view tests/test-paths.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

  $ hg init a
  $ hg clone a b
  updating to branch default
  0 files updated, 0 files merged, 0 files removed, 0 files unresolved
  $ cd a

with no paths:

  $ hg paths
  $ hg paths unknown
  not found!
  [1]
  $ hg paths -Tjson
  [
  ]

with paths:

  $ echo '[paths]' >> .hg/hgrc
  $ echo 'dupe = ../b#tip' >> .hg/hgrc
  $ echo 'expand = $SOMETHING/bar' >> .hg/hgrc
  $ hg in dupe
  comparing with $TESTTMP/b
  no changes found
  [1]
  $ cd ..
  $ hg -R a in dupe
  comparing with $TESTTMP/b
  no changes found
  [1]
  $ cd a
  $ hg paths
  dupe = $TESTTMP/b#tip
  expand = $TESTTMP/a/$SOMETHING/bar
  $ SOMETHING=foo hg paths
  dupe = $TESTTMP/b#tip
  expand = $TESTTMP/a/foo/bar
#if msys
  $ SOMETHING=//foo hg paths
  dupe = $TESTTMP/b#tip
  expand = /foo/bar
#else
  $ SOMETHING=/foo hg paths
  dupe = $TESTTMP/b#tip
  expand = /foo/bar
#endif
  $ hg paths -q
  dupe
  expand
  $ hg paths dupe
  $TESTTMP/b#tip
  $ hg paths -q dupe
  $ hg paths unknown
  not found!
  [1]
  $ hg paths -q unknown
  [1]

formatter output with paths:

  $ echo 'dupe:pushurl = https://example.com/dupe' >> .hg/hgrc
  $ hg paths -Tjson | sed 's|\\\\|\\|g'
  [
   {
    "name": "dupe",
    "pushurl": "https://example.com/dupe",
    "url": "$TESTTMP/b#tip"
   },
   {
    "name": "expand",
    "url": "$TESTTMP/a/$SOMETHING/bar"
   }
  ]
  $ hg paths -Tjson dupe | sed 's|\\\\|\\|g'
  [
   {
    "name": "dupe",
    "pushurl": "https://example.com/dupe",
    "url": "$TESTTMP/b#tip"
   }
  ]
  $ hg paths -Tjson -q unknown
  [
  ]
  [1]

log template:

 (behaves as a {name: path-string} dict by default)

  $ hg log -rnull -T '{peerurls}\n'
  dupe=$TESTTMP/b#tip expand=$TESTTMP/a/$SOMETHING/bar
  $ hg log -rnull -T '{join(peerurls, "\n")}\n'
  dupe=$TESTTMP/b#tip
  expand=$TESTTMP/a/$SOMETHING/bar
  $ hg log -rnull -T '{peerurls % "{name}: {url}\n"}'
  dupe: $TESTTMP/b#tip
  expand: $TESTTMP/a/$SOMETHING/bar
  $ hg log -rnull -T '{get(peerurls, "dupe")}\n'
  $TESTTMP/b#tip

 (sub options can be populated by map/dot operation)

  $ hg log -rnull \
  > -T '{get(peerurls, "dupe") % "url: {url}\npushurl: {pushurl}\n"}'
  url: $TESTTMP/b#tip
  pushurl: https://example.com/dupe
  $ hg log -rnull -T '{peerurls.dupe.pushurl}\n'
  https://example.com/dupe

 (in JSON, it's a dict of urls)

  $ hg log -rnull -T '{peerurls|json}\n' | sed 's|\\\\|/|g'
  {"dupe": "$TESTTMP/b#tip", "expand": "$TESTTMP/a/$SOMETHING/bar"}

password should be masked in plain output, but not in machine-readable/template
output:

  $ echo 'insecure = http://foo:[e-mail]/' >> .hg/hgrc
  $ hg paths insecure
  http://foo:***@example.com/
  $ hg paths -Tjson insecure
  [
   {
    "name": "insecure",
    "url": "http://foo:[e-mail]/"
   }
  ]
  $ hg log -rnull -T '{get(peerurls, "insecure")}\n'
  http://foo:[e-mail]/

zeroconf wraps ui.configitems(), which shouldn't crash at least:

  $ hg paths --config extensions.zeroconf=
  dupe = $TESTTMP/b#tip
  dupe:pushurl = https://example.com/dupe
  expand = $TESTTMP/a/$SOMETHING/bar
  insecure = http://foo:***@example.com/

  $ cd ..

sub-options for an undeclared path are ignored

  $ hg init suboptions
  $ cd suboptions

  $ cat > .hg/hgrc << EOF
  > [paths]
  > path0 = https://example.com/path0
  > path1:pushurl = https://example.com/path1
  > EOF
  $ hg paths
  path0 = https://example.com/path0

unknown sub-options aren't displayed

  $ cat > .hg/hgrc << EOF
  > [paths]
  > path0 = https://example.com/path0
  > path0:foo = https://example.com/path1
  > EOF

  $ hg paths
  path0 = https://example.com/path0

:pushurl must be a URL

  $ cat > .hg/hgrc << EOF
  > [paths]
  > default = /path/to/nothing
  > default:pushurl = /not/a/url
  > EOF

  $ hg paths
  (paths.default:pushurl not a URL; ignoring)
  default = /path/to/nothing

#fragment is not allowed in :pushurl

  $ cat > .hg/hgrc << EOF
  > [paths]
  > default = https://example.com/repo
  > invalid = https://example.com/repo
  > invalid:pushurl = https://example.com/repo#branch
  > EOF

  $ hg paths
  ("#fragment" in paths.invalid:pushurl not supported; ignoring)
  default = https://example.com/repo
  invalid = https://example.com/repo
  invalid:pushurl = https://example.com/repo

  $ cd ..

'file:' disables [paths] entries for clone destination

  $ cat >> $HGRCPATH <<EOF
  > [paths]
  > gpath1 = http://hg.example.com
  > EOF

  $ hg clone a gpath1
  abort: cannot create new http repository
  [255]

  $ hg clone a file:gpath1
  updating to branch default
  0 files updated, 0 files merged, 0 files removed, 0 files unresolved
  $ cd gpath1
  $ hg -q id
  000000000000

  $ cd ..