commands: make commit acquire locks before processing (issue4368)
Before this patch, "hg commit" (process A) executes steps below:
1. get current branch heads via 'repo.branchheads()'
- cache 'repo.changelog'
2. invoke 'repo.commit()'
3. acquire wlock
- invalidate 'repo.dirstate'
4. access 'repo.dirstate'
- re-read '.hg/dirstate'
- check validity of parent revisions with 'repo.changelog'
5. invoke 'repo.commitctx()'
6. acquire store lock (slock)
- invalidate 'repo.changelog'
7. do committing
8. release slock
9. release wlock
10. check new branch head (via 'cmdutil.commitstatus()')
If acquisition of wlock at (3) above waits for another "hg commit"
(process B) or so running parallelly to release wlock, process A
causes creating orphan revision, because:
- '.hg/dirstate' refers the revision, which is newly added by
process B, as its parent
- but already cached 'repo.changelog' doesn't contain such revision
- therefore, validating parents of '.hg/dirstate' at (4) above
replaces such revision with 'nullid'
Then, process A creates "orphan" revision, of which parent is "null"
revision.
In addition to it, "created new head" may be shown at the end of
process A unintentionally, if store is updated parallelly, because
both getting branch heads (1) and checking new branch head (10) are
executed outside slock scope.
To avoid this issue, this patch makes "hg commit" acquire wlock and
slock before processing.
This patch resolves the issue between "hg commit" processes, but not
one between "hg commit" and other commands. Subsequent patches resolve
the latter.
Even after this patch, there are still corner case problems below:
- filecache may overlook changes of '.hg/dirstate', and it causes
similar issue (see below for detail)
https://bz.mercurial-scm.org/show_bug.cgi?id=4368#c10
- 3rd party extension may cause similar issue, if it directly uses
'repo.commit()' without acquisition of wlock and slock
This can be fixed by acquisition of slock at the beginning of
'repo.commit()', but it seems suitable for "default" branch
In fact, acquisition of slock itself is already introduced at
"default" branch by 4414d500604f, but acquisition is not at the
beginning of 'repo.commit()'.
This patch also changes some tests:
- test-fncache.t needs this tricky wrapping, to release (= forced
failure of) wlock certainly
- order of "hg commit" output is changed by widening scope of locks,
because some hooks are fired after releasing wlock
$ catpatch() {
> cat $1 | sed -e "s/^\(# Parent \).*/\1/"
> }
$ echo "[extensions]" >> $HGRCPATH
$ echo "mq=" >> $HGRCPATH
$ runtest() {
> hg init mq
> cd mq
>
> echo a > a
> hg ci -Ama
>
> echo '% qnew should refuse bad patch names'
> hg qnew series
> hg qnew status
> hg qnew guards
> hg qnew .
> hg qnew ..
> hg qnew .hgignore
> hg qnew .mqfoo
> hg qnew 'foo#bar'
> hg qnew 'foo:bar'
> hg qnew "`echo foo; echo bar`"
>
> hg qinit -c
>
> echo '% qnew with name containing slash'
> hg qnew foo/
> hg qnew foo/bar.patch
> hg qnew foo
> hg qseries
> hg qpop
> hg qdelete foo/bar.patch
>
> echo '% qnew with uncommitted changes'
> echo a > somefile
> hg add somefile
> hg qnew uncommitted.patch
> hg st
> hg qseries
>
> echo '% qnew implies add'
> hg -R .hg/patches st
>
> echo '% qnew missing'
> hg qnew missing.patch missing
>
> echo '% qnew -m'
> hg qnew -m 'foo bar' mtest.patch
> catpatch .hg/patches/mtest.patch
>
> echo '% qnew twice'
> hg qnew first.patch
> hg qnew first.patch
>
> touch ../first.patch
> hg qimport ../first.patch
>
> echo '% qnew -f from a subdirectory'
> hg qpop -a
> mkdir d
> cd d
> echo b > b
> hg ci -Am t
> echo b >> b
> hg st
> hg qnew -g -f p
> catpatch ../.hg/patches/p
>
> echo '% qnew -u with no username configured'
> HGUSER= hg qnew -u blue red
> catpatch ../.hg/patches/red
>
> echo '% qnew -e -u with no username configured'
> HGUSER= hg qnew -e -u chartreuse fucsia
> catpatch ../.hg/patches/fucsia
>
> echo '% fail when trying to import a merge'
> hg init merge
> cd merge
> touch a
> hg ci -Am null
> echo a >> a
> hg ci -m a
> hg up -r 0
> echo b >> a
> hg ci -m b
> hg merge -f 1
> hg resolve --mark a
> hg qnew -f merge
>
> cd ../../..
> rm -r mq
> }
plain headers
$ echo "[mq]" >> $HGRCPATH
$ echo "plain=true" >> $HGRCPATH
$ mkdir sandbox
$ (cd sandbox ; runtest)
adding a
% qnew should refuse bad patch names
abort: "series" cannot be used as the name of a patch
abort: "status" cannot be used as the name of a patch
abort: "guards" cannot be used as the name of a patch
abort: "." cannot be used as the name of a patch
abort: ".." cannot be used as the name of a patch
abort: patch name cannot begin with ".hg"
abort: patch name cannot begin with ".mq"
abort: '#' cannot be used in the name of a patch
abort: ':' cannot be used in the name of a patch
abort: '\n' cannot be used in the name of a patch
% qnew with name containing slash
abort: path ends in directory separator: foo/ (glob)
abort: "foo" already exists as a directory
foo/bar.patch
popping foo/bar.patch
patch queue now empty
% qnew with uncommitted changes
uncommitted.patch
% qnew implies add
A .hgignore
A series
A uncommitted.patch
% qnew missing
abort: missing: * (glob)
% qnew -m
foo bar
% qnew twice
abort: patch "first.patch" already exists
abort: patch "first.patch" already exists
% qnew -f from a subdirectory
popping first.patch
popping mtest.patch
popping uncommitted.patch
patch queue now empty
adding d/b
M d/b
diff --git a/d/b b/d/b
--- a/d/b
+++ b/d/b
@@ -1,1 +1,2 @@
b
+b
% qnew -u with no username configured
From: blue
% qnew -e -u with no username configured
From: chartreuse
% fail when trying to import a merge
adding a
1 files updated, 0 files merged, 0 files removed, 0 files unresolved
created new head
merging a
warning: conflicts while merging a! (edit, then use 'hg resolve --mark')
0 files updated, 0 files merged, 0 files removed, 1 files unresolved
use 'hg resolve' to retry unresolved file merges or 'hg update -C .' to abandon
(no more unresolved files)
abort: cannot manage merge changesets
$ rm -r sandbox
hg headers
$ echo "plain=false" >> $HGRCPATH
$ mkdir sandbox
$ (cd sandbox ; runtest)
adding a
% qnew should refuse bad patch names
abort: "series" cannot be used as the name of a patch
abort: "status" cannot be used as the name of a patch
abort: "guards" cannot be used as the name of a patch
abort: "." cannot be used as the name of a patch
abort: ".." cannot be used as the name of a patch
abort: patch name cannot begin with ".hg"
abort: patch name cannot begin with ".mq"
abort: '#' cannot be used in the name of a patch
abort: ':' cannot be used in the name of a patch
abort: '\n' cannot be used in the name of a patch
% qnew with name containing slash
abort: path ends in directory separator: foo/ (glob)
abort: "foo" already exists as a directory
foo/bar.patch
popping foo/bar.patch
patch queue now empty
% qnew with uncommitted changes
uncommitted.patch
% qnew implies add
A .hgignore
A series
A uncommitted.patch
% qnew missing
abort: missing: * (glob)
% qnew -m
# HG changeset patch
# Parent
foo bar
% qnew twice
abort: patch "first.patch" already exists
abort: patch "first.patch" already exists
% qnew -f from a subdirectory
popping first.patch
popping mtest.patch
popping uncommitted.patch
patch queue now empty
adding d/b
M d/b
# HG changeset patch
# Parent
diff --git a/d/b b/d/b
--- a/d/b
+++ b/d/b
@@ -1,1 +1,2 @@
b
+b
% qnew -u with no username configured
# HG changeset patch
# User blue
# Parent
% qnew -e -u with no username configured
# HG changeset patch
# User chartreuse
# Parent
% fail when trying to import a merge
adding a
1 files updated, 0 files merged, 0 files removed, 0 files unresolved
created new head
merging a
warning: conflicts while merging a! (edit, then use 'hg resolve --mark')
0 files updated, 0 files merged, 0 files removed, 1 files unresolved
use 'hg resolve' to retry unresolved file merges or 'hg update -C .' to abandon
(no more unresolved files)
abort: cannot manage merge changesets
$ rm -r sandbox
Test saving last-message.txt
$ hg init repo
$ cd repo
$ cat > $TESTTMP/commitfailure.py <<EOF
> from mercurial import error
> def reposetup(ui, repo):
> class commitfailure(repo.__class__):
> def commit(self, *args, **kwargs):
> raise error.Abort('emulating unexpected abort')
> repo.__class__ = commitfailure
> EOF
$ cat >> .hg/hgrc <<EOF
> [extensions]
> # this failure occurs before editor invocation
> commitfailure = $TESTTMP/commitfailure.py
> EOF
$ cat > $TESTTMP/editor.sh << EOF
> echo "==== before editing"
> cat \$1
> echo "===="
> echo "test saving last-message.txt" >> \$1
> EOF
(test that editor is not invoked before transaction starting)
$ rm -f .hg/last-message.txt
$ HGEDITOR="sh $TESTTMP/editor.sh" hg qnew -e patch
abort: emulating unexpected abort
[255]
$ test -f .hg/last-message.txt
[1]
(test that editor is invoked and commit message is saved into
"last-message.txt")
$ cat >> .hg/hgrc <<EOF
> [extensions]
> commitfailure = !
> [hooks]
> # this failure occurs after editor invocation
> pretxncommit.unexpectedabort = false
> EOF
$ rm -f .hg/last-message.txt
$ hg status
$ HGEDITOR="sh $TESTTMP/editor.sh" hg qnew -e patch
==== before editing
HG: Enter commit message. Lines beginning with 'HG:' are removed.
HG: Leave message empty to use default message.
HG: --
HG: user: test
HG: branch 'default'
HG: no files changed
====
note: commit message saved in .hg/last-message.txt
transaction abort!
rollback completed
abort: pretxncommit.unexpectedabort hook exited with status 1
[255]
$ cat .hg/last-message.txt
test saving last-message.txt
$ cat >> .hg/hgrc <<EOF
> [hooks]
> pretxncommit.unexpectedabort =
> EOF
#if unix-permissions
Test handling default message with the patch filename with tail whitespaces
$ cat > $TESTTMP/editor.sh << EOF
> echo "==== before editing"
> cat \$1
> echo "===="
> echo "[mq]: patch " > \$1
> EOF
$ rm -f .hg/last-message.txt
$ hg status
$ HGEDITOR="sh $TESTTMP/editor.sh" hg qnew -e "patch "
==== before editing
HG: Enter commit message. Lines beginning with 'HG:' are removed.
HG: Leave message empty to use default message.
HG: --
HG: user: test
HG: branch 'default'
HG: no files changed
====
$ cat ".hg/patches/patch "
# HG changeset patch
# Parent 0000000000000000000000000000000000000000
$ cd ..
#endif