Many of our programs consider that it is OK to release dynamic
storage that is used throughout the life of the program by simply
exiting, but this makes it harder to leak detection tools to avoid
reporting false positives. Plug many existing leaks and introduce
a mechanism for developers to mark that the region of memory
pointed by a pointer is not lost/leaking to help these tools.
* jk/leak-checkers:
add UNLEAK annotation for reducing leak false positives
set_git_dir: handle feeding gitdir to itself
repository: free fields before overwriting them
reset: free allocated tree buffers
reset: make tree counting less confusing
config: plug user_config leak
update-index: fix cache entry leak in add_one_file()
add: free leaked pathspec after add_files_to_cache()
test-lib: set LSAN_OPTIONS to abort by default
test-lib: --valgrind should not override --verbose-log
Our hashmap implementation in hashmap.[ch] is not thread-safe when
adding a new item needs to expand the hashtable by rehashing; add
an API to disable the automatic rehashing to work it around.
* jh/hashmap-disable-counting:
hashmap: add API to disable item counting when threaded
The long-standing rule that an in-core lockfile instance, once it
is used, must not be freed, has been lifted and the lockfile and
tempfile APIs have been updated to reduce the chance of programming
errors.
* jk/incore-lockfile-removal:
stop leaking lock structs in some simple cases
ref_lock: stop leaking lock_files
lockfile: update lifetime requirements in documentation
tempfile: auto-allocate tempfiles on heap
tempfile: remove deactivated list entries
tempfile: use list.h for linked list
tempfile: release deactivated strbufs instead of resetting
tempfile: robustify cleanup handler
tempfile: factor out deactivation
tempfile: factor out activation
tempfile: replace die("BUG") with BUG()
tempfile: handle NULL tempfile pointers gracefully
tempfile: prefer is_tempfile_active to bare access
lockfile: do not rollback lock on failed close
tempfile: do not delete tempfile on failed close
always check return value of close_tempfile
verify_signed_buffer: prefer close_tempfile() to close()
setup_temporary_shallow: move tempfile struct into function
setup_temporary_shallow: avoid using inactive tempfile
write_index_as_tree: cleanup tempfile on error
"git gc" and friends when multiple worktrees are used off of a
single repository did not consider the index and per-worktree refs
of other worktrees as the root for reachability traversal, making
objects that are in use only in other worktrees to be subject to
garbage collection.
* nd/prune-in-worktree:
refs.c: reindent get_submodule_ref_store()
refs.c: remove fallback-to-main-store code get_submodule_ref_store()
rev-list: expose and document --single-worktree
revision.c: --reflog add HEAD reflog from all worktrees
files-backend: make reflog iterator go through per-worktree reflog
revision.c: --all adds HEAD from all worktrees
refs: remove dead for_each_*_submodule()
refs.c: move for_each_remote_ref_submodule() to submodule.c
revision.c: use refs_for_each*() instead of for_each_*_submodule()
refs: add refs_head_ref()
refs: move submodule slash stripping code to get_submodule_ref_store
refs.c: refactor get_submodule_ref_store(), share common free block
revision.c: --indexed-objects add objects from all worktrees
revision.c: refactor add_index_objects_to_pending()
refs.c: use is_dir_sep() in resolve_gitlink_ref()
revision.h: new flag in struct rev_info wrt. worktree-related refs
* ma/split-symref-update-fix:
refs/files-backend: add `refname`, not "HEAD", to list
refs/files-backend: correct return value in lock_ref_for_update
refs/files-backend: fix memory leak in lock_ref_for_update
refs/files-backend: add longer-scoped copy of string to list
* mh/notes-cleanup:
load_subtree(): check that `prefix_len` is in the expected range
load_subtree(): declare some variables to be `size_t`
hex_to_bytes(): simpler replacement for `get_oid_hex_segment()`
get_oid_hex_segment(): don't pad the rest of `oid`
load_subtree(): combine some common code
get_oid_hex_segment(): return 0 on success
load_subtree(): only consider blobs to be potential notes
load_subtree(): check earlier whether an internal node is a tree entry
load_subtree(): separate logic for internal vs. terminal entries
load_subtree(): fix incorrect comment
load_subtree(): reduce the scope of some local variables
load_subtree(): remove unnecessary conditional
notes: make GET_NIBBLE macro more robust
"git branch -M a b" while on a branch that is completely unrelated
to either branch a or branch b misbehaved when multiple worktree
was in use. This has been fixed.
* nd/worktree-kill-parse-ref:
branch: fix branch renaming not updating HEADs correctly
In addition to "cc: <a@dd.re.ss> # cruft", "cc: a@dd.re.ss # cruft"
was taught to "git send-email" as a valid way to tell it that it
needs to also send a carbon copy to <a@dd.re.ss> in the trailer
section.
* mm/send-email-cc-cruft:
send-email: don't use Mail::Address, even if available
send-email: fix garbage removal after address
Merge branch 'rs/archive-excluded-directory' into maint
"git archive" did not work well with pathspecs and the
export-ignore attribute.
We may want to resurrect the "we don't archive an empty directory"
bonus patch, but I do not mind merging the above early to 'next'
and leave it as a separate follow-up enhancement.
cf. <20170820090629.tumvqwzkromcykjf@sigill.intra.peff.net>
* rs/archive-excluded-directory:
archive: don't queue excluded directories
archive: factor out helper functions for handling attributes
t5001: add tests for export-ignore attributes and exclude pathspecs
Killing "git merge --edit" before the editor returns control left
the repository in a state with MERGE_MSG but without MERGE_HEAD,
which incorrectly tells the subsequent "git commit" that there was
a squash merge in progress. This has been fixed.
* mg/killed-merge:
merge: save merge state earlier
merge: split write_merge_state in two
merge: clarify call chain
Documentation/git-merge: explain --continue
"git apply" that is used as a better "patch -p1" failed to apply a
taken from a file with CRLF line endings to a file with CRLF line
endings. The root cause was because it misused convert_to_git()
that tried to do "safe-crlf" processing by looking at the index
entry at the same path, which is a nonsense---in that mode, "apply"
is not working on the data in (or derived from) the index at all.
This has been fixed.
* tb/apply-with-crlf:
apply: file commited with CRLF should roundtrip diff and apply
convert: add SAFE_CRLF_KEEP_CRLF
Merge branch 'cc/subprocess-handshake-missing-capabilities' into maint
When handshake with a subprocess filter notices that the process
asked for an unknown capability, Git did not report what program
the offending subprocess was running. This has been corrected.
We may want a follow-up fix to tighten the error checking, though.
* cc/subprocess-handshake-missing-capabilities:
sub-process: print the cmd when a capability is unsupported
"git svn" used with "--localtime" option did not compute the tz
offset for the timestamp in question and instead always used the
current time, which has been corrected.
* ur/svn-local-zone:
git svn fetch: Create correct commit timestamp when using --localtime
"git am -s" has been taught that some input may end with a trailer
block that is not Signed-off-by: and it should refrain from adding
an extra blank line before adding a new sign-off in such a case.
* pw/am-signoff:
am: fix signoff when other trailers are present
Merge branch 'ma/pager-per-subcommand-action' into maint
The "tag.pager" configuration variable was useless for those who
actually create tag objects, as it interfered with the use of an
editor. A new mechanism has been introduced for commands to enable
pager depending on what operation is being carried out to fix this,
and then "git tag -l" is made to run pager by default.
If this works out OK, I think there are low-hanging fruits in
other commands like "git branch" that outputs long list in one mode
while taking input in another.
* ma/pager-per-subcommand-action:
git.c: ignore pager.* when launching builtin as dashed external
tag: change default of `pager.tag` to "on"
tag: respect `pager.tag` in list-mode only
t7006: add tests for how git tag paginates
git.c: provide setup_auto_pager()
git.c: let builtins opt for handling `pager.foo` themselves
builtin.h: take over documentation from api-builtin.txt
"git log --tag=no-such-tag" showed log starting from HEAD, which
has been fixed---it now shows nothing.
* jk/rev-list-empty-input:
revision: do not fallback to default when rev_input_given is set
rev-list: don't show usage when we see empty ref patterns
revision: add rev_input_given flag
t6018: flesh out empty input/output rev-list tests
Merge branch 'st/lib-gpg-kill-stray-agent' into maint
Some versions of GnuPG fails to kill gpg-agent it auto-spawned
and such a left-over agent can interfere with a test. Work it
around by attempting to kill one before starting a new test.
* st/lib-gpg-kill-stray-agent:
t: lib-gpg: flush gpg agent on startup
refs/files-backend: add `refname`, not "HEAD", to list
An earlier patch rewrote `split_symref_update()` to add a copy of a
string to a string list instead of adding the original string. That was
so that the original string could be freed in a later patch, but it is
also conceptually cleaner, since now all calls to `string_list_insert()`
and `string_list_append()` add `update->refname`. --- Except a literal
"HEAD" is added in `split_head_update()`.
Restructure `split_head_update()` in the same way as the earlier patch
did for `split_symref_update()`. This does not correct any practical
problem, but makes things conceptually cleaner. The downside is a call
to `string_list_has_string()`, which should be relatively cheap.
Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Martin Ågren <martin.agren@gmail.com> Reviewed-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
refs/files-backend: correct return value in lock_ref_for_update
In one code path we return a literal -1 and not a symbolic constant. The
value -1 would be interpreted as TRANSACTION_NAME_CONFLICT, which is
wrong. Use TRANSACTION_GENERIC_ERROR instead (that is the only other
return value we have to choose from).
Noticed-by: Michael Haggerty <mhagger@alum.mit.edu> Reviewed-by: Michael Haggerty <mhagger@alum.mit.edu> Signed-off-by: Martin Ågren <martin.agren@gmail.com> Reviewed-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
refs/files-backend: fix memory leak in lock_ref_for_update
After the previous patch, none of the functions we call hold on to
`referent.buf`, so we can safely release the string buffer before
returning.
Reviewed-by: Michael Haggerty <mhagger@alum.mit.edu> Signed-off-by: Martin Ågren <martin.agren@gmail.com> Reviewed-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
refs/files-backend: add longer-scoped copy of string to list
split_symref_update() receives a string-pointer `referent` and adds it
to the list of `affected_refnames`. The list simply holds on to the
pointers it is given, it does not copy the strings and it does not ever
free them. The `referent` string in split_symref_update() belongs to a
string buffer in the caller. After we return, the string will be leaked.
In the next patch, we want to properly release the string buffer in the
caller, but we can't safely do so until we've made sure that
`affected_refnames` will not be holding on to a pointer to the string.
We could configure the list to handle its own resources, but it would
mean some alloc/free-churning. The list is already handling other
strings (through other code paths) which we do not need to worry about,
and we'd be memory-churning those strings too, completely unnecessary.
Observe that split_symref_update() creates a `new_update`-object through
ref_transaction_add_update(), after which `new_update->refname` is a
copy of `referent`. The difference is, this copy will be freed, and it
will be freed *after* `affected_refnames` has been cleared.
Rearrange the handling of `referent`, so that we don't add it directly
to `affected_refnames`. Instead, first just check whether `referent`
exists in the string list, and later add `new_update->refname`.
Helped-by: Michael Haggerty <mhagger@alum.mit.edu> Reviewed-by: Michael Haggerty <mhagger@alum.mit.edu> Signed-off-by: Martin Ågren <martin.agren@gmail.com> Reviewed-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
load_subtree(): check that `prefix_len` is in the expected range
This value, which is stashed in the last byte of an object_id hash,
gets handed around a lot. So add a sanity check before using it in
`load_subtree()`.
Signed-off-by: Michael Haggerty <mhagger@alum.mit.edu> Signed-off-by: Junio C Hamano <gitster@pobox.com>
add UNLEAK annotation for reducing leak false positives
It's a common pattern in git commands to allocate some
memory that should last for the lifetime of the program and
then not bother to free it, relying on the OS to throw it
away.
This keeps the code simple, and it's fast (we don't waste
time traversing structures or calling free at the end of the
program). But it also triggers warnings from memory-leak
checkers like valgrind or LSAN. They know that the memory
was still allocated at program exit, but they don't know
_when_ the leaked memory stopped being useful. If it was
early in the program, then it's probably a real and
important leak. But if it was used right up until program
exit, it's not an interesting leak and we'd like to suppress
it so that we can see the real leaks.
This patch introduces an UNLEAK() macro that lets us do so.
To understand its design, let's first look at some of the
alternatives.
Unfortunately the suppression systems offered by
leak-checking tools don't quite do what we want. A
leak-checker basically knows two things:
1. Which blocks were allocated via malloc, and the
callstack during the allocation.
2. Which blocks were left un-freed at the end of the
program (and which are unreachable, but more on that
later).
Their suppressions work by mentioning the function or
callstack of a particular allocation, and marking it as OK
to leak. So imagine you have code like this:
int cmd_foo(...)
{
/* this allocates some memory */
char *p = some_function();
printf("%s", p);
return 0;
}
You can say "ignore allocations from some_function(),
they're not leaks". But that's not right. That function may
be called elsewhere, too, and we would potentially want to
know about those leaks.
So you can say "ignore the callstack when main calls
some_function". That works, but your annotations are
brittle. In this case it's only two functions, but you can
imagine that the actual allocation is much deeper. If any of
the intermediate code changes, you have to update the
suppression.
What we _really_ want to say is that "the value assigned to
p at the end of the function is not a real leak". But
leak-checkers can't understand that; they don't know about
"p" in the first place.
However, we can do something a little bit tricky if we make
some assumptions about how leak-checkers work. They
generally don't just report all un-freed blocks. That would
report even globals which are still accessible when the
leak-check is run. Instead they take some set of memory
(like BSS) as a root and mark it as "reachable". Then they
scan the reachable blocks for anything that looks like a
pointer to a malloc'd block, and consider that block
reachable. And then they scan those blocks, and so on,
transitively marking anything reachable from a global as
"not leaked" (or at least leaked in a different category).
So we can mark the value of "p" as reachable by putting it
into a variable with program lifetime. One way to do that is
to just mark "p" as static. But that actually affects the
run-time behavior if the function is called twice (you
aren't likely to call main() twice, but some of our cmd_*()
functions are called from other commands).
Instead, we can trick the leak-checker by putting the value
into _any_ reachable bytes. This patch keeps a global
linked-list of bytes copied from "unleaked" variables. That
list is reachable even at program exit, which confers
recursive reachability on whatever values we unleak.
But wait, couldn't we just say "free(p)"? In this toy
example, yes. But UNLEAK()'s byte-copying strategy has
several advantages over actually freeing the memory:
1. It's recursive across structures. In many cases our "p"
is not just a pointer, but a complex struct whose
fields may have been allocated by a sub-function. And
in some cases (e.g., dir_struct) we don't even have a
function which knows how to free all of the struct
members.
By marking the struct itself as reachable, that confers
reachability on any pointers it contains (including those
found in embedded structs, or reachable by walking
heap blocks recursively.
2. It works on cases where we're not sure if the value is
allocated or not. For example:
char *p = argc > 1 ? argv[1] : some_function();
It's safe to use UNLEAK(p) here, because it's not
freeing any memory. In the case that we're pointing to
argv here, the reachability checker will just ignore
our bytes.
3. Likewise, it works even if the variable has _already_
been freed. We're just copying the pointer bytes. If
the block has been freed, the leak-checker will skip
over those bytes as uninteresting.
4. Because it's not actually freeing memory, you can
UNLEAK() before we are finished accessing the variable.
This is helpful in cases like this:
This patch adds the UNLEAK() macro and enables it
automatically when Git is compiled with SANITIZE=leak. In
normal builds it's a noop, so we pay no runtime cost.
It also adds some UNLEAK() annotations to show off how the
feature works. On top of other recent leak fixes, these are
enough to get t0000 and t0001 to pass when compiled with
LSAN.
Note the case in commit.c which actually converts a
strbuf_release() into an UNLEAK. This code was already
non-leaky, but the free didn't do anything useful, since
we're exiting. Converting it to an annotation means that
non-leak-checking builds pay no runtime cost. The cost is
minimal enough that it's probably not worth going on a
crusade to convert these kinds of frees to UNLEAKS. I did it
here for consistency with the "sb" leak (though it would
have been equally correct to go the other way, and turn them
both into strbuf_release() calls).
Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
"git pull" supports a --recurse-submodules option but does not parse the
submodule.recurse configuration item to set the default for that option.
Meanwhile "git fetch" does support submodule.recurse, producing
confusing behavior: when submodule.recurse is enabled, "git pull"
recursively fetches submodules but does not update them after fetch.
Handle submodule.recurse in "git pull" to fix this.
Reported-by: Magnus Homann <magnus@homann.se> Signed-off-by: Nicolas Morey-Chaisemartin <nicolas@morey-chaisemartin.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
pull parses first the cli options and then the config option.
The expected behavior is the other way around, so that config
options can not override the cli ones.
This patch changes the parsing order so config options are
parsed first.
Signed-off-by: Nicolas Morey-Chaisemartin <nicolas@morey-chaisemartin.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
hashmap: add API to disable item counting when threaded
This is to address concerns raised by ThreadSanitizer on the mailing list
about threaded unprotected R/W access to map.size with my previous "disallow
rehash" change (0607e10009ee4e37cb49b4cec8d28a9dda1656a4).
Add API to hashmap to disable item counting and thus automatic rehashing.
Also include API to later re-enable them.
When item counting is disabled, the map.size field is invalid. So to
prevent accidents, the field has been renamed and an accessor function
hashmap_get_size() has been added. All direct references to this
field have been been updated. And the name of the field changed
to map.private_size to communicate this.
Here is the relevant output from ThreadSanitizer showing the problem:
WARNING: ThreadSanitizer: data race (pid=10554)
Read of size 4 at 0x00000082d488 by thread T2 (mutexes: write M16):
#0 hashmap_add hashmap.c:209
#1 hash_dir_entry_with_parent_and_prefix name-hash.c:302
#2 handle_range_dir name-hash.c:347
#3 handle_range_1 name-hash.c:415
#4 lazy_dir_thread_proc name-hash.c:471
#5 <null> <null>
Martin gives instructions for running TSan on test t3008 in this post:
https://public-inbox.org/git/CAN0heSoJDL9pWELD6ciLTmWf-a=oyxe4EXXOmCKvsG5MSuzxsA@mail.gmail.com/
Signed-off-by: Jeff Hostetler <jeffhost@microsoft.com> Signed-off-by: Junio C Hamano <gitster@pobox.com>
git_extract_argv0_path: do nothing without RUNTIME_PREFIX
When the RUNTIME_PREFIX compile-time knob isn't set, we
never look at the argv0_path we extract. We can push its
declaration inside the #ifdef to make it more clear that the
extract code is effectively a noop.
This also un-confuses leak-checking of the argv0_path
variable when RUNTIME_PREFIX isn't set. The compiler is free
to drop this static variable that we set but never look at
(and "gcc -O2" does so). But the compiler still must call
strbuf_detach(), since it doesn't know whether that function
has side effects; it just throws away the result rather than
putting it into the global.
Leak-checkers which work by scanning the data segment for
pointers to heap blocks would normally consider the block
as reachable at program end. But if the compiler removes the
variable entirely, there's nothing to find.
Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
system_path: move RUNTIME_PREFIX to a sub-function
The system_path() function has an #ifdef in the middle of
it. Let's move the conditional logic into a sub-function.
This isolates it more, which will make it easier to change
and add to.
Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Documentation: mention that `eol` can change the dirty status of paths
When setting the `eol` attribute, paths can change their dirty status
without any change in the working directory. This can cause confusion
and should at least be mentioned with a remedy.
Signed-off-by: Ben Boeckel <mathstuf@gmail.com> Reviewed-by: Torsten Bögershausen <tboegi@web.de> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Using for_each_ref_in() with a full refname has always been
a questionable practice, but it became an error with b9c8e7f2fb (prefix_ref_iterator: don't trim too much,
2017-05-22), making "git rev-parse --bisect" pretty reliably
show a BUG.
Commit 03df567fbf (for_each_bisect_ref(): don't trim
refnames, 2017-06-18) fixed this case for revision.c, but
rev-parse handles this option on its own. We can use the
same solution here (and piggy-back on its test).
Signed-off-by: Jeff King <peff@peff.net> Acked-by: Michael Haggerty <mhagger@alum.mit.edu> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Ideally we'd free the existing gitdir field before assigning
the new one, to avoid a memory leak. But we can't do so
safely because some callers do the equivalent of:
set_git_dir(get_git_dir());
We can detect that case as a noop, but there are even more
complicated cases like:
where we really do need to do some work, but the original
string must remain valid.
Rather than put the burden on callers to make a copy of the
string (only to free it later, since we'll make a copy of it
ourselves), let's solve the problem inside set_git_dir(). We
can make a copy of the pointer for the old gitdir, and then
avoid freeing it until after we've made our new copy.
Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
It's possible that the repository data may be initialized
twice (e.g., after doing a chdir() to the top of the
worktree we may have to adjust a relative git_dir path). We
should free() any existing fields before assigning to them
to avoid leaks.
This should be safe, as the fields are set based on the
environment or on other strings like the gitdir or
commondir. That makes it impossible that we are feeding an
alias to the just-freed string.
Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
We read the tree objects with fill_tree_descriptor(), but
never actually free the resulting buffers, causing a memory
leak. This isn't a huge deal because we call this code at
most twice per program invocation. But it does potentially
double our heap usage if you have large root trees. Let's
free the trees before returning.
Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Depending on whether we're in --keep mode, git-reset may
feed one or two trees to unpack_trees(). We start a counter
at "1" and then increment it to "2" only for the two-tree
case. But that means we must always subtract one to find the
correct array slot to fill with each descriptor.
Instead, let's start at "0" and just increment our counter
after adding each tree. This skips the extra subtraction,
and will make things much easier when we start to actually
free our tree buffers.
While we're at it, let's make the first allocation use the
slot at "desc + nr", too, even though we know "nr" is 0 at
that point. It makes the two fill_tree_descriptor() calls
consistent (always "desc + nr", followed by always
incrementing "nr").
Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
We generate filenames for the user_config ("~/.gitconfig")
and the xdg config ("$XDG_CONFIG_HOME/git/config") and then
decide which to use by looking at the filesystem. But after
selecting one, the unused string is just leaked.
This is a tiny leak, but it creates noise in leak-checker
output.
Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
update-index: fix cache entry leak in add_one_file()
When we fail to add the cache entry to the index, we end up
just leaking the struct. We should follow the pattern of the
early-return above and free it.
Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
add: free leaked pathspec after add_files_to_cache()
After run_diff_files, we throw away the rev_info struct,
including the pathspec that we copied into it, leaking the
memory. this is probably not a big deal in practice. We
usually only run this once per process, and the leak is
proportional to the pathspec list we're already holding in
memory.
But it's still a leak, and it pollutes leak-checker output,
making it harder to find important leaks.
Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
We already set ASAN_OPTIONS to abort if it finds any errors.
As we start to experiment with LSAN, the leak sanitizer,
it's convenient if we give it the same treatment.
Note that ASAN is actually a superset of LSAN and can do the
leak detection itself. So this only has an effect if you
specifically build with "make SANITIZE=leak" (leak detection
but not the rest of ASAN). Building with just LSAN results
in a build that runs much faster. That makes the
build-test-fix cycle more pleasant.
In the long run, once we've fixed or suppressed all the
leaks, it will probably be worth turning leak-detection on
for ASAN and just using that (to check both leaks _and_
memory errors in a single test run). But there's still a lot
of work before we get there.
Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
test-lib: --valgrind should not override --verbose-log
The --verbose test option cannot be used with test harnesses
like "prove". Instead, you must use --verbose-log.
Since the --valgrind option implies --verbose, that means
that it cannot be used with prove. I.e., this does not work:
prove t0000-basic.sh :: --valgrind
You'd think it could be fixed by doing:
prove t0000-basic.sh :: --valgrind --verbose-log
but that doesn't work either, because the implied --verbose
takes precedence over --verbose-log. If the user has given
us a specific option, we should prefer that.
Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Now that it's safe to declare a "struct lock_file" on the
stack, we can do so (and avoid an intentional leak). These
leaks were found by running t0000 and t0001 under valgrind
(though certainly other similar leaks exist and just don't
happen to be exercised by those tests).
Initializing the lock_file's inner tempfile with NULL is not
strictly necessary in these cases, but it's a good practice
to model. It means that if we were to call a function like
rollback_lock_file() on a lock that was never taken in the
first place, it becomes a quiet noop (rather than undefined
behavior).
Likewise, it's always safe to rollback_lock_file() on a file
that has already been committed or deleted, since that
operation is a noop on an inactive lockfile (and that's why
the case in config.c can drop the "if (lock)" check as we
move away from using a pointer).
Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Since the tempfile code recently relaxed the rule that
tempfile structs (and thus locks) need to hang around
forever, we no longer have to leak our lock_file structs.
In fact, we don't even need to heap-allocate them anymore,
since their lifetime can just match that of the surrounding
ref_lock (and if we forget to delete a lock, the effect is
the same as before: it will eventually go away at program
exit).
Note that there is a check in unlock_ref() to only rollback
a lock file if it has been allocated. We don't need that
check anymore; we zero the ref_lock (and thus the
lock_file), so at worst we pass a NULL pointer to
delete_tempfile(), which considers that a noop.
Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
The previous commit taught the tempfile code to give up
ownership over tempfiles that have been renamed or deleted.
That makes it possible to use a stack variable like this:
struct tempfile t;
create_tempfile(&t, ...);
...
if (!err)
rename_tempfile(&t, ...);
else
delete_tempfile(&t);
But doing it this way has a high potential for creating
memory errors. The tempfile we pass to create_tempfile()
ends up on a global linked list, and it's not safe for it to
go out of scope until we've called one of those two
deactivation functions.
Imagine that we add an early return from the function that
forgets to call delete_tempfile(). With a static or heap
tempfile variable, the worst case is that the tempfile hangs
around until the program exits (and some functions like
setup_shallow_temporary rely on this intentionally, creating
a tempfile and then leaving it for later cleanup).
But with a stack variable as above, this is a serious memory
error: the variable goes out of scope and may be filled with
garbage by the time the tempfile code looks at it. Let's
see if we can make it harder to get this wrong.
Since many callers need to allocate arbitrary numbers of
tempfiles, we can't rely on static storage as a general
solution. So we need to turn to the heap. We could just ask
all callers to pass us a heap variable, but that puts the
burden on them to call free() at the right time.
Instead, let's have the tempfile code handle the heap
allocation _and_ the deallocation (when the tempfile is
deactivated and removed from the list).
This changes the return value of all of the creation
functions. For the cleanup functions (delete and rename),
we'll add one extra bit of safety: instead of taking a
tempfile pointer, we'll take a pointer-to-pointer and set it
to NULL after freeing the object. This makes it safe to
double-call functions like delete_tempfile(), as the second
call treats the NULL input as a noop. Several callsites
follow this pattern.
The resulting patch does have a fair bit of noise, as each
caller needs to be converted to handle:
1. Storing a pointer instead of the struct itself.
2. Passing the pointer instead of taking the struct
address.
3. Handling a "struct tempfile *" return instead of a file
descriptor.
We could play games to make this less noisy. For example, by
defining the tempfile like this:
Callers would continue to have a "struct tempfile", and it
would be "active" only when the inner pointer was non-NULL.
But that just makes things more awkward in the long run.
There aren't that many callers, so we can simply bite
the bullet and adjust all of them. And the compiler makes it
easy for us to find them all.
Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
Once a "struct tempfile" is added to the global cleanup
list, it is never removed. This means that its storage must
remain valid for the lifetime of the program. For single-use
tempfiles and locks, this isn't a big deal: we just declare
the struct static. But for library code which may take
multiple simultaneous locks (like the ref code), they're
forced to allocate a struct on the heap and leak it.
This is mostly OK in practice. The size of the leak is
bounded by the number of refs, and most programs exit after
operating on a fixed number of refs (and allocate
simultaneous memory proportional to the number of ref
updates in the first place). But:
1. It isn't hard to imagine a real leak: a program which
runs for a long time taking a series of ref update
instructions and fulfilling them one by one. I don't
think we have such a program now, but it's certainly
plausible.
2. The leaked entries appear as false positives to
tools like valgrind.
Let's relax this rule by keeping only "active" tempfiles on
the list. We can do this easily by moving the list-add
operation from prepare_tempfile_object to activate_tempfile,
and adding a deletion in deactivate_tempfile.
Existing callers do not need to be updated immediately.
They'll continue to leak any tempfile objects they may have
allocated, but that's no different than the status quo. We
can clean them up individually.
Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
The tempfile API keeps to-be-cleaned tempfiles in a
singly-linked list and never removes items from the list. A
future patch would like to start removing items, but removal
from a singly linked list is O(n), as we have to walk the
list to find the predecessor element. This means that a
process which takes "n" simultaneous lockfiles (for example,
an atomic transaction on "n" refs) may end up quadratic in
"n".
Before we start allowing items to be removed, it would be
nice to have a way to cover this case in linear time.
The simplest solution is to make an assumption about the
order in which tempfiles are added and removed from the
list. If both operations iterate over the tempfiles in the
same order, then by putting new items at the end of the list
our removal search will always find its items at the
beginning of the list. And indeed, that would work for the
case of refs. But it creates a hidden dependency between
unrelated parts of the code. If anybody changes the ref code
(or if we add a new caller that opens multiple simultaneous
tempfiles) they may unknowingly introduce a performance
regression.
Another solution is to use a better data structure. A
doubly-linked list works fine, and we already have an
implementation in list.h. But there's one snag: the elements
of "struct tempfile" are all marked as "volatile", since a
signal handler may interrupt us and iterate over the list at
any moment (even if we were in the middle of adding a new
entry).
We can declare a "volatile struct list_head", but we can't
actually use it with the normal list functions. The compiler
complains about passing a pointer-to-volatile via a regular
pointer argument. And rightfully so, as the sub-function
would potentially need different code to deal with the
volatile case.
That leaves us with a few options:
1. Drop the "volatile" modifier for the list items.
This is probably a bad idea. I checked the assembly
output from "gcc -O2", and the "volatile" really does
impact the order in which it updates memory.
2. Use macros instead of inline functions. The irony here
is that list.h is entirely implemented as trivial
inline functions. So we basically are already
generating custom code for each call. But sadly there's no
way in C to declare the inline function to take a more
generic type.
We could do so by switching the inline functions to
macros, but it does make the end result harder to read.
And it doesn't fully solve the problem (for instance,
the declaration of list_head needs to change so that
its "prev" and "next" pointers point to other volatile
structs).
3. Don't use list.h, and just make our own ad-hoc
doubly-linked list. It's not that much code to
implement the basics that we need here. But if we're
going to do so, why not add the few extra lines
required to model it after the actual list.h interface?
We can even reuse a few of the macro helpers.
So this patch takes option 3, but actually implements a
parallel "volatile list" interface in list.h, where it could
potentially be reused by other code. This implements just
enough for tempfile.c's use, though we could easily port
other functions later if need be.
Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
tempfile: release deactivated strbufs instead of resetting
When a tempfile is deactivated, we reset its strbuf to the
empty string, which means we hold onto the memory for later
reuse.
Since we'd like to move to a system where tempfile structs
can actually be freed, deactivating one should drop all
resources it is currently using. And thus "release" rather
than "reset" is the appropriate function to call.
In theory the reset may have saved a malloc() when a
tempfile (or a lockfile) is reused multiple times. But in
practice this happened rarely. Most of our tempfiles are
single-use, since in cases where we might actually use many
(like ref locking) we xcalloc() a fresh one for each ref. In
fact, we leak those locks (to appease the rule that tempfile
storage can never be freed). Which means that using reset is
actively hurting us: instead of leaking just the tempfile
struct, we're leaking the extra heap chunk for the filename,
too.
Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
We may call remove_tempfiles() from an atexit handler, or
from a signal handler. In the latter case we must take care
to avoid functions which may deadlock if the process is in
an unknown state, including looking at any stdio handles
(which may be in the middle of doing I/O and locked) or
calling malloc() or free().
The current implementation calls delete_tempfile(). We unset
the tempfile's stdio handle (if any) to avoid deadlocking
there. But delete_tempfile() still calls unlink_or_warn(),
which can deadlock writing to stderr if the unlink fails.
Since delete_tempfile() isn't very long, let's just
open-code our own simple conservative version of the same
thing. Notably:
1. The "skip_fclose" flag is now called "in_signal_handler",
because it should inform more decisions than just the
fclose handling.
2. We can replace close_tempfile() with just close(fd).
That skips the fclose() question altogether. This is
fine for the atexit() case, too; there's no point
flushing data to a file which we're about to delete
anyway.
3. We can choose between unlink/unlink_or_warn based on
whether it's safe to use stderr.
4. We can replace the deactivate_tempfile() call with a
simple setting of the active flag. There's no need to
do any further cleanup since we know the program is
exiting. And even though the current deactivation code
is safe in a signal handler, this frees us up in future
patches to make non-signal deactivation more
complicated (e.g., by freeing resources).
5. There's no need to remove items from the tempfile_list.
The "active" flag is the ultimate answer to whether an
entry has been handled or not. Manipulating the list
just introduces more chance of recursive signals
stomping on each other, and the whole list will go away
when the program exits anyway. Less is more.
Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
When we deactivate a tempfile, we also have to clean up the
"filename" strbuf. Let's pull this out into its own function
to keep the logic in one place (which will become more
important when a future patch makes it more complicated).
Note that we can use the same function when deactivating an
object that _isn't_ actually active yet (like when we hit an
error creating a tempfile). These callsites don't currently
reset the "active" flag to 0, but it's OK to do so (it's
just a noop for these cases).
Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>
There are a few steps required to "activate" a tempfile
struct. Let's pull these out into a function. That saves a
few repeated lines now, but more importantly will make it
easier to change the activation scheme later.
Signed-off-by: Jeff King <peff@peff.net> Signed-off-by: Junio C Hamano <gitster@pobox.com>