.

wcfs/notes.txt

@@ -57,7 +57,7 @@ Trees _and_ Buckets nodes - would be required.
 -> we took the approach where we send invalidation to client about a block
 lazily only when the block is actually accessed.
 
-XXX building δFtail lazily along serving fuse reads during scope of one
+XXX building δFtail lazily along serving fuse reads during scope of one         XXX kill - fixed
 transaction is not trivial and creates concurrency bottlenecks if simple
 locking scheme is used. With the main difficulty being to populate tracking set
 of δBtree lazily. However as the first approach we can still build complete
@@ -107,7 +107,7 @@ We can change a mapping while a page from it is under pagefault:
   * https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/mm/filemap.c?id=v4.20-rc3-83-g06e68fed3282#n2457
   * https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/tree/mm/filemap.c?id=v4.20-rc3-83-g06e68fed3282#n1301
 
-- the filesystem server upon receiving the read request can manipulate
+- the filesystem server, upon receiving the read request, can manipulate
   client's address space. This requires to write-lock client->mm->mmap_sem,
   but we can be sure it won't deadlock because the kernel releases it
   before waiting (see previous point).

wcfs/wcfs.go

@@ -261,7 +261,7 @@ package main
 //
 //    The invariant helps on invalidation: when δFtail (see below) sees a
 //    changed oid, it is guaranteed that if the change affects block that was
-//    ever provided to OS, δFtail will detect that this block has changed.	XXX review
+//    ever provided to OS, δFtail will detect that this block has changed.
 //    And if oid relates to a file block but is not in δFtail's tracking set -
 //    we know that block is not cached and will trigger ZODB load on a future
 //    file read.
@@ -308,7 +308,7 @@ package main
 //	  cache for @<rev>/bigfile/file[blk], where
 //
 //	    # see below about file.δtail
-//	    # XXX -> file.BlkRevAt(#blk, zhead.at)
+//	    # XXX -> δFtail.BlkRevAt(file, #blk, zhead.at)
 //	    rev = max(file.δtail.by(#blk)) || min(rev ∈ file.δtail) || zhead.at
 //
 //	- invalidate head/bigfile/file[blk] in OS file cache.
@@ -339,7 +339,7 @@ package main
 //    XXX δtail can miss ...
 //
 //    to support initial openings with @at being slightly in the past, we also
-//    make sure that min(rev) is enough to cover last 10 minutes of history
+//    make sure that min(rev) is enough to cover last 10 minutes of history	XXX 10m -> 1m ?
 //    from head/at.
 //
 // 7) when we receive a FUSE read(#blk) request to a head/bigfile/file, we process it as follows:
@@ -368,7 +368,7 @@ package main
 //	  rev(blk) ≤ rev'(blk)		rev'(blk) = min(^^^)
 //
 //
-//	XXX we delay recomputing δFtail.BlkRevAt(file, #blk, head) because
+//	Note: we delay recomputing δFtail.BlkRevAt(file, #blk, head) because
 //	using just cheap revmax estimate can frequently result in all watches
 //	being skipped.
 //
@@ -425,6 +425,7 @@ package main
 // (^) see notes.txt -> "Client cannot be ptraced while under pagefault"
 // (%) no need to keep track of ZData - ZBlk1 is always marked as changed on blk data change.
 
+
 // Wcfs locking organization
 //
 // As it was said processing ZODB invalidations (see "4") and serving file
@@ -463,6 +464,7 @@ package main
 //	WatchLink.byfileMu	>  BigFileDir.fileMu
 //	WatchLink.byfileMu	>  Watch.atMu
 
+
 // Notation used
 //
 // δZ    - change in ZODB space
@@ -600,7 +602,7 @@ type BigFileDir struct {
 
 	// δ tail of tracked BTree nodes of all BigFiles + -> which file
 	// (used only for head/, not revX/)
-	δFmu   sync.RWMutex		// zheadMu.W  |  zheadMu.R + δFmu.X
+	δFmu   sync.RWMutex		// zheadMu.W  |  zheadMu.R + δFmu.X	XXX kill δFmu
 	δFtail *zdata.ΔFtail
 }
 
@@ -624,15 +626,9 @@ type BigFile struct {
 			 //     while populating δFtail lazily
 			 // XXX or then it is not "constant during lifetime of current txn"
 
-//	// tail change history of this file.
-//	//
-//	// XXX computationally expensive to start - see "Invalidations to wcfs
-//	// clients are delayed ..." in notes.txt
-//	δtail *ΔTailI64 // [](rev↑, []#blk)
-
 	// blocks that were ever read-accessed (head/ only)	XXX locking by bfdir.δFmu ?
 	// XXX = δFtail.Tracked(f) ?
-	// XXX goes away if δFtail query returns only tracked blocks
+	// XXX goes away if δFtail query returns only tracked blocks  (XXX that's not the case)
 	accessed setI64
 
 	// inflight loadings of ZBigFile from ZODB.