Re: [PATCH v2] refs/files: skip lock files during consistency checks
- From
Karthik Nayak <karthik.188@gmail.com>
- Date
- Apr 22, 2026, 12:04 UTC
- Message-ID
- <CAOLa=ZT1zE+MLeaYE_5jWmNzSvtTTBw3ZAopai+2Ei27kmYm2g@mail.gmail.com>
- In-Reply-To
- <aeil2Q_Mh_fKCwGa@pks.im>
Patrick Steinhardt <ps@pks.im> writes:
Show 14 quoted lines
> On Wed, Apr 22, 2026 at 11:49:58AM +0200, Karthik Nayak wrote: >> Consistency checks in the files reference backend involve two steps: >> >> 1. Iterate over all entries within the 'refs/' directory and call >> `files_fsck_ref()` on each. >> 2. Iterate over all root refs via `for_each_root_ref()` and call >> `files_fsck_ref()` on each. >> >> `files_fsck_ref()` then runs all fsck checks defined in >> `fsck_refs_fn[]`. Step 2 goes through the refs API and only sees valid >> refs, but step 1 iterates the directory directly and will also encounter > > Nit, obviously not worth a reroll: maybe do s/will/may/? >
I thought will would go with 'intermediate' better, but 'may' is the right choice I guess. Will add it in locally.
Show 32 quoted lines
>> intermediate '*.lock' files. >> >> Currently, `files_fsck_refs_name()`, one of the functions in >> `fsck_refs_fn[]`, filters out lock files itself. The other function, >> `files_fsck_refs_content()`, has no such check and would parse the lock >> file. Any new function added to `fsck_refs_fn[]` would have the same >> problem. >> >> Move the filter up into `files_fsck_refs_dir()`, where the directory >> iteration happens. Since step 2 cannot produce lock files, this is the >> only site where the filter is needed, and individual checks no longer >> have to re-implement it. > > Makes sense. > >> diff --git a/refs/files-backend.c b/refs/files-backend.c >> index b3b0c25f84..1504a1e2f3 100644 >> --- a/refs/files-backend.c >> +++ b/refs/files-backend.c >> @@ -3962,6 +3953,15 @@ static int files_fsck_refs_dir(struct ref_store *ref_store, >> strbuf_addf(&refname, "worktrees/%s/", wt->id); >> strbuf_addf(&refname, "refs/%s", iter->relative_path); >> >> + filename = basename((char *) iter->path.buf); > > Not a new issue, but this cast made me wonder. As it turns out, > basename(3p) is documented as "may modify the string pointed to by > path". I assume that this can happen if the path itself ends with a > slash for example, as in that case the basename should of course not > include the slash itself. So maybe it modifies the caller-provided path > directly in that case? >
I guess it depends on the implementation, the glibc for example doesn't seem to [1].
Show 5 quoted lines
> In any case, it shouldn't be much of an issue as we only use this on > discovered path names, and those cannot contain contain a trailing > slash. > > Patrick
Yeah we should be fine here. Thanks for the review. I'll avoid re-rolling for now and see if there are other changes needed.
[1]: https://sourceware.org/git/?p=glibc.git;a=blob;f=string/basename.c;h=1658ba98d3ff89e8257b36219599184866798d0d;hb=refs/heads/master