git/builtin
Jeff King 993d38a066 index-pack: spawn threads atomically
The t5309 script triggers a racy false positive with SANITIZE=leak on a
multi-core system. Running with "--stress --run=6" usually fails within
10 seconds or so for me, complaining with something like:

    + git index-pack --fix-thin --stdin
    fatal: REF_DELTA at offset 46 already resolved (duplicate base 01d7713666f4de822776c7622c10f1b07de280dc?)

    =================================================================
    ==3904583==ERROR: LeakSanitizer: detected memory leaks

    Direct leak of 32 byte(s) in 1 object(s) allocated from:
        #0 0x7fa790d01986 in __interceptor_realloc ../../../../src/libsanitizer/lsan/lsan_interceptors.cpp:98
        #1 0x7fa790add769 in __pthread_getattr_np nptl/pthread_getattr_np.c:180
        #2 0x7fa790d117c5 in __sanitizer::GetThreadStackTopAndBottom(bool, unsigned long*, unsigned long*) ../../../../src/libsanitizer/sanitizer_common/sanitizer_linux_libcdep.cpp:150
        #3 0x7fa790d11957 in __sanitizer::GetThreadStackAndTls(bool, unsigned long*, unsigned long*, unsigned long*, unsigned long*) ../../../../src/libsanitizer/sanitizer_common/sanitizer_linux_libcdep.cpp:598
        #4 0x7fa790d03fe8 in __lsan::ThreadStart(unsigned int, unsigned long long, __sanitizer::ThreadType) ../../../../src/libsanitizer/lsan/lsan_posix.cpp:51
        #5 0x7fa790d013fd in __lsan_thread_start_func ../../../../src/libsanitizer/lsan/lsan_interceptors.cpp:440
        #6 0x7fa790adc3eb in start_thread nptl/pthread_create.c:444
        #7 0x7fa790b5ca5b in clone3 ../sysdeps/unix/sysv/linux/x86_64/clone3.S:81

    SUMMARY: LeakSanitizer: 32 byte(s) leaked in 1 allocation(s).
    Aborted

What happens is this:

  0. We construct a bogus pack with a duplicate object in it and trigger
     index-pack.

  1. We spawn a bunch of worker threads to resolve deltas (on my system
     it is 16 threads).

  2. One of the threads sees the duplicate object and bails by calling
     exit(), taking down all of the threads. This is expected and is the
     point of the test.

  3. At the time exit() is called, we may still be spawning threads from
     the main process via pthread_create(). LSan hooks thread creation
     to update its book-keeping; it has to know where each thread's
     stack is (so it can find entry points for reachable memory). So it
     calls pthread_getattr_np() to get information about the new thread.
     That may allocate memory that must be freed with a matching call to
     pthread_attr_destroy(). Probably LSan does that immediately, but
     if you're unlucky enough, the exit() will happen while it's between
     those two calls, and the allocated pthread_attr_t appears as a
     leak.

This isn't a real leak. It's not even in our code, but rather in the
LSan instrumentation code. So we could just ignore it. But the false
positive can cause people to waste time tracking it down.

It's possibly something that LSan could protect against (e.g., cover the
getattr/destroy pair with a mutex, and then in the final post-exit()
check for leaks try to take the same mutex). But I don't know enough
about LSan to say if that's a reasonable approach or not (or if my
analysis is even completely correct).

In the meantime, it's pretty easy to avoid the race by making creation
of the worker threads "atomic". That is, we'll spawn all of them before
letting any of them start to work. That's easy to do because we already
have a work_lock() mutex for handing out that work. If the main process
takes it, then all of the threads will immediately block until we've
finished spawning and released it.

This shouldn't make any practical difference for non-LSan runs. The
thread spawning is quick, and could happen before any worker thread gets
scheduled anyway.

Probably other spots that use threads are subject to the same issues.
But since we have to manually insert locking (and since this really is
kind of a hack), let's not bother with them unless somebody experiences
a similar racy false-positive in practice.

Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
2024-01-05 08:40:56 -08:00
..
add.c Merge branch 'ab/plug-leak-in-revisions' 2022-06-07 14:10:56 -07:00
am.c git-compat-util.h: use "UNUSED", not "UNUSED(var)" 2022-09-01 10:49:48 -07:00
annotate.c strvec: rename struct fields 2020-07-30 19:18:06 -07:00
apply.c apply.c: remove unnecessary include 2022-04-06 09:42:14 -07:00
archive.c parse-options: PARSE_OPT_KEEP_UNKNOWN only applies to --options 2022-08-19 11:13:14 -07:00
bisect--helper.c Merge branch 'rs/bisect-start-leakfix' into maint-2.38 2022-10-25 17:11:37 -07:00
blame.c parse-options: add support for parsing subcommands 2022-08-19 11:13:14 -07:00
branch.c Merge branch 'rj/branch-edit-desc-unborn' into maint-2.38 2022-10-27 15:24:13 -07:00
bugreport.c builtin/bugreport.c: create '--diagnose' option 2022-08-12 13:20:02 -07:00
bundle.c builtin/bundle.c: let parse-options parse subcommands 2022-08-19 11:13:15 -07:00
cat-file.c Merge branch 'tb/cat-file-z' 2022-08-05 15:52:14 -07:00
check-attr.c
check-ignore.c dir.[ch]: replace dir_init() with DIR_INIT 2021-07-01 12:32:22 -07:00
check-mailmap.c shortlog: remove unused(?) "repo-abbrev" feature 2021-01-12 14:04:42 -08:00
check-ref-format.c check-ref-format: fix trivial memory leak 2022-07-01 11:43:42 -07:00
checkout--worker.c pkt-line.[ch]: remove unused packet_read_line_buf() 2021-10-15 13:09:40 -07:00
checkout-index.c checkout-index: integrate with sparse index 2022-01-13 13:49:45 -08:00
checkout.c git-compat-util.h: use "UNUSED", not "UNUSED(var)" 2022-09-01 10:49:48 -07:00
clean.c Merge branch 'vd/sparse-clean-etc' 2022-02-17 16:25:05 -08:00
clone.c Sync with 2.37.6 2023-02-06 09:43:28 +01:00
column.c column: fix parsing of the '--nl' option 2021-08-26 14:36:27 -07:00
commit-graph.c Merge branch 'ab/unused-annotation' 2022-09-14 12:56:39 -07:00
commit-tree.c use xopen() to handle fatal open(2) failures 2021-08-25 14:39:08 -07:00
commit.c commit: avoid writing to global in option callback 2022-10-06 09:58:06 -07:00
config.c git-compat-util.h: use "UNUSED", not "UNUSED(var)" 2022-09-01 10:49:48 -07:00
count-objects.c i18n: remove from i18n strings that do not hold translatable parts 2022-02-04 13:58:28 -08:00
credential-cache--daemon.c unix-socket: add backlog size option to unix_stream_listen() 2021-03-15 14:32:51 -07:00
credential-cache.c credential-cache: check for windows specific errors 2021-09-14 09:30:54 -07:00
credential-store.c Use a better name for the function interpolating paths 2021-07-26 12:17:16 -07:00
credential.c doc: fix git credential synopsis 2021-10-28 09:57:09 -07:00
describe.c git-compat-util.h: use "UNUSED", not "UNUSED(var)" 2022-09-01 10:49:48 -07:00
diagnose.c builtin/diagnose.c: don't translate the two mode values 2022-09-21 11:53:35 -07:00
diff-files.c diff-files: move misplaced cleanup label 2022-07-12 07:17:28 -07:00
diff-index.c revisions API: call diff_free(&revs->pruning) in revisions_release() 2022-04-13 23:56:10 -07:00
diff-tree.c 2.36 gitk/diff-tree --stdin regression fix 2022-04-26 09:26:35 -07:00
diff.c Merge branch 'ab/plug-leak-in-revisions' 2022-06-07 14:10:56 -07:00
difftool.c Merge branch 'ab/unused-annotation' 2022-09-14 12:56:39 -07:00
env--helper.c parse-options: PARSE_OPT_KEEP_UNKNOWN only applies to --options 2022-08-19 11:13:14 -07:00
fast-export.c Merge branch 'ab/unused-annotation' 2022-09-14 12:56:39 -07:00
fast-import.c git-compat-util.h: use "UNUSED", not "UNUSED(var)" 2022-09-01 10:49:48 -07:00
fetch-pack.c list-objects-filter: add and use initializers 2022-09-12 08:38:59 -07:00
fetch.c Merge branch 'jk/list-objects-filter-cleanup' 2022-09-19 14:35:24 -07:00
fmt-merge-msg.c merge: allow to pretend a merge is made into a different branch 2021-12-20 14:55:02 -08:00
for-each-ref.c for-each-ref: delay parsing of --sort=<atom> options 2021-10-20 14:33:07 -07:00
for-each-repo.c builtin/for-each-repo: remove unnecessary argv copy to plug leak 2021-07-26 12:19:20 -07:00
fsck.c Merge branch 'jk/fsck-on-diet' into maint-2.38 2022-10-25 17:11:33 -07:00
fsmonitor--daemon.c Merge branch 'ah/fsmonitor-daemon-usage-non-l10n' into maint-2.38 2022-10-25 17:11:33 -07:00
gc.c gc: simplify maintenance_task_pack_refs() 2022-10-05 12:46:27 -07:00
get-tar-commit-id.c
grep.c grep: add --max-count command line option 2022-06-22 13:23:29 -07:00
hash-object.c Merge branch 'ab/object-file-api-updates' 2022-03-16 17:53:08 -07:00
help.c git help: special-case scalar 2022-09-02 10:02:56 -07:00
hook.c builtin/hook.c: let parse-options parse subcommands 2022-08-19 11:13:15 -07:00
index-pack.c index-pack: spawn threads atomically 2024-01-05 08:40:56 -08:00
init-db.c i18n: refactor "foo and bar are mutually exclusive" 2022-01-05 13:29:23 -08:00
interpret-trailers.c Use OPT_CALLBACK and OPT_CALLBACK_F 2020-04-28 10:47:10 -07:00
log.c Merge branch 'ab/unused-annotation' 2022-09-14 12:56:39 -07:00
ls-files.c ls-files: fix black space in error message 2022-09-12 09:25:40 -07:00
ls-remote.c Merge branch 'ep/maint-equals-null-cocci' 2022-05-20 15:26:59 -07:00
ls-tree.c git-compat-util.h: use "UNUSED", not "UNUSED(var)" 2022-09-01 10:49:48 -07:00
mailinfo.c mailinfo: allow squelching quoted CRLF warning 2021-05-10 15:06:22 +09:00
mailsplit.c Merge branch 'ep/maint-equals-null-cocci' 2022-05-20 15:26:59 -07:00
merge-base.c merge-base: free() allocated "struct commit **" list 2022-03-04 13:24:17 -08:00
merge-file.c merge-file: fix memory leaks on error path 2022-07-01 11:43:43 -07:00
merge-index.c merge-index: ensure full index 2021-04-14 13:47:21 -07:00
merge-ours.c builtins + test helpers: use return instead of exit() in cmd_* 2021-06-09 09:15:58 +09:00
merge-recursive.c gettext API users: don't explicitly cast ngettext()'s "n" 2022-03-07 11:57:52 -08:00
merge-tree.c merge-tree: add a --allow-unrelated-histories flag 2022-06-22 16:10:06 -07:00
merge.c Merge branch 'en/merge-unstash-only-on-clean-merge' into maint 2022-09-13 12:21:11 -07:00
mktag.c Merge branch 'ab/object-file-api-updates' 2022-03-16 17:53:08 -07:00
mktree.c mktree: do not check type of remote objects 2022-06-21 10:12:15 -07:00
multi-pack-index.c multi-pack-index: avoid writing to global in option callback 2022-10-06 09:56:51 -07:00
mv.c Merge branch 'sy/mv-out-of-cone' 2022-09-19 14:35:23 -07:00
name-rev.c git-compat-util.h: use "UNUSED", not "UNUSED(var)" 2022-09-01 10:49:48 -07:00
notes.c notes, remote: show unknown subcommands between `' 2022-09-07 12:06:12 -07:00
pack-objects.c Merge branch 'ab/unused-annotation' 2022-09-14 12:56:39 -07:00
pack-redundant.c tree-wide: apply equals-null.cocci 2022-05-02 09:50:37 -07:00
pack-refs.c
patch-id.c patch-id: fix scan_hunk_header on diffs with 1 line of before/after 2022-02-02 11:24:23 -08:00
prune-packed.c i18n: remove from i18n strings that do not hold translatable parts 2022-02-04 13:58:28 -08:00
prune.c revisions API users: add straightforward release_revisions() 2022-04-13 23:56:08 -07:00
pull.c pull: fix a "struct oid_array" memory leak 2022-07-01 11:43:43 -07:00
push.c push: fix capitalisation of the option name autoSetupMerge 2022-06-15 11:45:46 -07:00
range-diff.c range-diff: optionally accept pathspecs 2022-08-26 09:49:26 -07:00
read-tree.c read-tree: make three-way merge sparse-aware 2022-03-01 12:36:01 -08:00
rebase.c rebase: add rebase.updateRefs config option 2022-07-19 12:49:04 -07:00
receive-pack.c git-compat-util.h: use "UNUSED", not "UNUSED(var)" 2022-09-01 10:49:48 -07:00
reflog.c Merge branch 'ab/unused-annotation' 2022-09-14 12:56:39 -07:00
remote-ext.c strvec: convert builtin/ callers away from argv_array name 2020-07-28 15:02:18 -07:00
remote-fd.c
remote.c Merge branch 'jk/remote-rename-without-fetch-refspec' into maint-2.38 2022-10-25 17:11:32 -07:00
repack.c git-compat-util.h: use "UNUSED", not "UNUSED(var)" 2022-09-01 10:49:48 -07:00
replace.c refs: use ref_namespaces for replace refs base 2022-08-05 14:13:12 -07:00
rerere.c xdiff users: use designated initializers for out_line 2021-05-11 12:47:31 +09:00
reset.c pathspec.h: move pathspec_needs_expanded_index() from reset.c to here 2022-08-08 13:23:26 -07:00
rev-list.c rev-list: support human-readable output for --disk-usage 2022-08-11 13:45:23 -07:00
rev-parse.c Merge branch 'ab/unused-annotation' 2022-09-14 12:56:39 -07:00
revert.c parse-options: PARSE_OPT_KEEP_UNKNOWN only applies to --options 2022-08-19 11:13:14 -07:00
rm.c rm: integrate with sparse-index 2022-08-08 13:23:26 -07:00
send-pack.c i18n: factorize "invalid value" messages 2022-02-04 13:58:28 -08:00
shortlog.c parse-options: add support for parsing subcommands 2022-08-19 11:13:14 -07:00
show-branch.c git-compat-util.h: use "UNUSED", not "UNUSED(var)" 2022-09-01 10:49:48 -07:00
show-index.c builtin/show-index: set the algorithm for object IDs 2021-04-27 16:31:39 +09:00
show-ref.c git-compat-util.h: use "UNUSED", not "UNUSED(var)" 2022-09-01 10:49:48 -07:00
sparse-checkout.c pass subcommand "prefix" arguments to parse_options() 2022-08-25 09:43:29 -07:00
stash.c Merge branch 'ab/unused-annotation' 2022-09-14 12:56:39 -07:00
stripspace.c i18n: remove from i18n strings that do not hold translatable parts 2022-02-04 13:58:28 -08:00
submodule--helper.c Merge branch 'jk/list-objects-filter-cleanup' 2022-09-19 14:35:24 -07:00
symbolic-ref.c symbolic-ref: refuse to set syntactically invalid target 2022-08-01 12:17:13 -07:00
tag.c Merge branch 'ep/maint-equals-null-cocci' 2022-05-20 15:26:59 -07:00
unpack-file.c
unpack-objects.c unpack-objects: use stream_loose_object() to unpack large objects 2022-06-13 10:22:36 -07:00
update-index.c Merge branch 'jh/builtin-fsmonitor-part3' 2022-06-10 15:04:15 -07:00
update-ref.c update-ref: fix streaming of status updates 2021-09-03 11:35:15 -07:00
update-server-info.c i18n: remove from i18n strings that do not hold translatable parts 2022-02-04 13:58:28 -08:00
upload-archive.c upload-archive: use regular "struct child_process" pattern 2021-11-25 22:15:07 -08:00
upload-pack.c upload-pack: document and rename --advertise-refs 2021-08-05 08:59:37 -07:00
var.c var: add GIT_DEFAULT_BRANCH variable 2021-11-03 13:25:36 -07:00
verify-commit.c
verify-pack.c Merge branch 'bc/sha-256-part-3' 2020-08-11 18:04:11 -07:00
verify-tag.c
worktree.c builtin/worktree.c: let parse-options parse subcommands 2022-08-19 11:13:16 -07:00
write-tree.c