Commit d5f2eec0 authored by Bradley C. Kuszmaul's avatar Bradley C. Kuszmaul

Add the list of txns. Addresses #735. Fixes #749.

git-svn-id: file:///svn/tokudb@3659 c7de825b-a66e-492c-adef-691d508d4ae1
parent 4d059818
...@@ -1810,6 +1810,8 @@ int toku_brt_create(BRT *brt_ptr) { ...@@ -1810,6 +1810,8 @@ int toku_brt_create(BRT *brt_ptr) {
brt->nodesize = BRT_DEFAULT_NODE_SIZE; brt->nodesize = BRT_DEFAULT_NODE_SIZE;
brt->compare_fun = toku_default_compare_fun; brt->compare_fun = toku_default_compare_fun;
brt->dup_compare = toku_default_compare_fun; brt->dup_compare = toku_default_compare_fun;
int r = toku_omt_create(&brt->txns);
if (r!=0) { toku_free(brt); return r; }
*brt_ptr = brt; *brt_ptr = brt;
return 0; return 0;
} }
...@@ -2114,6 +2116,10 @@ int toku_close_brt (BRT brt, TOKULOGGER logger) { ...@@ -2114,6 +2116,10 @@ int toku_close_brt (BRT brt, TOKULOGGER logger) {
if (brt->fname) toku_free(brt->fname); if (brt->fname) toku_free(brt->fname);
if (brt->skey) { toku_free(brt->skey); } if (brt->skey) { toku_free(brt->skey); }
if (brt->sval) { toku_free(brt->sval); } if (brt->sval) { toku_free(brt->sval); }
assert(toku_omt_size(brt->txns)==0);
r=toku_txn_note_close_brt(brt);
assert(r==0);
toku_omt_destroy(&brt->txns);
toku_free(brt); toku_free(brt);
return 0; return 0;
} }
...@@ -2284,6 +2290,8 @@ int toku_brt_insert (BRT brt, DBT *key, DBT *val, TOKUTXN txn) { ...@@ -2284,6 +2290,8 @@ int toku_brt_insert (BRT brt, DBT *key, DBT *val, TOKUTXN txn) {
toku_cachefile_refup(brt->cf); toku_cachefile_refup(brt->cf);
r = toku_logger_save_rollback_cmdinsert(txn, toku_txn_get_txnid(txn), toku_cachefile_filenum(brt->cf), keybs, databs); r = toku_logger_save_rollback_cmdinsert(txn, toku_txn_get_txnid(txn), toku_cachefile_filenum(brt->cf), keybs, databs);
if (r!=0) return r; if (r!=0) return r;
r = toku_txn_note_brt(txn, brt);
if (r!=0) return r;
} }
BRT_CMD_S brtcmd = { BRT_INSERT, toku_txn_get_txnid(txn), .u.id={key,val}}; BRT_CMD_S brtcmd = { BRT_INSERT, toku_txn_get_txnid(txn), .u.id={key,val}};
r = toku_brt_root_put_cmd(brt, &brtcmd, toku_txn_logger(txn)); r = toku_brt_root_put_cmd(brt, &brtcmd, toku_txn_logger(txn));
...@@ -2313,6 +2321,8 @@ int toku_brt_delete(BRT brt, DBT *key, TOKUTXN txn) { ...@@ -2313,6 +2321,8 @@ int toku_brt_delete(BRT brt, DBT *key, TOKUTXN txn) {
toku_cachefile_refup(brt->cf); toku_cachefile_refup(brt->cf);
r = toku_logger_save_rollback_cmddelete(txn, toku_txn_get_txnid(txn), toku_cachefile_filenum(brt->cf), keybs); r = toku_logger_save_rollback_cmddelete(txn, toku_txn_get_txnid(txn), toku_cachefile_filenum(brt->cf), keybs);
if (r!=0) return r; if (r!=0) return r;
r = toku_txn_note_brt(txn, brt);
if (r!=0) return r;
} }
DBT val; DBT val;
BRT_CMD_S brtcmd = { BRT_DELETE_ANY, toku_txn_get_txnid(txn), .u.id={key, toku_init_dbt(&val)}}; BRT_CMD_S brtcmd = { BRT_DELETE_ANY, toku_txn_get_txnid(txn), .u.id={key, toku_init_dbt(&val)}};
...@@ -2328,6 +2338,8 @@ int toku_brt_delete_both(BRT brt, DBT *key, DBT *val, TOKUTXN txn) { ...@@ -2328,6 +2338,8 @@ int toku_brt_delete_both(BRT brt, DBT *key, DBT *val, TOKUTXN txn) {
toku_cachefile_refup(brt->cf); toku_cachefile_refup(brt->cf);
r = toku_logger_save_rollback_cmddeleteboth(txn, toku_txn_get_txnid(txn), toku_cachefile_filenum(brt->cf), keybs, databs); r = toku_logger_save_rollback_cmddeleteboth(txn, toku_txn_get_txnid(txn), toku_cachefile_filenum(brt->cf), keybs, databs);
if (r!=0) return r; if (r!=0) return r;
r = toku_txn_note_brt(txn, brt);
if (r!=0) return r;
} }
BRT_CMD_S brtcmd = { BRT_DELETE_BOTH, toku_txn_get_txnid(txn), .u.id={key,val}}; BRT_CMD_S brtcmd = { BRT_DELETE_BOTH, toku_txn_get_txnid(txn), .u.id={key,val}};
r = toku_brt_root_put_cmd(brt, &brtcmd, toku_txn_logger(txn)); r = toku_brt_root_put_cmd(brt, &brtcmd, toku_txn_logger(txn));
......
/* -*- mode: C; c-basic-offset: 4 -*- */
#ident "Copyright (c) 2007, 2008 Tokutek Inc. All rights reserved."
/*
* We always write nodes to a new location on disk.
* The nodes themselves contain the information about the tree structure.
* Q: During recovery, how do we find the root node without looking at every block on disk?
* A: The root node is either the designated root near the front of the freelist.
* The freelist is updated infrequently. Before updating the stable copy of the freelist, we make sure that
* the root is up-to-date. We can make the freelist-and-root update be an arbitrarily small fraction of disk bandwidth.
*
*/
#include <arpa/inet.h>
#include <errno.h>
#include <inttypes.h>
#include <stdio.h>
#include <stdlib.h>
#include <string.h>
#include <unistd.h>
#include "toku_assert.h"
#include "brt-internal.h"
#include "key.h"
#include "log_header.h"
typedef struct weakstrong { char ignore; } *WS;
#define WEAK ((WS)1)
#define STRONG ((WS)0)
extern long long n_items_malloced;
static void verify_local_fingerprint_nonleaf (BRTNODE node);
#ifdef FOO
static int malloc_diskblock (DISKOFF *res, BRT brt, int size, TOKULOGGER);
/* Frees a node, including all the stuff in the hash table. */
void toku_brtnode_free (BRTNODE *nodep) {
BRTNODE node=*nodep;
int i;
//printf("%s:%d %p->mdict[0]=%p\n", __FILE__, __LINE__, node, node->mdicts[0]);
if (node->height>0) {
for (i=0; i<node->u.n.n_children-1; i++) {
toku_free((void*)node->u.n.childkeys[i]);
}
for (i=0; i<node->u.n.n_children; i++) {
if (BNC_BUFFER(node,i)) {
toku_fifo_free(&BNC_BUFFER(node,i));
}
}
} else {
if (node->u.l.buffer) // The buffer may have been freed already, in some cases.
toku_pma_free(&node->u.l.buffer);
}
toku_free(node);
*nodep=0;
}
#endif
static long brtnode_size(BRTNODE node) {
return toku_serialize_brtnode_size(node);
}
static void toku_update_brtnode_loggerlsn(BRTNODE node, TOKULOGGER logger) {
if (logger) {
node->log_lsn = toku_logger_last_lsn(logger);
}
}
static void fixup_child_fingerprint(BRTNODE node, int childnum_of_node, BRTNODE child, BRT brt, TOKULOGGER logger) {
u_int32_t old_fingerprint = BNC_SUBTREE_FINGERPRINT(node,childnum_of_node);
u_int32_t sum = child->local_fingerprint;
if (child->height>0) {
int i;
for (i=0; i<child->u.n.n_children; i++) {
sum += BNC_SUBTREE_FINGERPRINT(child,i);
}
}
// Don't try to get fancy about not modifying the fingerprint if it didn't change.
// We only call this function if we have reason to believe that the child's fingerprint did change.
BNC_SUBTREE_FINGERPRINT(node,childnum_of_node)=sum;
node->dirty=1;
toku_log_changechildfingerprint(logger, toku_cachefile_filenum(brt->cf), node->thisnodename, childnum_of_node, old_fingerprint, sum);
toku_update_brtnode_loggerlsn(node, logger);
}
// If you pass in data==0 then it only compares the key, not the data (even if is a DUPSORT database)
static int brt_compare_pivot(BRT brt, DBT *key, DBT *data, bytevec ck) {
int cmp;
DBT mydbt;
struct kv_pair *kv = (struct kv_pair *) ck;
if (brt->flags & TOKU_DB_DUPSORT) {
cmp = brt->compare_fun(brt->db, key, toku_fill_dbt(&mydbt, kv_pair_key(kv), kv_pair_keylen(kv)));
if (cmp == 0 && data != 0)
cmp = brt->dup_compare(brt->db, data, toku_fill_dbt(&mydbt, kv_pair_val(kv), kv_pair_vallen(kv)));
} else {
cmp = brt->compare_fun(brt->db, key, toku_fill_dbt(&mydbt, kv_pair_key(kv), kv_pair_keylen(kv)));
}
return cmp;
}
#ifdef FOO
void toku_brtnode_flush_callback (CACHEFILE cachefile, DISKOFF nodename, void *brtnode_v, long size __attribute((unused)), BOOL write_me, BOOL keep_me, LSN modified_lsn __attribute__((__unused__)) , BOOL rename_p __attribute__((__unused__))) {
BRTNODE brtnode = brtnode_v;
if (0) {
printf("%s:%d toku_brtnode_flush_callback %p keep_me=%d height=%d", __FILE__, __LINE__, brtnode, keep_me, brtnode->height);
if (brtnode->height==0) printf(" pma=%p", brtnode->u.l.buffer);
printf("\n");
}
assert(brtnode->thisnodename==nodename);
if (write_me) {
toku_serialize_brtnode_to(toku_cachefile_fd(cachefile), brtnode->thisnodename, brtnode->nodesize, brtnode);
}
if (!keep_me) {
toku_brtnode_free(&brtnode);
}
}
int toku_brtnode_fetch_callback (CACHEFILE cachefile, DISKOFF nodename, void **brtnode_pv, long *sizep, void*extraargs, LSN *written_lsn) {
BRT t =(BRT)extraargs;
BRTNODE *result=(BRTNODE*)brtnode_pv;
int r = toku_deserialize_brtnode_from(toku_cachefile_fd(cachefile), nodename, result, t->flags, t->nodesize,
t->compare_fun, t->dup_compare, t->db, toku_cachefile_filenum(t->cf));
if (r == 0) {
*sizep = brtnode_size(*result);
*written_lsn = (*result)->disk_lsn;
}
return r;
}
void toku_brtheader_flush_callback (CACHEFILE cachefile, DISKOFF nodename, void *header_v, long size __attribute((unused)), BOOL write_me, BOOL keep_me, LSN lsn __attribute__((__unused__)), BOOL rename_p __attribute__((__unused__))) {
struct brt_header *h = header_v;
assert(nodename==0);
assert(!h->dirty); // shouldn't be dirty once it is unpinned.
if (write_me) {
toku_serialize_brt_header_to(toku_cachefile_fd(cachefile), h);
}
if (!keep_me) {
if (h->n_named_roots>0) {
int i;
for (i=0; i<h->n_named_roots; i++) {
toku_free(h->names[i]);
}
toku_free(h->names);
toku_free(h->roots);
}
toku_free(h);
}
}
int toku_brtheader_fetch_callback (CACHEFILE cachefile, DISKOFF nodename, void **headerp_v, long *sizep __attribute__((unused)), void*extraargs __attribute__((__unused__)), LSN *written_lsn) {
struct brt_header **h = (struct brt_header **)headerp_v;
assert(nodename==0);
int r = toku_deserialize_brtheader_from(toku_cachefile_fd(cachefile), nodename, h);
written_lsn->lsn = 0; // !!! WRONG. This should be stored or kept redundantly or something.
return r;
}
int toku_read_and_pin_brt_header (CACHEFILE cf, struct brt_header **header) {
void *header_p;
//fprintf(stderr, "%s:%d read_and_pin_brt_header(...)\n", __FILE__, __LINE__);
int r = toku_cachetable_get_and_pin(cf, 0, &header_p, NULL,
toku_brtheader_flush_callback, toku_brtheader_fetch_callback, 0);
if (r!=0) return r;
*header = header_p;
return 0;
}
int toku_unpin_brt_header (BRT brt) {
int r = toku_cachetable_unpin(brt->cf, 0, brt->h->dirty, 0);
brt->h->dirty=0;
brt->h=0;
return r;
}
#endif
static int unpin_brtnode (BRT brt, BRTNODE node) {
return toku_cachetable_unpin(brt->cf, node->thisnodename, node->dirty, brtnode_size(node));
}
#ifdef FOO
typedef struct kvpair {
bytevec key;
unsigned int keylen;
bytevec val;
unsigned int vallen;
} *KVPAIR;
/* Forgot to handle the case where there is something in the freelist. */
static int malloc_diskblock_header_is_in_memory (DISKOFF *res, BRT brt, int size, TOKULOGGER logger) {
DISKOFF result = brt->h->unused_memory;
brt->h->unused_memory+=size;
brt->h->dirty = 1;
int r = toku_log_changeunusedmemory(logger, toku_cachefile_filenum(brt->cf), result, brt->h->unused_memory);
*res = result;
return r;
}
int malloc_diskblock (DISKOFF *res, BRT brt, int size, TOKULOGGER logger) {
#if 0
int r = read_and_pin_brt_header(brt->fd, &brt->h);
assert(r==0);
{
DISKOFF result = malloc_diskblock_header_is_in_memory(brt, size);
r = write_brt_header(brt->fd, &brt->h);
assert(r==0);
return result;
}
#else
return malloc_diskblock_header_is_in_memory(res, brt,size, logger);
#endif
}
static void initialize_brtnode (BRT t, BRTNODE n, DISKOFF nodename, int height) {
int i;
n->tag = TYP_BRTNODE;
n->nodesize = t->h->nodesize;
n->flags = t->h->flags;
n->thisnodename = nodename;
n->disk_lsn.lsn = 0; // a new one can always be 0.
n->log_lsn = n->disk_lsn;
n->layout_version = BRT_LAYOUT_VERSION;
n->height = height;
n->rand4fingerprint = random();
n->local_fingerprint = 0;
n->dirty = 1;
assert(height>=0);
if (height>0) {
n->u.n.n_children = 0;
for (i=0; i<TREE_FANOUT; i++) {
// n->u.n.childkeys[i] = 0;
// n->u.n.childkeylens[i] = 0;
}
n->u.n.totalchildkeylens = 0;
for (i=0; i<TREE_FANOUT+1; i++) {
BNC_SUBTREE_FINGERPRINT(n, i) = 0;
// n->u.n.children[i] = 0;
// n->u.n.buffers[i] = 0;
BNC_NBYTESINBUF(n,i) = 0;
}
n->u.n.n_bytes_in_buffers = 0;
} else {
int r = toku_pma_create(&n->u.l.buffer, t->compare_fun, t->db, toku_cachefile_filenum(t->cf), n->nodesize);
assert(r==0);
toku_pma_set_dup_mode(n->u.l.buffer, t->flags & (TOKU_DB_DUP+TOKU_DB_DUPSORT));
toku_pma_set_dup_compare(n->u.l.buffer, t->dup_compare);
static int rcount=0;
//printf("%s:%d n PMA= %p (rcount=%d)\n", __FILE__, __LINE__, n->u.l.buffer, rcount);
rcount++;
n->u.l.n_bytes_in_buffer = 0;
}
}
int toku_create_new_brtnode (BRT t, BRTNODE *result, int height, TOKULOGGER logger) {
TAGMALLOC(BRTNODE, n);
int r;
DISKOFF name;
if ((r = malloc_diskblock(&name, t, t->h->nodesize, logger))) return r;
assert(n);
assert(t->h->nodesize>0);
//printf("%s:%d malloced %lld (and malloc again=%lld)\n", __FILE__, __LINE__, name, malloc_diskblock(t, t->nodesize));
initialize_brtnode(t, n, name, height);
*result = n;
assert(n->nodesize>0);
// n->brt = t;
//printf("%s:%d putting %p (%lld) parent=%p\n", __FILE__, __LINE__, n, n->thisnodename, parent_brtnode);
if ((r = toku_cachetable_put(t->cf, n->thisnodename, n, brtnode_size(n), toku_brtnode_flush_callback, toku_brtnode_fetch_callback, t)))
return r;
if ((r = toku_log_newbrtnode(logger, toku_cachefile_filenum(t->cf), n->thisnodename, height, n->nodesize, (t->flags&TOKU_DB_DUPSORT)!=0, n->rand4fingerprint)))
return r;
toku_update_brtnode_loggerlsn(n, logger);
return 0;
}
#ifdef FOO
static int insert_to_buffer_in_nonleaf (BRTNODE node, int childnum, BRT_CMD cmd) {
unsigned int n_bytes_added = BRT_CMD_OVERHEAD + KEY_VALUE_OVERHEAD + cmd->u.id.key->size + cmd->u.id.val->size;
int r = toku_fifo_enq_cmdstruct(BNC_BUFFER(node,childnum), cmd);
if (r!=0) return r;
node->local_fingerprint += node->rand4fingerprint*toku_calccrc32_cmdstruct(cmd);
BNC_NBYTESINBUF(node,childnum) += n_bytes_added;
node->u.n.n_bytes_in_buffers += n_bytes_added;
node->dirty = 1;
return 0;
}
#endif
// Split a leaf node, reusing it in new_nodes (as the last element)
static int split_leaf_node (BRT t, TOKULOGGER logger, BRTNODE node, int *n_new_nodes, BRTNODE **new_nodes, DBT **splitks) {
assert(node->height==0);
int r;
int n_children=1; // Initially we have the node itself.
BRTNODE *result_nodes=toku_malloc(sizeof(*result_nodes));
if (errno!=0) { r=errno; if (0) { died0: toku_free(result_nodes); } return r; }
DBT *result_splitks=toku_malloc(sizeof(*result_splitks));
if (errno!=0) { r=errno; if (0) { died1: toku_free(result_splitks); } goto died0; }
while (toku_serialize_brtnode_size(node)>node->nodesize) {
BRTNODE B;
DBT splitk;
if ((r = toku_create_new_brtnode(t, &B, 0, logger))) return r;
// Split so that B is at least 1/2 full
// The stuff in B goes *before* node
if ((r = toku_pma_split(logger, toku_cachefile_filenum(t->cf),
node->thisnodename, node->u.l.buffer, &node->u.l.n_bytes_in_buffer, node->rand4fingerprint, &node->local_fingerprint, &node->log_lsn,
&splitk,
B->thisnodename, B->u.l.buffer, &B->u.l.n_bytes_in_buffer, B->rand4fingerprint, &B->local_fingerprint, &B->log_lsn)))
goto died1;
n_children++;
result_nodes = toku_realloc(result_nodes, n_children*sizeof(*result_nodes));
result_nodes[n_children-2] = B;
result_splitks = toku_realloc(result_nodes, (n_children-1)*sizeof(*result_splitks));
result_splitks[n_children-2] = splitk;
}
result_nodes[n_children-1]=node;
*n_new_nodes = n_children;
*new_nodes = result_nodes;
*splitks = result_splitks;
return 0;
}
#endif
static void find_heaviest_child (BRTNODE node, int *childnum) {
int max_child = 0;
int max_weight = BNC_NBYTESINBUF(node, 0);
int i;
if (0) printf("%s:%d weights: %d", __FILE__, __LINE__, max_weight);
assert(node->u.n.n_children>0);
for (i=1; i<node->u.n.n_children; i++) {
int this_weight = BNC_NBYTESINBUF(node,i);
if (0) printf(" %d", this_weight);
if (max_weight < this_weight) {
max_child = i;
max_weight = this_weight;
}
}
*childnum = max_child;
if (0) printf("\n");
}
/* find the leftmost child that may contain the key */
static unsigned int brtnode_which_child (BRTNODE node , DBT *k, DBT *d, BRT t) {
int i;
assert(node->height>0);
for (i=0; i<node->u.n.n_children-1; i++) {
int cmp = brt_compare_pivot(t, k, d, node->u.n.childkeys[i]);
if (cmp > 0) continue;
if (cmp < 0) return i;
return i;
}
return node->u.n.n_children-1;
}
static int brtnode_put (BRT t, BRTNODE node, BRT_CMD cmd, TOKULOGGER logger, WS weak_p);
// If CHILD is too wide, split it, and create a new node with the new children. Unpin CHILD or the new children (even if something goes wrong).
// If it does split, unpin the new root node also.
static int maybe_split_root(BRT brt, BRTNODE child, CACHEKEY *rootp, TOKULOGGER logger);
// if CHILD is too wide, split it, and fix up NODE. Either way, unpin the child or resulting children (even if it fails do the unpin)
static int maybe_split_nonroot (BRT brt, BRTNODE node, int childnum, BRTNODE child, int *n_children_replacing_child, TOKULOGGER logger);
// Push stuff into a child weakly. (That is don't cause any I/O or cause the child to get too big.)
static int weak_push_to_child (BRT brt, BRTNODE node, int childnum, TOKULOGGER logger) {
void *child_v;
int r = toku_cachetable_maybe_get_and_pin(brt->cf, BNC_DISKOFF(node, childnum), &child_v);
if (r!=0) return 0;
BRTNODE child = child_v;
DBT key,val;
BRT_CMD_S cmd;
while (0 == toku_fifo_peek_cmdstruct(BNC_BUFFER(node, childnum), &cmd, &key, &val)) {
r = brtnode_put(brt, child, &cmd, logger, WEAK);
if (r==EAGAIN) break;
if (r!=0) goto died;
r=toku_fifo_deq(BNC_BUFFER(node, childnum));
if (r!=0) goto died;
}
return unpin_brtnode(brt, child);
died:
unpin_brtnode(brt, child);
return r;
}
// If the buffers are too big, push stuff down. The subchild may need to be split, in which case our fanout may get too large.
// When are done, this node is has little enough stuff in its buffers (but the fanout may be too large), and all the descendant
// nodes are properly sized (the buffer sizes and fanouts are all small enough).
static int push_down_if_buffers_too_full(BRT brt, BRTNODE node, TOKULOGGER logger) {
if (node->height==0) return 0; // can't push down for leaf nodes
while (node->u.n.n_bytes_in_buffers > 0 && toku_serialize_brtnode_size(node)>node->nodesize) {
int childnum;
find_heaviest_child(node, &childnum);
void *child_v;
int r = toku_cachetable_get_and_pin(brt->cf, BNC_DISKOFF(node, childnum), &child_v, NULL,
toku_brtnode_flush_callback, toku_brtnode_fetch_callback, brt);
if (r!=0) return r;
BRTNODE child=child_v;
if (0) { died: unpin_brtnode(brt, child); return r; }
BRT_CMD_S cmd;
DBT key,val;
while (0==toku_fifo_peek_cmdstruct(BNC_BUFFER(node, childnum), &cmd, &key, &val)) {
r=toku_fifo_deq(BNC_BUFFER(node, childnum));
assert(r==0); // we just did a peek, so the buffer must be nonempty
r=brtnode_put(brt, child, &cmd, logger, WEAK);
if (r!=EAGAIN && r!=0) goto died;
if (r==EAGAIN) {
// Weak pushes ran out of steam. Now do a strong push if there is still something in the buffer.
if (0==toku_fifo_peek_cmdstruct(BNC_BUFFER(node, childnum), &cmd, &key, &val)) {
r=brtnode_put(brt, child, &cmd, logger, STRONG);
if (r!=0) goto died;
r=toku_fifo_deq(BNC_BUFFER(node, childnum));
if (r!=0) goto died;
// Now it's possible that the child must be split. (Or maybe the child managed to flush stuff to our grandchildren)
int n_children_replacing_child;
r=maybe_split_nonroot(brt, node, childnum, child, &n_children_replacing_child, logger);
if (r!=0) return r; // don't go to died since that unpins
int i;
for (i=0; i<n_children_replacing_child; i++) {
r=weak_push_to_child(brt, node, childnum+i, logger);
if (r!=0) return r;
}
// we basically pushed as much as we could to that child
}
}
}
}
return 0;
}
/* Side effect: sets splitk->data pointer to a malloc'd value */
static int brt_nonleaf_split (BRT t, BRTNODE node, BRTNODE *nodeb, DBT *splitk, TOKULOGGER logger) {
int old_n_children = node->u.n.n_children;
int n_children_in_a = old_n_children/2;
int n_children_in_b = old_n_children-n_children_in_a;
BRTNODE B;
FILENUM fnum = toku_cachefile_filenum(t->cf);
assert(node->height>0);
assert(node->u.n.n_children>=2); // Otherwise, how do we split? We need at least two children to split. */
assert(t->h->nodesize>=node->nodesize); /* otherwise we might be in trouble because the nodesize shrank. */
toku_create_new_brtnode(t, &B, node->height, logger);
B->u.n.n_children =n_children_in_b;
//printf("%s:%d %p (%lld) becomes %p and %p\n", __FILE__, __LINE__, node, node->thisnodename, A, B);
//printf("%s:%d A is at %lld\n", __FILE__, __LINE__, A->thisnodename);
{
/* The first n_children_in_a go into node a.
* That means that the first n_children_in_a-1 keys go into node a.
* The splitter key is key number n_children_in_a */
int i;
for (i=0; i<n_children_in_b; i++) {
int r = toku_fifo_create(&BNC_BUFFER(B,i));
if (r!=0) return r;
}
for (i=n_children_in_a; i<old_n_children; i++) {
int targchild = i-n_children_in_a;
FIFO from_htab = BNC_BUFFER(node,i);
FIFO to_htab = BNC_BUFFER(B, targchild);
DISKOFF thischilddiskoff = BNC_DISKOFF(node, i);
BNC_DISKOFF(B, targchild) = thischilddiskoff;
int r = toku_log_addchild(logger, fnum, B->thisnodename, targchild, thischilddiskoff, BNC_SUBTREE_FINGERPRINT(node, i));
if (r!=0) return r;
while (1) {
bytevec key, data;
unsigned int keylen, datalen;
int type;
TXNID xid;
int fr = toku_fifo_peek(from_htab, &key, &keylen, &data, &datalen, &type, &xid);
if (fr!=0) break;
int n_bytes_moved = keylen+datalen + KEY_VALUE_OVERHEAD + BRT_CMD_OVERHEAD;
BYTESTRING keybs = { .len = keylen, .data = (char*)key };
BYTESTRING databs = { .len = datalen, .data = (char*)data };
u_int32_t old_from_fingerprint = node->local_fingerprint;
u_int32_t old_to_fingerprint = B->local_fingerprint;
u_int32_t delta = toku_calccrc32_cmd(type, xid, key, keylen, data, datalen);
u_int32_t new_from_fingerprint = old_from_fingerprint - node->rand4fingerprint*delta;
u_int32_t new_to_fingerprint = old_to_fingerprint + B->rand4fingerprint *delta;
if (r!=0) return r;
r = toku_log_brtdeq(logger, fnum, node->thisnodename, n_children_in_a, xid, type, keybs, databs, old_from_fingerprint, new_from_fingerprint);
if (r!=0) return r;
r = toku_log_brtenq(logger, fnum, B->thisnodename, targchild, xid, type, keybs, databs, old_to_fingerprint, new_to_fingerprint);
r = toku_fifo_enq(to_htab, key, keylen, data, datalen, type, xid);
if (r!=0) return r;
toku_fifo_deq(from_htab);
// key and data will no longer be valid
node->local_fingerprint = new_from_fingerprint;
B->local_fingerprint = new_to_fingerprint;
B->u.n.n_bytes_in_buffers += n_bytes_moved;
BNC_NBYTESINBUF(B, targchild) += n_bytes_moved;
node->u.n.n_bytes_in_buffers -= n_bytes_moved;
BNC_NBYTESINBUF(node, i) -= n_bytes_moved;
// verify_local_fingerprint_nonleaf(B);
// verify_local_fingerprint_nonleaf(node);
}
// Delete a child, removing it's fingerprint, and also the preceeding pivot key. The child number must be > 0
{
BYTESTRING bs = { .len = kv_pair_keylen(node->u.n.childkeys[i-1]),
.data = kv_pair_key(node->u.n.childkeys[i-1]) };
assert(i>0);
r = toku_log_delchild(logger, fnum, node->thisnodename, n_children_in_a, thischilddiskoff, BNC_SUBTREE_FINGERPRINT(node, i), bs);
if (r!=0) return r;
if (i>n_children_in_a) {
r = toku_log_setpivot(logger, fnum, B->thisnodename, targchild-1, bs);
if (r!=0) return r;
B->u.n.childkeys[targchild-1] = node->u.n.childkeys[i-1];
B->u.n.totalchildkeylens += toku_brt_pivot_key_len(t, node->u.n.childkeys[i-1]);
node->u.n.totalchildkeylens -= toku_brt_pivot_key_len(t, node->u.n.childkeys[i-1]);
node->u.n.childkeys[i-1] = 0;
}
}
BNC_DISKOFF(node, i) = 0;
BNC_SUBTREE_FINGERPRINT(B, targchild) = BNC_SUBTREE_FINGERPRINT(node, i);
BNC_SUBTREE_FINGERPRINT(node, i) = 0;
assert(BNC_NBYTESINBUF(node, i) == 0);
}
// Drop the n_children now (not earlier) so that we can do the fingerprint verification at any time.
node->u.n.n_children=n_children_in_a;
for (i=n_children_in_a; i<old_n_children; i++) {
toku_fifo_free(&BNC_BUFFER(node,i));
}
splitk->data = (void*)(node->u.n.childkeys[n_children_in_a-1]);
splitk->size = toku_brt_pivot_key_len(t, node->u.n.childkeys[n_children_in_a-1]);
node->u.n.totalchildkeylens -= toku_brt_pivot_key_len(t, node->u.n.childkeys[n_children_in_a-1]);
node->u.n.childkeys[n_children_in_a-1]=0;
verify_local_fingerprint_nonleaf(node);
verify_local_fingerprint_nonleaf(B);
}
*nodeb = B;
assert(toku_serialize_brtnode_size(node)<node->nodesize);
assert(toku_serialize_brtnode_size(B)<B->nodesize);
return 0;
}
static int nonleaf_node_is_too_wide (BRTNODE node) {
assert(node->height>0);
return node->u.n.n_children > TREE_FANOUT;
}
static int maybe_fixup_fat_child(BRT brt, BRTNODE node, int childnum, BRTNODE child, TOKULOGGER logger) // If the node is too big then deal with it. Unpin the child (or children if it splits) NODE may be too big at the end
{
int r = push_down_if_buffers_too_full(brt, child, logger);
if (r!=0) return r;
// now the child may have too much fanout.
if (child->height>0) {
if (nonleaf_node_is_too_wide(child)) {
BRTNODE newchild;
DBT splitk;
if ((r=brt_nonleaf_split(brt, child, &newchild, &splitk, logger))) return r;
int old_n_children = node->u.n.n_children;
FIFO old_fifo = BNC_BUFFER(node, childnum);
// slide the children over
REALLOC_N(old_n_children+1, node->u.n.childinfos);
memmove(&node->u.n.childinfos[childnum+1], &node->u.n.childinfos[childnum+2], (old_n_children-childnum-1)*sizeof(node->u.n.childinfos[0]));
// fill in the new children
{
struct brtnode_nonleaf_childinfo *ci = &node->u.n.childinfos[childnum+1];
ci->subtree_fingerprint = 0;
ci->diskoff = newchild->thisnodename;
ci->n_bytes_in_buffer = 0;
r=toku_fifo_create(&ci->buffer);
if (r!=0) return r;
}
// replace the fifo in the old child
r=toku_fifo_create(&BNC_BUFFER(node, childnum));
if (r!=0) return r;
// slide the keys over
REALLOC_N(old_n_children, node->u.n.childkeys);
memmove(&node->u.n.childkeys[childnum], &node->u.n.childkeys[childnum+1], (old_n_children-childnum-1)*sizeof(node->u.n.childkeys[0]));
{
struct kv_pair *pivot = splitk.data;
BYTESTRING bs = { .len = splitk.size,
.data = kv_pair_key(pivot) };
r = toku_log_setpivot(logger, toku_cachefile_filenum(brt->cf), node->thisnodename, childnum, bs);
if (r!=0) return r;
node->u.n.childkeys[childnum] = pivot;
node->u.n.totalchildkeylens += toku_brt_pivot_key_len(brt, pivot);
}
node->u.n.n_children++;
// fix up fingerprints
fixup_child_fingerprint(node, childnum, child, brt, logger);
fixup_child_fingerprint(node, childnum+1, newchild, brt, logger);
// now everything in the fifo must be put again into one of the two fifos
BRT_CMD_S cmd;
DBT key,val;
while (0==toku_fifo_peek_deq_cmdstruct(old_fifo, &cmd, &key, &val)) {
int cmp = brt_compare_pivot(brt, cmd.u.id.key, 0, node->u.n.childkeys[childnum]);
if (cmp<=0) {
r=toku_fifo_enq_cmdstruct(BNC_BUFFER(node, childnum), &cmd);
if (r!=0) return r;
}
if (cmp==0 && cmd.type==BRT_DELETE && brt->flags&TOKU_DB_DUPSORT) {
r=toku_fifo_enq_cmdstruct(BNC_BUFFER(node, childnum+1), &cmd);
if (r!=0) return r;
}
}
toku_fifo_free(&old_fifo);
if (r!=0) return r;
}
} else {
abort(); // if a leaf is too fat need to split it.
}
return 0;
}
// There are two kinds of puts:
// A "weak" put that is guaranteed to trigger no I/O, and will not leaf the node overfull.
// A weak put may not actually perform the put, however (in which case it returns EAGAIN instead of 0)
// A "strong" put that is guaranteed to do the put. However, it may trigger I/O and the resulting node may be too big.
static int brt_leaf_put (BRT t, BRTNODE node, BRT_CMD cmd, TOKULOGGER logger, WS weak_p) {
FILENUM filenum = toku_cachefile_filenum(t->cf);
switch (cmd->type) {
case BRT_INSERT: {
int r = toku_pma_insert_or_replace_ws(node->u.l.buffer,
cmd->u.id.key, cmd->u.id.val,
logger, cmd->xid,
filenum, node->thisnodename, node->rand4fingerprint, &node->local_fingerprint,
&node->log_lsn, &node->u.l.n_bytes_in_buffer,
weak_p==WEAK);
if (r==EAGAIN) return EAGAIN;
assert(r==0);
node->dirty=1;
return r;
}
case BRT_DELETE: {
int r = toku_pma_delete_fixupsize(node->u.l.buffer, cmd->u.id.key, (DBT*)0,
logger, cmd->xid, node->thisnodename,
node->rand4fingerprint, &node->local_fingerprint, &node->log_lsn, &node->u.l.n_bytes_in_buffer);
if (r==0) node->dirty=1;
return r;
}
case BRT_DELETE_BOTH: {
int r = toku_pma_delete_fixupsize(node->u.l.buffer, cmd->u.id.key, cmd->u.id.val,
logger, cmd->xid, node->thisnodename,
node->rand4fingerprint, &node->local_fingerprint, &node->log_lsn, &node->u.l.n_bytes_in_buffer);
if (r == 0) node->dirty = 1;
return r;
}
case BRT_NONE: return 0;
}
return EINVAL; // if none of the cases match, then the command is messed up.
}
// Put an command in a particular child's fifo.
// If weak_p then do it without doing I/O or overfilling the child.
// If the child is in main memory and we can do a weak put on the child, then push into the child.
// Otherwise we return EAGAIN.
// If not weak_p then we are willing to overfill the child.
static int brt_nonleaf_put_cmd_to_child (BRT t, BRTNODE node, int childnum, BRT_CMD cmd, TOKULOGGER logger, WS weak_p) {
DBT *k = cmd->u.id.key;
DBT *v = cmd->u.id.val;
int r;
if (toku_fifo_n_entries(BNC_BUFFER(node,childnum))==0) {
void *child_v;
r = toku_cachetable_maybe_get_and_pin(t->cf, BNC_DISKOFF(node, childnum), &child_v);
if (r==0) {
BRTNODE child=child_v;
r = brtnode_put(t, child, cmd, logger, weak_p);
if (r==EAGAIN) {
r = unpin_brtnode(t, child);
if (r!=0) return r; // node is still OK
} else if (r==0) {
return maybe_fixup_fat_child(t, node, childnum, child, logger); // If the node is too big then deal with it. Unpin the child. NODE may be too big. I think the only way a node can get fat is if weak_p==STRONG.
} else {
unpin_brtnode(t, child);
return r; // node is still OK
}
}
}
// For some reason we didn't put it into the child, so we must put it in the fifo.
int diff = k->size + v->size + KEY_VALUE_OVERHEAD + BRT_CMD_OVERHEAD;
if (diff+toku_serialize_brtnode_size(node)>node->nodesize) return EAGAIN; // And it doesn't fit here.
r=toku_fifo_enq_cmdstruct(BNC_BUFFER(node,childnum), cmd);
if (r!=0) return r;
node->local_fingerprint += node->rand4fingerprint * toku_calccrc32_cmdstruct(cmd);
node->u.n.n_bytes_in_buffers += diff;
BNC_NBYTESINBUF(node, childnum) += diff;
node->dirty = 1;
return 0; // node may be too big
}
static void determine_which_children_to_push_delete (BRT t, BRTNODE node, BRT_CMD cmd, int *n_children_to_push, int *children_to_push) {
int i;
*n_children_to_push=0;
for (i=0; i<node->u.n.n_children-1; i++) {
int cmp = brt_compare_pivot(t, cmd->u.id.key, 0, node->u.n.childkeys[i]);
if (cmp>0) continue; // the cmd is bigger than the pivot, so it doesn't go here.
else if (cmp<0) {
// the cmd is smaller than the pivot, so it goes here, and goes nowhere else to the right
children_to_push[(*n_children_to_push)++] = i;
return;
} else if (t->flags & TOKU_DB_DUPSORT) {
// the cmd is equal and we are in a dupsort, so push and and go around to push additional ones.
children_to_push[(*n_children_to_push)++] = i;
continue;
} else {
// the cmd is equal but we are not in a dupsort, so we save i, but there is no saving the next one.
children_to_push[(*n_children_to_push)++] = i;
return;
}
}
// if we fell off the bottom, which means we must include the last one.
children_to_push[(*n_children_to_push)++] = i;
}
// Put the cmd into all the subtrees that it belong in. (Deletes can end up in several subtrees.)
// If weak_p then
// Don't do any I/O and the node will not be overfull.
// To guarantee that no I/O will occur, we must make sure we can insert everything before inserting anything.
// else put it regardless, possibly overflowing the node.
static int brt_nonleaf_put_delete (BRT t, BRTNODE node, BRT_CMD cmd, TOKULOGGER logger, WS weak_p) {
int singlediff = cmd->u.id.key->size + cmd->u.id.val->size + KEY_VALUE_OVERHEAD + BRT_CMD_OVERHEAD;
int n_children_to_push = 0;
int children_to_push[node->u.n.n_children];
determine_which_children_to_push_delete(t, node, cmd, &n_children_to_push, children_to_push);
int totaldiff = singlediff * n_children_to_push;
if (weak_p && (totaldiff + toku_serialize_brtnode_size(node) > node->nodesize)) return EAGAIN;
// Now we know it will fit, so do all the weak pushes. We are being a little bit conservative,
// since a soft push might succeed, in getting data to a child without using up the local storage.
int i;
for (i=0; i<n_children_to_push; i++) {
int r=brt_nonleaf_put_cmd_to_child(t, node, children_to_push[i], cmd, logger, WEAK);
if (r==EAGAIN) {
r = toku_fifo_enq_cmdstruct(BNC_BUFFER(node, children_to_push[i]), cmd);
if (r!=0) return r;
} else if (r!=0) return r;
}
// We did we weak pushes to the children, but if that didn't work we put it in the buffer. The node could be overfull now.
return 0;
}
// a DELETE could be replicating in a dupsort database. Everything else is non replicating.
static int brt_nonleaf_put_nonreplicating_cmd (BRT t, BRTNODE node, BRT_CMD cmd, TOKULOGGER logger, WS weak_p) {
return brt_nonleaf_put_cmd_to_child(t, node,
brtnode_which_child(node, cmd->u.id.key, cmd->u.id.val, t),
cmd, logger,
weak_p);
}
// Put the cmd into the node. Possibly results in the node being overfull. (But not if weak_p is set, in which case EAGAIN is returned instead)
// The command could get pushed into the appropriate child if the child is in main memory and has space to hold the command.
static int brt_nonleaf_put (BRT t, BRTNODE node, BRT_CMD cmd, TOKULOGGER logger, WS weak_p) {
if (cmd->type == BRT_INSERT || cmd->type == BRT_DELETE_BOTH) {
return brt_nonleaf_put_nonreplicating_cmd(t, node, cmd, logger, weak_p);
} else if (cmd->type == BRT_DELETE) {
return brt_nonleaf_put_delete(t, node, cmd, logger, weak_p);
} else
return EINVAL;
}
// Put the command into the node.
// If weak_p is set then neither the node nor any descendants will get too big, and no I/O will occur.
// if !weak_p then I/O could occur and the node could end up with too much fanout. (But the children will all be properly sized)
static int brtnode_put (BRT t, BRTNODE node, BRT_CMD cmd, TOKULOGGER logger, WS weak_p) {
if (node->height==0) {
return brt_leaf_put(t, node, cmd, logger, weak_p);
} else {
return brt_nonleaf_put(t, node, cmd, logger, weak_p);
}
}
static void verify_local_fingerprint_nonleaf (BRTNODE node) {
u_int32_t fp=0;
int i;
if (node->height==0) return;
for (i=0; i<node->u.n.n_children; i++)
FIFO_ITERATE(BNC_BUFFER(node,i), key, keylen, data, datalen, type, xid,
({
fp += node->rand4fingerprint * toku_calccrc32_cmd(type, xid, key, keylen, data, datalen);
}));
assert(fp==node->local_fingerprint);
}
#ifdef FOO
static int setup_initial_brt_root_node (BRT t, DISKOFF offset, TOKULOGGER logger) {
int r;
TAGMALLOC(BRTNODE, node);
assert(node);
initialize_brtnode(t, node,
offset, /* the location is one nodesize offset from 0. */
0);
// node->brt = t;
if (0) {
printf("%s:%d for tree %p node %p mdict_create--> %p\n", __FILE__, __LINE__, t, node, node->u.l.buffer);
printf("%s:%d put root at %lld\n", __FILE__, __LINE__, offset);
}
r=toku_cachetable_put(t->cf, offset, node, brtnode_size(node),
toku_brtnode_flush_callback, toku_brtnode_fetch_callback, t);
if (r!=0) {
toku_free(node);
return r;
}
toku_verify_counts(node);
toku_log_newbrtnode(logger, toku_cachefile_filenum(t->cf), offset, 0, t->h->nodesize, (t->flags&TOKU_DB_DUPSORT)!=0, node->rand4fingerprint);
toku_update_brtnode_loggerlsn(node, logger);
r=unpin_brtnode(t, node);
if (r!=0) {
toku_free(node);
return r;
}
return 0;
}
int toku_brt_create(BRT *brt_ptr) {
BRT brt = toku_malloc(sizeof *brt);
if (brt == 0)
return ENOMEM;
memset(brt, 0, sizeof *brt);
list_init(&brt->cursors);
brt->flags = 0;
brt->nodesize = BRT_DEFAULT_NODE_SIZE;
brt->compare_fun = toku_default_compare_fun;
brt->dup_compare = toku_default_compare_fun;
*brt_ptr = brt;
return 0;
}
int toku_brt_set_flags(BRT brt, unsigned int flags) {
brt->flags = flags;
return 0;
}
int toku_brt_get_flags(BRT brt, unsigned int *flags) {
*flags = brt->flags;
return 0;
}
int toku_brt_set_nodesize(BRT brt, unsigned int nodesize) {
brt->nodesize = nodesize;
return 0;
}
int toku_brt_get_nodesize(BRT brt, unsigned int *nodesize) {
*nodesize = brt->nodesize;
return 0;
}
int toku_brt_set_bt_compare(BRT brt, int (*bt_compare)(DB *, const DBT*, const DBT*)) {
brt->compare_fun = bt_compare;
return 0;
}
int toku_brt_set_dup_compare(BRT brt, int (*dup_compare)(DB *, const DBT*, const DBT*)) {
brt->dup_compare = dup_compare;
return 0;
}
int toku_brt_get_fd(BRT brt, int *fdp) {
*fdp = toku_cachefile_fd(brt->cf);
return 0;
}
enum { UNDO_COUNTER_LIMIT=10 };
typedef void(*undo_fun)(void*);
struct undo_rec { undo_fun f; void *v; };
struct undo {
int undo_counter;
struct undo_rec undos[UNDO_COUNTER_LIMIT];
};
#define INITUNDO(u) struct undo u = (struct undo){.undo_counter=0}
void push_undo(struct undo *undos, undo_fun f, void *v) {
assert(undos->undo_counter<UNDO_COUNTER_LIMIT);
undos->undos[undos->undo_counter++]=(struct undo_rec){f,v};
}
void do_undos(struct undo *undos) {
while (undos->undo_counter>0) {
struct undo_rec *r = &undos->undos[--undos->undo_counter];
r->f(r->v);
}
}
void undo_free (void *v) {
void **ptr=v;
toku_free(*ptr);
*ptr=0;
}
// tbou means "toku_brt_open undo"
void tbou_close_cachefile (void *v) {
BRT t = v;
toku_cachefile_close(&t->cf);
}
struct maybe_unpin_info {
int is_pinned;
CACHEFILE cf;
CACHEKEY ckey;
};
void tbou_maybe_unpin (void *v) {
struct maybe_unpin_info *mui = v;
if (mui->is_pinned)
toku_cachetable_unpin(mui->cf, mui->ckey, 0, 0);
mui->is_pinned=0;
}
int toku_brt_open(BRT t, const char *fname, const char *fname_in_env, const char *dbname, int is_create, int only_create, int load_flags, CACHETABLE cachetable, TOKUTXN txn, DB *db) {
/* If dbname is NULL then we setup to hold a single tree. Otherwise we setup an array. */
int r;
struct maybe_unpin_info mui = {.is_pinned=0};
INITUNDO(undos);
push_undo(&undos, tbou_maybe_unpin, &mui); // if we pin a cf, then we put it into the maybe_undo_info so it will get undone on error.
assert(is_create || !only_create);
assert(!load_flags || !only_create);
if (0) {
died:
do_undos(&undos);
return r;
}
{
if (dbname) {
char *malloced_name = toku_strdup(dbname);
if (malloced_name==0) { r = errno; goto died; }
push_undo(&undos, undo_free, &t->database_name);
t->database_name = malloced_name;
} else {
t->database_name = 0;
}
}
t->db = db;
{
int fd = open(fname, O_RDWR, 0777);
r = errno;
if (fd==-1) {
if (r==ENOENT) {
if (!is_create) { goto died; }
fd = open(fname, O_RDWR | O_CREAT, 0777);
if (fd==-1) { r=errno; goto died; }
r = toku_logger_log_fcreate(txn, fname_in_env, 0777);
if (r!=0) goto died;
} else
goto died;
}
if ((r = toku_cachetable_openfd(&t->cf, cachetable, fd, t))) goto died;
push_undo(&undos, tbou_close_cachefile, t);
}
if ((r = toku_logger_log_fopen(txn, fname_in_env, toku_cachefile_filenum(t->cf)))) goto died;
// no undo action for log_fopen
assert(t->nodesize>0);
if (is_create) {
r = toku_read_and_pin_brt_header(t->cf, &t->h);
if (r!=0 && r!=-1) goto died;
if (r==0) {
mui=(struct maybe_unpin_info){.is_pinned=1, .cf=t->cf, .ckey=0}; // remember to unpin it
int i;
assert(r==0);
assert(dbname);
if (t->h->unnamed_root!=-1) { r=EINVAL; goto died; } // Cannot create a subdb in a file that is not enabled for subdbs
assert(t->h->n_named_roots>=0);
for (i=0; i<t->h->n_named_roots; i++) {
if (strcmp(t->h->names[i], dbname)==0) {
if (only_create) {
r = EEXIST;
goto died;
}
else goto found_it;
}
}
if ((t->h->names = toku_realloc(t->h->names, (1+t->h->n_named_roots)*sizeof(*t->h->names))) == 0) { r=errno; goto died; }
if ((t->h->roots = toku_realloc(t->h->roots, (1+t->h->n_named_roots)*sizeof(*t->h->roots))) == 0) { r=errno; goto died; }
t->h->n_named_roots++;
if ((t->h->names[t->h->n_named_roots-1] = toku_strdup(dbname)) == 0) { r=errno; goto died; }
push_undo(&undos, undo_free, &t->h->names[t->h->n_named_roots-1]);
r = malloc_diskblock_header_is_in_memory(&t->h->roots[t->h->n_named_roots-1], t, t->h->nodesize, toku_txn_logger(txn));
if (r!=0) goto died;
t->h->dirty = 1;
if ((r=setup_initial_brt_root_node(t, t->h->roots[t->h->n_named_roots-1], toku_txn_logger(txn)))!=0) goto died;
} else {
assert(r==-1); // the pin failed because no data was present
/* construct a new header. */
if ((MALLOC(t->h))==0) { r = errno; goto died; }
t->h->dirty=1;
t->h->flags = t->flags;
t->h->nodesize=t->nodesize;
t->h->freelist=-1;
t->h->unused_memory=2*t->nodesize;
if (dbname) {
t->h->unnamed_root = -1;
t->h->n_named_roots = 1;
if ((MALLOC_N(1, t->h->names))==0) { r=errno; goto died; } push_undo(&undos, undo_free, &t->h->names);
if ((MALLOC_N(1, t->h->roots))==0) { r=errno; goto died; } push_undo(&undos, undo_free, &t->h->roots);
if ((t->h->names[0] = toku_strdup(dbname))==0) { r=errno; goto died; } push_undo(&undos, undo_free, &t->h->names[0]);
t->h->roots[0] = t->nodesize;
} else {
t->h->unnamed_root = t->nodesize;
t->h->n_named_roots = -1;
t->h->names=0;
t->h->roots=0;
}
if ((r=toku_logger_log_header(txn, toku_cachefile_filenum(t->cf), t->h))) goto died;
if ((r=setup_initial_brt_root_node(t, t->nodesize, toku_txn_logger(txn)))!=0) goto died;
if ((r=toku_cachetable_put(t->cf, 0, t->h, 0, toku_brtheader_flush_callback, toku_brtheader_fetch_callback, 0))) goto died;
mui=(struct maybe_unpin_info){.is_pinned=1, .cf=t->cf, .ckey=0}; // remember to unpin it
}
} else {
if ((r = toku_read_and_pin_brt_header(t->cf, &t->h))!=0) goto died;
mui=(struct maybe_unpin_info){.is_pinned=1, .cf=t->cf, .ckey=0}; // remember to unpin it
if (!dbname) {
if (t->h->n_named_roots!=-1) { r = EINVAL; goto died; } // requires a subdb
} else {
int i;
if (t->h->n_named_roots==-1) { r=EINVAL; goto died; } // no suddbs in the db
// printf("%s:%d n_roots=%d\n", __FILE__, __LINE__, t->h->n_named_roots);
for (i=0; i<t->h->n_named_roots; i++) {
if (strcmp(t->h->names[i], dbname)==0) {
goto found_it;
}
}
r=ENOENT; /* the database doesn't exist */
goto died;
}
found_it:
t->nodesize = t->h->nodesize; /* inherit the pagesize from the file */
if (t->flags != t->h->flags) { /* flags must match */
if (load_flags) t->flags = t->h->flags;
else { r = EINVAL; goto died; }
}
}
assert(t->h);
if ((r = toku_unpin_brt_header(t)) !=0) goto died; // it's unpinned
mui.is_pinned=0;
assert(t->h==0);
return 0;
}
int toku_brt_remove_subdb(BRT brt, const char *dbname, u_int32_t flags) {
int r;
int i;
int found = -1;
assert(flags == 0);
r = toku_read_and_pin_brt_header(brt->cf, &brt->h);
if (r!=0) return r;
assert(brt->h->unnamed_root==-1);
assert(brt->h->n_named_roots>=0);
for (i = 0; i < brt->h->n_named_roots; i++) {
if (strcmp(brt->h->names[i], dbname) == 0) {
found = i;
break;
}
}
if (found == -1) {
//Should not be possible.
r = ENOENT;
goto error;
}
//Free old db name
toku_free(brt->h->names[found]);
//TODO: Free Diskblocks including root
for (i = found + 1; i < brt->h->n_named_roots; i++) {
brt->h->names[i - 1] = brt->h->names[i];
brt->h->roots[i - 1] = brt->h->roots[i];
}
brt->h->n_named_roots--;
brt->h->dirty = 1;
//TODO: What if n_named_roots becomes 0? Should we handle it specially? Should we delete the file?
if ((brt->h->names = toku_realloc(brt->h->names, (brt->h->n_named_roots)*sizeof(*brt->h->names))) == 0) { r=errno; goto error; }
if ((brt->h->roots = toku_realloc(brt->h->roots, (brt->h->n_named_roots)*sizeof(*brt->h->roots))) == 0) { r=errno; goto error; }
r = toku_unpin_brt_header(brt);
return r;
error:
toku_unpin_brt_header(brt);
return r;
}
// This one has no env
int toku_open_brt (const char *fname, const char *dbname, int is_create, BRT *newbrt, int nodesize, CACHETABLE cachetable, TOKUTXN txn,
int (*compare_fun)(DB*,const DBT*,const DBT*), DB *db) {
BRT brt;
int r;
const int only_create = 0;
const int load_flags = 0;
r = toku_brt_create(&brt);
if (r != 0) return r;
toku_brt_set_nodesize(brt, nodesize);
toku_brt_set_bt_compare(brt, compare_fun);
r = toku_brt_open(brt, fname, fname, dbname, is_create, only_create, load_flags, cachetable, txn, db);
if (r != 0) {
toku_free(brt);
return r;
}
*newbrt = brt;
return 0;
}
int toku_close_brt (BRT brt) {
int r;
while (!list_empty(&brt->cursors)) {
BRT_CURSOR c = list_struct(list_pop(&brt->cursors), struct brt_cursor, cursors_link);
r=toku_brt_cursor_close(c);
if (r!=0) return r;
}
if (brt->cf) {
assert(0==toku_cachefile_count_pinned(brt->cf, 1)); // For the brt, the pinned count should be zero.
//printf("%s:%d closing cachetable\n", __FILE__, __LINE__);
if ((r = toku_cachefile_close(&brt->cf))!=0) return r;
}
if (brt->database_name) toku_free(brt->database_name);
if (brt->skey) { toku_free(brt->skey); }
if (brt->sval) { toku_free(brt->sval); }
toku_free(brt);
return 0;
}
CACHEKEY* toku_calculate_root_offset_pointer (BRT brt) {
if (brt->database_name==0) {
return &brt->h->unnamed_root;
} else {
int i;
for (i=0; i<brt->h->n_named_roots; i++) {
if (strcmp(brt->database_name, brt->h->names[i])==0) {
return &brt->h->roots[i];
}
}
}
abort();
}
static int brt_init_new_root(BRT brt, int n_new_nodes, BRTNODE *new_nodes, DBT *splitks, CACHEKEY *rootp, TOKULOGGER logger, BRTNODE *newrootp) {
assert(n_new_nodes>0);
TAGMALLOC(BRTNODE, newroot);
int r;
int new_height = new_nodes[0]->height+1;
int new_nodesize = brt->h->nodesize;
DISKOFF newroot_diskoff;
if ((r=malloc_diskblock(&newroot_diskoff, brt, new_nodesize, logger))) return r;
assert(newroot);
if (brt->database_name==0) {
toku_log_changeunnamedroot(logger, toku_cachefile_filenum(brt->cf), *rootp, newroot_diskoff);
} else {
BYTESTRING bs;
bs.len = 1+strlen(brt->database_name);
bs.data = brt->database_name;
toku_log_changenamedroot(logger, toku_cachefile_filenum(brt->cf), bs, *rootp, newroot_diskoff);
}
*rootp=newroot_diskoff;
brt->h->dirty=1;
initialize_brtnode (brt, newroot, newroot_diskoff, new_height);
newroot->u.n.n_children=n_new_nodes;
r=toku_log_newbrtnode(logger, toku_cachefile_filenum(brt->cf), newroot_diskoff, new_height, new_nodesize, (brt->flags&TOKU_DB_DUPSORT)!=0, newroot->rand4fingerprint);
if (r!=0) return r;
int i;
for (i=0; i<n_new_nodes; i++) {
BNC_DISKOFF(newroot, i)=new_nodes[i]->thisnodename;
r=toku_fifo_create(&BNC_BUFFER(newroot,i)); if (r!=0) return r;
r=toku_log_addchild(logger, toku_cachefile_filenum(brt->cf), newroot_diskoff, 0, new_nodes[i]->thisnodename, 0);
if (r!=0) return r;
fixup_child_fingerprint(newroot, i, new_nodes[i], brt, logger);
}
toku_verify_counts(newroot);
int sum_splitk_sizes=0;
for (i=0; i+1<n_new_nodes; i++) {
sum_splitk_sizes += splitks[i].size;
newroot->u.n.childkeys[i] = splitks[i].data;
BYTESTRING bs = { .len = kv_pair_keylen(newroot->u.n.childkeys[0]),
.data = kv_pair_key(newroot->u.n.childkeys[0]) };
r=toku_log_setpivot(logger, toku_cachefile_filenum(brt->cf), newroot_diskoff, 0, bs);
if (r!=0) return r;
toku_update_brtnode_loggerlsn(newroot, logger);
}
newroot->u.n.totalchildkeylens=sum_splitk_sizes;
for (i=0; i<n_new_nodes; i++) {
r=unpin_brtnode(brt, new_nodes[i]);
if (r!=0) return r;
}
toku_cachetable_put(brt->cf, newroot_diskoff, newroot, brtnode_size(newroot),
toku_brtnode_flush_callback, toku_brtnode_fetch_callback, brt);
*newrootp = newroot;
return 0;
}
static int split_nonleaf_node(BRT, int *n_new_nodes, BRTNODE **new_nodes, DBT **splitks);
static int leaf_node_is_too_full (BRT, BRTNODE);
// push things down into node's children (and into their children and so forth) but don't make any descendant too big.
static int push_down_without_overfilling (BRT brt, BRTNODE node, TOKULOGGER logger);
// Push data toward a child. If the child gets too big then the child will push down or split.
// If a split happens, then return immediately so that we can check to see if NODE needs to be split
static int flush_toward_child (BRT brt, BRTNODE node, int childnum, TOKULOGGER logger);
static int maybe_fixup_root (BRT brt, BRTNODE node, CACHEKEY *rootp, TOKULOGGER logger) {
int r;
if (node->height>0) {
// internal nodes can be too wide, but if too full, they did a push down
maybe_reshape_internal_node:
while (nonleaf_node_is_too_wide(brt, node)) {
int n_new_nodes; BRTNODE *new_nodes; DBT *splitks;
if ((r=split_nonleaf_node(brt, node, &n_new_nodes, &new_nodes, &splitks))) return r;
if ((r=brt_init_new_root(brt, n_new_nodes, new_nodes, splitks, rootp, logger, &node))) return r; // unpins all the new nodes, which are all small enough
// now node is still possibly too wide, hence the loop
}
} else {
// leaf nodes can be too full
if (leaf_node_is_too_full(brt, node)) {
int n_new_nodes; BRTNODE *new_nodes; DBT *splitks;
if ((r==split_leaf_node(brt, logger, node, &n_new_nodes, &new_nodes, &splitks))) return r;
if ((r==brt_init_new_root(brt, n_new_nodes, new_nodes, splitks, rootp, logger, &node))) return r; // unpins all the new nodes, which are all small enough
assert(node->height>0);
goto maybe_reshape_internal_node;
}
}
return 0;
}
#endif
static int brt_root_put_cmd(BRT brt, BRT_CMD cmd, TOKULOGGER logger) {
void *node_v;
BRTNODE node;
CACHEKEY *rootp;
int r;
//assert(0==toku_cachetable_assert_all_unpinned(brt->cachetable));
if ((r = toku_read_and_pin_brt_header(brt->cf, &brt->h))) {
if (0) { died0: toku_unpin_brt_header(brt); }
return r;
}
rootp = toku_calculate_root_offset_pointer(brt);
if ((r=toku_cachetable_get_and_pin(brt->cf, *rootp, &node_v, NULL,
toku_brtnode_flush_callback, toku_brtnode_fetch_callback, brt))) {
if (0) { died1: unpin_brtnode(brt, node); goto died0; }
goto died0;
}
//printf("%s:%d pin %p\n", __FILE__, __LINE__, node_v);
node=node_v;
if ((r = brtnode_put(brt, node, cmd, logger, STRONG))) goto died1; // put stuff in, possibly causing the buffers to get too big
if ((r = push_down_if_buffers_too_full(brt, node, logger))) goto died1; // if the buffers are too big, push stuff down
if ((r = maybe_split_root(brt, node, rootp, logger))) goto died1; // now the node might have to split (leaf nodes can't push down, and internal nodes have too much fanout) This will change node.
// Now the node is OK,
brt->h->dirty=1;
return toku_unpin_brt_header(brt);
}
int toku_brt_insert (BRT brt, DBT *key, DBT *val, TOKUTXN txn) {
int r;
BRT_CMD_S brtcmd = { BRT_INSERT, toku_txn_get_txnid(txn), .u.id={key,val}};
r = brt_root_put_cmd(brt, &brtcmd, toku_txn_logger(txn));
return r;
}
int toku_brt_lookup (BRT brt, DBT *k, DBT *v) {
int r, rr;
BRT_CURSOR cursor;
rr = toku_brt_cursor(brt, &cursor);
if (rr != 0) return rr;
int op = brt->flags & TOKU_DB_DUPSORT ? DB_GET_BOTH : DB_SET;
r = toku_brt_cursor_get(cursor, k, v, op, 0);
rr = toku_brt_cursor_close(cursor); assert(rr == 0);
return r;
}
int toku_brt_delete(BRT brt, DBT *key, TOKUTXN txn) {
int r;
DBT val;
BRT_CMD_S brtcmd = { BRT_DELETE, toku_txn_get_txnid(txn), .u.id={key, toku_init_dbt(&val)}};
r = brt_root_put_cmd(brt, &brtcmd, toku_txn_logger(txn));
return r;
}
int toku_brt_delete_both(BRT brt, DBT *key, DBT *val, TOKUTXN txn) {
int r;
BRT_CMD_S brtcmd = { BRT_DELETE_BOTH, toku_txn_get_txnid(txn), .u.id={key,val}};
r = brt_root_put_cmd(brt, &brtcmd, toku_txn_logger(txn));
return r;
}
int toku_verify_brtnode (BRT brt, DISKOFF off, bytevec lorange, ITEMLEN lolen, bytevec hirange, ITEMLEN hilen, int recurse, BRTNODE parent_brtnode);
int toku_dump_brtnode (BRT brt, DISKOFF off, int depth, bytevec lorange, ITEMLEN lolen, bytevec hirange, ITEMLEN hilen, BRTNODE parent_brtnode) {
int result=0;
BRTNODE node;
void *node_v;
int r = toku_cachetable_get_and_pin(brt->cf, off, &node_v, NULL,
toku_brtnode_flush_callback, toku_brtnode_fetch_callback, brt);
assert(r==0);
printf("%s:%d pin %p\n", __FILE__, __LINE__, node_v);
node=node_v;
result=toku_verify_brtnode(brt, off, lorange, lolen, hirange, hilen, 0, parent_brtnode);
printf("%*sNode=%p\n", depth, "", node);
if (node->height>0) {
printf("%*sNode %lld nodesize=%d height=%d n_children=%d n_bytes_in_buffers=%d keyrange=%s %s\n",
depth, "", off, node->nodesize, node->height, node->u.n.n_children, node->u.n.n_bytes_in_buffers, (char*)lorange, (char*)hirange);
//printf("%s %s\n", lorange ? lorange : "NULL", hirange ? hirange : "NULL");
{
int i;
for (i=0; i< node->u.n.n_children; i++) {
printf("%*schild %d buffered (%d entries):\n", depth+1, "", i, toku_fifo_n_entries(BNC_BUFFER(node,i)));
FIFO_ITERATE(BNC_BUFFER(node,i), key, keylen, data, datalen, type, xid,
({
data=data; datalen=datalen; keylen=keylen;
printf("%*s xid=%"PRId64" %d (type=%d)\n", depth+2, "", xid, ntohl(*(int*)key), type);
//assert(strlen((char*)key)+1==keylen);
//assert(strlen((char*)data)+1==datalen);
}));
}
for (i=0; i<node->u.n.n_children; i++) {
printf("%*schild %d\n", depth, "", i);
if (i>0) {
printf("%*spivot %d len=%d %d\n", depth+1, "", i-1, node->u.n.childkeys[i-1]->keylen, ntohl(*(int*)&node->u.n.childkeys[i-1]->key));
}
toku_dump_brtnode(brt, BNC_DISKOFF(node, i), depth+4,
(i==0) ? lorange : node->u.n.childkeys[i-1],
(i==0) ? lolen : toku_brt_pivot_key_len(brt, node->u.n.childkeys[i-1]),
(i==node->u.n.n_children-1) ? hirange : node->u.n.childkeys[i],
(i==node->u.n.n_children-1) ? hilen : toku_brt_pivot_key_len(brt, node->u.n.childkeys[i]),
node
);
}
}
} else {
printf("%*sNode %lld nodesize=%d height=%d n_bytes_in_buffer=%d keyrange=%d %d\n",
depth, "", off, node->nodesize, node->height, node->u.l.n_bytes_in_buffer, lorange ? ntohl(*(int*)lorange) : 0, hirange ? ntohl(*(int*)hirange) : 0);
PMA_ITERATE(node->u.l.buffer, key, keylen, val __attribute__((__unused__)), vallen,
( keylen=keylen, vallen=vallen, printf(" (%d)%d ", keylen, ntohl(*(int*)key))));
printf("\n");
}
r = toku_cachetable_unpin(brt->cf, off, 0, 0);
assert(r==0);
return result;
}
int toku_dump_brt (BRT brt) {
int r;
CACHEKEY *rootp;
struct brt_header *prev_header = brt->h;
if ((r = toku_read_and_pin_brt_header(brt->cf, &brt->h))) {
if (0) { died0: toku_unpin_brt_header(brt); }
return r;
}
rootp = toku_calculate_root_offset_pointer(brt);
if ((r = toku_dump_brtnode(brt, *rootp, 0, 0, 0, 0, 0, null_brtnode))) goto died0;
if ((r = toku_unpin_brt_header(brt))!=0) return r;
brt->h = prev_header;
return 0;
}
static int show_brtnode_blocknumbers (BRT brt, DISKOFF off) {
BRTNODE node;
void *node_v;
int i,r;
assert(off%brt->h->nodesize==0);
if ((r = toku_cachetable_get_and_pin(brt->cf, off, &node_v, NULL,
toku_brtnode_flush_callback, toku_brtnode_fetch_callback, brt))) {
if (0) { died0: toku_cachetable_unpin(brt->cf, off, 0, 0); }
return r;
}
printf("%s:%d pin %p\n", __FILE__, __LINE__, node_v);
node=node_v;
printf(" %lld", off/brt->h->nodesize);
if (node->height>0) {
for (i=0; i<node->u.n.n_children; i++) {
if ((r=show_brtnode_blocknumbers(brt, BNC_DISKOFF(node, i)))) goto died0;
}
}
r = toku_cachetable_unpin(brt->cf, off, 0, 0);
return r;
}
#if 0
int show_brt_blocknumbers (BRT brt) {
int r;
CACHEKEY *rootp;
if ((r = toku_read_and_pin_brt_header(brt->cf, &brt->h))) {
if (0) { died0: toku_unpin_brt_header(brt); }
return r;
}
rootp = toku_calculate_root_offset_pointer(brt);
printf("BRT %p has blocks:", brt);
if ((r=show_brtnode_blocknumbers (brt, *rootp, 0))) goto died0;
printf("\n");
if ((r = toku_unpin_brt_header(brt))!=0) return r;
return 0;
}
#endif
int toku_brt_dbt_set_key(BRT brt, DBT *ybt, bytevec val, ITEMLEN vallen) {
int r = toku_dbt_set_value(ybt, val, vallen, &brt->skey);
return r;
}
int toku_brt_dbt_set_value(BRT brt, DBT *ybt, bytevec val, ITEMLEN vallen) {
int r = toku_dbt_set_value(ybt, val, vallen, &brt->sval);
return r;
}
#ifdef FOO
/* search in a node's child */
static int brt_search_child(BRT brt, BRTNODE node, int childnum, brt_search_t *search, DBT *newkey, DBT *newval, TOKULOGGER logger) {
int r, rr;
/* if the child's buffer is not empty then try to empty it */
if (BNC_NBYTESINBUF(node, childnum) > 0) {
rr = maybe_push_some_brt_cmds_down(brt, node, childnum, logger);
if (rr!=0) return rr;
/* push down may cause a child split, so childnum may not be appropriate, and the node itself may split, so retry */
return EAGAIN;
}
void *node_v;
rr = toku_cachetable_get_and_pin(brt->cf, BNC_DISKOFF(node,childnum), &node_v, NULL, toku_brtnode_flush_callback, toku_brtnode_fetch_callback, brt);
assert(rr == 0);
for (;;) {
BRTNODE childnode = node_v;
BRT_SPLIT childsplit; brt_split_init(&childsplit);
r = brt_search_node(brt, childnode, search, newkey, newval, &childsplit, logger);
if (childsplit.did_split) {
rr = handle_split_of_child(brt, node, childnum, childsplit.nodea, childsplit.nodeb, &childsplit.splitk,
&split->did_split, &split->nodea, &split->nodeb, &split->splitk, logger);
assert(rr == 0);
break;
} else {
if (r == EAGAIN)
continue;
rr = toku_cachetable_unpin(brt->cf, childnode->thisnodename, childnode->dirty, brtnode_size(childnode));
assert(rr == 0);
break;
}
}
return r;
}
static int brt_search_nonleaf_node(BRT brt, BRTNODE node, brt_search_t *search, DBT *newkey, DBT *newval, TOKULOGGER logger) {
int c;
restart:
{
/* binary search is overkill for a small array */
int child[node->u.n.n_children];
/* scan left to right or right to left depending on the search direction */
for (c = 0; c < node->u.n.n_children; c++)
child[c] = search->direction & BRT_SEARCH_LEFT ? c : node->u.n.n_children - 1 - c;
for (c = 0; c < node->u.n.n_children-1; c++) {
int p = search->direction & BRT_SEARCH_LEFT ? child[c] : child[c] - 1;
struct kv_pair *pivot = node->u.n.childkeys[p];
DBT pivotkey, pivotval;
if (search->compare(search,
toku_fill_dbt(&pivotkey, kv_pair_key(pivot), kv_pair_keylen(pivot)),
brt->flags & TOKU_DB_DUPSORT ? toku_fill_dbt(&pivotval, kv_pair_val(pivot), kv_pair_vallen(pivot)): 0)) {
// We know which child we want to search. First make sure the buffer is empty.
r = flush_toward_child(brt, node, child[c], logger, &did_split);
if (did_split) goto restart;
// If we didn't split, then the buffer is empty, so search that child
r=search_that_child();
// Now that child may be bent out of shape
???
int r = brt_search_child(brt, node, child[c], search, newkey, newval, logger);
// searching the child can cause it to get bent out of shape
int rr = maybe_fixup_nonroot(brt, node, child[c], logger);
if (rr!=0) return rr;
if (r == 0) return r;
}
}
/* check the first (left) or last (right) node if nothing has been found */
if (r == DB_NOTFOUND && c == node->u.n.n_children-1)
r = brt_search_child(brt, node, child[c], search, newkey, newval, split, logger);
return r;
}
static int brt_search_leaf_node(BRTNODE node, brt_search_t *search, DBT *newkey, DBT *newval) {
PMA pma = node->u.l.buffer;
int r = toku_pma_search(pma, search, newkey, newval);
return r;
}
static int brt_search_node(BRT brt, BRTNODE node, brt_search_t *search, DBT *newkey, DBT *newval, TOKULOGGER logger) {
if (node->height > 0)
return brt_search_nonleaf_node(brt, node, search, newkey, newval, logger);
else
return brt_search_leaf_node(node, search, newkey, newval);
}
int toku_brt_search(BRT brt, brt_search_t *search, DBT *newkey, DBT *newval, TOKULOGGER logger) {
int r, rr;
rr = toku_read_and_pin_brt_header(brt->cf, &brt->h);
if (rr!=0) {
if (0) { died0: toku_unpin_brt_header(brt); }
return rr;
}
CACHEKEY *rootp = toku_calculate_root_offset_pointer(brt);
void *node_v;
BRTNODE node;
rr = toku_cachetable_get_and_pin(brt->cf, *rootp, &node_v, NULL, toku_brtnode_flush_callback, toku_brtnode_fetch_callback, brt);
if (rr!=0) {
if (0) { died1: unpin_brtnode(brt, node); }
goto died0;
}
node = node_v;
r = brt_search_node(brt, node, search, newkey, newval, logger);
rr = maybe_fixup_root(brt, node, rootp, logger);
if (rr!=0) { goto died1; }
rr = unpin_brtnode(brt, node);
if (rr!=0) { goto died0; }
rr = toku_unpin_brt_header(brt);
if (rr!=0) return rr;
return r;
}
static inline void dbt_cleanup(DBT *dbt) {
if (dbt->data && (dbt->flags & DB_DBT_MALLOC)) {
toku_free_n(dbt->data, dbt->size); dbt->data = 0;
}
}
static inline void brt_cursor_cleanup(BRT_CURSOR cursor) {
dbt_cleanup(&cursor->key);
dbt_cleanup(&cursor->val);
}
static inline int brt_cursor_not_set(BRT_CURSOR cursor) {
return cursor->key.data == 0 || cursor->val.data == 0;
}
BOOL toku_brt_cursor_uninitialized(BRT_CURSOR c) {
return brt_cursor_not_set(c);
}
static inline void brt_cursor_set_key_val(BRT_CURSOR cursor, DBT *newkey, DBT *newval) {
brt_cursor_cleanup(cursor);
cursor->key = *newkey; memset(newkey, 0, sizeof *newkey);
cursor->val = *newval; memset(newval, 0, sizeof *newval);
}
int toku_brt_cursor(BRT brt, BRT_CURSOR *cursorptr) {
BRT_CURSOR cursor = toku_malloc(sizeof *cursor);
if (cursor == 0)
return ENOMEM;
cursor->brt = brt;
toku_init_dbt(&cursor->key);
toku_init_dbt(&cursor->val);
list_push(&brt->cursors, &cursor->cursors_link);
*cursorptr = cursor;
return 0;
}
int toku_brt_cursor_close(BRT_CURSOR cursor) {
brt_cursor_cleanup(cursor);
list_remove(&cursor->cursors_link);
toku_free_n(cursor, sizeof *cursor);
return 0;
}
static inline int compare_k_x(BRT brt, DBT *k, DBT *x) {
return brt->compare_fun(brt->db, k, x);
}
static inline int compare_v_y(BRT brt, DBT *v, DBT *y) {
return brt->dup_compare(brt->db, v, y);
}
static inline int compare_kv_xy(BRT brt, DBT *k, DBT *v, DBT *x, DBT *y) {
int cmp = brt->compare_fun(brt->db, k, x);
if (cmp == 0 && v && y)
cmp = brt->dup_compare(brt->db, v, y);
return cmp;
}
static inline int brt_cursor_copyout(BRT_CURSOR cursor, DBT *key, DBT *val) {
int r = 0;
if (key)
r = toku_dbt_set_value(key, cursor->key.data, cursor->key.size, &cursor->brt->skey);
if (r == 0 && val)
r = toku_dbt_set_value(val, cursor->val.data, cursor->val.size, &cursor->brt->sval);
return r;
}
static int brt_cursor_compare_set(brt_search_t *search, DBT *x, DBT *y) {
BRT brt = search->context;
return compare_kv_xy(brt, search->k, search->v, x, y) <= 0; /* return min xy: kv <= xy */
}
static int brt_cursor_current(BRT_CURSOR cursor, int op, DBT *outkey, DBT *outval, TOKULOGGER logger) {
if (brt_cursor_not_set(cursor))
return EINVAL;
if (op == DB_CURRENT) {
DBT newkey; toku_init_dbt(&newkey);
DBT newval; toku_init_dbt(&newval);
brt_search_t search; brt_search_init(&search, brt_cursor_compare_set, BRT_SEARCH_LEFT, &cursor->key, &cursor->val, cursor->brt);
int r = toku_brt_search(cursor->brt, &search, &newkey, &newval, logger);
if (r != 0 || compare_kv_xy(cursor->brt, &cursor->key, &cursor->val, &newkey, &newval) != 0)
return DB_KEYEMPTY;
}
return brt_cursor_copyout(cursor, outkey, outval);
}
/* search for the first kv pair that matches the search object */
static int brt_cursor_search(BRT_CURSOR cursor, brt_search_t *search, DBT *outkey, DBT *outval, TOKULOGGER logger) {
DBT newkey; toku_init_dbt(&newkey); newkey.flags = DB_DBT_MALLOC;
DBT newval; toku_init_dbt(&newval); newval.flags = DB_DBT_MALLOC;
int r = toku_brt_search(cursor->brt, search, &newkey, &newval, logger);
if (r == 0) {
brt_cursor_set_key_val(cursor, &newkey, &newval);
r = brt_cursor_copyout(cursor, outkey, outval);
}
dbt_cleanup(&newkey);
dbt_cleanup(&newval);
return r;
}
/* search for the kv pair that matches the search object and is equal to kv */
static int brt_cursor_search_eq_kv_xy(BRT_CURSOR cursor, brt_search_t *search, DBT *outkey, DBT *outval, TOKULOGGER logger) {
DBT newkey; toku_init_dbt(&newkey); newkey.flags = DB_DBT_MALLOC;
DBT newval; toku_init_dbt(&newval); newval.flags = DB_DBT_MALLOC;
int r = toku_brt_search(cursor->brt, search, &newkey, &newval, logger);
if (r == 0) {
if (compare_kv_xy(cursor->brt, search->k, search->v, &newkey, &newval) == 0) {
brt_cursor_set_key_val(cursor, &newkey, &newval);
r = brt_cursor_copyout(cursor, outkey, outval);
} else
r = DB_NOTFOUND;
}
dbt_cleanup(&newkey);
dbt_cleanup(&newval);
return r;
}
/* search for the kv pair that matches the search object and is equal to k */
static int brt_cursor_search_eq_k_x(BRT_CURSOR cursor, brt_search_t *search, DBT *outkey, DBT *outval, TOKULOGGER logger) {
DBT newkey; toku_init_dbt(&newkey); newkey.flags = DB_DBT_MALLOC;
DBT newval; toku_init_dbt(&newval); newval.flags = DB_DBT_MALLOC;
int r = toku_brt_search(cursor->brt, search, &newkey, &newval, logger);
if (r == 0) {
if (compare_k_x(cursor->brt, search->k, &newkey) == 0) {
brt_cursor_set_key_val(cursor, &newkey, &newval);
r = brt_cursor_copyout(cursor, outkey, outval);
} else
r = DB_NOTFOUND;
}
dbt_cleanup(&newkey);
dbt_cleanup(&newval);
return r;
}
static int brt_cursor_compare_one(brt_search_t *search, DBT *x, DBT *y) {
search = search; x = x; y = y;
return 1;
}
static int brt_cursor_first(BRT_CURSOR cursor, DBT *outkey, DBT *outval, TOKULOGGER logger) {
brt_search_t search; brt_search_init(&search, brt_cursor_compare_one, BRT_SEARCH_LEFT, 0, 0, cursor->brt);
return brt_cursor_search(cursor, &search, outkey, outval, logger);
}
static int brt_cursor_last(BRT_CURSOR cursor, DBT *outkey, DBT *outval, TOKULOGGER logger) {
brt_search_t search; brt_search_init(&search, brt_cursor_compare_one, BRT_SEARCH_RIGHT, 0, 0, cursor->brt);
return brt_cursor_search(cursor, &search, outkey, outval, logger);
}
static int brt_cursor_compare_next(brt_search_t *search, DBT *x, DBT *y) {
BRT brt = search->context;
return compare_kv_xy(brt, search->k, search->v, x, y) < 0; /* return min xy: kv < xy */
}
static int brt_cursor_next(BRT_CURSOR cursor, DBT *outkey, DBT *outval, TOKULOGGER logger) {
brt_search_t search; brt_search_init(&search, brt_cursor_compare_next, BRT_SEARCH_LEFT, &cursor->key, &cursor->val, cursor->brt);
return brt_cursor_search(cursor, &search, outkey, outval, logger);
}
static int brt_cursor_compare_next_nodup(brt_search_t *search, DBT *x, DBT *y) {
BRT brt = search->context; y = y;
return compare_k_x(brt, search->k, x) < 0; /* return min x: k < x */
}
static int brt_cursor_next_nodup(BRT_CURSOR cursor, DBT *outkey, DBT *outval, TOKULOGGER logger) {
brt_search_t search; brt_search_init(&search, brt_cursor_compare_next_nodup, BRT_SEARCH_LEFT, &cursor->key, &cursor->val, cursor->brt);
return brt_cursor_search(cursor, &search, outkey, outval, logger);
}
static int brt_cursor_compare_next_dup(brt_search_t *search, DBT *x, DBT *y) {
BRT brt = search->context;
int keycmp = compare_k_x(brt, search->k, x);
if (keycmp < 0)
return 1;
else
return keycmp == 0 && y && compare_v_y(brt, search->v, y) < 0; /* return min xy: k <= x && v < y */
}
static int brt_cursor_next_dup(BRT_CURSOR cursor, DBT *outkey, DBT *outval, TOKULOGGER logger) {
brt_search_t search; brt_search_init(&search, brt_cursor_compare_next_dup, BRT_SEARCH_LEFT, &cursor->key, &cursor->val, cursor->brt);
return brt_cursor_search_eq_k_x(cursor, &search, outkey, outval, logger);
}
static int brt_cursor_compare_get_both_range(brt_search_t *search, DBT *x, DBT *y) {
BRT brt = search->context;
int keycmp = compare_k_x(brt, search->k, x);
if (keycmp < 0)
return 1;
else
return keycmp == 0 && (y == 0 || compare_v_y(brt, search->v, y) <= 0); /* return min xy: k <= x && v <= y */
}
static int brt_cursor_get_both_range(BRT_CURSOR cursor, DBT *key, DBT *val, DBT *outkey, DBT *outval, TOKULOGGER logger) {
brt_search_t search; brt_search_init(&search, brt_cursor_compare_get_both_range, BRT_SEARCH_LEFT, key, val, cursor->brt);
return brt_cursor_search_eq_k_x(cursor, &search, outkey, outval, logger);
}
static int brt_cursor_compare_prev(brt_search_t *search, DBT *x, DBT *y) {
BRT brt = search->context;
return compare_kv_xy(brt, search->k, search->v, x, y) > 0; /* return max xy: kv > xy */
}
static int brt_cursor_prev(BRT_CURSOR cursor, DBT *outkey, DBT *outval, TOKULOGGER logger) {
brt_search_t search; brt_search_init(&search, brt_cursor_compare_prev, BRT_SEARCH_RIGHT, &cursor->key, &cursor->val, cursor->brt);
return brt_cursor_search(cursor, &search, outkey, outval, logger);
}
static int brt_cursor_compare_prev_nodup(brt_search_t *search, DBT *x, DBT *y) {
BRT brt = search->context; y = y;
return compare_k_x(brt, search->k, x) > 0; /* return max x: k > x */
}
static int brt_cursor_prev_nodup(BRT_CURSOR cursor, DBT *outkey, DBT *outval, TOKULOGGER logger) {
brt_search_t search; brt_search_init(&search, brt_cursor_compare_prev_nodup, BRT_SEARCH_RIGHT, &cursor->key, &cursor->val, cursor->brt);
return brt_cursor_search(cursor, &search, outkey, outval, logger);
}
#ifdef DB_PREV_DUP
static int brt_cursor_compare_prev_dup(brt_search_t *search, DBT *x, DBT *y) {
BRT brt = search->context;
int keycmp = compare_k_x(brt, search->k, x);
if (keycmp > 0)
return 1;
else
return keycmp == 0 && y && compare_v_y(brt, search->v, y) > 0; /* return max xy: k >= x && v > y */
}
static int brt_cursor_prev_dup(BRT_CURSOR cursor, DBT *outkey, DBT *outval, TOKULOGGER logger) {
brt_search_t search; brt_search_init(&search, brt_cursor_compare_prev_dup, BRT_SEARCH_RIGHT, &cursor->key, &cursor->val, cursor->brt);
return brt_cursor_search_eq_k_x(cursor, &search, outkey, outval, logger);
}
#endif
static int brt_cursor_compare_set_range(brt_search_t *search, DBT *x, DBT *y) {
BRT brt = search->context;
return compare_kv_xy(brt, search->k, search->v, x, y) <= 0; /* return kv <= xy */
}
static int brt_cursor_set(BRT_CURSOR cursor, DBT *key, DBT *val, DBT *outkey, DBT *outval, TOKULOGGER logger) {
brt_search_t search; brt_search_init(&search, brt_cursor_compare_set_range, BRT_SEARCH_LEFT, key, val, cursor->brt);
return brt_cursor_search_eq_kv_xy(cursor, &search, outkey, outval, logger);
}
static int brt_cursor_set_range(BRT_CURSOR cursor, DBT *key, DBT *outkey, DBT *outval, TOKULOGGER logger) {
brt_search_t search; brt_search_init(&search, brt_cursor_compare_set_range, BRT_SEARCH_LEFT, key, 0, cursor->brt);
return brt_cursor_search(cursor, &search, outkey, outval, logger);
}
int toku_brt_cursor_get (BRT_CURSOR cursor, DBT *key, DBT *val, int get_flags, TOKUTXN txn) {
int r;
int op = get_flags & DB_OPFLAGS_MASK;
TOKULOGGER logger = toku_txn_logger(txn);
if (get_flags & ~DB_OPFLAGS_MASK)
return EINVAL;
switch (op) {
case DB_CURRENT:
case DB_CURRENT_BINDING:
r = brt_cursor_current(cursor, op, key, val, logger);
break;
case DB_FIRST:
r = brt_cursor_first(cursor, key, val, logger);
break;
case DB_LAST:
r = brt_cursor_last(cursor, key, val, logger);
break;
case DB_NEXT:
if (brt_cursor_not_set(cursor))
r = brt_cursor_first(cursor, key, val, logger);
else
r = brt_cursor_next(cursor, key, val, logger);
break;
case DB_NEXT_DUP:
if (brt_cursor_not_set(cursor))
r = EINVAL;
else
r = brt_cursor_next_dup(cursor, key, val, logger);
break;
case DB_NEXT_NODUP:
if (brt_cursor_not_set(cursor))
r = brt_cursor_first(cursor, key, val, logger);
else
r = brt_cursor_next_nodup(cursor, key, val, logger);
break;
case DB_PREV:
if (brt_cursor_not_set(cursor))
r = brt_cursor_last(cursor, key, val, logger);
else
r = brt_cursor_prev(cursor, key, val, logger);
break;
#ifdef DB_PREV_DUP
case DB_PREV_DUP:
if (brt_cursor_not_set(cursor))
r = EINVAL;
else
r = brt_cursor_prev_dup(cursor, key, val, logger);
break;
#endif
case DB_PREV_NODUP:
if (brt_cursor_not_set(cursor))
r = brt_cursor_last(cursor, key, val, logger);
else
r = brt_cursor_prev_nodup(cursor, key, val, logger);
break;
case DB_SET:
r = brt_cursor_set(cursor, key, 0, 0, val, logger);
break;
case DB_SET_RANGE:
r = brt_cursor_set_range(cursor, key, key, val, logger);
break;
case DB_GET_BOTH:
r = brt_cursor_set(cursor, key, val, 0, 0, logger);
break;
case DB_GET_BOTH_RANGE:
r = brt_cursor_get_both_range(cursor, key, val, 0, val, logger);
break;
default:
r = EINVAL;
break;
}
return r;
}
int toku_brt_cursor_delete(BRT_CURSOR cursor, int flags, TOKUTXN txn) {
if ((flags & ~DB_DELETE_ANY) != 0)
return EINVAL;
if (brt_cursor_not_set(cursor))
return EINVAL;
int r = 0;
if (!(flags & DB_DELETE_ANY))
r = brt_cursor_current(cursor, DB_CURRENT, 0, 0, toku_txn_logger(txn));
if (r == 0)
r = toku_brt_delete_both(cursor->brt, &cursor->key, &cursor->val, txn);
return r;
}
int toku_brt_height_of_root(BRT brt, int *height) {
// for an open brt, return the current height.
int r;
if ((r = toku_read_and_pin_brt_header(brt->cf, &brt->h))) {
if (0) { died0: toku_unpin_brt_header(brt); }
return r;
}
CACHEKEY *rootp = toku_calculate_root_offset_pointer(brt);
void *node_v;
if ((r=toku_cachetable_get_and_pin(brt->cf, *rootp, &node_v, NULL,
toku_brtnode_flush_callback, toku_brtnode_fetch_callback, brt))) {
goto died0;
}
BRTNODE node = node_v;
*height = node->height;
r = unpin_brtnode(brt, node); assert(r==0);
r = toku_unpin_brt_header(brt); assert(r==0);
return 0;
}
#endif
...@@ -92,7 +92,7 @@ struct tokutxn { ...@@ -92,7 +92,7 @@ struct tokutxn {
char *rollentry_filename; char *rollentry_filename;
int rollentry_fd; // If we spill the roll_entries, we write them into this fd. int rollentry_fd; // If we spill the roll_entries, we write them into this fd.
off_t rollentry_filesize; // How many bytes are in the rollentry. off_t rollentry_filesize; // How many bytes are in the rollentry.
OMT used_open_brtcachefile_pairs; // a collection of the brts that we touched and which are still open. OMT open_brts; // a collection of the brts that we touched. Indexed by filenum.
}; };
int toku_logger_finish (TOKULOGGER logger, struct logbytes *logbytes, struct wbuf *wbuf, int do_fsync); int toku_logger_finish (TOKULOGGER logger, struct logbytes *logbytes, struct wbuf *wbuf, int do_fsync);
......
...@@ -343,6 +343,8 @@ int toku_logger_finish (TOKULOGGER logger, struct logbytes *logbytes, struct wbu ...@@ -343,6 +343,8 @@ int toku_logger_finish (TOKULOGGER logger, struct logbytes *logbytes, struct wbu
return toku_logger_log_bytes(logger, logbytes, do_fsync); return toku_logger_log_bytes(logger, logbytes, do_fsync);
} }
static void note_txn_closing (TOKUTXN txn);
static void cleanup_txn (TOKUTXN txn) { static void cleanup_txn (TOKUTXN txn) {
struct roll_entry *item; struct roll_entry *item;
while ((item=txn->newest_logentry)) { while ((item=txn->newest_logentry)) {
...@@ -362,10 +364,8 @@ static void cleanup_txn (TOKUTXN txn) { ...@@ -362,10 +364,8 @@ static void cleanup_txn (TOKUTXN txn) {
} }
list_remove(&txn->live_txns_link); list_remove(&txn->live_txns_link);
assert(toku_omt_size(txn->used_open_brtcachefile_pairs)==0); note_txn_closing(txn);
toku_omt_destroy(&txn->used_open_brtcachefile_pairs);
toku_free(txn); toku_free(txn);
return; return;
} }
...@@ -393,6 +393,10 @@ int toku_abort_rollback_item (TOKUTXN txn, struct roll_entry *item) { ...@@ -393,6 +393,10 @@ int toku_abort_rollback_item (TOKUTXN txn, struct roll_entry *item) {
return 0; return 0;
} }
static int note_brt_used_in_parent_txn(OMTVALUE brtv, u_int32_t UU(index), void*parentv) {
return toku_txn_note_brt(parentv, brtv);
}
int toku_logger_commit (TOKUTXN txn, int nosync) { int toku_logger_commit (TOKUTXN txn, int nosync) {
// printf("%s:%d committing\n", __FILE__, __LINE__); // printf("%s:%d committing\n", __FILE__, __LINE__);
// panic handled in log_commit // panic handled in log_commit
...@@ -433,6 +437,10 @@ int toku_logger_commit (TOKUTXN txn, int nosync) { ...@@ -433,6 +437,10 @@ int toku_logger_commit (TOKUTXN txn, int nosync) {
} }
txn->newest_logentry = txn->oldest_logentry = 0; txn->newest_logentry = txn->oldest_logentry = 0;
// Note the open brts, the omts must be merged
r = toku_omt_iterate(txn->open_brts, note_brt_used_in_parent_txn, txn->parent);
assert(r==0);
} else { } else {
// do the commit calls and free everything // do the commit calls and free everything
// we do the commit calls in reverse order too. // we do the commit calls in reverse order too.
...@@ -470,7 +478,7 @@ int toku_logger_txn_begin (TOKUTXN parent_tokutxn, TOKUTXN *tokutxn, TOKULOGGER ...@@ -470,7 +478,7 @@ int toku_logger_txn_begin (TOKUTXN parent_tokutxn, TOKUTXN *tokutxn, TOKULOGGER
if (result==0) return errno; if (result==0) return errno;
int r =toku_log_xbegin(logger, &result->first_lsn, 0, parent_tokutxn ? parent_tokutxn->txnid64 : 0); int r =toku_log_xbegin(logger, &result->first_lsn, 0, parent_tokutxn ? parent_tokutxn->txnid64 : 0);
if (r!=0) { toku_logger_panic(logger, r); toku_free(result); return r; } if (r!=0) { toku_logger_panic(logger, r); toku_free(result); return r; }
if ((r=toku_omt_create(&result->used_open_brtcachefile_pairs))!=0) { if ((r=toku_omt_create(&result->open_brts))!=0) {
toku_logger_panic(logger, r); toku_logger_panic(logger, r);
toku_free(result); toku_free(result);
return r; return r;
...@@ -969,3 +977,76 @@ int toku_read_rollback_backwards(int fd, off_t at, struct roll_entry **item, off ...@@ -969,3 +977,76 @@ int toku_read_rollback_backwards(int fd, off_t at, struct roll_entry **item, off
toku_free(buf); toku_free(buf);
return 0; return 0;
} }
static int find_ptr (OMTVALUE v, void *vfind) {
if (v<vfind) return -1;
if (v>vfind) return +1;
return 0;
}
static int find_filenum (OMTVALUE v, void *brtv) {
BRT brt = v;
BRT brtfind = brtv;
FILENUM fnum = toku_cachefile_filenum(brt ->cf);
FILENUM fnumfind = toku_cachefile_filenum(brtfind->cf);
if (fnum.fileid<fnumfind.fileid) return -1;
if (fnum.fileid>fnumfind.fileid) return +1;
return 0;
}
int toku_txn_note_brt (TOKUTXN txn, BRT brt) {
OMTVALUE txnv;
u_int32_t index;
int r = toku_omt_find_zero(brt->txns, find_ptr, txn, &txnv, &index);
if (r==0) {
// It's already there.
assert((TOKUTXN)txnv==txn);
return 0;
}
// Otherwise it's not there.
r = toku_omt_insert_at(brt->txns, txn, index);
assert(r==0);
r = toku_omt_insert(txn->open_brts, brt, find_filenum, brt, 0);
assert(r==0);
return 0;
}
static int remove_brt (OMTVALUE txnv, u_int32_t UU(idx), void *brtv) {
TOKUTXN txn = txnv;
BRT brt = brtv;
OMTVALUE brtv_again;
u_int32_t index;
int r = toku_omt_find_zero(txn->open_brts, find_filenum, brt, &brtv_again, &index);
assert(r==0);
assert((void*)brtv_again==brtv);
r = toku_omt_delete_at(txn->open_brts, index);
assert(r==0);
return 0;
}
int toku_txn_note_close_brt (BRT brt) {
int r = toku_omt_iterate(brt->txns, remove_brt, brt);
assert(r==0);
return 0;
}
static int remove_txn (OMTVALUE brtv, u_int32_t UU(idx), void *txnv) {
BRT brt = brtv;
TOKUTXN txn = txnv;
OMTVALUE txnv_again;
u_int32_t index;
int r = toku_omt_find_zero(brt->txns, find_ptr, txn, &txnv_again, &index);
assert(r==0);
assert((void*)txnv_again==txnv);
r = toku_omt_delete_at(brt->txns, index);
assert(r==0);
return 0;
}
// for every BRT in txn, remove it.
static void note_txn_closing (TOKUTXN txn) {
toku_omt_iterate(txn->open_brts, remove_txn, txn);
toku_omt_destroy(&txn->open_brts);
}
...@@ -165,4 +165,7 @@ int toku_commit_fileentries (int fd, off_t filesize, TOKUTXN txn); ...@@ -165,4 +165,7 @@ int toku_commit_fileentries (int fd, off_t filesize, TOKUTXN txn);
int toku_commit_rollback_item (TOKUTXN txn, struct roll_entry *item); int toku_commit_rollback_item (TOKUTXN txn, struct roll_entry *item);
int toku_abort_rollback_item (TOKUTXN txn, struct roll_entry *item); int toku_abort_rollback_item (TOKUTXN txn, struct roll_entry *item);
int toku_txn_note_brt (TOKUTXN txn, BRT brt);
int toku_txn_note_close_brt (BRT brt);
#endif #endif
Markdown is supported
0%
or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment