 4234 4235 4236 4237 4238 4239
        struct toku_fifo_entry_key_msn_heaviside_extra lbi_extra;
        ZERO_STRUCT(lbi_extra);
        lbi_extra.desc = desc;
        lbi_extra.cmp = cmp;
        lbi_extra.fifo = buffer;
        lbi_extra.key = bounds->lower_bound_exclusive;
        lbi_extra.msn = MAX_MSN;
        int32_t found_lb;
        r = message_tree.template find<struct toku_fifo_entry_key_msn_heaviside_extra, toku_fifo_entry_key_msn_heaviside>(lbi_extra, +1, &found_lb, lbi);
4240
        if (r == DB_NOTFOUND) {
4241 4242 4243
            // There is no relevant data (the lower bound is bigger than
            // any message in this tree), so we have no range and we're
            // done.
Zardosht Kasheff's avatar
Zardosht Kasheff committed
4244 4245
            *lbi = 0;
            *ube = 0;
4246
            return;
4247 4248
        }
        if (bounds->upper_bound_inclusive) {
4249
            // Check if what we found for lbi is greater than the upper
4250 4251
            // bound inclusive that we have.  If so, there are no relevant
            // messages between these bounds.
4252
            const DBT *ubi = bounds->upper_bound_inclusive;
4253
            const int32_t offset = found_lb;
Zardosht Kasheff's avatar
Zardosht Kasheff committed
4254 4255
            DBT found_lbidbt;
            fill_dbt_for_fifo_entry(&found_lbidbt, toku_fifo_get_entry(buffer, offset));
Zardosht Kasheff's avatar
Zardosht Kasheff committed
4256
            FAKE_DB(db, desc);
4257
            int c = cmp(&db, &found_lbidbt, ubi);
4258 4259 4260 4261 4262
            // These DBTs really are both inclusive bounds, so we need
            // strict inequality in order to determine that there's
            // nothing between them.  If they're equal, then we actually
            // need to apply the message pointed to by lbi, and also
            // anything with the same key but a bigger msn.
4263
            if (c > 0) {
Zardosht Kasheff's avatar
Zardosht Kasheff committed
4264 4265
                *lbi = 0;
                *ube = 0;
4266
                return;
4267 4268
            }
        }
4269
    } else {
4270 4271
        // No lower bound given, it's negative infinity, so we start at
        // the first message in the OMT.
Zardosht Kasheff's avatar
Zardosht Kasheff committed
4272
        *lbi = 0;
4273 4274
    }
    if (bounds->upper_bound_inclusive) {
4275 4276 4277 4278
        // Again, we use an msn of MAX_MSN and a direction of +1 to get
        // the first thing bigger than the upper_bound_inclusive key.
        // This is therefore the smallest thing we don't want to apply,
        // and toku_omt_iterate_on_range will not examine it.
4279 4280 4281 4282 4283 4284 4285 4286
        struct toku_fifo_entry_key_msn_heaviside_extra ube_extra;
        ZERO_STRUCT(ube_extra);
        ube_extra.desc = desc;
        ube_extra.cmp = cmp;
        ube_extra.fifo = buffer;
        ube_extra.key = bounds->upper_bound_inclusive;
        ube_extra.msn = MAX_MSN;
        r = message_tree.template find<struct toku_fifo_entry_key_msn_heaviside_extra, toku_fifo_entry_key_msn_heaviside>(ube_extra, +1, nullptr, ube);
4287
        if (r == DB_NOTFOUND) {
4288 4289 4290
            // Couldn't find anything in the buffer bigger than our key,
            // so we need to look at everything up to the end of
            // message_tree.
4291
            *ube = message_tree.size();
4292
        }
4293
    } else {
4294 4295
        // No upper bound given, it's positive infinity, so we need to go
        // through the end of the OMT.
4296
        *ube = message_tree.size();
4297 4298 4299
    }
}

4300 4301 4302 4303 4304 4305 4306
/**
 * For each message in the ancestor's buffer (determined by childnum) that
 * is key-wise between lower_bound_exclusive and upper_bound_inclusive,
 * apply the message to the basement node.  We treat the bounds as minus
 * or plus infinity respectively if they are NULL.  Do not mark the node
 * as dirty (preserve previous state of 'dirty' bit).
 */
4307
static void
4308
bnc_apply_messages_to_basement_node(
4309
    FT_HANDLE t,             // used for comparison function
4310
    BASEMENTNODE bn,   // where to apply messages
4311
    FTNODE ancestor,  // the ancestor node where we can find messages to apply
4312
    int childnum,      // which child buffer of ancestor contains messages we want
Zardosht Kasheff's avatar
Zardosht Kasheff committed
4313
    struct pivot_bounds const * const bounds,  // contains pivot key bounds of this basement node
4314
    TXNID oldest_referenced_xid, // may be younger than what's in ancestor, we should grab the value from the highest node we have
Yoni Fogel's avatar
Yoni Fogel committed
4315
    bool* msgs_applied
4316 4317 4318 4319
    )
{
    int r;
    NONLEAF_CHILDINFO bnc = BNC(ancestor, childnum);
4320

4321 4322
    // Determine the offsets in the message trees between which we need to
    // apply messages from this buffer
4323
    STAT64INFO_S stats_delta = {0,0};
4324 4325
    uint64_t workdone_this_ancestor = 0;

4326
    uint32_t stale_lbi, stale_ube;
4327
    if (!bn->stale_ancestor_messages_applied) {
4328
        find_bounds_within_message_tree(&t->ft->cmp_descriptor, t->ft->compare_fun, bnc->stale_message_tree, bnc->buffer, bounds, &stale_lbi, &stale_ube);
4329
    } else {
Zardosht Kasheff's avatar
Zardosht Kasheff committed
4330 4331
        stale_lbi = 0;
        stale_ube = 0;
4332
    }
Yoni Fogel's avatar
Yoni Fogel committed
4333
    uint32_t fresh_lbi, fresh_ube;
4334
    find_bounds_within_message_tree(&t->ft->cmp_descriptor, t->ft->compare_fun, bnc->fresh_message_tree, bnc->buffer, bounds, &fresh_lbi, &fresh_ube);
4335

4336 4337
    // We now know where all the messages we must apply are, so one of the
    // following 4 cases will do the application, depending on which of
4338 4339
    // the lists contains relevant messages:
    //
4340
    // 1. broadcast messages and anything else, or a mix of fresh and stale
4341 4342
    // 2. only fresh messages
    // 3. only stale messages
4343 4344 4345
    if (bnc->broadcast_list.size() > 0 ||
        (stale_lbi != stale_ube && fresh_lbi != fresh_ube)) {
        // We have messages in multiple trees, so we grab all
4346 4347
        // the relevant messages' offsets and sort them by MSN, then apply
        // them in MSN order.
4348
        const int buffer_size = ((stale_ube - stale_lbi) + (fresh_ube - fresh_lbi) + bnc->broadcast_list.size());
4349
        int32_t *XMALLOC_N(buffer_size, offsets);
4350
        struct store_fifo_offset_extra sfo_extra = { .offsets = offsets, .i = 0 };
4351

4352
        // Populate offsets array with offsets to stale messages
4353
        r = bnc->stale_message_tree.iterate_on_range<struct store_fifo_offset_extra, store_fifo_offset>(stale_lbi, stale_ube, &sfo_extra);
4354
        assert_zero(r);
4355

4356 4357
        // Then store fresh offsets, and mark them to be moved to stale later.
        r = bnc->fresh_message_tree.iterate_and_mark_range<struct store_fifo_offset_extra, store_fifo_offset>(fresh_lbi, fresh_ube, &sfo_extra);
4358 4359
        assert_zero(r);

4360
        // Store offsets of all broadcast messages.
4361
        r = bnc->broadcast_list.iterate<struct store_fifo_offset_extra, store_fifo_offset>(&sfo_extra);
4362
        assert_zero(r);
4363
        invariant(sfo_extra.i == buffer_size);
4364 4365

        // Sort by MSN.
4366
        r = toku::sort<int32_t, FIFO, fifo_offset_msn_cmp>::mergesort_r(offsets, buffer_size, bnc->buffer);
4367 4368
        assert_zero(r);

4369
        // Apply the messages in MSN order.
4370
        for (int i = 0; i < buffer_size; ++i) {
Yoni Fogel's avatar
Yoni Fogel committed
4371
            *msgs_applied = true;
4372
            struct fifo_entry *entry = toku_fifo_get_entry(bnc->buffer, offsets[i]);
4373
            do_bn_apply_cmd(t, bn, entry, oldest_referenced_xid, &workdone_this_ancestor, &stats_delta);
4374
        }
4375

4376
        toku_free(offsets);
4377
    } else if (stale_lbi == stale_ube) {
4378
        // No stale messages to apply, we just apply fresh messages, and mark them to be moved to stale later.
4379
        struct iterate_do_bn_apply_cmd_extra iter_extra = { .t = t, .bn = bn, .bnc = bnc, .oldest_referenced_xid = oldest_referenced_xid, .workdone = &workdone_this_ancestor, .stats_to_update = &stats_delta };
Yoni Fogel's avatar
Yoni Fogel committed
4380
        if (fresh_ube - fresh_lbi > 0) *msgs_applied = true;
4381
        r = bnc->fresh_message_tree.iterate_and_mark_range<struct iterate_do_bn_apply_cmd_extra, iterate_do_bn_apply_cmd>(fresh_lbi, fresh_ube, &iter_extra);
4382
        assert_zero(r);
4383 4384
    } else {
        invariant(fresh_lbi == fresh_ube);
4385 4386
        // No fresh messages to apply, we just apply stale messages.

Yoni Fogel's avatar
Yoni Fogel committed
4387
        if (stale_ube - stale_lbi > 0) *msgs_applied = true;
4388
        struct iterate_do_bn_apply_cmd_extra iter_extra = { .t = t, .bn = bn, .bnc = bnc, .oldest_referenced_xid = oldest_referenced_xid, .workdone = &workdone_this_ancestor, .stats_to_update = &stats_delta };
4389

4390
        r = bnc->stale_message_tree.iterate_on_range<struct iterate_do_bn_apply_cmd_extra, iterate_do_bn_apply_cmd>(stale_lbi, stale_ube, &iter_extra);
4391
        assert_zero(r);
4392
    }
4393 4394 4395
    //
    // update stats
    //
4396 4397 4398
    if (workdone_this_ancestor > 0) {
        (void) toku_sync_fetch_and_add(&BP_WORKDONE(ancestor, childnum), workdone_this_ancestor);
    }
4399
    if (stats_delta.numbytes || stats_delta.numrows) {
4400
        toku_ft_update_stats(&t->ft->in_memory_stats, stats_delta);
4401
    }
4402 4403
}

Zardosht Kasheff's avatar
Zardosht Kasheff committed
4404
void
4405
toku_apply_ancestors_messages_to_node (FT_HANDLE t, FTNODE node, ANCESTORS ancestors, struct pivot_bounds const * const bounds, bool* msgs_applied)
4406
// Effect:
4407
//   Bring a leaf node up-to-date according to all the messages in the ancestors.
4408
//   If the leaf node is already up-to-date then do nothing.
4409 4410 4411 4412 4413
//   If the leaf node is not already up-to-date, then record the work done
//   for that leaf in each ancestor.
// Requires:
//   This is being called when pinning a leaf node for the query path.
//   The entire root-to-leaf path is pinned and appears in the ancestors list.
4414 4415
{
    VERIFY_NODE(t, node);
4416
    invariant(node->height == 0);
4417

4418
    TXNID oldest_referenced_xid = ancestors->node->oldest_referenced_xid_known;
4419
    for (ANCESTORS curr_ancestors = ancestors; curr_ancestors; curr_ancestors = curr_ancestors->next) {
4420 4421
        if (curr_ancestors->node->oldest_referenced_xid_known > oldest_referenced_xid) {
            oldest_referenced_xid = curr_ancestors->node->oldest_referenced_xid_known;
4422 4423 4424
        }
    }

4425
    // know we are a leaf node
4426 4427 4428 4429 4430
    // An important invariant:
    // We MUST bring every available basement node up to date.
    // flushing on the cleaner thread depends on this. This invariant
    // allows the cleaner thread to just pick an internal node and flush it
    // as opposed to being forced to start from the root.
4431
    for (int i = 0; i < node->n_children; i++) {
4432
        if (BP_STATE(node, i) != PT_AVAIL) { continue; }
4433 4434 4435 4436
        BASEMENTNODE curr_bn = BLB(node, i);
        struct pivot_bounds curr_bounds = next_pivot_keys(node, i, bounds);
        for (ANCESTORS curr_ancestors = ancestors; curr_ancestors; curr_ancestors = curr_ancestors->next) {
            if (curr_ancestors->node->max_msn_applied_to_node_on_disk.msn > curr_bn->max_msn_applied.msn) {
4437
                paranoid_invariant(BP_STATE(curr_ancestors->node, curr_ancestors->childnum) == PT_AVAIL);
4438
                bnc_apply_messages_to_basement_node(
4439 4440 4441 4442
                    t,
                    curr_bn,
                    curr_ancestors->node,
                    curr_ancestors->childnum,
Zardosht Kasheff's avatar
Zardosht Kasheff committed
4443
                    &curr_bounds,
4444
                    oldest_referenced_xid,
Zardosht Kasheff's avatar
Zardosht Kasheff committed
4445
                    msgs_applied
4446
                    );
4447 4448
                // We don't want to check this ancestor node again if the
                // next time we query it, the msn hasn't changed.
4449 4450 4451
                curr_bn->max_msn_applied = curr_ancestors->node->max_msn_applied_to_node_on_disk;
            }
        }
4452 4453 4454 4455 4456
        // At this point, we know all the stale messages above this
        // basement node have been applied, and any new messages will be
        // fresh, so we don't need to look at stale messages for this
        // basement node, unless it gets evicted (and this field becomes
        // false when it's read in again).
4457
        curr_bn->stale_ancestor_messages_applied = true;
4458 4459
    }
    VERIFY_NODE(t, node);
4460 4461
}

Leif Walsh's avatar
Leif Walsh committed
4462 4463 4464 4465 4466 4467 4468 4469 4470 4471 4472 4473 4474 4475 4476 4477 4478 4479 4480 4481 4482 4483 4484 4485 4486 4487 4488 4489
bool toku_ft_leaf_needs_ancestors_messages(FT ft, FTNODE node, ANCESTORS ancestors, struct pivot_bounds const * const bounds, MSN *const max_msn_in_path)
// Effect: Determine whether there are messages in a node's ancestors
//  which must be applied to it.  These messages are in the correct
//  keyrange for any available basement nodes, and are in nodes with the
//  correct max_msn_applied_to_node_on_disk.
// Notes:
//  This is an approximate query.
// Output:
//  max_msn_in_path: max of "max_msn_applied_to_node_on_disk" over
//    ancestors.  This is used later to update basement nodes'
//    max_msn_applied values in case we don't do the full algorithm.
// Returns:
//  true if there may be some such messages
//  false only if there are definitely no such messages
// Rationale:
//  When we pin a node with a read lock, we want to quickly determine if
//  we should exchange it for a write lock in preparation for applying
//  messages.  If there are no messages, we don't need the write lock.
{
    invariant(node->height == 0);
    MSN max_msn_applied = ZERO_MSN;
    bool needs_ancestors_messages = false;
    for (int i = 0; i < node->n_children; ++i) {
        if (BP_STATE(node, i) != PT_AVAIL) { continue; }
        BASEMENTNODE bn = BLB(node, i);
        struct pivot_bounds curr_bounds = next_pivot_keys(node, i, bounds);
        for (ANCESTORS curr_ancestors = ancestors; curr_ancestors; curr_ancestors = curr_ancestors->next) {
            if (curr_ancestors->node->max_msn_applied_to_node_on_disk.msn > bn->max_msn_applied.msn) {
4490
                paranoid_invariant(BP_STATE(curr_ancestors->node, curr_ancestors->childnum) == PT_AVAIL);
Leif Walsh's avatar
Leif Walsh committed
4491 4492 4493 4494 4495 4496 4497 4498 4499 4500 4501 4502 4503 4504 4505 4506 4507 4508 4509 4510 4511 4512 4513 4514 4515 4516 4517 4518 4519 4520 4521 4522 4523 4524 4525 4526 4527 4528 4529 4530 4531 4532 4533 4534 4535 4536 4537 4538
                NONLEAF_CHILDINFO bnc = BNC(curr_ancestors->node, curr_ancestors->childnum);
                if (bnc->broadcast_list.size() > 0) {
                    needs_ancestors_messages = true;
                    goto cleanup;
                }
                if (!bn->stale_ancestor_messages_applied) {
                    uint32_t stale_lbi, stale_ube;
                    find_bounds_within_message_tree(&ft->cmp_descriptor,
                                                    ft->compare_fun,
                                                    bnc->stale_message_tree,
                                                    bnc->buffer,
                                                    &curr_bounds,
                                                    &stale_lbi,
                                                    &stale_ube);
                    if (stale_lbi < stale_ube) {
                        needs_ancestors_messages = true;
                        goto cleanup;
                    }
                }
                uint32_t fresh_lbi, fresh_ube;
                find_bounds_within_message_tree(&ft->cmp_descriptor,
                                                ft->compare_fun,
                                                bnc->fresh_message_tree,
                                                bnc->buffer,
                                                &curr_bounds,
                                                &fresh_lbi,
                                                &fresh_ube);
                if (fresh_lbi < fresh_ube) {
                    needs_ancestors_messages = true;
                    goto cleanup;
                }
                if (curr_ancestors->node->max_msn_applied_to_node_on_disk.msn > max_msn_applied.msn) {
                    max_msn_applied = curr_ancestors->node->max_msn_applied_to_node_on_disk;
                }
            }
        }
    }
    *max_msn_in_path = max_msn_applied;
cleanup:
    return needs_ancestors_messages;
}

void toku_ft_bn_update_max_msn(FTNODE node, MSN max_msn_applied) {
    invariant(node->height == 0);
    for (int i = 0; i < node->n_children; ++i) {
        if (BP_STATE(node, i) != PT_AVAIL) { continue; }
        BASEMENTNODE bn = BLB(node, i);
        if (max_msn_applied.msn > bn->max_msn_applied.msn) {
4539 4540 4541 4542 4543
            // This function runs in a shared access context, so to silence tools
            // like DRD, we use a CAS and ignore the result.
            // Any threads trying to update these basement nodes should be
            // updating them to the same thing (since they all have a read lock on
            // the same root-to-leaf path) so this is safe.
4544
            (void) toku_sync_val_compare_and_swap(&bn->max_msn_applied.msn, bn->max_msn_applied.msn, max_msn_applied.msn);
Leif Walsh's avatar
Leif Walsh committed
4545 4546 4547 4548
        }
    }
}

4549 4550 4551 4552 4553
struct copy_to_stale_extra {
    FT ft;
    NONLEAF_CHILDINFO bnc;
};

4554 4555
int copy_to_stale(const int32_t &offset, const uint32_t UU(idx), struct copy_to_stale_extra *const extra) __attribute__((nonnull(3)));
int copy_to_stale(const int32_t &offset, const uint32_t UU(idx), struct copy_to_stale_extra *const extra)
4556 4557 4558 4559 4560 4561 4562 4563 4564 4565 4566 4567 4568 4569 4570 4571 4572 4573 4574 4575 4576 4577 4578 4579 4580 4581 4582 4583 4584
{
    struct fifo_entry *entry = toku_fifo_get_entry(extra->bnc->buffer, offset);
    DBT keydbt;
    DBT *key = fill_dbt_for_fifo_entry(&keydbt, entry);
    struct toku_fifo_entry_key_msn_heaviside_extra heaviside_extra = { .desc = &extra->ft->cmp_descriptor, .cmp = extra->ft->compare_fun, .fifo = extra->bnc->buffer, .key = key, .msn = entry->msn };
    int r = extra->bnc->stale_message_tree.insert<struct toku_fifo_entry_key_msn_heaviside_extra, toku_fifo_entry_key_msn_heaviside>(offset, heaviside_extra, nullptr);
    invariant_zero(r);
    return 0;
}

__attribute__((nonnull))
void
toku_move_ftnode_messages_to_stale(FT ft, FTNODE node) {
    invariant(node->height > 0);
    for (int i = 0; i < node->n_children; ++i) {
        if (BP_STATE(node, i) != PT_AVAIL) {
            continue;
        }
        NONLEAF_CHILDINFO bnc = BNC(node, i);
        // We can't delete things out of the fresh tree inside the above
        // procedures because we're still looking at the fresh tree.  Instead
        // we have to move messages after we're done looking at it.
        struct copy_to_stale_extra cts_extra = { .ft = ft, .bnc = bnc };
        int r = bnc->fresh_message_tree.iterate_over_marked<struct copy_to_stale_extra, copy_to_stale>(&cts_extra);
        invariant_zero(r);
        bnc->fresh_message_tree.delete_all_marked();
    }
}

4585
static int
4586 4587
ft_cursor_shortcut (
    FT_CURSOR cursor,
4588
    int direction,
4589
    FT_GET_CALLBACK_FUNCTION getf,
4590
    void *getf_v,
Yoni Fogel's avatar
Yoni Fogel committed
4591
    uint32_t *keylen,
4592
    void **key,
Yoni Fogel's avatar
Yoni Fogel committed
4593
    uint32_t *vallen,
4594 4595 4596
    void **val
    );

4597 4598
// This is a bottom layer of the search functions.
static int
4599
ft_search_basement_node(
4600
    BASEMENTNODE bn,
4601 4602
    ft_search_t *search,
    FT_GET_CALLBACK_FUNCTION getf,
4603
    void *getf_v,
Yoni Fogel's avatar
Yoni Fogel committed
4604
    bool *doprefetch,
4605
    FT_CURSOR ftcursor,
Yoni Fogel's avatar
Yoni Fogel committed
4606
    bool can_bulk_fetch
4607
    )
4608
{
4609
    // Now we have to convert from ft_search_t to the heaviside function with a direction.  What a pain...
4610

4611 4612
    int direction;
    switch (search->direction) {
4613 4614
    case FT_SEARCH_LEFT:   direction = +1; goto ok;
    case FT_SEARCH_RIGHT:  direction = -1; goto ok;
4615 4616
    }
    return EINVAL;  // This return and the goto are a hack to get both compile-time and run-time checking on enum
4617
ok: ;
4618
    OMTVALUE datav;
Yoni Fogel's avatar
Yoni Fogel committed
4619
    uint32_t idx = 0;
4620
    int r = toku_omt_find(bn->buffer,
4621 4622 4623
                          heaviside_from_search_t,
                          search,
                          direction,
4624
                          &datav, &idx);
4625 4626
    if (r!=0) return r;

4627
    LEAFENTRY CAST_FROM_VOIDP(le, datav);
4628
    if (toku_ft_cursor_is_leaf_mode(ftcursor))
4629
        goto got_a_good_value;        // leaf mode cursors see all leaf entries
4630
    if (is_le_val_del(le,ftcursor)) {
4631 4632 4633 4634
        // Provisionally deleted stuff is gone.
        // So we need to scan in the direction to see if we can find something
        while (1) {
            switch (search->direction) {
4635
            case FT_SEARCH_LEFT:
4636
                idx++;
4637 4638
                if (idx >= toku_omt_size(bn->buffer))
                    return DB_NOTFOUND;
4639
                break;
4640
            case FT_SEARCH_RIGHT:
4641
                if (idx == 0)
4642
                    return DB_NOTFOUND;
4643 4644 4645
                idx--;
                break;
            default:
4646
                abort();
4647
            }
4648
            r = toku_omt_fetch(bn->buffer, idx, &datav);
4649
            assert_zero(r); // we just validated the index
4650
            CAST_FROM_VOIDP(le, datav);
4651
            if (!is_le_val_del(le,ftcursor)) goto got_a_good_value;
4652
        }
4653 4654
    }
got_a_good_value:
4655
    {
Yoni Fogel's avatar
Yoni Fogel committed
4656
        uint32_t keylen;
4657
        void *key;
Yoni Fogel's avatar
Yoni Fogel committed
4658
        uint32_t vallen;
4659 4660
        void *val;

4661 4662
        ft_cursor_extract_key_and_val(le,
                                       ftcursor,
4663 4664 4665 4666
                                       &keylen,
                                       &key,
                                       &vallen,
                                       &val
4667
            );
4668

4669
        r = getf(keylen, key, vallen, val, getf_v, false);
4670
        if (r==0 || r == TOKUDB_CURSOR_CONTINUE) {
4671 4672
            ftcursor->leaf_info.to_be.omt   = bn->buffer;
            ftcursor->leaf_info.to_be.index = idx;
4673

4674
            if (r == TOKUDB_CURSOR_CONTINUE && can_bulk_fetch) {
4675 4676
                r = ft_cursor_shortcut(
                    ftcursor,
4677 4678 4679 4680 4681 4682 4683 4684 4685 4686
                    direction,
                    getf,
                    getf_v,
                    &keylen,
                    &key,
                    &vallen,
                    &val
                    );
            }

4687 4688
            ft_cursor_cleanup_dbts(ftcursor);
            if (!ftcursor->is_temporary) {
4689 4690
                toku_memdup_dbt(&ftcursor->key, key, keylen);
                toku_memdup_dbt(&ftcursor->val, val, vallen);
4691
            }
4692
            //The search was successful.  Prefetching can continue.
Yoni Fogel's avatar
Yoni Fogel committed
4693
            *doprefetch = true;
4694
        }
4695
    }
4696
    if (r == TOKUDB_CURSOR_CONTINUE) r = 0;
4697
    return r;
4698 4699 4700
}

static int
4701 4702 4703 4704
ft_search_node (
    FT_HANDLE brt,
    FTNODE node,
    ft_search_t *search,
4705
    int child_to_search,
4706
    FT_GET_CALLBACK_FUNCTION getf,
4707
    void *getf_v,
Yoni Fogel's avatar
Yoni Fogel committed
4708
    bool *doprefetch,
4709
    FT_CURSOR ftcursor,
4710 4711
    UNLOCKERS unlockers,
    ANCESTORS,
4712
    struct pivot_bounds const * const bounds,
Yoni Fogel's avatar
Yoni Fogel committed
4713
    bool can_bulk_fetch
4714
    );
4715

4716
static int
Zardosht Kasheff's avatar
Zardosht Kasheff committed
4717
ftnode_fetch_callback_and_free_bfe(CACHEFILE cf, PAIR p, int fd, BLOCKNUM nodename, uint32_t fullhash, void **ftnode_pv, void** UU(disk_data), PAIR_ATTR *sizep, int *dirtyp, void *extraargs)
4718
{
Zardosht Kasheff's avatar
Zardosht Kasheff committed
4719
    int r = toku_ftnode_fetch_callback(cf, p, fd, nodename, fullhash, ftnode_pv, disk_data, sizep, dirtyp, extraargs);
4720
    struct ftnode_fetch_extra *CAST_FROM_VOIDP(ffe, extraargs);
4721 4722
    destroy_bfe_for_prefetch(ffe);
    toku_free(ffe);
4723 4724 4725 4726
    return r;
}

static int
4727
ftnode_pf_callback_and_free_bfe(void *ftnode_pv, void* disk_data, void *read_extraargs, int fd, PAIR_ATTR *sizep)
4728
{
4729
    int r = toku_ftnode_pf_callback(ftnode_pv, disk_data, read_extraargs, fd, sizep);
4730
    struct ftnode_fetch_extra *CAST_FROM_VOIDP(ffe, read_extraargs);
4731 4732
    destroy_bfe_for_prefetch(ffe);
    toku_free(ffe);
4733 4734 4735
    return r;
}

4736
static void
Yoni Fogel's avatar
Yoni Fogel committed
4737
ft_node_maybe_prefetch(FT_HANDLE brt, FTNODE node, int childnum, FT_CURSOR ftcursor, bool *doprefetch) {
4738 4739
    // the number of nodes to prefetch
    const int num_nodes_to_prefetch = 1;
Rich Prohaska's avatar
Rich Prohaska committed
4740

4741
    // if we want to prefetch in the tree
4742
    // then prefetch the next children if there are any
4743 4744
    if (*doprefetch && ft_cursor_prefetching(ftcursor) && !ftcursor->disable_prefetching) {
        int rc = ft_cursor_rightmost_child_wanted(ftcursor, brt, node);
4745
        for (int i = childnum + 1; (i <= childnum + num_nodes_to_prefetch) && (i <= rc); i++) {
4746
            BLOCKNUM nextchildblocknum = BP_BLOCKNUM(node, i);
Yoni Fogel's avatar
Yoni Fogel committed
4747
            uint32_t nextfullhash = compute_child_fullhash(brt->ft->cf, node, i);
4748
            struct ftnode_fetch_extra *MALLOC(bfe);
4749
            fill_bfe_for_prefetch(bfe, brt->ft, ftcursor);
Yoni Fogel's avatar
Yoni Fogel committed
4750
            bool doing_prefetch = false;
4751
            toku_cachefile_prefetch(
4752
                brt->ft->cf,
4753 4754
                nextchildblocknum,
                nextfullhash,
4755
                get_write_callbacks_for_node(brt->ft),
4756 4757 4758
                ftnode_fetch_callback_and_free_bfe,
                toku_ftnode_pf_req_callback,
                ftnode_pf_callback_and_free_bfe,
4759 4760 4761 4762 4763 4764 4765
                bfe,
                &doing_prefetch
                );
            if (!doing_prefetch) {
                destroy_bfe_for_prefetch(bfe);
                toku_free(bfe);
            }
Yoni Fogel's avatar
Yoni Fogel committed
4766
            *doprefetch = false;
4767
        }
4768 4769 4770
    }
}

4771 4772 4773
struct unlock_ftnode_extra {
    FT_HANDLE ft_handle;
    FTNODE node;
Yoni Fogel's avatar
Yoni Fogel committed
4774
    bool msgs_applied;
4775 4776 4777
};
// When this is called, the cachetable lock is held
static void
4778
unlock_ftnode_fun (PAIR p, void *v) {
Zardosht Kasheff's avatar
Zardosht Kasheff committed
4779 4780
    struct unlock_ftnode_extra *x = NULL;
    CAST_FROM_VOIDP(x, v);
4781 4782
    FT_HANDLE brt = x->ft_handle;
    FTNODE node = x->node;
4783
    // CT lock is held
Zardosht Kasheff's avatar
Zardosht Kasheff committed
4784
    int r = toku_cachetable_unpin_ct_prelocked_no_flush(
4785
        p,
4786
        brt->ft->cf,
Zardosht Kasheff's avatar
Zardosht Kasheff committed
4787
        node->ct_pair,
4788
        (enum cachetable_dirty) node->dirty,
4789
        x->msgs_applied ? make_ftnode_pair_attr(node) : make_invalid_pair_attr()
Zardosht Kasheff's avatar
Zardosht Kasheff committed
4790
        );
4791
    assert_zero(r);
4792 4793
}

4794
/* search in a node's child */
4795
static int
Yoni Fogel's avatar
Yoni Fogel committed
4796 4797
ft_search_child(FT_HANDLE brt, FTNODE node, int childnum, ft_search_t *search, FT_GET_CALLBACK_FUNCTION getf, void *getf_v, bool *doprefetch, FT_CURSOR ftcursor, UNLOCKERS unlockers,
                 ANCESTORS ancestors, struct pivot_bounds const * const bounds, bool can_bulk_fetch)
4798
// Effect: Search in a node's child.  Searches are read-only now (at least as far as the hardcopy is concerned).
4799
{
4800
    struct ancestors next_ancestors = {node, childnum, ancestors};
4801

4802
    BLOCKNUM childblocknum = BP_BLOCKNUM(node,childnum);
Yoni Fogel's avatar
Yoni Fogel committed
4803
    uint32_t fullhash = compute_child_fullhash(brt->ft->cf, node, childnum);
4804
    FTNODE childnode;
4805

4806 4807 4808
    // If the current node's height is greater than 1, then its child is an internal node.
    // Therefore, to warm the cache better (#5798), we want to read all the partitions off disk in one shot.
    bool read_all_partitions = node->height > 1;
4809
    struct ftnode_fetch_extra bfe;
4810
    fill_bfe_for_subset_read(
4811
        &bfe,
4812
        brt->ft,
4813
        search,
4814 4815 4816 4817
        &ftcursor->range_lock_left_key,
        &ftcursor->range_lock_right_key,
        ftcursor->left_is_neg_infty,
        ftcursor->right_is_pos_infty,
4818 4819
        ftcursor->disable_prefetching,
        read_all_partitions
4820
        );
Yoni Fogel's avatar
Yoni Fogel committed
4821
    bool msgs_applied = false;
4822
    {
Zardosht Kasheff's avatar
Zardosht Kasheff committed
4823 4824 4825 4826
        int rr = toku_pin_ftnode_batched(brt, childblocknum, fullhash,
                                         unlockers,
                                         &next_ancestors, bounds,
                                         &bfe,
Leif Walsh's avatar
Leif Walsh committed
4827
                                         PL_READ, // we try to get a read lock, but we may upgrade to a write lock on a leaf for message application.
Zardosht Kasheff's avatar
Zardosht Kasheff committed
4828 4829 4830 4831 4832 4833 4834 4835 4836
                                         true,
                                         &childnode,
                                         &msgs_applied);
        if (rr==TOKUDB_TRY_AGAIN) {
            return rr;
        }
        // We end the batch before applying ancestor messages if we get
        // all the way to a leaf.
        invariant_zero(rr);
4837 4838
    }

4839
    struct unlock_ftnode_extra unlock_extra   = {brt,childnode,msgs_applied};
Yoni Fogel's avatar
Yoni Fogel committed
4840
    struct unlockers next_unlockers = {true, unlock_ftnode_fun, (void*)&unlock_extra, unlockers};
4841

4842
    int r = ft_search_node(brt, childnode, search, bfe.child_to_read, getf, getf_v, doprefetch, ftcursor, &next_unlockers, &next_ancestors, bounds, can_bulk_fetch);
4843
    if (r!=TOKUDB_TRY_AGAIN) {
4844 4845
        // maybe prefetch the next child
        if (r == 0 && node->height == 1) {
4846
            ft_node_maybe_prefetch(brt, node, childnum, ftcursor, doprefetch);
4847
        }
4848

4849
        assert(next_unlockers.locked);
Zardosht Kasheff's avatar
Zardosht Kasheff committed
4850
        if (msgs_applied) {
4851
            toku_unpin_ftnode(brt->ft, childnode);
Zardosht Kasheff's avatar
Zardosht Kasheff committed
4852 4853
        }
        else {
4854
            toku_unpin_ftnode_read_only(brt->ft, childnode);
Zardosht Kasheff's avatar
Zardosht Kasheff committed
4855
        }
4856
    } else {
4857
        // try again.
4858 4859

        // there are two cases where we get TOKUDB_TRY_AGAIN
4860
        //  case 1 is when some later call to toku_pin_ftnode returned
4861
        //  that value and unpinned all the nodes anyway. case 2
4862
        //  is when ft_search_node had to stop its search because
4863 4864
        //  some piece of a node that it needed was not in memory. In this case,
        //  the node was not unpinned, so we unpin it here
4865
        if (next_unlockers.locked) {
Zardosht Kasheff's avatar
Zardosht Kasheff committed
4866
            if (msgs_applied) {
4867
                toku_unpin_ftnode(brt->ft, childnode);
Zardosht Kasheff's avatar
Zardosht Kasheff committed
4868 4869
            }
            else {
4870
                toku_unpin_ftnode_read_only(brt->ft, childnode);
Zardosht Kasheff's avatar
Zardosht Kasheff committed
4871
            }
4872
        }
4873
    }
4874

4875 4876 4877
    return r;
}

4878
static inline int
4879
search_which_child_cmp_with_bound(DB *db, ft_compare_func cmp, FTNODE node, int childnum, ft_search_t *search, DBT *dbt)
4880
{
4881
    return cmp(db, toku_copy_dbt(dbt, node->childkeys[childnum]), &search->pivot_bound);
4882 4883
}

4884
int
4885
toku_ft_search_which_child(
Zardosht Kasheff's avatar
Zardosht Kasheff committed
4886
    DESCRIPTOR desc,
4887 4888 4889
    ft_compare_func cmp,
    FTNODE node,
    ft_search_t *search
4890
    )
4891
{
4892 4893 4894 4895 4896 4897 4898 4899 4900
    if (node->n_children <= 1) return 0;

    DBT pivotkey;
    toku_init_dbt(&pivotkey);
    int lo = 0;
    int hi = node->n_children - 1;
    int mi;
    while (lo < hi) {
        mi = (lo + hi) / 2;
4901
        toku_copy_dbt(&pivotkey, node->childkeys[mi]);
4902 4903 4904 4905 4906 4907 4908 4909 4910 4911 4912 4913
        // search->compare is really strange, and only works well with a
        // linear search, it makes binary search a pita.
        //
        // if you are searching left to right, it returns
        //   "0" for pivots that are < the target, and
        //   "1" for pivots that are >= the target
        // if you are searching right to left, it's the opposite.
        //
        // so if we're searching from the left and search->compare says
        // "1", we want to go left from here, if it says "0" we want to go
        // right.  searching from the right does the opposite.
        bool c = search->compare(search, &pivotkey);
4914 4915
        if (((search->direction == FT_SEARCH_LEFT) && c) ||
            ((search->direction == FT_SEARCH_RIGHT) && !c)) {
4916 4917
            hi = mi;
        } else {
4918 4919
            assert(((search->direction == FT_SEARCH_LEFT) && !c) ||
                   ((search->direction == FT_SEARCH_RIGHT) && c));
4920 4921 4922 4923 4924
            lo = mi + 1;
        }
    }
    // ready to return something, if the pivot is bounded, we have to move
    // over a bit to get away from what we've already searched
4925
    if (search->pivot_bound.data != nullptr) {
Zardosht Kasheff's avatar
Zardosht Kasheff committed
4926
        FAKE_DB(db, desc);
4927
        if (search->direction == FT_SEARCH_LEFT) {
4928 4929 4930 4931 4932 4933 4934 4935 4936 4937 4938 4939 4940 4941 4942 4943 4944 4945 4946
            while (lo < node->n_children - 1 &&
                   search_which_child_cmp_with_bound(&db, cmp, node, lo, search, &pivotkey) <= 0) {
                // searching left to right, if the comparison says the
                // current pivot (lo) is left of or equal to our bound,
                // don't search that child again
                lo++;
            }
        } else {
            while (lo > 0 &&
                   search_which_child_cmp_with_bound(&db, cmp, node, lo - 1, search, &pivotkey) >= 0) {
                // searching right to left, same argument as just above
                // (but we had to pass lo - 1 because the pivot between lo
                // and the thing just less than it is at that position in
                // the childkeys array)
                lo--;
            }
        }
    }
    return lo;
4947
}
4948

4949 4950
static void
maybe_search_save_bound(
4951
    FTNODE node,
4952
    int child_searched,
4953
    ft_search_t *search)
4954
{
4955
    int p = (search->direction == FT_SEARCH_LEFT) ? child_searched : child_searched - 1;
4956
    if (p >= 0 && p < node->n_children-1) {
4957 4958
        toku_destroy_dbt(&search->pivot_bound);
        toku_clone_dbt(&search->pivot_bound, node->childkeys[p]);
4959 4960 4961 4962
    }
}

static int
4963 4964 4965 4966
ft_search_node(
    FT_HANDLE brt,
    FTNODE node,
    ft_search_t *search,
4967
    int child_to_search,
4968
    FT_GET_CALLBACK_FUNCTION getf,
4969
    void *getf_v,
Yoni Fogel's avatar
Yoni Fogel committed
4970
    bool *doprefetch,
4971
    FT_CURSOR ftcursor,
4972
    UNLOCKERS unlockers,
4973
    ANCESTORS ancestors,
4974
    struct pivot_bounds const * const bounds,
Yoni Fogel's avatar
Yoni Fogel committed
4975
    bool can_bulk_fetch
4976
    )
Zardosht Kasheff's avatar
Zardosht Kasheff committed
4977 4978
{
    int r = 0;
4979
    // assert that we got a valid child_to_search
4980 4981
    invariant(child_to_search >= 0);
    invariant(child_to_search < node->n_children);
4982 4983 4984 4985 4986
    //
    // At this point, we must have the necessary partition available to continue the search
    //
    assert(BP_STATE(node,child_to_search) == PT_AVAIL);
    while (child_to_search >= 0 && child_to_search < node->n_children) {
4987 4988
        //
        // Normally, the child we want to use is available, as we checked
4989
        // before entering this while loop. However, if we pass through
4990
        // the loop once, getting DB_NOTFOUND for this first value
4991
        // of child_to_search, we enter the while loop again with a
4992 4993
        // child_to_search that may not be in memory. If it is not,
        // we need to return TOKUDB_TRY_AGAIN so the query can
Zardosht Kasheff's avatar
Zardosht Kasheff committed
4994
        // read the appropriate partition into memory
4995 4996 4997 4998
        //
        if (BP_STATE(node,child_to_search) != PT_AVAIL) {
            return TOKUDB_TRY_AGAIN;
        }
4999 5000
        const struct pivot_bounds next_bounds = next_pivot_keys(node, child_to_search, bounds);
        if (node->height > 0) {
5001
            r = ft_search_child(
5002 5003 5004 5005 5006 5007 5008
                brt,
                node,
                child_to_search,
                search,
                getf,
                getf_v,
                doprefetch,
5009
                ftcursor,
5010 5011
                unlockers,
                ancestors,
5012 5013
                &next_bounds,
                can_bulk_fetch
5014 5015 5016
                );
        }
        else {
5017
            r = ft_search_basement_node(
5018
                BLB(node, child_to_search),
5019 5020 5021 5022
                search,
                getf,
                getf_v,
                doprefetch,
5023
                ftcursor,
5024
                can_bulk_fetch
5025 5026 5027
                );
        }
        if (r == 0) return r; //Success
5028

5029 5030
        if (r != DB_NOTFOUND) {
            return r; //Error (or message to quit early, such as TOKUDB_FOUND_BUT_REJECTED or TOKUDB_TRY_AGAIN)
5031
        }
5032 5033 5034 5035
        // not really necessary, just put this here so that reading the
        // code becomes simpler. The point is at this point in the code,
        // we know that we got DB_NOTFOUND and we have to continue
        assert(r == DB_NOTFOUND);
Zardosht Kasheff's avatar
Zardosht Kasheff committed
5036 5037 5038 5039 5040 5041 5042 5043 5044 5045 5046 5047 5048 5049 5050 5051 5052 5053 5054 5055 5056 5057 5058 5059
        // we have a new pivotkey
        if (node->height == 0) {
            // when we run off the end of a basement, try to lock the range up to the pivot. solves #3529
            const DBT *pivot = NULL;
            if (search->direction == FT_SEARCH_LEFT)
                pivot = next_bounds.upper_bound_inclusive; // left -> right
            else
                pivot = next_bounds.lower_bound_exclusive; // right -> left
            if (pivot) {
                int rr = getf(pivot->size, pivot->data, 0, NULL, getf_v, true);
                if (rr != 0)
                    return rr; // lock was not granted
            }
        }

        // If we got a DB_NOTFOUND then we have to search the next record.        Possibly everything present is not visible.
        // This way of doing DB_NOTFOUND is a kludge, and ought to be simplified.  Something like this is needed for DB_NEXT, but
        //        for point queries, it's overkill.  If we got a DB_NOTFOUND on a point query then we should just stop looking.
        // When releasing locks on I/O we must not search the same subtree again, or we won't be guaranteed to make forward progress.
        // If we got a DB_NOTFOUND, then the pivot is too small if searching from left to right (too large if searching from right to left).
        // So save the pivot key in the search object.
        maybe_search_save_bound(node, child_to_search, search);

        // We're about to pin some more nodes, but we thought we were done before.
5060
        if (search->direction == FT_SEARCH_LEFT) {
5061 5062 5063 5064 5065
            child_to_search++;
        }
        else {
            child_to_search--;
        }
5066
    }
5067
    return r;
5068 5069 5070
}

static int
Yoni Fogel's avatar
Yoni Fogel committed
5071
toku_ft_search (FT_HANDLE brt, ft_search_t *search, FT_GET_CALLBACK_FUNCTION getf, void *getf_v, FT_CURSOR ftcursor, bool can_bulk_fetch)
5072 5073 5074
// Effect: Perform a search.  Associate cursor with a leaf if possible.
// All searches are performed through this function.
{
5075
    int r;
5076
    uint trycount = 0;     // How many tries did it take to get the result?
Zardosht Kasheff's avatar
Zardosht Kasheff committed
5077
    FT ft = brt->ft;
5078

Zardosht Kasheff's avatar
Zardosht Kasheff committed
5079
try_again:
5080

5081
    trycount++;
5082

Zardosht Kasheff's avatar
Zardosht Kasheff committed
5083 5084 5085 5086 5087 5088 5089 5090 5091 5092
    //
    // Here is how searches work
    // At a high level, we descend down the tree, using the search parameter
    // to guide us towards where to look. But the search parameter is not
    // used here to determine which child of a node to read (regardless
    // of whether that child is another node or a basement node)
    // The search parameter is used while we are pinning the node into
    // memory, because that is when the system needs to ensure that
    // the appropriate partition of the child we are using is in memory.
    // So, here are the steps for a search (and this applies to this function
5093 5094 5095
    // as well as ft_search_child:
    //  - Take the search parameter, and create a ftnode_fetch_extra, that will be used by toku_pin_ftnode(_holding_lock)
    //  - Call toku_pin_ftnode(_holding_lock) with the bfe as the extra for the fetch callback (in case the node is not at all in memory)
Zardosht Kasheff's avatar
Zardosht Kasheff committed
5096
    //       and the partial fetch callback (in case the node is perhaps partially in memory) to the fetch the node
5097
    //  - This eventually calls either toku_ftnode_fetch_callback or  toku_ftnode_pf_req_callback depending on whether the node is in
Zardosht Kasheff's avatar
Zardosht Kasheff committed
5098
    //     memory at all or not.
5099 5100 5101
    //  - Within these functions, the "ft_search_t search" parameter is used to evaluate which child the search is interested in.
    //     If the node is not in memory at all, toku_ftnode_fetch_callback will read the node and decompress only the partition for the
    //     relevant child, be it a message buffer or basement node. If the node is in memory, then toku_ftnode_pf_req_callback
Zardosht Kasheff's avatar
Zardosht Kasheff committed
5102
    //     will tell the cachetable that a partial fetch is required if and only if the relevant child is not in memory. If the relevant child
5103
    //     is not in memory, then toku_ftnode_pf_callback is called to fetch the partition.
Zardosht Kasheff's avatar
Zardosht Kasheff committed
5104
    //  - These functions set bfe->child_to_read so that the search code does not need to reevaluate it.
5105 5106 5107
    //  - Just to reiterate, all of the last item happens within toku_ftnode_pin(_holding_lock)
    //  - At this point, toku_ftnode_pin_holding_lock has returned, with bfe.child_to_read set,
    //  - ft_search_node is called, assuming that the node and its relevant partition are in memory.
Zardosht Kasheff's avatar
Zardosht Kasheff committed
5108
    //
5109
    struct ftnode_fetch_extra bfe;
Zardosht Kasheff's avatar
Zardosht Kasheff committed
5110 5111
    fill_bfe_for_subset_read(
        &bfe,
Zardosht Kasheff's avatar
Zardosht Kasheff committed
5112
        ft,
Zardosht Kasheff's avatar
Zardosht Kasheff committed
5113
        search,
5114 5115 5116 5117
        &ftcursor->range_lock_left_key,
        &ftcursor->range_lock_right_key,
        ftcursor->left_is_neg_infty,
        ftcursor->right_is_pos_infty,
5118 5119
        ftcursor->disable_prefetching,
        true // We may as well always read the whole root into memory, if it's a leaf node it's a tiny tree anyway.
Zardosht Kasheff's avatar
Zardosht Kasheff committed
5120
        );
5121
    FTNODE node = NULL;
Zardosht Kasheff's avatar
Zardosht Kasheff committed
5122
    {
Yoni Fogel's avatar
Yoni Fogel committed
5123
        uint32_t fullhash;
5124
        CACHEKEY root_key;
Zardosht Kasheff's avatar
Zardosht Kasheff committed
5125 5126 5127
        toku_calculate_root_offset_pointer(ft, &root_key, &fullhash);
        toku_pin_ftnode_off_client_thread_batched(
            ft,
5128
            root_key,
5129
            fullhash,
5130
            &bfe,
Zardosht Kasheff's avatar
Zardosht Kasheff committed
5131
            PL_READ, // may_modify_node set to false, because root cannot change during search
5132 5133 5134 5135
            0,
            NULL,
            &node
            );
Zardosht Kasheff's avatar
Zardosht Kasheff committed
5136
    }
Zardosht Kasheff's avatar
Zardosht Kasheff committed
5137

5138
    uint tree_height = node->height + 1;  // How high is the tree?  This is the height of the root node plus one (leaf is at height 0).
5139

Zardosht Kasheff's avatar
Zardosht Kasheff committed
5140

Yoni Fogel's avatar
Yoni Fogel committed
5141 5142
    struct unlock_ftnode_extra unlock_extra   = {brt,node,false};
    struct unlockers                unlockers      = {true, unlock_ftnode_fun, (void*)&unlock_extra, (UNLOCKERS)NULL};
5143

5144
    {
Yoni Fogel's avatar
Yoni Fogel committed
5145
        bool doprefetch = false;
5146
        //static int counter = 0;         counter++;
5147
        r = ft_search_node(brt, node, search, bfe.child_to_read, getf, getf_v, &doprefetch, ftcursor, &unlockers, (ANCESTORS)NULL, &infinite_bounds, can_bulk_fetch);
5148
        if (r==TOKUDB_TRY_AGAIN) {
5149
            // there are two cases where we get TOKUDB_TRY_AGAIN
5150
            //  case 1 is when some later call to toku_pin_ftnode returned
5151
            //  that value and unpinned all the nodes anyway. case 2
5152
            //  is when ft_search_node had to stop its search because
5153
            //  some piece of a node that it needed was not in memory.
Zardosht Kasheff's avatar
Zardosht Kasheff committed
5154
            //  In this case, the node was not unpinned, so we unpin it here
5155
            if (unlockers.locked) {
5156
                toku_unpin_ftnode_read_only(brt->ft, node);
5157
            }
5158 5159 5160 5161
            goto try_again;
        } else {
            assert(unlockers.locked);
        }
5162 5163
    }

5164
    assert(unlockers.locked);
5165
    toku_unpin_ftnode_read_only(brt->ft, node);
5166

5167

5168 5169
    //Heaviside function (+direction) queries define only a lower or upper
    //bound.  Some queries require both an upper and lower bound.
5170
    //They do this by wrapping the FT_GET_CALLBACK_FUNCTION with another
5171 5172 5173 5174 5175 5176
    //test that checks for the other bound.  If the other bound fails,
    //it returns TOKUDB_FOUND_BUT_REJECTED which means not found, but
    //stop searching immediately, as opposed to DB_NOTFOUND
    //which can mean not found, but keep looking in another leaf.
    if (r==TOKUDB_FOUND_BUT_REJECTED) r = DB_NOTFOUND;
    else if (r==DB_NOTFOUND) {
5177
        //We truly did not find an answer to the query.
5178
        //Therefore, the FT_GET_CALLBACK_FUNCTION has NOT been called.
5179 5180 5181 5182 5183 5184 5185
        //The contract specifies that the callback function must be called
        //for 'r= (0|DB_NOTFOUND|TOKUDB_FOUND_BUT_REJECTED)'
        //TODO: #1378 This is not the ultimate location of this call to the
        //callback.  It is surely wrong for node-level locking, and probably
        //wrong for the STRADDLE callback for heaviside function(two sets of key/vals)
        int r2 = getf(0,NULL, 0,NULL, getf_v, false);
        if (r2!=0) r = r2;
5186
    }
5187
    {   // accounting (to detect and measure thrashing)
5188
        uint retrycount = trycount - 1;         // how many retries were needed?
5189 5190 5191
        if (retrycount) {
            STATUS_INC(FT_TOTAL_RETRIES, retrycount);
        }
5192
        if (retrycount > tree_height) {         // if at least one node was read from disk more than once
5193
            STATUS_INC(FT_SEARCH_TRIES_GT_HEIGHT, 1);
5194
            if (retrycount > (tree_height+3))
5195
                STATUS_INC(FT_SEARCH_TRIES_GT_HEIGHTPLUS3, 1);
5196
        }
5197
    }
5198 5199 5200
    return r;
}

5201 5202
struct ft_cursor_search_struct {
    FT_GET_CALLBACK_FUNCTION getf;
5203
    void *getf_v;
5204 5205
    FT_CURSOR cursor;
    ft_search_t *search;
5206
};
5207

5208 5209
/* search for the first kv pair that matches the search object */
static int
Yoni Fogel's avatar
Yoni Fogel committed
5210
ft_cursor_search(FT_CURSOR cursor, ft_search_t *search, FT_GET_CALLBACK_FUNCTION getf, void *getf_v, bool can_bulk_fetch)
5211
{
5212
    int r = toku_ft_search(cursor->ft_handle, search, getf, getf_v, cursor, can_bulk_fetch);
5213 5214 5215
    return r;
}

5216
static inline int compare_k_x(FT_HANDLE brt, const DBT *k, const DBT *x) {
5217 5218
    FAKE_DB(db, &brt->ft->cmp_descriptor);
    return brt->ft->compare_fun(&db, k, x);
5219 5220
}

5221
static int
5222
ft_cursor_compare_one(ft_search_t *search __attribute__((__unused__)), DBT *x __attribute__((__unused__)))
5223
{
5224 5225 5226
    return 1;
}

5227
static int ft_cursor_compare_set(ft_search_t *search, DBT *x) {
5228
    FT_HANDLE CAST_FROM_VOIDP(brt, search->context);
5229
    return compare_k_x(brt, search->k, x) <= 0; /* return min xy: kv <= xy */
5230 5231
}

5232
static int
5233
ft_cursor_current_getf(ITEMLEN keylen,                 bytevec key,
5234 5235
                        ITEMLEN vallen,                 bytevec val,
                        void *v, bool lock_only) {
5236
    struct ft_cursor_search_struct *CAST_FROM_VOIDP(bcss, v);
5237 5238
    int r;
    if (key==NULL) {
5239
        r = bcss->getf(0, NULL, 0, NULL, bcss->getf_v, lock_only);
5240
    } else {
5241
        FT_CURSOR cursor = bcss->cursor;
5242 5243
        DBT newkey;
        toku_fill_dbt(&newkey, key, keylen);
5244
        if (compare_k_x(cursor->ft_handle, &cursor->key, &newkey) != 0) {
5245 5246 5247 5248 5249
            r = bcss->getf(0, NULL, 0, NULL, bcss->getf_v, lock_only); // This was once DB_KEYEMPTY
            if (r==0) r = TOKUDB_FOUND_BUT_REJECTED;
        }
        else
            r = bcss->getf(keylen, key, vallen, val, bcss->getf_v, lock_only);
5250 5251 5252 5253 5254
    }
    return r;
}

int
5255
toku_ft_cursor_current(FT_CURSOR cursor, int op, FT_GET_CALLBACK_FUNCTION getf, void *getf_v)
5256
{
5257
    if (ft_cursor_not_set(cursor))
5258
        return EINVAL;
5259
    if (op == DB_CURRENT) {
5260 5261
        struct ft_cursor_search_struct bcss = {getf, getf_v, cursor, 0};
        ft_search_t search; ft_search_init(&search, ft_cursor_compare_set, FT_SEARCH_LEFT, &cursor->key, cursor->ft_handle);
Yoni Fogel's avatar
Yoni Fogel committed
5262
        int r = toku_ft_search(cursor->ft_handle, &search, ft_cursor_current_getf, &bcss, cursor, false);
5263
        ft_search_finish(&search);
5264
        return r;
5265
    }
5266
    return getf(cursor->key.size, cursor->key.data, cursor->val.size, cursor->val.data, getf_v, false); // ft_cursor_copyout(cursor, outkey, outval);
5267 5268
}

5269
int
5270
toku_ft_cursor_first(FT_CURSOR cursor, FT_GET_CALLBACK_FUNCTION getf, void *getf_v)
5271
{
5272
    ft_search_t search; ft_search_init(&search, ft_cursor_compare_one, FT_SEARCH_LEFT, 0, cursor->ft_handle);
Yoni Fogel's avatar
Yoni Fogel committed
5273
    int r = ft_cursor_search(cursor, &search, getf, getf_v, false);
5274
    ft_search_finish(&search);
5275
    return r;
5276 5277
}

5278
int
5279
toku_ft_cursor_last(FT_CURSOR cursor, FT_GET_CALLBACK_FUNCTION getf, void *getf_v)
5280
{
5281
    ft_search_t search; ft_search_init(&search, ft_cursor_compare_one, FT_SEARCH_RIGHT, 0, cursor->ft_handle);
Yoni Fogel's avatar
Yoni Fogel committed
5282
    int r = ft_cursor_search(cursor, &search, getf, getf_v, false);
5283
    ft_search_finish(&search);
5284
    return r;
5285 5286
}

5287
static int ft_cursor_compare_next(ft_search_t *search, DBT *x) {
5288
    FT_HANDLE CAST_FROM_VOIDP(brt, search->context);
5289
    return compare_k_x(brt, search->k, x) < 0; /* return min xy: kv < xy */
5290 5291
}

5292

5293
static int
5294 5295
ft_cursor_shortcut (
    FT_CURSOR cursor,
5296
    int direction,
5297
    FT_GET_CALLBACK_FUNCTION getf,
5298
    void *getf_v,
Yoni Fogel's avatar
Yoni Fogel committed
5299
    uint32_t *keylen,
5300
    void **key,
Yoni Fogel's avatar
Yoni Fogel committed
5301
    uint32_t *vallen,
5302
    void **val
5303
    )
5304
{
5305
    int r = 0;
Yoni Fogel's avatar
Yoni Fogel committed
5306
    uint32_t index = cursor->leaf_info.to_be.index;
5307 5308 5309
    OMT omt = cursor->leaf_info.to_be.omt;
    // if we are searching towards the end, limit is last element
    // if we are searching towards the beginning, limit is the first element
Yoni Fogel's avatar
Yoni Fogel committed
5310
    uint32_t limit = (direction > 0) ? (toku_omt_size(omt) - 1) : 0;
5311 5312

    //Starting with the prev, find the first real (non-provdel) leafentry.
5313
    OMTVALUE lev = NULL;
5314 5315
    while (index != limit) {
        index += direction;
5316
        r = toku_omt_fetch(omt, index, &lev);
5317
        assert_zero(r);
5318
        LEAFENTRY CAST_FROM_VOIDP(le, lev);
5319

5320
        if (toku_ft_cursor_is_leaf_mode(cursor) || !is_le_val_del(le, cursor)) {
5321

5322
            ft_cursor_extract_key_and_val(
5323 5324 5325 5326 5327 5328 5329
                le,
                cursor,
                keylen,
                key,
                vallen,
                val
                );
5330

5331 5332
            r = getf(*keylen, *key, *vallen, *val, getf_v, false);
            if (r == 0 || r == TOKUDB_CURSOR_CONTINUE) {
5333 5334 5335
                //Update cursor.
                cursor->leaf_info.to_be.index = index;
            }
5336
            if (r == TOKUDB_CURSOR_CONTINUE) {
5337 5338 5339 5340 5341
                continue;
            }
            else {
                break;
            }
5342
        }
5343
    }
5344

5345
    return r;
5346 5347 5348
}

int
5349
toku_ft_cursor_next(FT_CURSOR cursor, FT_GET_CALLBACK_FUNCTION getf, void *getf_v)
5350
{
5351
    ft_search_t search; ft_search_init(&search, ft_cursor_compare_next, FT_SEARCH_LEFT, &cursor->key, cursor->ft_handle);
Yoni Fogel's avatar
Yoni Fogel committed
5352
    int r = ft_cursor_search(cursor, &search, getf, getf_v, true);
5353 5354
    ft_search_finish(&search);
    if (r == 0) ft_cursor_set_prefetching(cursor);
5355
    return r;
5356 5357
}

5358
static int
5359
ft_cursor_search_eq_k_x_getf(ITEMLEN keylen,               bytevec key,
5360 5361
                              ITEMLEN vallen,               bytevec val,
                              void *v, bool lock_only) {
5362
    struct ft_cursor_search_struct *CAST_FROM_VOIDP(bcss, v);
5363 5364
    int r;
    if (key==NULL) {
5365
        r = bcss->getf(0, NULL, 0, NULL, bcss->getf_v, false);
5366
    } else {
5367
        FT_CURSOR cursor = bcss->cursor;
5368 5369
        DBT newkey;
        toku_fill_dbt(&newkey, key, keylen);
5370
        if (compare_k_x(cursor->ft_handle, bcss->search->k, &newkey) == 0) {
5371 5372 5373 5374 5375
            r = bcss->getf(keylen, key, vallen, val, bcss->getf_v, lock_only);
        } else {
            r = bcss->getf(0, NULL, 0, NULL, bcss->getf_v, lock_only);
            if (r==0) r = TOKUDB_FOUND_BUT_REJECTED;
        }
5376 5377 5378
    }
    return r;
}
5379 5380 5381

/* search for the kv pair that matches the search object and is equal to k */
static int
5382
ft_cursor_search_eq_k_x(FT_CURSOR cursor, ft_search_t *search, FT_GET_CALLBACK_FUNCTION getf, void *getf_v)
5383
{
5384
    struct ft_cursor_search_struct bcss = {getf, getf_v, cursor, search};
Yoni Fogel's avatar
Yoni Fogel committed
5385
    int r = toku_ft_search(cursor->ft_handle, search, ft_cursor_search_eq_k_x_getf, &bcss, cursor, false);
5386
    return r;
5387 5388
}

5389
static int ft_cursor_compare_prev(ft_search_t *search, DBT *x) {
5390
    FT_HANDLE CAST_FROM_VOIDP(brt, search->context);
5391
    return compare_k_x(brt, search->k, x) > 0; /* return max xy: kv > xy */
5392 5393
}

5394
int
5395
toku_ft_cursor_prev(FT_CURSOR cursor, FT_GET_CALLBACK_FUNCTION getf, void *getf_v)
5396
{
5397
    ft_search_t search; ft_search_init(&search, ft_cursor_compare_prev, FT_SEARCH_RIGHT, &cursor->key, cursor->ft_handle);
Yoni Fogel's avatar
Yoni Fogel committed
5398
    int r = ft_cursor_search(cursor, &search, getf, getf_v, true);
5399
    ft_search_finish(&search);
5400
    return r;
5401 5402
}

5403
static int ft_cursor_compare_set_range(ft_search_t *search, DBT *x) {
5404
    FT_HANDLE CAST_FROM_VOIDP(brt, search->context);
5405
    return compare_k_x(brt, search->k,        x) <= 0; /* return kv <= xy */
5406 5407 5408
}

int
5409
toku_ft_cursor_set(FT_CURSOR cursor, DBT *key, FT_GET_CALLBACK_FUNCTION getf, void *getf_v)
5410
{
5411 5412 5413
    ft_search_t search; ft_search_init(&search, ft_cursor_compare_set_range, FT_SEARCH_LEFT, key, cursor->ft_handle);
    int r = ft_cursor_search_eq_k_x(cursor, &search, getf, getf_v);
    ft_search_finish(&search);
5414
    return r;
5415 5416 5417
}

int
5418
toku_ft_cursor_set_range(FT_CURSOR cursor, DBT *key, FT_GET_CALLBACK_FUNCTION getf, void *getf_v)
5419
{
5420
    ft_search_t search; ft_search_init(&search, ft_cursor_compare_set_range, FT_SEARCH_LEFT, key, cursor->ft_handle);
Yoni Fogel's avatar
Yoni Fogel committed
5421
    int r = ft_cursor_search(cursor, &search, getf, getf_v, false);
5422
    ft_search_finish(&search);
5423
    return r;
5424 5425
}

5426
static int ft_cursor_compare_set_range_reverse(ft_search_t *search, DBT *x) {
5427
    FT_HANDLE CAST_FROM_VOIDP(brt, search->context);
5428
    return compare_k_x(brt, search->k, x) >= 0; /* return kv >= xy */
5429 5430 5431
}

int
5432
toku_ft_cursor_set_range_reverse(FT_CURSOR cursor, DBT *key, FT_GET_CALLBACK_FUNCTION getf, void *getf_v)
5433
{
5434
    ft_search_t search; ft_search_init(&search, ft_cursor_compare_set_range_reverse, FT_SEARCH_RIGHT, key, cursor->ft_handle);
Yoni Fogel's avatar
Yoni Fogel committed
5435
    int r = ft_cursor_search(cursor, &search, getf, getf_v, false);
5436
    ft_search_finish(&search);
5437
    return r;
5438 5439 5440
}


5441 5442 5443
//TODO: When tests have been rewritten, get rid of this function.
//Only used by tests.
int
5444
toku_ft_cursor_get (FT_CURSOR cursor, DBT *key, FT_GET_CALLBACK_FUNCTION getf, void *getf_v, int get_flags)
5445
{
5446
    int op = get_flags & DB_OPFLAGS_MASK;
5447
    if (get_flags & ~DB_OPFLAGS_MASK)
5448
        return EINVAL;
5449 5450 5451 5452

    switch (op) {
    case DB_CURRENT:
    case DB_CURRENT_BINDING:
5453
        return toku_ft_cursor_current(cursor, op, getf, getf_v);
5454
    case DB_FIRST:
5455
        return toku_ft_cursor_first(cursor, getf, getf_v);
5456
    case DB_LAST:
5457
        return toku_ft_cursor_last(cursor, getf, getf_v);
5458
    case DB_NEXT:
John Esmet's avatar
John Esmet committed
5459
        if (ft_cursor_not_set(cursor)) {
5460
            return toku_ft_cursor_first(cursor, getf, getf_v);
John Esmet's avatar
John Esmet committed
5461
        } else {
5462
            return toku_ft_cursor_next(cursor, getf, getf_v);
John Esmet's avatar
John Esmet committed
5463
        }
5464
    case DB_PREV:
John Esmet's avatar
John Esmet committed
5465
        if (ft_cursor_not_set(cursor)) {
5466
            return toku_ft_cursor_last(cursor, getf, getf_v);
John Esmet's avatar
John Esmet committed
5467
        } else {
5468
            return toku_ft_cursor_prev(cursor, getf, getf_v);
John Esmet's avatar
John Esmet committed
5469
        }
5470
    case DB_SET:
5471
        return toku_ft_cursor_set(cursor, key, getf, getf_v);
5472
    case DB_SET_RANGE:
5473
        return toku_ft_cursor_set_range(cursor, key, getf, getf_v);
5474
    default: ;// Fall through
5475
    }
5476
    return EINVAL;
5477 5478
}

5479
void
5480
toku_ft_cursor_peek(FT_CURSOR cursor, const DBT **pkey, const DBT **pval)
5481 5482
// Effect: Retrieves a pointer to the DBTs for the current key and value.
// Requires:  The caller may not modify the DBTs or the memory at which they points.
5483
// Requires:  The caller must be in the context of a
5484
// FT_GET_(STRADDLE_)CALLBACK_FUNCTION
5485 5486 5487 5488 5489 5490
{
    *pkey = &cursor->key;
    *pval = &cursor->val;
}

//We pass in toku_dbt_fake to the search functions, since it will not pass the
5491
//key(or val) to the heaviside function if key(or val) is NULL.
5492 5493 5494
//It is not used for anything else,
//the actual 'extra' information for the heaviside function is inside the
//wrapper.
5495
static const DBT __toku_dbt_fake = {};
5496 5497
static const DBT* const toku_dbt_fake = &__toku_dbt_fake;

Yoni Fogel's avatar
Yoni Fogel committed
5498
bool toku_ft_cursor_uninitialized(FT_CURSOR c) {
5499
    return ft_cursor_not_set(c);
5500 5501 5502
}


5503 5504
/* ********************************* lookup **************************************/

5505
int
5506
toku_ft_lookup (FT_HANDLE brt, DBT *k, FT_GET_CALLBACK_FUNCTION getf, void *getf_v)
5507
{
5508
    int r, rr;
5509
    FT_CURSOR cursor;
5510

Yoni Fogel's avatar
Yoni Fogel committed
5511
    rr = toku_ft_cursor(brt, &cursor, NULL, false, false);
5512 5513
    if (rr != 0) return rr;

5514
    int op = DB_SET;
5515
    r = toku_ft_cursor_get(cursor, k, getf, getf_v, op);
5516

5517
    toku_ft_cursor_close(cursor);
5518 5519 5520 5521

    return r;
}

5522
/* ********************************* delete **************************************/
5523
static int
5524
getf_nothing (ITEMLEN UU(keylen), bytevec UU(key), ITEMLEN UU(vallen), bytevec UU(val), void *UU(pair_v), bool UU(lock_only)) {
5525 5526 5527 5528
    return 0;
}

int
5529
toku_ft_cursor_delete(FT_CURSOR cursor, int flags, TOKUTXN txn) {
5530 5531 5532
    int r;

    int unchecked_flags = flags;
Yoni Fogel's avatar
Yoni Fogel committed
5533
    bool error_if_missing = (bool) !(flags&DB_DELETE_ANY);
5534 5535
    unchecked_flags &= ~DB_DELETE_ANY;
    if (unchecked_flags!=0) r = EINVAL;
5536
    else if (ft_cursor_not_set(cursor)) r = EINVAL;
5537
    else {
5538 5539
        r = 0;
        if (error_if_missing) {
5540
            r = toku_ft_cursor_current(cursor, DB_CURRENT, getf_nothing, NULL);
5541 5542
        }
        if (r == 0) {
5543
            toku_ft_delete(cursor->ft_handle, &cursor->key, txn);
5544
        }
5545 5546 5547 5548 5549 5550 5551
    }
    return r;
}

/* ********************* keyrange ************************ */


5552
struct keyrange_compare_s {
5553
    FT_HANDLE ft_handle;
5554 5555 5556
    DBT *key;
};

5557
static int
5558
keyrange_compare (OMTVALUE lev, void *extra) {
5559
    LEAFENTRY CAST_FROM_VOIDP(le, lev);
Yoni Fogel's avatar
Yoni Fogel committed
5560
    uint32_t keylen;
5561 5562 5563
    void* key = le_key_and_len(le, &keylen);
    DBT   omt_dbt;
    toku_fill_dbt(&omt_dbt, key, keylen);
5564
    struct keyrange_compare_s *CAST_FROM_VOIDP(s, extra);
5565
    // TODO: maybe put a const fake_db in the header
5566 5567
    FAKE_DB(db, &s->ft_handle->ft->cmp_descriptor);
    return s->ft_handle->ft->compare_fun(&db, &omt_dbt, s->key);
5568 5569
}

5570
static void
Yoni Fogel's avatar
Yoni Fogel committed
5571 5572
keyrange_in_leaf_partition (FT_HANDLE brt, FTNODE node, DBT *key, int child_number, uint64_t estimated_num_rows,
                            uint64_t *less, uint64_t *equal, uint64_t *greater)
5573
// If the partition is in main memory then estimate the number
5574
// If KEY==NULL then use an arbitrary key (leftmost or zero)
5575
{
5576
    paranoid_invariant(node->height == 0); // we are in a leaf
5577
    if (BP_STATE(node, child_number) == PT_AVAIL) {
5578 5579 5580 5581
        // If the partition is in main memory then get an exact count.
        struct keyrange_compare_s s = {brt,key};
        BASEMENTNODE bn = BLB(node, child_number);
        OMTVALUE datav;
Yoni Fogel's avatar
Yoni Fogel committed
5582
        uint32_t idx = 0;
5583 5584 5585 5586 5587 5588 5589 5590 5591 5592 5593 5594
        // if key is NULL then set r==-1 and idx==0.
        int r = key ? toku_omt_find_zero(bn->buffer, keyrange_compare, &s, &datav, &idx) : -1;
        if (r==0) {
            *less    = idx;
            *equal   = 1;
            *greater = toku_omt_size(bn->buffer)-idx-1;
        } else {
            // If not found, then the idx says where it's between.
            *less    = idx;
            *equal   = 0;
            *greater = toku_omt_size(bn->buffer)-idx;
        }
5595
    } else {
5596 5597 5598
        *less    = estimated_num_rows / 2;
        *equal   = 0;
        *greater = *less;
5599 5600 5601
    }
}

5602
static int
5603
toku_ft_keyrange_internal (FT_HANDLE brt, FTNODE node,
Yoni Fogel's avatar
Yoni Fogel committed
5604 5605
                            DBT *key, uint64_t *less, uint64_t *equal, uint64_t *greater,
                            uint64_t estimated_num_rows,
5606
                            struct ftnode_fetch_extra *bfe, // set up to read a minimal read.
5607
                            struct unlockers *unlockers, ANCESTORS ancestors, struct pivot_bounds const * const bounds)
5608 5609 5610
// Implementation note: Assign values to less, equal, and greater, and then on the way out (returning up the stack) we add more values in.
{
    int r = 0;
5611
    // if KEY is NULL then use the leftmost key.
5612
    int child_number = key ? toku_ftnode_which_child (node, key, &brt->ft->cmp_descriptor, brt->ft->compare_fun) : 0;
5613 5614 5615
    uint64_t rows_per_child = estimated_num_rows / node->n_children;
    if (node->height == 0) {

5616
        keyrange_in_leaf_partition(brt, node, key, child_number, rows_per_child, less, equal, greater);
5617 5618 5619

        *less    += rows_per_child * child_number;
        *greater += rows_per_child * (node->n_children - child_number - 1);
5620

5621
    } else {
5622 5623 5624
        // do the child.
        struct ancestors next_ancestors = {node, child_number, ancestors};
        BLOCKNUM childblocknum = BP_BLOCKNUM(node, child_number);
Yoni Fogel's avatar
Yoni Fogel committed
5625
        uint32_t fullhash = compute_child_fullhash(brt->ft->cf, node, child_number);
5626
        FTNODE childnode;
Yoni Fogel's avatar
Yoni Fogel committed
5627
        bool msgs_applied = false;
5628
        r = toku_pin_ftnode_batched(
5629 5630 5631 5632 5633 5634
            brt,
            childblocknum,
            fullhash,
            unlockers,
            &next_ancestors,
            bounds,
Zardosht Kasheff's avatar
Zardosht Kasheff committed
5635
            bfe,
Zardosht Kasheff's avatar
Zardosht Kasheff committed
5636
            PL_READ, // may_modify_node is false, because node guaranteed to not change
Yoni Fogel's avatar
Yoni Fogel committed
5637
            false,
5638
            &childnode,
Zardosht Kasheff's avatar
Zardosht Kasheff committed
5639 5640
            &msgs_applied
            );
5641
        paranoid_invariant(!msgs_applied);
5642
        if (r != TOKUDB_TRY_AGAIN) {
5643
            assert_zero(r);
5644

Yoni Fogel's avatar
Yoni Fogel committed
5645 5646
            struct unlock_ftnode_extra unlock_extra   = {brt,childnode,false};
            struct unlockers next_unlockers = {true, unlock_ftnode_fun, (void*)&unlock_extra, unlockers};
5647
            const struct pivot_bounds next_bounds = next_pivot_keys(node, child_number, bounds);
5648

5649
            r = toku_ft_keyrange_internal(brt, childnode, key, less, equal, greater, rows_per_child,
5650
                                           bfe, &next_unlockers, &next_ancestors, &next_bounds);
5651
            if (r != TOKUDB_TRY_AGAIN) {
5652
                assert_zero(r);
5653

5654 5655
                *less    += rows_per_child * child_number;
                *greater += rows_per_child * (node->n_children - child_number - 1);
5656

5657
                assert(unlockers->locked);
5658
                toku_unpin_ftnode_read_only(brt->ft, childnode);
5659 5660
            }
        }
5661
    }
5662
    return r;
5663 5664
}

5665
void toku_ft_keyrange(FT_HANDLE brt, DBT *key, uint64_t *less_p, uint64_t *equal_p, uint64_t *greater_p)
5666 5667
// Effect: Return an estimate  of the number of keys to the left, the number equal, and the number to the right of the key.
//   The values are an estimate.
5668
//   If you perform a keyrange on two keys that are in the same in-memory and uncompressed basement,
5669 5670 5671
//   you can use the keys_right numbers (or the keys_left) numbers to get an exact number keys in the range,
//   if the basement does not change between the keyrange queries.
//   TODO 4184: What to do with a NULL key?
5672
//   If KEY is NULL then the system picks an arbitrary key and returns it.
5673
{
5674
    struct ftnode_fetch_extra bfe;
5675
    fill_bfe_for_min_read(&bfe, brt->ft);  // read pivot keys but not message buffers
5676
try_again:
5677
    {
Yoni Fogel's avatar
Yoni Fogel committed
5678
        uint64_t less = 0, equal = 0, greater = 0;
5679
        FTNODE node = NULL;
Zardosht Kasheff's avatar
Zardosht Kasheff committed
5680
        {
Yoni Fogel's avatar
Yoni Fogel committed
5681
            uint32_t fullhash;
5682
            CACHEKEY root_key;
5683
            toku_calculate_root_offset_pointer(brt->ft, &root_key, &fullhash);
5684
            toku_pin_ftnode_off_client_thread_batched(
5685
                brt->ft,
5686
                root_key,
5687
                fullhash,
5688
                &bfe,
Zardosht Kasheff's avatar
Zardosht Kasheff committed
5689
                PL_READ, // may_modify_node, cannot change root during keyrange
5690 5691 5692 5693
                0,
                NULL,
                &node
                );
Zardosht Kasheff's avatar
Zardosht Kasheff committed
5694
        }
5695

Yoni Fogel's avatar
Yoni Fogel committed
5696 5697
        struct unlock_ftnode_extra unlock_extra = {brt,node,false};
        struct unlockers unlockers = {true, unlock_ftnode_fun, (void*)&unlock_extra, (UNLOCKERS)NULL};
5698 5699

        {
5700
            int64_t numrows = brt->ft->in_memory_stats.numrows;
5701 5702
            if (numrows < 0)
                numrows = 0;  // prevent appearance of a negative number
5703
            int r = toku_ft_keyrange_internal (brt, node, key,
5704 5705 5706 5707 5708 5709 5710 5711 5712 5713
                                                &less, &equal, &greater,
                                                numrows,
                                                &bfe, &unlockers, (ANCESTORS)NULL, &infinite_bounds);
            assert(r == 0 || r == TOKUDB_TRY_AGAIN);
            if (r == TOKUDB_TRY_AGAIN) {
                assert(!unlockers.locked);
                goto try_again;
            }
        }
        assert(unlockers.locked);
5714
        toku_unpin_ftnode_read_only(brt->ft, node);
5715 5716 5717
        *less_p    = less;
        *equal_p   = equal;
        *greater_p = greater;
5718
    }
5719 5720
}

5721
void toku_ft_handle_stat64 (FT_HANDLE brt, TOKUTXN UU(txn), struct ftstat64_s *s) {
5722
    toku_ft_stat64(brt->ft, s);
5723 5724
}

5725 5726 5727 5728 5729 5730 5731 5732
void toku_ft_handle_get_fractal_tree_info64(FT_HANDLE ft_h, struct ftinfo64 *s) {
    toku_ft_get_fractal_tree_info64(ft_h->ft, s);
}

int toku_ft_handle_iterate_fractal_tree_block_map(FT_HANDLE ft_h, int (*iter)(uint64_t,int64_t,int64_t,int64_t,int64_t,void*), void *iter_extra) {
    return toku_ft_iterate_fractal_tree_block_map(ft_h->ft, iter, iter_extra);
}

5733 5734
/* ********************* debugging dump ************************ */
static int
5735
toku_dump_ftnode (FILE *file, FT_HANDLE brt, BLOCKNUM blocknum, int depth, const DBT *lorange, const DBT *hirange) {
5736
    int result=0;
5737
    FTNODE node;
5738
    toku_get_node_for_verify(blocknum, brt, &node);
5739
    result=toku_verify_ftnode(brt, brt->ft->h->max_msn_in_ft, brt->ft->h->max_msn_in_ft, false, node, -1, lorange, hirange, NULL, NULL, 0, 1, 0);
Yoni Fogel's avatar
Yoni Fogel committed
5740
    uint32_t fullhash = toku_cachetable_hash(brt->ft->cf, blocknum);
5741
    struct ftnode_fetch_extra bfe;
5742
    fill_bfe_for_full_read(&bfe, brt->ft);
Zardosht Kasheff's avatar
Zardosht Kasheff committed
5743 5744
    toku_pin_ftnode_off_client_thread(
        brt->ft,
5745
        blocknum,
5746
        fullhash,
Zardosht Kasheff's avatar
Zardosht Kasheff committed
5747 5748 5749
        &bfe,
        PL_WRITE_EXPENSIVE,
        0,
5750
        NULL,
Zardosht Kasheff's avatar
Zardosht Kasheff committed
5751
        &node
5752
        );
5753
    assert(node->fullhash==fullhash);
5754
    fprintf(file, "%*sNode=%p\n", depth, "", node);
5755

Zardosht Kasheff's avatar
Zardosht Kasheff committed
5756 5757
    fprintf(file, "%*sNode %" PRId64 " height=%d n_children=%d  keyrange=%s %s\n",
            depth, "", blocknum.b, node->height, node->n_children, (char*)(lorange ? lorange->data : 0), (char*)(hirange ? hirange->data : 0));
5758
    {
5759 5760 5761
        int i;
        for (i=0; i+1< node->n_children; i++) {
            fprintf(file, "%*spivotkey %d =", depth+1, "", i);
5762
            toku_print_BYTESTRING(file, node->childkeys[i].size, (char *) node->childkeys[i].data);
5763 5764 5765 5766
            fprintf(file, "\n");
        }
        for (i=0; i< node->n_children; i++) {
            if (node->height > 0) {
5767
                NONLEAF_CHILDINFO bnc = BNC(node, i);
5768 5769 5770 5771
                fprintf(file, "%*schild %d buffered (%d entries):", depth+1, "", i, toku_bnc_n_entries(bnc));
                FIFO_ITERATE(bnc->buffer, key, keylen, data, datalen, type, msn, xids, UU(is_fresh),
                             {
                                 data=data; datalen=datalen; keylen=keylen;
5772
                                 fprintf(file, "%*s xid=%" PRIu64 " %u (type=%d) msn=0x%" PRIu64 "\n", depth+2, "", xids_get_innermost_xid(xids), (unsigned)toku_dtoh32(*(int*)key), type, msn.msn);
5773 5774 5775 5776 5777 5778 5779 5780 5781
                                 //assert(strlen((char*)key)+1==keylen);
                                 //assert(strlen((char*)data)+1==datalen);
                             });
            }
            else {
                int size = toku_omt_size(BLB_BUFFER(node, i));
                if (0)
                    for (int j=0; j<size; j++) {
                        OMTVALUE v = 0;
Zardosht Kasheff's avatar
Zardosht Kasheff committed
5782
                        int r = toku_omt_fetch(BLB_BUFFER(node, i), j, &v);
5783
                        assert_zero(r);
5784
                        LEAFENTRY CAST_FROM_VOIDP(le, v);
5785
                        fprintf(file, " [%d]=", j);
5786
                        print_leafentry(file, le);
5787 5788 5789 5790 5791 5792 5793 5794 5795 5796
                        fprintf(file, "\n");
                    }
                //               printf(" (%d)%u ", len, *(int*)le_key(data)));
                fprintf(file, "\n");
            }
        }
        if (node->height > 0) {
            for (i=0; i<node->n_children; i++) {
                fprintf(file, "%*schild %d\n", depth, "", i);
                if (i>0) {
5797
                    char *CAST_FROM_VOIDP(key, node->childkeys[i-1].data);
5798 5799
                    fprintf(file, "%*spivot %d len=%u %u\n", depth+1, "", i-1, node->childkeys[i-1].size, (unsigned)toku_dtoh32(*(int*)key));
                }
5800
                toku_dump_ftnode(file, brt, BP_BLOCKNUM(node, i), depth+4,
5801 5802 5803 5804
                                  (i==0) ? lorange : &node->childkeys[i-1],
                                  (i==node->n_children-1) ? hirange : &node->childkeys[i]);
            }
        }
5805
    }
Zardosht Kasheff's avatar
Zardosht Kasheff committed
5806
    toku_unpin_ftnode_off_client_thread(brt->ft, node);
5807 5808 5809
    return result;
}

5810
int toku_dump_ft (FILE *f, FT_HANDLE brt) {
Zardosht Kasheff's avatar
Zardosht Kasheff committed
5811
    int r;
5812 5813
    assert(brt->ft);
    toku_dump_translation_table(f, brt->ft->blocktable);
Zardosht Kasheff's avatar
Zardosht Kasheff committed
5814
    {
Yoni Fogel's avatar
Yoni Fogel committed
5815
        uint32_t fullhash = 0;
5816
        CACHEKEY root_key;
5817
        toku_calculate_root_offset_pointer(brt->ft, &root_key, &fullhash);
5818
        r = toku_dump_ftnode(f, brt, root_key, 0, 0, 0);
Zardosht Kasheff's avatar
Zardosht Kasheff committed
5819 5820
    }
    return r;
5821
}
5822

5823
int toku_ft_layer_init(void) {
Yoni Fogel's avatar
Yoni Fogel committed
5824 5825
    int r = 0;
    //Portability must be initialized first
5826 5827 5828
    r = toku_portability_init();
    if (r) { goto exit; }

5829
    partitioned_counters_init();
5830
    status_init();
5831
    txn_status_init();
5832 5833
    toku_checkpoint_init();
    toku_ft_serialize_layer_init();
5834 5835
    toku_mutex_init(&ft_open_close_lock, NULL);
exit:
Yoni Fogel's avatar
Yoni Fogel committed
5836
    return r;
Yoni Fogel's avatar
Yoni Fogel committed
5837 5838
}

5839
void toku_ft_layer_destroy(void) {
5840
    toku_mutex_destroy(&ft_open_close_lock);
5841 5842
    toku_ft_serialize_layer_destroy();
    toku_checkpoint_destroy();
5843
    status_destroy();
5844
    txn_status_destroy();
5845
    partitioned_counters_destroy();
Yoni Fogel's avatar
Yoni Fogel committed
5846
    //Portability must be cleaned up last
5847
    toku_portability_destroy();
Yoni Fogel's avatar
Yoni Fogel committed
5848 5849
}

5850 5851
// This lock serializes all opens and closes because the cachetable requires that clients do not try to open or close a cachefile in parallel.  We made
// it coarser by not allowing any cachefiles to be open or closed in parallel.
5852 5853 5854 5855 5856 5857 5858 5859
void toku_ft_open_close_lock(void) {
    toku_mutex_lock(&ft_open_close_lock);
}

void toku_ft_open_close_unlock(void) {
    toku_mutex_unlock(&ft_open_close_lock);
}

5860 5861
// Prepare to remove a dictionary from the database when this transaction is committed:
//  - mark transaction as NEED fsync on commit
5862
//  - make entry in rollback log
5863
//  - make fdelete entry in recovery log
5864 5865 5866 5867 5868 5869 5870 5871 5872 5873 5874 5875
//
// Effect: when the txn commits, the ft's cachefile will be marked as unlink
//         on close. see toku_commit_fdelete and how unlink on close works
//         in toku_cachefile_close();
// Requires: serialized with begin checkpoint
//           this does not need to take the open close lock because
//           1.) the ft/cf cannot go away because we have a live handle.
//           2.) we're not setting the unlink on close bit _here_. that
//           happens on txn commit (as the name suggests).
//           3.) we're already holding the multi operation lock to 
//           synchronize with begin checkpoint.
// Contract: the iname of the ft should never be reused.
5876
void toku_ft_unlink_on_commit(FT_HANDLE handle, TOKUTXN txn) {
5877
    assert(txn);
5878 5879

    CACHEFILE cf = handle->ft->cf;
5880
    FT CAST_FROM_VOIDP(ft, toku_cachefile_get_userdata(cf));
5881

5882 5883
    toku_txn_maybe_note_ft(txn, ft);

5884 5885 5886 5887
    // If the txn commits, the commit MUST be in the log before the file is actually unlinked
    toku_txn_force_fsync_on_commit(txn); 
    // make entry in rollback log
    FILENUM filenum = toku_cachefile_filenum(cf);
5888
    toku_logger_save_rollback_fdelete(txn, filenum);
5889
    // make entry in recovery log
5890
    toku_logger_log_fdelete(txn, filenum);
5891 5892
}

5893 5894 5895 5896 5897 5898
// Non-transactional version of fdelete
//
// Effect: The ft file is unlinked when the handle closes and it's ft is not
//         pinned by checkpoint. see toku_remove_ft_ref() and how unlink on
//         close works in toku_cachefile_close();
// Requires: serialized with begin checkpoint
5899
void toku_ft_unlink(FT_HANDLE handle) {
5900
    CACHEFILE cf;
5901 5902
    cf = handle->ft->cf;
    toku_cachefile_unlink_on_close(cf);
5903 5904
}

5905
int
5906
toku_ft_get_fragmentation(FT_HANDLE brt, TOKU_DB_FRAGMENTATION report) {
5907 5908
    int r;

5909
    int fd = toku_cachefile_get_fd(brt->ft->cf);
5910
    toku_ft_lock(brt->ft);
5911 5912

    int64_t file_size;
5913
    r = toku_os_get_file_size(fd, &file_size);
5914
    if (r==0) {
5915
        report->file_size_bytes = file_size;
5916
        toku_block_table_get_fragmentation_unlocked(brt->ft->blocktable, report);
5917
    }
5918
    toku_ft_unlock(brt->ft);
5919 5920
    return r;
}
5921

Yoni Fogel's avatar
Yoni Fogel committed
5922
static bool is_empty_fast_iter (FT_HANDLE brt, FTNODE node) {
5923
    if (node->height > 0) {
5924
        for (int childnum=0; childnum<node->n_children; childnum++) {
5925
            if (toku_bnc_nbytesinbuf(BNC(node, childnum)) != 0) {
5926 5927
                return 0; // it's not empty if there are bytes in buffers
            }
5928
            FTNODE childnode;
5929 5930
            {
                BLOCKNUM childblocknum = BP_BLOCKNUM(node,childnum);
Yoni Fogel's avatar
Yoni Fogel committed
5931
                uint32_t fullhash =  compute_child_fullhash(brt->ft->cf, node, childnum);
5932
                struct ftnode_fetch_extra bfe;
5933
                fill_bfe_for_full_read(&bfe, brt->ft);
5934
                // don't need to pass in dependent nodes as we are not
5935
                // modifying nodes we are pinning
5936
                toku_pin_ftnode_off_client_thread(
5937
                    brt->ft,
5938 5939 5940
                    childblocknum,
                    fullhash,
                    &bfe,
Zardosht Kasheff's avatar
Zardosht Kasheff committed
5941
                    PL_READ, // may_modify_node set to false, as nodes not modified
5942 5943 5944
                    0,
                    NULL,
                    &childnode
5945
                    );
5946 5947
            }
            int child_is_empty = is_empty_fast_iter(brt, childnode);
5948
            toku_unpin_ftnode(brt->ft, childnode);
5949 5950 5951
            if (!child_is_empty) return 0;
        }
        return 1;
5952
    } else {
5953 5954 5955
        // leaf:  If the omt is empty, we are happy.
        for (int i = 0; i < node->n_children; i++) {
            if (toku_omt_size(BLB_BUFFER(node, i))) {
Yoni Fogel's avatar
Yoni Fogel committed
5956
                return false;
5957 5958
            }
        }
Yoni Fogel's avatar
Yoni Fogel committed
5959
        return true;
5960 5961 5962
    }
}

Yoni Fogel's avatar
Yoni Fogel committed
5963
bool toku_ft_is_empty_fast (FT_HANDLE brt)
5964 5965 5966
// A fast check to see if the tree is empty.  If there are any messages or leafentries, we consider the tree to be nonempty.  It's possible that those
// messages and leafentries would all optimize away and that the tree is empty, but we'll say it is nonempty.
{
Yoni Fogel's avatar
Yoni Fogel committed
5967
    uint32_t fullhash;
5968
    FTNODE node;
5969
    {
5970
        CACHEKEY root_key;
5971
        toku_calculate_root_offset_pointer(brt->ft, &root_key, &fullhash);
5972
        struct ftnode_fetch_extra bfe;
5973
        fill_bfe_for_full_read(&bfe, brt->ft);
5974
        toku_pin_ftnode_off_client_thread(
5975
            brt->ft,
5976
            root_key,
5977
            fullhash,
5978
            &bfe,
Zardosht Kasheff's avatar
Zardosht Kasheff committed
5979
            PL_READ, // may_modify_node set to false, node does not change
5980 5981 5982
            0,
            NULL,
            &node
5983
            );
5984
    }
Yoni Fogel's avatar
Yoni Fogel committed
5985
    bool r = is_empty_fast_iter(brt, node);
5986
    toku_unpin_ftnode(brt->ft, node);
5987 5988
    return r;
}
5989

5990
// test-only
5991
int toku_ft_strerror_r(int error, char *buf, size_t buflen)
5992 5993
{
    if (error>=0) {
5994
        return (long) strerror_r(error, buf, buflen);
5995
    } else {
5996 5997 5998 5999 6000 6001 6002 6003 6004
        switch (error) {
        case DB_KEYEXIST:
            snprintf(buf, buflen, "Key exists");
            return 0;
        case TOKUDB_CANCELED:
            snprintf(buf, buflen, "User canceled operation");
            return 0;
        default:
            snprintf(buf, buflen, "Unknown error %d", error);
6005
            return EINVAL;
6006
        }
6007 6008
    }
}
6009

6010
#include <toku_race_tools.h>
6011
void __attribute__((__constructor__)) toku_ft_helgrind_ignore(void);
6012
void
6013
toku_ft_helgrind_ignore(void) {
6014
    TOKU_VALGRIND_HG_DISABLE_CHECKING(&ft_status, sizeof ft_status);
6015
}
6016

6017
#undef STATUS_INC