view tests/test-fncache.t @ 25757:4d1382fd96ff

context: write dirstate out explicitly at the end of markcommitted

To detect change of a file without redundant comparison of file
content, dirstate recognizes a file as certainly clean, if:

  (1) it is already known as "normal",
  (2) dirstate entry for it has valid (= not "-1") timestamp, and
  (3) mode, size and timestamp of it on the filesystem are as same as
      ones expected in dirstate

This works as expected in many cases, but doesn't in the corner case
that changing a file keeps mode, size and timestamp of it on the
filesystem.

The timetable below shows steps in one of typical such situations:

  ---- ----------------------------------- ----------------
                                           timestamp of "f"
                                           ----------------
                                           dirstate   file-
  time          action                     mem  file  system
  ---- ----------------------------------- ---- ----- -----
   *                                             ***    ***
       - 'hg transplant REV1 REV2 ...'
         - transplanting REV1
           ....
   N
           - change "f", but keep size                   N
             (via 'patch.patch()')
           - 'dirstate.normal("f")'          N   ***
             (via 'repo.commit()')

         - transplanting REV2
           - change "f", but keep size                   N
             (via 'patch.patch()')
           - aborted while patching
  N+1
         - release wlock
           - 'dirstate.write()'              N    N      N

       - 'hg status' shows "r1" as "clean"   N    N      N
  ---- ----------------------------------- ---- ----- -----

The most important point is that 'dirstate.write()' is executed at N+1
or later. This causes writing dirstate timestamp N of "f" out
successfully. If it is executed at N, 'parsers.pack_dirstate()'
replaces timestamp N with "-1" before actual writing dirstate out.

This issue can occur when 'hg transplant' satisfies conditions below:

  - multiple revisions to be transplanted change the same file
  - those revisions don't change mode and size of the file, and
  - the 2nd or later revision of them fails after changing the file

The root cause of this issue is that files are changed without
flushing in-memory dirstate changes via 'repo.commit()' (even though
omitting 'dirstate.normallookup()' on files changed by 'patch.patch()'
for efficiency also causes this issue).

To detect changes of files correctly, this patch writes in-memory
dirstate changes out explicitly after marking files as clean in
'committablectx.markcommitted()', which is invoked via
'repo.commit()'.

After this change, timetable is changed as below:

  ---- ----------------------------------- ----------------
                                           timestamp of "f"
                                           ----------------
                                           dirstate   file-
  time          action                     mem  file  system
  ---- ----------------------------------- ---- ----- -----
   *                                             ***    ***
       - 'hg transplant REV1 REV2 ...'
         - transplanting REV1
           ....
   N
           - change "f", but keep size                   N
             (via 'patch.patch()')
           - 'dirstate.normal("f")'          N    ***
             (via 'repo.commit()')
       ----------------------------------- ---- ----- -----
           - 'dirsttate.write()'            -1    -1
       ----------------------------------- ---- ----- -----
         - transplanting REV2
           - change "f", but keep size                   N
             (via 'patch.patch()')
           - aborted while patching
  N+1
         - release wlock
           - 'dirstate.write()'              -1   -1     N

       - 'hg status' shows "r1" as "clean"   -1   -1     N
  ---- ----------------------------------- ---- ----- -----

To reproduce this issue in tests certainly, this patch emulates some
timing critical actions as below:

  - change "f" at N

    'patch.patch()' with 'fakepatchtime.py' explicitly changes mtime
    of patched files to "2000-01-01 00:00" (= N).

  - 'dirstate.write()' via 'repo.commit()' at N

    'fakedirstatewritetime.py' forces 'pack_dirstate()' to use
    "2000-01-01 00:00" as "now", only if 'pack_dirstate()' is invoked
    via 'committablectx.markcommitted()'.

  - 'dirstate.write()' via releasing wlock at N+1 (or "not at N")

    'pack_dirstate()' via releasing wlock uses actual timestamp at
    runtime as "now", and it should be different from the "2000-01-01
    00:00" of "f".

BTW, this patch doesn't test cases below, even though 'patch.patch()'
is used similarly in these cases:

  1. failure of 'hg import' or 'hg qpush'

  2. success of 'hg import', 'hg qpush' or 'hg transplant'

Case (1) above doesn't cause this kind of issue, because:

  - if patching is aborted by conflicts, changed files are committed

    changed files are marked as CLEAN, even though they are partially
    patched.

  - otherwise, dirstate are fully restored by 'dirstateguard'

    For example in timetable above, timestamp of "f" in .hg/dirstate
    is restored to -1 (or less than N), and subsequent 'hg status' can
    detect changes correctly.

Case (2) always causes 'repo.status()' invocation via 'repo.commit()'
just after changing files inside same wlock scope.

  ---- ----------------------------------- ----------------
                                           timestamp of "f"
                                           ----------------
                                           dirstate   file-
  time          action                     mem  file  system
  ---- ----------------------------------- ---- ----- -----
  N                                              ***   ***
       - make file "f" clean                            N

       - execute 'hg foobar'
         ....
         - 'dirstate.normal("f")'           N    ***
           (e.g. via dirty check
            or previous 'repo.commit()')

         - change "f", but keep size                    N
         - 'repo.status()' (*1)
           (via 'repo.commit()')
  ---- ----------------------------------- ---- ----- -----

At a glance, 'repo.status()' at (*1) seems to cause similar issue (=
"changed files are treated as clean"), but actually doesn't.

'dirstate._lastnormaltime' should be N at (*1) above, because
'dirstate.normal()' via dirty check is finished at N.

Therefore, "f" changed at N (= 'dirstate._lastnormaltime') is forcibly
treated as "unsure" at (*1), and changes are detected as expected (see
'dirstate.status()' for detail).

If 'hg import' is executed with '--no-commit', 'repo.status()' isn't
invoked just after changing files inside same wlock scope.

But preceding 'dirstate.normal()' is invoked inside another wlock
scope via 'cmdutil.bailifchanged()', and in-memory changes should be
flushed at the end of that scope.

Therefore, timestamp N of clean "f" should be replaced by -1, if
'dirstate.write()' is invoked at N. It means that condition of this
issue isn't satisfied.

Init repo1:

  $ hg init repo1
  $ cd repo1
  $ echo "some text" > a
  $ hg add
  adding a
  $ hg ci -m first
  $ cat .hg/store/fncache | sort
  data/a.i

Testing a.i/b:

  $ mkdir a.i
  $ echo "some other text" > a.i/b
  $ hg add
  adding a.i/b (glob)
  $ hg ci -m second
  $ cat .hg/store/fncache | sort
  data/a.i
  data/a.i.hg/b.i

Testing a.i.hg/c:

  $ mkdir a.i.hg
  $ echo "yet another text" > a.i.hg/c
  $ hg add
  adding a.i.hg/c (glob)
  $ hg ci -m third
  $ cat .hg/store/fncache | sort
  data/a.i
  data/a.i.hg.hg/c.i
  data/a.i.hg/b.i

Testing verify:

  $ hg verify
  checking changesets
  checking manifests
  crosschecking files in changesets and manifests
  checking files
  3 files, 3 changesets, 3 total revisions

  $ rm .hg/store/fncache

  $ hg verify
  checking changesets
  checking manifests
  crosschecking files in changesets and manifests
  checking files
   warning: revlog 'data/a.i' not in fncache!
   warning: revlog 'data/a.i.hg/c.i' not in fncache!
   warning: revlog 'data/a.i/b.i' not in fncache!
  3 files, 3 changesets, 3 total revisions
  3 warnings encountered!
  hint: run "hg debugrebuildfncache" to recover from corrupt fncache

Follow the hint to make sure it works

  $ hg debugrebuildfncache
  adding data/a.i
  adding data/a.i.hg/c.i
  adding data/a.i/b.i
  3 items added, 0 removed from fncache

  $ hg verify
  checking changesets
  checking manifests
  crosschecking files in changesets and manifests
  checking files
  3 files, 3 changesets, 3 total revisions

  $ cd ..

Non store repo:

  $ hg --config format.usestore=False init foo
  $ cd foo
  $ mkdir tst.d
  $ echo foo > tst.d/foo
  $ hg ci -Amfoo
  adding tst.d/foo
  $ find .hg | sort
  .hg
  .hg/00changelog.i
  .hg/00manifest.i
  .hg/cache
  .hg/cache/branch2-served
  .hg/cache/rbc-names-v1
  .hg/cache/rbc-revs-v1
  .hg/data
  .hg/data/tst.d.hg
  .hg/data/tst.d.hg/foo.i
  .hg/dirstate
  .hg/last-message.txt
  .hg/phaseroots
  .hg/requires
  .hg/undo
  .hg/undo.backupfiles
  .hg/undo.bookmarks
  .hg/undo.branch
  .hg/undo.desc
  .hg/undo.dirstate
  .hg/undo.phaseroots
  $ cd ..

Non fncache repo:

  $ hg --config format.usefncache=False init bar
  $ cd bar
  $ mkdir tst.d
  $ echo foo > tst.d/Foo
  $ hg ci -Amfoo
  adding tst.d/Foo
  $ find .hg | sort
  .hg
  .hg/00changelog.i
  .hg/cache
  .hg/cache/branch2-served
  .hg/cache/rbc-names-v1
  .hg/cache/rbc-revs-v1
  .hg/dirstate
  .hg/last-message.txt
  .hg/requires
  .hg/store
  .hg/store/00changelog.i
  .hg/store/00manifest.i
  .hg/store/data
  .hg/store/data/tst.d.hg
  .hg/store/data/tst.d.hg/_foo.i
  .hg/store/phaseroots
  .hg/store/undo
  .hg/store/undo.backupfiles
  .hg/store/undo.phaseroots
  .hg/undo.bookmarks
  .hg/undo.branch
  .hg/undo.desc
  .hg/undo.dirstate
  $ cd ..

Encoding of reserved / long paths in the store

  $ hg init r2
  $ cd r2
  $ cat <<EOF > .hg/hgrc
  > [ui]
  > portablefilenames = ignore
  > EOF

  $ hg import -q --bypass - <<EOF
  > # HG changeset patch
  > # User test
  > # Date 0 0
  > # Node ID 1c7a2f7cb77be1a0def34e4c7cabc562ad98fbd7
  > # Parent  0000000000000000000000000000000000000000
  > 1
  > 
  > diff --git a/12345678/12345678/12345678/12345678/12345678/12345678/12345678/12345/xxxxxxxxx-xxxxxxxxx-xxxxxxxxx-123456789-12.3456789-12345-ABCDEFGHIJKLMNOPRSTUVWXYZ-abcdefghjiklmnopqrstuvwxyz b/12345678/12345678/12345678/12345678/12345678/12345678/12345678/12345/xxxxxxxxx-xxxxxxxxx-xxxxxxxxx-123456789-12.3456789-12345-ABCDEFGHIJKLMNOPRSTUVWXYZ-abcdefghjiklmnopqrstuvwxyz
  > new file mode 100644
  > --- /dev/null
  > +++ b/12345678/12345678/12345678/12345678/12345678/12345678/12345678/12345/xxxxxxxxx-xxxxxxxxx-xxxxxxxxx-123456789-12.3456789-12345-ABCDEFGHIJKLMNOPRSTUVWXYZ-abcdefghjiklmnopqrstuvwxyz
  > @@ -0,0 +1,1 @@
  > +foo
  > diff --git a/AUX/SECOND/X.PRN/FOURTH/FI:FTH/SIXTH/SEVENTH/EIGHTH/NINETH/TENTH/ELEVENTH/LOREMIPSUM.TXT b/AUX/SECOND/X.PRN/FOURTH/FI:FTH/SIXTH/SEVENTH/EIGHTH/NINETH/TENTH/ELEVENTH/LOREMIPSUM.TXT
  > new file mode 100644
  > --- /dev/null
  > +++ b/AUX/SECOND/X.PRN/FOURTH/FI:FTH/SIXTH/SEVENTH/EIGHTH/NINETH/TENTH/ELEVENTH/LOREMIPSUM.TXT
  > @@ -0,0 +1,1 @@
  > +foo
  > diff --git a/Project Planning/Resources/AnotherLongDirectoryName/Followedbyanother/AndAnother/AndThenAnExtremelyLongFileName.txt b/Project Planning/Resources/AnotherLongDirectoryName/Followedbyanother/AndAnother/AndThenAnExtremelyLongFileName.txt
  > new file mode 100644
  > --- /dev/null
  > +++ b/Project Planning/Resources/AnotherLongDirectoryName/Followedbyanother/AndAnother/AndThenAnExtremelyLongFileName.txt	
  > @@ -0,0 +1,1 @@
  > +foo
  > diff --git a/bla.aux/prn/PRN/lpt/com3/nul/coma/foo.NUL/normal.c b/bla.aux/prn/PRN/lpt/com3/nul/coma/foo.NUL/normal.c
  > new file mode 100644
  > --- /dev/null
  > +++ b/bla.aux/prn/PRN/lpt/com3/nul/coma/foo.NUL/normal.c
  > @@ -0,0 +1,1 @@
  > +foo
  > diff --git a/enterprise/openesbaddons/contrib-imola/corba-bc/netbeansplugin/wsdlExtension/src/main/java/META-INF/services/org.netbeans.modules.xml.wsdl.bindingsupport.spi.ExtensibilityElementTemplateProvider b/enterprise/openesbaddons/contrib-imola/corba-bc/netbeansplugin/wsdlExtension/src/main/java/META-INF/services/org.netbeans.modules.xml.wsdl.bindingsupport.spi.ExtensibilityElementTemplateProvider
  > new file mode 100644
  > --- /dev/null
  > +++ b/enterprise/openesbaddons/contrib-imola/corba-bc/netbeansplugin/wsdlExtension/src/main/java/META-INF/services/org.netbeans.modules.xml.wsdl.bindingsupport.spi.ExtensibilityElementTemplateProvider
  > @@ -0,0 +1,1 @@
  > +foo
  > EOF

  $ find .hg/store -name *.i  | sort
  .hg/store/00changelog.i
  .hg/store/00manifest.i
  .hg/store/data/bla.aux/pr~6e/_p_r_n/lpt/co~6d3/nu~6c/coma/foo._n_u_l/normal.c.i
  .hg/store/dh/12345678/12345678/12345678/12345678/12345678/12345678/12345678/12345/xxxxxx168e07b38e65eff86ab579afaaa8e30bfbe0f35f.i
  .hg/store/dh/au~78/second/x.prn/fourth/fi~3afth/sixth/seventh/eighth/nineth/tenth/loremia20419e358ddff1bf8751e38288aff1d7c32ec05.i
  .hg/store/dh/enterpri/openesba/contrib-/corba-bc/netbeans/wsdlexte/src/main/java/org.net7018f27961fdf338a598a40c4683429e7ffb9743.i
  .hg/store/dh/project_/resource/anotherl/followed/andanoth/andthenanextremelylongfilename0d8e1f4187c650e2f1fdca9fd90f786bc0976b6b.i

  $ cd ..

Aborting lock does not prevent fncache writes

  $ cat > exceptionext.py <<EOF
  > import os
  > from mercurial import commands, util
  > from mercurial.extensions import wrapfunction
  > 
  > def lockexception(orig, vfs, lockname, wait, releasefn, acquirefn, desc):
  >     def releasewrap():
  >         raise util.Abort("forced lock failure")
  >     return orig(vfs, lockname, wait, releasewrap, acquirefn, desc)
  > 
  > def reposetup(ui, repo):
  >     wrapfunction(repo, '_lock', lockexception)
  > 
  > cmdtable = {}
  > 
  > EOF
  $ extpath=`pwd`/exceptionext.py
  $ hg init fncachetxn
  $ cd fncachetxn
  $ printf "[extensions]\nexceptionext=$extpath\n" >> .hg/hgrc
  $ touch y
  $ hg ci -qAm y
  abort: forced lock failure
  [255]
  $ cat .hg/store/fncache
  data/y.i

Aborting transaction prevents fncache change

  $ cat > ../exceptionext.py <<EOF
  > import os
  > from mercurial import commands, util, localrepo
  > from mercurial.extensions import wrapfunction
  > 
  > def wrapper(orig, self, *args, **kwargs):
  >     tr = orig(self, *args, **kwargs)
  >     def fail(tr):
  >         raise util.Abort("forced transaction failure")
  >     # zzz prefix to ensure it sorted after store.write
  >     tr.addfinalize('zzz-forcefails', fail)
  >     return tr
  > 
  > def uisetup(ui):
  >     wrapfunction(localrepo.localrepository, 'transaction', wrapper)
  > 
  > cmdtable = {}
  > 
  > EOF
  $ rm -f "${extpath}c"
  $ touch z
  $ hg ci -qAm z
  transaction abort!
  rollback completed
  abort: forced transaction failure
  [255]
  $ cat .hg/store/fncache
  data/y.i

Aborted transactions can be recovered later

  $ cat > ../exceptionext.py <<EOF
  > import os
  > from mercurial import commands, util, transaction, localrepo
  > from mercurial.extensions import wrapfunction
  > 
  > def trwrapper(orig, self, *args, **kwargs):
  >     tr = orig(self, *args, **kwargs)
  >     def fail(tr):
  >         raise util.Abort("forced transaction failure")
  >     # zzz prefix to ensure it sorted after store.write
  >     tr.addfinalize('zzz-forcefails', fail)
  >     return tr
  > 
  > def abortwrapper(orig, self, *args, **kwargs):
  >     raise util.Abort("forced transaction failure")
  > 
  > def uisetup(ui):
  >     wrapfunction(localrepo.localrepository, 'transaction', trwrapper)
  >     wrapfunction(transaction.transaction, '_abort', abortwrapper)
  > 
  > cmdtable = {}
  > 
  > EOF
  $ rm -f "${extpath}c"
  $ hg up -q 1
  $ touch z
  $ hg ci -qAm z 2>/dev/null
  [255]
  $ cat .hg/store/fncache | sort
  data/y.i
  data/z.i
  $ hg recover
  rolling back interrupted transaction
  checking changesets
  checking manifests
  crosschecking files in changesets and manifests
  checking files
  1 files, 1 changesets, 1 total revisions
  $ cat .hg/store/fncache
  data/y.i

  $ cd ..

debugrebuildfncache does nothing unless repo has fncache requirement

  $ hg --config format.usefncache=false init nofncache
  $ cd nofncache
  $ hg debugrebuildfncache
  (not rebuilding fncache because repository does not support fncache

  $ cd ..

debugrebuildfncache works on empty repository

  $ hg init empty
  $ cd empty
  $ hg debugrebuildfncache
  fncache already up to date
  $ cd ..

debugrebuildfncache on an up to date repository no-ops

  $ hg init repo
  $ cd repo
  $ echo initial > foo
  $ echo initial > .bar
  $ hg commit -A -m initial
  adding .bar
  adding foo

  $ cat .hg/store/fncache | sort
  data/.bar.i
  data/foo.i

  $ hg debugrebuildfncache
  fncache already up to date

debugrebuildfncache restores deleted fncache file

  $ rm -f .hg/store/fncache
  $ hg debugrebuildfncache
  adding data/.bar.i
  adding data/foo.i
  2 items added, 0 removed from fncache

  $ cat .hg/store/fncache | sort
  data/.bar.i
  data/foo.i

Rebuild after rebuild should no-op

  $ hg debugrebuildfncache
  fncache already up to date

A single missing file should get restored, an extra file should be removed

  $ cat > .hg/store/fncache << EOF
  > data/foo.i
  > data/bad-entry.i
  > EOF

  $ hg debugrebuildfncache
  removing data/bad-entry.i
  adding data/.bar.i
  1 items added, 1 removed from fncache

  $ cat .hg/store/fncache | sort
  data/.bar.i
  data/foo.i

  $ cd ..

Try a simple variation without dotencode to ensure fncache is ignorant of encoding

  $ hg --config format.dotencode=false init nodotencode
  $ cd nodotencode
  $ echo initial > foo
  $ echo initial > .bar
  $ hg commit -A -m initial
  adding .bar
  adding foo

  $ cat .hg/store/fncache | sort
  data/.bar.i
  data/foo.i

  $ rm .hg/store/fncache
  $ hg debugrebuildfncache
  adding data/.bar.i
  adding data/foo.i
  2 items added, 0 removed from fncache

  $ cat .hg/store/fncache | sort
  data/.bar.i
  data/foo.i