Mercurial > hg
view tests/test-casefolding.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 | 803b7569c9ea |
children | 7ce8b4d2bd55 |
line wrap: on
line source
#require icasefs $ hg debugfs | grep 'case-sensitive:' case-sensitive: no test file addition with bad case $ hg init repo1 $ cd repo1 $ echo a > a $ hg add A $ hg st A a $ hg ci -m adda $ hg manifest a $ cd .. test case collision on rename (issue750) $ hg init repo2 $ cd repo2 $ echo a > a $ hg --debug ci -Am adda adding a committing files: a committing manifest committing changelog updating the branch cache committed changeset 0:07f4944404050f47db2e5c5071e0e84e7a27bba9 Case-changing renames should work: $ hg mv a A $ hg mv A a $ hg st addremove after case-changing rename has no effect (issue4590) $ hg mv a A $ hg addremove recording removal of a as rename to A (100% similar) $ hg revert --all forgetting A undeleting a test changing case of path components $ mkdir D $ echo b > D/b $ hg ci -Am addb D/b $ hg mv D/b d/b D/b: not overwriting - file already committed ('hg rename --force' to replace the file by recording a rename) [1] $ hg mv D/b d/c $ hg st A D/c R D/b $ mv D temp $ mv temp d $ hg st A D/c R D/b $ hg revert -aq $ rm d/c $ echo c > D/c $ hg add D/c $ hg st A D/c $ hg ci -m addc D/c $ hg mv d/b d/e $ hg st A D/e R D/b $ hg revert -aq $ rm d/e $ hg mv d/b D/B $ hg st A D/B R D/b $ cd .. test case collision between revisions (issue912) $ hg init repo3 $ cd repo3 $ echo a > a $ hg ci -Am adda adding a $ hg rm a $ hg ci -Am removea $ echo A > A on linux hfs keeps the old case stored, force it $ mv a aa $ mv aa A $ hg ci -Am addA adding A used to fail under case insensitive fs $ hg up -C 0 1 files updated, 0 files merged, 1 files removed, 0 files unresolved $ hg up -C 1 files updated, 0 files merged, 1 files removed, 0 files unresolved no clobbering of untracked files with wrong casing $ hg up -r null 0 files updated, 0 files merged, 1 files removed, 0 files unresolved $ echo gold > a $ hg up A: untracked file differs abort: untracked files in working directory differ from files in requested revision [255] $ cat a gold $ rm a test that normal file in different case on target context is not unlinked by largefiles extension. $ cat >> .hg/hgrc <<EOF > [extensions] > largefiles= > EOF $ hg update -q -C 1 $ hg status -A $ echo 'A as largefiles' > A $ hg add --large A $ hg commit -m '#3' created new head $ hg manifest -r 3 .hglf/A $ hg manifest -r 0 a $ hg update -q -C 0 $ hg status -A C a $ hg update -q -C 3 $ hg update -q 0 $ hg up -C -r 2 1 files updated, 0 files merged, 0 files removed, 0 files unresolved $ hg mv A a $ hg diff -g > rename.diff $ hg ci -m 'A -> a' $ hg up -q '.^' $ hg import rename.diff -m "import rename A -> a" applying rename.diff $ hg st ? rename.diff $ hg files a $ find * | sort a rename.diff $ rm rename.diff $ cd .. issue 3342: file in nested directory causes unexpected abort $ hg init issue3342 $ cd issue3342 $ mkdir -p a/B/c/D $ echo e > a/B/c/D/e $ hg add a/B/c/D/e $ hg ci -m 'add e' issue 4481: revert across case only renames $ hg mv a/B/c/D/e a/B/c/d/E $ hg ci -m "uppercase E" $ echo 'foo' > a/B/c/D/E $ hg ci -m 'e content change' $ hg revert --all -r 0 removing a/B/c/D/E adding a/B/c/D/e $ find * | sort a a/B a/B/c a/B/c/D a/B/c/D/e a/B/c/D/e.orig $ cd .. issue 3340: mq does not handle case changes correctly in addition to reported case, 'hg qrefresh' is also tested against case changes. $ echo "[extensions]" >> $HGRCPATH $ echo "mq=" >> $HGRCPATH $ hg init issue3340 $ cd issue3340 $ echo a > mIxEdCaSe $ hg add mIxEdCaSe $ hg commit -m '#0' $ hg rename mIxEdCaSe tmp $ hg rename tmp MiXeDcAsE $ hg status -A A MiXeDcAsE mIxEdCaSe R mIxEdCaSe $ hg qnew changecase $ hg status -A C MiXeDcAsE $ hg qpop -a popping changecase patch queue now empty $ hg qnew refresh-casechange $ hg status -A C mIxEdCaSe $ hg rename mIxEdCaSe tmp $ hg rename tmp MiXeDcAsE $ hg status -A A MiXeDcAsE mIxEdCaSe R mIxEdCaSe $ hg qrefresh $ hg status -A C MiXeDcAsE $ hg qpop -a popping refresh-casechange patch queue now empty $ hg qnew refresh-pattern $ hg status $ echo A > A $ hg add adding A $ hg qrefresh a # issue 3271, qrefresh with file handled case wrong $ hg status # empty status means the qrefresh worked #if osx We assume anyone running the tests on a case-insensitive volume on OS X will be using HFS+. If that's not true, this test will fail. $ rm A >>> open(u'a\u200c'.encode('utf-8'), 'w').write('unicode is fun') and None $ hg status M A #endif $ cd ..