Mercurial > hg
view tests/test-rebase-named-branches.t @ 46607:e9901d01d135
revlog: add a mechanism to verify expected file position before appending
If someone uses `hg debuglocks`, or some non-hg process writes to the .hg
directory without respecting the locks, or if the repo's on a networked
filesystem, it's possible for the revlog code to write out corrupted data.
The form of this corruption can vary depending on what data was written and how
that happened. We are in the "networked filesystem" case (though I've had users
also do this to themselves with the "`hg debuglocks`" scenario), and most often
see this with the changelog. What ends up happening is we produce two items
(let's call them rev1 and rev2) in the .i file that have the same linkrev,
baserev, and offset into the .d file, while the data in the .d file is appended
properly. rev2's compressed_size is accurate for rev2, but when we go to
decompress the data in the .d file, we use the offset that's recorded in the
index file, which is the same as rev1, and attempt to decompress
rev2.compressed_size bytes of rev1's data. This usually does not succeed. :)
When using inline data, this also fails, though I haven't investigated why too
closely. This shows up as a "patch decode" error. I believe what's happening
there is that we're basically ignoring the offset field, getting the data
properly, but since baserev != rev, it thinks this is a delta based on rev
(instead of a full text) and can't actually apply it as such.
For now, I'm going to make this an optional component and default it to entirely
off. I may increase the default severity of this in the future, once I've
enabled it for my users and we gain more experience with it. Luckily, most of my
users have a versioned filesystem and can roll back to before the corruption has
been written, it's just a hassle to do so and not everyone knows how (so it's a
support burden). Users on other filesystems will not have that luxury, and this
can cause them to have a corrupted repository that they are unlikely to know how
to resolve, and they'll see this as a data-loss event. Refusing to create the
corruption is a much better user experience.
This mechanism is not perfect. There may be false-negatives (racy writes that
are not detected). There should not be any false-positives (non-racy writes that
are detected as such). This is not a mechanism that makes putting a repo on a
networked filesystem "safe" or "supported", just *less* likely to cause
corruption.
Differential Revision: https://phab.mercurial-scm.org/D9952
author | Kyle Lippincott <spectral@google.com> |
---|---|
date | Wed, 03 Feb 2021 16:33:10 -0800 |
parents | 527ce85c2e60 |
children | 9989a276712f |
line wrap: on
line source
$ cat >> $HGRCPATH <<EOF > [extensions] > rebase= > > [phases] > publish=False > > [alias] > tglog = log -G --template "{rev}: {node|short} '{desc}' {branches}\n" > EOF $ hg init a $ cd a $ hg unbundle "$TESTDIR/bundles/rebase.hg" adding changesets adding manifests adding file changes added 8 changesets with 7 changes to 7 files (+2 heads) new changesets cd010b8cd998:02de42196ebe (8 drafts) (run 'hg heads' to see heads, 'hg merge' to merge) $ hg up tip 3 files updated, 0 files merged, 0 files removed, 0 files unresolved $ cd .. $ hg clone -q -u . a a1 $ cd a1 $ hg update 3 3 files updated, 0 files merged, 2 files removed, 0 files unresolved $ hg branch dev-one marked working directory as branch dev-one (branches are permanent and global, did you want a bookmark?) $ hg ci -m 'dev-one named branch' $ hg update 7 2 files updated, 0 files merged, 3 files removed, 0 files unresolved $ hg branch dev-two marked working directory as branch dev-two $ echo x > x $ hg add x $ hg ci -m 'dev-two named branch' $ hg tglog @ 9: cb039b7cae8e 'dev-two named branch' dev-two | | o 8: 643fc9128048 'dev-one named branch' dev-one | | o | 7: 02de42196ebe 'H' | | +---o 6: eea13746799a 'G' | | | o | | 5: 24b6387c8c8c 'F' | | | +---o 4: 9520eea781bc 'E' | | | o 3: 32af7686d403 'D' | | | o 2: 5fddd98957c8 'C' | | | o 1: 42ccdea3bb16 'B' |/ o 0: cd010b8cd998 'A' Branch name containing a dash (issue3181) $ hg rebase -b dev-two -d dev-one --keepbranches rebasing 5:24b6387c8c8c "F" rebasing 6:eea13746799a "G" rebasing 7:02de42196ebe "H" rebasing 9:cb039b7cae8e tip "dev-two named branch" saved backup bundle to $TESTTMP/a1/.hg/strip-backup/24b6387c8c8c-24cb8001-rebase.hg $ hg tglog @ 9: 9e70cd31750f 'dev-two named branch' dev-two | o 8: 31d0e4ba75e6 'H' | | o 7: 4b988a958030 'G' |/| o | 6: 24de4aff8e28 'F' | | o | 5: 643fc9128048 'dev-one named branch' dev-one | | | o 4: 9520eea781bc 'E' | | o | 3: 32af7686d403 'D' | | o | 2: 5fddd98957c8 'C' | | o | 1: 42ccdea3bb16 'B' |/ o 0: cd010b8cd998 'A' $ hg rebase -s dev-one -d 0 --keepbranches rebasing 5:643fc9128048 "dev-one named branch" rebasing 6:24de4aff8e28 "F" rebasing 7:4b988a958030 "G" rebasing 8:31d0e4ba75e6 "H" rebasing 9:9e70cd31750f tip "dev-two named branch" saved backup bundle to $TESTTMP/a1/.hg/strip-backup/643fc9128048-c4ee9ef5-rebase.hg $ hg tglog @ 9: 59c2e59309fe 'dev-two named branch' dev-two | o 8: 904590360559 'H' | | o 7: 1a1e6f72ec38 'G' |/| o | 6: 42aa3cf0fa7a 'F' | | o | 5: bc8139ee757c 'dev-one named branch' dev-one | | | o 4: 9520eea781bc 'E' |/ | o 3: 32af7686d403 'D' | | | o 2: 5fddd98957c8 'C' | | | o 1: 42ccdea3bb16 'B' |/ o 0: cd010b8cd998 'A' $ hg update 3 3 files updated, 0 files merged, 3 files removed, 0 files unresolved $ hg branch -f dev-one marked working directory as branch dev-one $ hg ci -m 'dev-one named branch' created new head $ hg tglog @ 10: 643fc9128048 'dev-one named branch' dev-one | | o 9: 59c2e59309fe 'dev-two named branch' dev-two | | | o 8: 904590360559 'H' | | | | o 7: 1a1e6f72ec38 'G' | |/| | o | 6: 42aa3cf0fa7a 'F' | | | | o | 5: bc8139ee757c 'dev-one named branch' dev-one | | | | | o 4: 9520eea781bc 'E' | |/ o | 3: 32af7686d403 'D' | | o | 2: 5fddd98957c8 'C' | | o | 1: 42ccdea3bb16 'B' |/ o 0: cd010b8cd998 'A' $ hg rebase -b 'max(branch("dev-two"))' -d dev-one --keepbranches rebasing 5:bc8139ee757c "dev-one named branch" note: not rebasing 5:bc8139ee757c "dev-one named branch", its destination already has all its changes rebasing 6:42aa3cf0fa7a "F" rebasing 7:1a1e6f72ec38 "G" rebasing 8:904590360559 "H" rebasing 9:59c2e59309fe "dev-two named branch" saved backup bundle to $TESTTMP/a1/.hg/strip-backup/bc8139ee757c-f11c1080-rebase.hg $ hg tglog o 9: 71325f8bc082 'dev-two named branch' dev-two | o 8: 12b2bc666e20 'H' | | o 7: 549f007a9f5f 'G' |/| o | 6: 679f28760620 'F' | | @ | 5: 643fc9128048 'dev-one named branch' dev-one | | | o 4: 9520eea781bc 'E' | | o | 3: 32af7686d403 'D' | | o | 2: 5fddd98957c8 'C' | | o | 1: 42ccdea3bb16 'B' |/ o 0: cd010b8cd998 'A' $ hg rebase -s 'max(branch("dev-one"))' -d 0 --keepbranches rebasing 5:643fc9128048 "dev-one named branch" rebasing 6:679f28760620 "F" rebasing 7:549f007a9f5f "G" rebasing 8:12b2bc666e20 "H" rebasing 9:71325f8bc082 tip "dev-two named branch" saved backup bundle to $TESTTMP/a1/.hg/strip-backup/643fc9128048-6cdd1a52-rebase.hg $ hg tglog o 9: 3944801ae4ea 'dev-two named branch' dev-two | o 8: 8e279d293175 'H' | | o 7: aeefee77ab01 'G' |/| o | 6: e908b85f3729 'F' | | @ | 5: bc8139ee757c 'dev-one named branch' dev-one | | | o 4: 9520eea781bc 'E' |/ | o 3: 32af7686d403 'D' | | | o 2: 5fddd98957c8 'C' | | | o 1: 42ccdea3bb16 'B' |/ o 0: cd010b8cd998 'A' $ hg up -r 0 > /dev/null Rebasing descendant onto ancestor across different named branches $ hg rebase -s 1 -d 9 --keepbranches rebasing 1:42ccdea3bb16 "B" rebasing 2:5fddd98957c8 "C" rebasing 3:32af7686d403 "D" saved backup bundle to $TESTTMP/a1/.hg/strip-backup/42ccdea3bb16-3cb021d3-rebase.hg $ hg tglog o 9: e9f862ce8bad 'D' | o 8: a0d543090fa4 'C' | o 7: 3bdb949809d9 'B' | o 6: 3944801ae4ea 'dev-two named branch' dev-two | o 5: 8e279d293175 'H' | | o 4: aeefee77ab01 'G' |/| o | 3: e908b85f3729 'F' | | o | 2: bc8139ee757c 'dev-one named branch' dev-one | | | o 1: 9520eea781bc 'E' |/ @ 0: cd010b8cd998 'A' $ hg rebase -s 5 -d 6 abort: source and destination form a cycle [255] $ hg rebase -s 6 -d 5 rebasing 6:3944801ae4ea "dev-two named branch" rebasing 7:3bdb949809d9 "B" rebasing 8:a0d543090fa4 "C" rebasing 9:e9f862ce8bad tip "D" saved backup bundle to $TESTTMP/a1/.hg/strip-backup/3944801ae4ea-fb46ed74-rebase.hg $ hg tglog o 9: e522577ccdbd 'D' | o 8: 810110211f50 'C' | o 7: 160b0930ccc6 'B' | o 6: c57724c84928 'dev-two named branch' | o 5: 8e279d293175 'H' | | o 4: aeefee77ab01 'G' |/| o | 3: e908b85f3729 'F' | | o | 2: bc8139ee757c 'dev-one named branch' dev-one | | | o 1: 9520eea781bc 'E' |/ @ 0: cd010b8cd998 'A' Reopen branch by rebase $ hg up -qr3 $ hg branch -q b $ hg ci -m 'create b' $ hg ci -m 'close b' --close $ hg rebase -b 8 -d b reopening closed branch head 2b586e70108d rebasing 5:8e279d293175 "H" rebasing 6:c57724c84928 "dev-two named branch" rebasing 7:160b0930ccc6 "B" rebasing 8:810110211f50 "C" rebasing 9:e522577ccdbd "D" saved backup bundle to $TESTTMP/a1/.hg/strip-backup/8e279d293175-b023e27c-rebase.hg $ hg log -G -Tcompact o 11[tip] be1dea60f2a6 2011-04-30 15:24 +0200 nicdumz | D | o 10 ac34ce92632a 2011-04-30 15:24 +0200 nicdumz | C | o 9 7bd665b6ce12 2011-04-30 15:24 +0200 nicdumz | B | o 8 58e7c36e77f7 1970-01-01 00:00 +0000 test | dev-two named branch | o 7 8e5a320651f3 2011-04-30 15:24 +0200 nicdumz | H | @ 6 2b586e70108d 1970-01-01 00:00 +0000 test | close b | o 5:3 3f9d5df8a707 1970-01-01 00:00 +0000 test | create b | | o 4:3,1 aeefee77ab01 2011-04-30 15:24 +0200 nicdumz |/| G | | o | 3 e908b85f3729 2011-04-30 15:24 +0200 nicdumz | | F | | o | 2:0 bc8139ee757c 1970-01-01 00:00 +0000 test | | dev-one named branch | | | o 1 9520eea781bc 2011-04-30 15:24 +0200 nicdumz |/ E | o 0 cd010b8cd998 2011-04-30 15:24 +0200 nicdumz A $ echo A-mod > A $ hg diff diff -r 2b586e70108d A --- a/A Thu Jan 01 00:00:00 1970 +0000 +++ b/A Thu Jan 01 00:00:00 1970 +0000 @@ -1,1 +1,1 @@ -A +A-mod --dry-run doesn't affect a dirty working directory that is unrelated to the source or destination. $ hg rebase -s tip -d 4 --dry-run starting dry-run rebase; repository will not be changed rebasing 11:be1dea60f2a6 tip "D" dry-run rebase completed successfully; run without -n/--dry-run to perform this rebase $ hg diff diff -r 2b586e70108d A --- a/A Thu Jan 01 00:00:00 1970 +0000 +++ b/A Thu Jan 01 00:00:00 1970 +0000 @@ -1,1 +1,1 @@ -A +A-mod Bailing out on --confirm doesn't affect a dirty working directory that is unrelated to the source or destination. $ echo A-mod > A $ echo n | hg rebase -s tip -d 4 --confirm --config ui.interactive=True starting in-memory rebase rebasing 11:be1dea60f2a6 tip "D" rebase completed successfully apply changes (yn)? n $ hg diff diff -r 2b586e70108d A --- a/A Thu Jan 01 00:00:00 1970 +0000 +++ b/A Thu Jan 01 00:00:00 1970 +0000 @@ -1,1 +1,1 @@ -A +A-mod $ echo A-mod > A $ hg rebase -s tip -d 4 --confirm starting in-memory rebase rebasing 11:be1dea60f2a6 tip "D" rebase completed successfully apply changes (yn)? y saved backup bundle to $TESTTMP/a1/.hg/strip-backup/be1dea60f2a6-ca6d2dac-rebase.hg $ hg diff diff -r 2b586e70108d A --- a/A Thu Jan 01 00:00:00 1970 +0000 +++ b/A Thu Jan 01 00:00:00 1970 +0000 @@ -1,1 +1,1 @@ -A +A-mod Attempting to rebase the parent of a dirty working directory will abort, without mangling the working directory... $ hg rebase -s 5 -d 4 --dry-run starting dry-run rebase; repository will not be changed abort: uncommitted changes [20] $ hg diff diff -r 2b586e70108d A --- a/A Thu Jan 01 00:00:00 1970 +0000 +++ b/A Thu Jan 01 00:00:00 1970 +0000 @@ -1,1 +1,1 @@ -A +A-mod ... ditto for --confirm $ echo n | hg rebase -s 5 -d 4 --confirm --config ui.interactive=True starting in-memory rebase abort: uncommitted changes [20] $ hg diff diff -r 2b586e70108d A --- a/A Thu Jan 01 00:00:00 1970 +0000 +++ b/A Thu Jan 01 00:00:00 1970 +0000 @@ -1,1 +1,1 @@ -A +A-mod $ hg rebase -s 5 -d 4 --confirm starting in-memory rebase abort: uncommitted changes [20] $ hg diff diff -r 2b586e70108d A --- a/A Thu Jan 01 00:00:00 1970 +0000 +++ b/A Thu Jan 01 00:00:00 1970 +0000 @@ -1,1 +1,1 @@ -A +A-mod $ cd .. Rebase to other head on branch Set up a case: $ hg init case1 $ cd case1 $ touch f $ hg ci -qAm0 $ hg branch -q b $ echo >> f $ hg ci -qAm 'b1' $ hg up -qr -2 $ hg branch -qf b $ hg ci -qm 'b2' $ hg up -qr -3 $ hg branch -q c $ hg ci -m 'c1' $ hg tglog @ 3: c062e3ecd6c6 'c1' c | | o 2: 792845bb77ee 'b2' b |/ | o 1: 40039acb7ca5 'b1' b |/ o 0: d681519c3ea7 '0' $ hg clone -q . ../case2 rebase 'b2' to another lower branch head $ hg up -qr 2 $ hg rebase rebasing 2:792845bb77ee "b2" note: not rebasing 2:792845bb77ee "b2", its destination already has all its changes saved backup bundle to $TESTTMP/case1/.hg/strip-backup/792845bb77ee-627120ee-rebase.hg $ hg tglog o 2: c062e3ecd6c6 'c1' c | | @ 1: 40039acb7ca5 'b1' b |/ o 0: d681519c3ea7 '0' rebase 'b1' on top of the tip of the branch ('b2') - ignoring the tip branch ('c1') $ cd ../case2 $ hg up -qr 1 $ hg rebase rebasing 1:40039acb7ca5 "b1" saved backup bundle to $TESTTMP/case2/.hg/strip-backup/40039acb7ca5-342b72d1-rebase.hg $ hg tglog @ 3: 76abc1c6f8c7 'b1' b | | o 2: c062e3ecd6c6 'c1' c | | o | 1: 792845bb77ee 'b2' b |/ o 0: d681519c3ea7 '0' rebase 'c1' to the branch head 'c2' that is closed $ hg branch -qf c $ hg ci -qm 'c2 closed' --close $ hg up -qr 2 $ hg tglog _ 4: 8427af5d86f2 'c2 closed' c | o 3: 76abc1c6f8c7 'b1' b | | @ 2: c062e3ecd6c6 'c1' c | | o | 1: 792845bb77ee 'b2' b |/ o 0: d681519c3ea7 '0' $ hg rebase abort: branch 'c' has one head - please rebase to an explicit rev (run 'hg heads' to see all heads, specify destination with -d) [255] $ hg tglog _ 4: 8427af5d86f2 'c2 closed' c | o 3: 76abc1c6f8c7 'b1' b | | @ 2: c062e3ecd6c6 'c1' c | | o | 1: 792845bb77ee 'b2' b |/ o 0: d681519c3ea7 '0' $ hg up -cr 1 0 files updated, 0 files merged, 0 files removed, 0 files unresolved $ hg branch x marked working directory as branch x $ hg rebase -r 3:: -d . rebasing 3:76abc1c6f8c7 "b1" rebasing 4:8427af5d86f2 tip "c2 closed" note: not rebasing 4:8427af5d86f2 tip "c2 closed", its destination already has all its changes saved backup bundle to $TESTTMP/case2/.hg/strip-backup/76abc1c6f8c7-cd698d13-rebase.hg $ hg tglog o 3: 117b0ed08075 'b1' x | | o 2: c062e3ecd6c6 'c1' c | | @ | 1: 792845bb77ee 'b2' b |/ o 0: d681519c3ea7 '0' $ cd ..