view tests/test-update-atomic.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

#require execbit unix-permissions no-chg

Checking that experimental.atomic-file works.

  $ cat > $TESTTMP/show_mode.py <<EOF
  > from __future__ import print_function
  > import os
  > import stat
  > import sys
  > ST_MODE = stat.ST_MODE
  > 
  > for file_path in sys.argv[1:]:
  >     file_stat = os.stat(file_path)
  >     octal_mode = oct(file_stat[ST_MODE] & 0o777).replace('o', '')
  >     print("%s:%s" % (file_path, octal_mode))
  > 
  > EOF

  $ hg init repo
  $ cd repo

  $ cat > .hg/showwrites.py <<EOF
  > from __future__ import print_function
  > from mercurial import pycompat
  > from mercurial.utils import stringutil
  > def uisetup(ui):
  >   from mercurial import vfs
  >   class newvfs(vfs.vfs):
  >     def __call__(self, *args, **kwargs):
  >       print(pycompat.sysstr(stringutil.pprint(
  >           ('vfs open', args, sorted(list(kwargs.items()))))))
  >       return super(newvfs, self).__call__(*args, **kwargs)
  >   vfs.vfs = newvfs
  > EOF

  $ for v in a1 a2 b1 b2 c ro; do echo $v > $v; done
  $ chmod +x b*
  $ hg commit -Aqm _

# We check that
# - the changes are actually atomic
# - that permissions are correct (all 4 cases of (executable before) * (executable after))
# - that renames work, though they should be atomic anyway
# - that it works when source files are read-only (but directories are read-write still)

  $ for v in a1 a2 b1 b2 ro; do echo changed-$v > $v; done
  $ chmod -x *1; chmod +x *2
  $ hg rename c d
  $ hg commit -qm _

Check behavior without update.atomic-file

  $ hg update -r 0 -q
  $ hg update -r 1 --config extensions.showwrites=.hg/showwrites.py 2>&1 | grep "a1'.*wb"
  ('vfs open', ('a1', 'wb'), [('atomictemp', False), ('backgroundclose', True)])

  $ $PYTHON $TESTTMP/show_mode.py *
  a1:0644
  a2:0755
  b1:0644
  b2:0755
  d:0644
  ro:0644

Add a second revision for the ro file so we can test update when the file is
present or not

  $ echo "ro" > ro

  $ hg commit -qm _

Check behavior without update.atomic-file first

  $ hg update -C -r 0 -q

  $ hg update -r 1
  6 files updated, 0 files merged, 1 files removed, 0 files unresolved

  $ $PYTHON $TESTTMP/show_mode.py *
  a1:0644
  a2:0755
  b1:0644
  b2:0755
  d:0644
  ro:0644

Manually reset the mode of the read-only file

  $ chmod a-w ro

  $ $PYTHON $TESTTMP/show_mode.py ro
  ro:0444

Now the file is present, try to update and check the permissions of the file

  $ hg up -r 2
  1 files updated, 0 files merged, 0 files removed, 0 files unresolved

  $ $PYTHON $TESTTMP/show_mode.py ro
  ro:0644

# The file which was read-only is now writable in the default behavior

Check behavior with update.atomic-files


  $ cat >> .hg/hgrc <<EOF
  > [experimental]
  > update.atomic-file = true
  > EOF

  $ hg update -C -r 0 -q
  $ hg update -r 1 --config extensions.showwrites=.hg/showwrites.py 2>&1 | grep "a1'.*wb"
  ('vfs open', ('a1', 'wb'), [('atomictemp', True), ('backgroundclose', True)])
  $ hg st -A --rev 1
  C a1
  C a2
  C b1
  C b2
  C d
  C ro

Check the file permission after update
  $ $PYTHON $TESTTMP/show_mode.py *
  a1:0644
  a2:0755
  b1:0644
  b2:0755
  d:0644
  ro:0644

Manually reset the mode of the read-only file

  $ chmod a-w ro

  $ $PYTHON $TESTTMP/show_mode.py ro
  ro:0444

Now the file is present, try to update and check the permissions of the file

  $ hg update -r 2 --traceback
  1 files updated, 0 files merged, 0 files removed, 0 files unresolved

  $ $PYTHON $TESTTMP/show_mode.py ro
  ro:0644

# The behavior is the same as without atomic update