From: "Ævar Arnfjörð Bjarmason" <avarab@gmail.com>
To: phillip.wood@dunelm.org.uk
Cc: git@vger.kernel.org, Junio C Hamano <gitster@pobox.com>,
Phillip Wood <phillip.wood123@gmail.com>,
Jeff King <peff@peff.net>,
Derrick Stolee <derrickstolee@github.com>,
Elijah Newren <newren@gmail.com>
Subject: Re: [PATCH 10/17] sequencer.c: fix "opts->strategy" leak in read_strategy_opts()
Date: Tue, 08 Nov 2022 16:26:37 +0100 [thread overview]
Message-ID: <221108.86zgd1qxac.gmgdl@evledraar.gmail.com> (raw)
In-Reply-To: <66835bfb-2815-4321-9d06-52f85a0c72f9@dunelm.org.uk>
On Tue, Nov 08 2022, Phillip Wood wrote:
> On 04/11/2022 14:50, Phillip Wood wrote:
>> On 03/11/2022 17:06, Ævar Arnfjörð Bjarmason wrote:
>>> When "read_strategy_opts()" is called we may have populated the
>>> "opts->strategy" before, so we'll need to free() it to avoid leaking
>>> memory.
>> Where is the previous value coming from? I guess it may be the
>> config but otherwise I'm confused.
>
> Having looked a bit more at this I think this is the wrong fix. The
> reason we're overwriting the existing value is that we're reading some
> of the state files twice. I think the only way to get to this point is
> to go through a code path that calls rebase.c:read_basic_state() which
> already populates these options via a later call to
> rebase.c:get_replay_opts(). I think the correct fixes looks something
> like the diff below.
>
> I have also looked at the cherry-pick/revert case and I think that
> we're leaking opts->strategy (and probably some others) when running
>
> git cherry-pick --continue
>
> after
>
> git -c pull.twohead=recursive cherry-pick -s ort <some commits>
Related: My just-sent [1].
For this one though, I have a v2 re-roll prepared, in which I explained
why the leak is happening (I'll send this soon to the list, but for
now): https://github.com/avar/git/commit/243ab74120b
> I'm not sure what your strategy has been with the fixes in this series
> but we're never going to have 100% coverage of all the option
> combinations for rebase & cherry-pick so I think it is helpful to
> treat these LSAN reports as a starting point for looking into why the
> leak is occurring and also look for similar leaks.
For this series I'm aiming to solve some common leaks, and increase our
test coverage.
Per [1] the destruction in this area is quite messy, and we should do it
better. But that's a much larger fix. I think in the meantime adding
this free() is the bets way to make incremental progress here, and (an
the commit linked above notes) it's consistent with existing patterns.
1. https://lore.kernel.org/git/221108.864jv9sc9r.gmgdl@evledraar.gmail.com/
next prev parent reply other threads:[~2022-11-08 15:30 UTC|newest]
Thread overview: 45+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-11-03 17:05 [PATCH 00/17] leak fixes: use existing constructors & other trivia Ævar Arnfjörð Bjarmason
2022-11-03 17:06 ` [PATCH 01/17] tests: mark tests as passing with SANITIZE=leak Ævar Arnfjörð Bjarmason
2022-11-03 17:06 ` [PATCH 02/17] {reset,merge}: call discard_index() before returning Ævar Arnfjörð Bjarmason
2022-11-03 17:06 ` [PATCH 03/17] commit: discard partial cache before (re-)reading it Ævar Arnfjörð Bjarmason
2022-11-03 17:06 ` [PATCH 04/17] read-cache.c: clear and free "sparse_checkout_patterns" Ævar Arnfjörð Bjarmason
2022-11-03 17:06 ` [PATCH 05/17] dir.c: free "ident" and "exclude_per_dir" in "struct untracked_cache" Ævar Arnfjörð Bjarmason
2022-11-03 17:06 ` [PATCH 06/17] built-ins & libs & helpers: add/move destructors, fix leaks Ævar Arnfjörð Bjarmason
2022-11-03 17:06 ` [PATCH 07/17] unpack-file: fix ancient leak in create_temp_file() Ævar Arnfjörð Bjarmason
2022-11-03 17:06 ` [PATCH 08/17] revision API: call graph_clear() in release_revisions() Ævar Arnfjörð Bjarmason
2022-11-03 17:06 ` [PATCH 09/17] ls-files: fix a --with-tree memory leak Ævar Arnfjörð Bjarmason
2022-11-03 17:06 ` [PATCH 10/17] sequencer.c: fix "opts->strategy" leak in read_strategy_opts() Ævar Arnfjörð Bjarmason
2022-11-04 14:50 ` Phillip Wood
2022-11-04 21:44 ` Taylor Blau
2022-11-05 12:43 ` Ævar Arnfjörð Bjarmason
2022-11-08 15:00 ` Phillip Wood
2022-11-08 15:26 ` Ævar Arnfjörð Bjarmason [this message]
2022-11-03 17:06 ` [PATCH 11/17] connected.c: free the "struct packed_git" Ævar Arnfjörð Bjarmason
2022-11-03 17:06 ` [PATCH 12/17] sequencer.c: fix a pick_commits() leak Ævar Arnfjörð Bjarmason
2022-11-03 17:06 ` [PATCH 13/17] rebase: don't leak on "--abort" Ævar Arnfjörð Bjarmason
2022-11-04 14:42 ` Phillip Wood
2022-11-05 12:01 ` Ævar Arnfjörð Bjarmason
2022-11-03 17:06 ` [PATCH 14/17] sequencer.c: fix sequencer_continue() leak Ævar Arnfjörð Bjarmason
2022-11-03 17:06 ` [PATCH 15/17] cherry-pick: free "struct replay_opts" members Ævar Arnfjörð Bjarmason
2022-11-03 17:06 ` [PATCH 16/17] revert: fix parse_options_concat() leak Ævar Arnfjörð Bjarmason
2022-11-03 17:06 ` [PATCH 17/17] built-ins: use free() not UNLEAK() if trivial, rm dead code Ævar Arnfjörð Bjarmason
2022-11-04 15:20 ` [PATCH 00/17] leak fixes: use existing constructors & other trivia Phillip Wood
2022-11-05 12:46 ` Ævar Arnfjörð Bjarmason
2022-11-07 9:46 ` Phillip Wood
2022-11-08 18:17 ` [PATCH v2 00/15] " Ævar Arnfjörð Bjarmason
2022-11-08 18:17 ` [PATCH v2 01/15] tests: mark tests as passing with SANITIZE=leak Ævar Arnfjörð Bjarmason
2022-11-08 18:17 ` [PATCH v2 02/15] {reset,merge}: call discard_index() before returning Ævar Arnfjörð Bjarmason
2022-11-08 18:17 ` [PATCH v2 03/15] commit: discard partial cache before (re-)reading it Ævar Arnfjörð Bjarmason
2022-11-08 18:17 ` [PATCH v2 04/15] read-cache.c: clear and free "sparse_checkout_patterns" Ævar Arnfjörð Bjarmason
2022-11-08 18:17 ` [PATCH v2 05/15] dir.c: free "ident" and "exclude_per_dir" in "struct untracked_cache" Ævar Arnfjörð Bjarmason
2022-11-08 18:17 ` [PATCH v2 06/15] built-ins & libs & helpers: add/move destructors, fix leaks Ævar Arnfjörð Bjarmason
2022-11-08 18:17 ` [PATCH v2 07/15] unpack-file: fix ancient leak in create_temp_file() Ævar Arnfjörð Bjarmason
2022-11-08 18:17 ` [PATCH v2 08/15] revision API: call graph_clear() in release_revisions() Ævar Arnfjörð Bjarmason
2022-11-08 18:17 ` [PATCH v2 09/15] ls-files: fix a --with-tree memory leak Ævar Arnfjörð Bjarmason
2022-11-08 18:17 ` [PATCH v2 10/15] sequencer.c: fix "opts->strategy" leak in read_strategy_opts() Ævar Arnfjörð Bjarmason
2022-11-08 18:17 ` [PATCH v2 11/15] connected.c: free the "struct packed_git" Ævar Arnfjörð Bjarmason
2022-11-08 18:17 ` [PATCH v2 12/15] rebase: don't leak on "--abort" Ævar Arnfjörð Bjarmason
2022-11-08 18:17 ` [PATCH v2 13/15] cherry-pick: free "struct replay_opts" members Ævar Arnfjörð Bjarmason
2022-11-08 18:17 ` [PATCH v2 14/15] revert: fix parse_options_concat() leak Ævar Arnfjörð Bjarmason
2022-11-08 18:17 ` [PATCH v2 15/15] built-ins: use free() not UNLEAK() if trivial, rm dead code Ævar Arnfjörð Bjarmason
2022-11-08 20:54 ` [PATCH v2 00/15] leak fixes: use existing constructors & other trivia Taylor Blau
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=221108.86zgd1qxac.gmgdl@evledraar.gmail.com \
--to=avarab@gmail.com \
--cc=derrickstolee@github.com \
--cc=git@vger.kernel.org \
--cc=gitster@pobox.com \
--cc=newren@gmail.com \
--cc=peff@peff.net \
--cc=phillip.wood123@gmail.com \
--cc=phillip.wood@dunelm.org.uk \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).