# HG changeset patch # User Pierre-Yves David # Date 2023-12-02 01:13:23 # Node ID 3e2a878fb96f5d3d16214f0e4925bd5cadbd6e43 # Parent e9eac01c57f3872cff1376631ce66844a625f4cd censor: fix things around inlining The temporary revlog cannot go through the inline → split process as this would break at transaction commit. (that might be fixable, but lets keep things simple for now). We introduce a cleaner way to enforce this as the previous one was broken in 6.6 On the way we remove multiple weird, fragile and broken overwrite of revlog attributes and we focus on passing the configuration across. We also had to update the test to actually create a non-inline revlog. diff --git a/mercurial/revlogutils/rewrite.py b/mercurial/revlogutils/rewrite.py --- a/mercurial/revlogutils/rewrite.py +++ b/mercurial/revlogutils/rewrite.py @@ -72,11 +72,16 @@ def v1_censor(rl, tr, censornode, tombst radix=rl.radix, postfix=b'tmpcensored', censorable=True, + data_config=rl.data_config, + delta_config=rl.delta_config, + feature_config=rl.feature_config, + may_inline=rl._inline, ) - newrl._format_version = rl._format_version - newrl._format_flags = rl._format_flags - newrl.delta_config.general_delta = rl.delta_config.general_delta - newrl._parse_index = rl._parse_index + # inline splitting will prepare some transaction work that will get + # confused by the final file move. So if there is a risk of not being + # inline at the end, we prevent the new revlog to be inline in the first + # place. + assert not (newrl._inline and not rl._inline) for rev in rl.revs(): node = rl.node(rev) @@ -122,7 +127,10 @@ def v1_censor(rl, tr, censornode, tombst tr.addbackup(rl._datafile, location=b'store') rl.opener.rename(newrl._indexfile, rl._indexfile) - if not rl._inline: + if newrl._inline: + assert rl._inline + else: + assert not rl._inline rl.opener.rename(newrl._datafile, rl._datafile) rl.clearcaches() diff --git a/tests/test-censor.t b/tests/test-censor.t --- a/tests/test-censor.t +++ b/tests/test-censor.t @@ -298,12 +298,9 @@ Can censor enough revision to move back $ hg debugrevlogstats | grep target rev-count data-size inl type target - 8 ????????? no file target (glob) (revlogv2 !) - 8 ????????? yes file target (glob) (revlogv1 !) - $ for x in `"$PYTHON" $TESTDIR/seq.py 0 50000` - > do - > echo "Password: hunter$x" >> target - > done + 8 ??? no file target (glob) (revlogv2 !) + 8 ??? yes file target (glob) (revlogv1 !) + $ cat /dev/rand?m | dd status=none count=200 | f --hexdump > target $ hg ci -m 'add 100k passwords' $ H2=`hg id --debug -i` $ C5=$H2 @@ -312,17 +309,25 @@ Can censor enough revision to move back $ H2=`hg id --debug -i` $ hg debugrevlogstats | grep target rev-count data-size inl type target - 10 ????????? no file target (glob) (revlogv2 !) - 10 ????????? no file target (glob) (missing-correct-output revlogv1 !) - 10 ????????? yes file target (glob) (known-bad-output revlogv1 !) + 10 ?????? no file target (glob) $ hg --config extensions.censor= censor -r $C5 target + +The important part is for the censor operation to not crash and the repository +to not be corrupted. Right now this involve keeping the revlog split. + $ hg debugrevlogstats | grep target rev-count data-size inl type target - 10 ????????? no file target (glob) (revlogv2 !) - 10 ????????? yes file target (glob) (revlogv1 !) + 10 ??? no file target (glob) $ hg cat -r $C5 target | head -n 10 $ hg cat -r $H2 target | head -n 10 fresh start + $ hg verify + checking changesets + checking manifests + crosschecking files in changesets and manifests + checking files + checking dirstate + checked 12 changesets with 13 changes to 2 files Repo with censored nodes can be cloned and cloned nodes are censored @@ -354,7 +359,7 @@ Repo cloned before tainted content intro adding manifests adding file changes added 11 changesets with 11 changes to 2 files (+1 heads) - new changesets 186fb27560c3:683e4645fded + new changesets * (glob) (run 'hg heads' to see heads, 'hg merge' to merge) $ hg update 4 2 files updated, 0 files merged, 0 files removed, 0 files unresolved @@ -414,7 +419,7 @@ Censored nodes can be bundled up and unb adding manifests adding file changes added 2 changesets with 2 changes to 2 files (+1 heads) - new changesets 075be80ac777:dcbaf17bf3a1 (2 drafts) + new changesets * (glob) (run 'hg heads .' to see heads, 'hg merge' to merge) $ hg cat -r $REV target | head -n 10 $ hg cat -r $CLEANREV target | head -n 10 @@ -471,7 +476,7 @@ Censored nodes can be imported on top of adding manifests adding file changes added 6 changesets with 5 changes to 2 files (+1 heads) - new changesets efbe78065929:683e4645fded (6 drafts) + new changesets * (glob) (run 'hg heads .' to see heads, 'hg merge' to merge) $ hg update $H2 2 files updated, 0 files merged, 0 files removed, 0 files unresolved