If we attempt to grep non-ascii log message text with an ascii pattern, we
run into the following issue:
$ git log --color --author='.var.*Bjar' -1 origin/master | grep ^Author
grep: (standard input): binary file matches
So, to fix this teach the grep code to use PCRE2_UTF, as long as the log
output is encoded in UTF-8.
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Hamza Mahfooz <someguy@effective-light.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
These changes are made in preparation of, the colorization support for the
"git log" subcommands that, rely on regex functionality (i.e. "--author",
"--committer" and "--grep"). These changes are necessary primarily because
match_one_pattern() expects header lines to be prefixed, however, in
pretty, the prefixes are stripped from the lines because the name-email
pairs need to go through additional parsing, before they can be printed and
because next_match() doesn't handle the case of
"ctx == GREP_CONTEXT_HEAD" at all. So, teach next_match() how to handle the
new case and move match_one_pattern()'s core logic to
headerless_match_one_pattern() while preserving match_one_pattern()'s uses
that depend on the additional processing.
Signed-off-by: Hamza Mahfooz <someguy@effective-light.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Our grep_buffer() function takes a non-const buffer, which is confusing:
we don't take ownership of nor write to the buffer.
This mostly comes from the fact that the underlying grep_source struct
in which we store the buffer uses non-const pointer. The memory pointed
to by the struct is sometimes owned by us (for FILE or OID sources), and
sometimes not (for BUF sources).
Let's store it as const, which lets us err on the side of caution (i.e.,
the compiler will warn us if any of our code writes to or tries to free
it).
As a result, we must annotate the one place where we do free it by
casting away the constness. But that's a small price to pay for the
extra safety and clarity elsewhere (and indeed, it already had a comment
explaining why GREP_SOURCE_BUF _didn't_ free it).
And then we can mark grep_buffer() as taking a const buffer.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When we're grepping in a buffer, we don't need to modify it. So we can
take "const char *" buffers, rather than "char *". This can avoid some
awkward casts in our callers, and make our expectations more clear (we
will not take ownership of the memory, nor will we ever write to it).
These spots don't all necessarily have to be converted at the same time,
but some of them are dependent on others, because we pass
pointers-to-pointers in a few cases. And none of this should change any
behavior, since we're just adding "const" qualifiers (and likewise, the
compiler will let us know if we missed any spots). So it's relatively
low-risk to just do this all at once.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
We find the end of each matching line of a buffer, and then temporarily
write a NUL to turn it into a regular C string. But we don't need to do
so, because the only thing we do in the interim is pass the line and its
length (via an "eol" pointer) to match_line(). And that function should
only look at the bytes we passed it, whether it has a terminating NUL or
not.
We can drop this temporary write in order to simplify the code and make
it easier to use const buffers in more of grep.c.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When showing lines via grep (or looking for funcnames), we call
show_line() on a multi-line buffer. It finds the end of line and marks
it with a NUL. However, we don't need to do so, as the resulting line is
only used along with its "eol" marker:
- we pass both to next_match(), which takes care to look at only the
bytes we specified
- we pass the line to output_color() without its matching eol marker.
However, we do use the "match" struct we got from next_match() to
tell it how many bytes to look at (which can never exceed the string
we passed it).
So we can stop setting and restoring this NUL marker. That makes the
code simpler, and will allow us to take a const buffer in a future
patch.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When grepping for headers in commit objects, we receive individual
lines (e.g., "author Name <email> 1234 -0000"), and then strip off the
timestamp to do our match. We do so by writing a NUL byte over the
whitespace separator, and then remembering to restore it later.
We had to do it this way when this was added back in a4d7d2c6db (log
--author/--committer: really match only with name part, 2008-09-04),
because we fed the result directly to regexec(), which expects a
NUL-terminated string. But since b7d36ffca0 (regex: use regexec_buf(),
2016-09-21), we have a function which can match part of a buffer.
So instead of modifying the string, we can instead just move the "eol"
pointer, and the rest of the code will do the right thing. This will let
further patches mark more buffers as "const".
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Record the repository whenever an OID grep source is created, and teach
the worker threads to explicitly provide the repository when accessing
objects.
Signed-off-by: Jonathan Tan <jonathantanmy@google.com>
Reviewed-by: Matheus Tavares <matheus.bernardino@usp.br>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
grep_source_init() can create "struct grep_source" objects and,
depending on the value of the type passed, some void-pointer parameters have
different meanings. Because one of these types (GREP_SOURCE_OID) will
require an additional parameter in a subsequent patch, take the
opportunity to increase clarity and type safety by replacing this
function with individual functions for each type.
Signed-off-by: Jonathan Tan <jonathantanmy@google.com>
Reviewed-by: Matheus Tavares <matheus.bernardino@usp.br>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
"git grep --and -e foo" ought to have been diagnosed as an error
but instead segfaulted, which has been corrected.
* rs/grep-parser-fix:
grep: report missing left operand of --and
Git grep allows combining two patterns with --and. It checks and
reports if the second pattern is missing when compiling the expression.
A missing first pattern, however, is only reported later at match time.
Thus no error is returned if no matching is done, e.g. because no file
matches the also given pathspec.
When that happens we get an expression tree with an GREP_NODE_AND node
and a NULL pointer to the missing left child. free_pattern_expr()
tries to dereference it during the cleanup at the end, which results
in a segmentation fault.
Fix this by verifying the presence of the left operand at expression
compilation time.
Reported-by: Matthew Hughes <matthewhughes934@gmail.com>
Helped-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: René Scharfe <l.s.r@web.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Up until recently, object IDs did not have an algorithm member, only a
hash. Consequently, it was possible to share one null (all-zeros)
object ID among all hash algorithms. Now that we're going to be
handling objects from multiple hash algorithms, it's important to make
sure that all object IDs have a correct algorithm field.
Introduce a per-algorithm null OID, and add it to struct hash_algo.
Introduce a wrapper function as well, and use it everywhere we used to
use the null_oid constant.
Signed-off-by: brian m. carlson <sandals@crustytoothpaste.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Updates to memory allocation code around the use of pcre2 library.
* ab/grep-pcre2-allocfix:
grep/pcre2: move definitions of pcre2_{malloc,free}
grep/pcre2: move back to thread-only PCREv2 structures
grep/pcre2: actually make pcre2 use custom allocator
grep/pcre2: use pcre2_maketables_free() function
grep/pcre2: use compile-time PCREv2 version test
grep/pcre2: add GREP_PCRE2_DEBUG_MALLOC debug mode
grep/pcre2: prepare to add debugging to pcre2_malloc()
grep/pcre2: correct reference to grep_init() in comment
grep/pcre2: drop needless assignment to NULL
grep/pcre2: drop needless assignment + assert() on opt->pcre2
Add and apply a semantic patch for converting code that open-codes
CALLOC_ARRAY to use it instead. It shortens the code and infers the
element size automatically.
Signed-off-by: René Scharfe <l.s.r@web.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Move the definitions of the pcre2_{malloc,free} functions above the
compile_pcre2_pattern() function they're used in.
Before the preceding commit they used to be needed earlier, but now we
can move them to be adjacent to the other PCREv2 functions.
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Change the setup of the "pcre2_general_context" to happen per-thread
in compile_pcre2_pattern() instead of in grep_init().
This change brings it in line with how the rest of the pcre2_* members
in the grep_pat structure are set up.
As noted in the preceding commit the approach 513f2b0bbd (grep: make
PCRE2 aware of custom allocator, 2019-10-16) took to allocate the
pcre2_general_context seems to have been initially based on a
misunderstanding of how PCREv2 memory allocation works.
The approach of creating a global context in grep_init() is just added
complexity for almost zero gain. On my system it's 24 bytes saved
per-thread. For comparison PCREv2 will then go on to allocate at least
a kilobyte for its own thread-local state.
As noted in 6d423dd542 (grep: don't redundantly compile throwaway
patterns under threading, 2017-05-25) the grep code is intentionally
not trying to micro-optimize allocations by e.g. sharing some PCREv2
structures globally, while making others thread-local.
So let's remove this special case and make all of them thread-local
again for simplicity. With this change we could move the
pcre2_{malloc,free} functions around to live closer to their current
use. I'm not doing that here to keep this change small, that cleanup
will be done in a follow-up commit.
See also the discussion in 94da9193a6 (grep: add support for PCRE v2,
2017-06-01) about thread safety, and Johannes's comments[1] to the
effect that we should be doing what this patch is doing.
1. https://lore.kernel.org/git/nycvar.QRO.7.76.6.1908052120302.46@tvgsbejvaqbjf.bet/
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Continue work started in 513f2b0bbd (grep: make PCRE2 aware of custom
allocator, 2019-10-16) and make PCREv2 use our pcre2_{malloc,free}().
functions for allocation. We'll now use it for all PCREv2 allocations.
The reason 513f2b0bbd worked as a bugfix for the USE_NED_ALLOCATOR
issue is because it targeted the allocation freed via free(), as
opposed to by a pcre2_*free() function. I.e. the pcre2_maketables()
and pcre2_maketables_free() pair.
For most of the rest we continued allocating with stock malloc()
inside PCREv2 itself, but didn't segfault because we'd use its
corresponding free().
In a preceding commit of mine I changed the free() to
pcre2_maketables_free() on versions of PCREv2 10.34 and newer. So as
far as fixing the segfault goes we could revert 513f2b0bbd. But then
we wouldn't use the desired allocator, let's just use it instead.
Before this patch we'd on e.g.:
grep --threads=1 -iP æ.*var.*xyz
Only use pcre2_{malloc,free}() for 2 malloc() calls and 2
corresponding free() calls. Now it's 12 calls to each. This can be
observed with the GREP_PCRE2_DEBUG_MALLOC debug mode.
Reading the history of how this bug got introduced it wasn't present
in Johannes's original patch[1] to fix the issue.
My reading of that thread is that the approach the follow-up patches
to Johannes's original pursued were based on misunderstanding of how
the PCREv2 API works. In particular this part of [2]:
"most of the time (like when using UTF-8) the chartable (and
therefore the global context) is not needed (even when using
alternate allocators)"
That's simply not how PCREv2 memory allocation works. It's easy to see
how the misunderstanding came about. It's because (as noted above) the
issue was noticed because of our use of free() in our own grep.c for
freeing the memory allocated by pcre2_maketables().
Thus the misunderstanding that PCREv2's compile context is something
only needed for pcre2_maketables(), and e.g. an aborted earlier
attempt[3] to only set it up when we ourselves called
pcre2_maketables().
That's not what PCREv2's compile context is. To quote PCREv2's
documentation:
"This context just contains pointers to (and data for) external
memory management functions that are called from several places in
the PCRE2 library."
Thus the failed attempts to go down the route of only creating the
general context in cases where we ourselves call pcre2_maketables(),
before finally settling on the approach 513f2b0bbd took of always
creating it, but then mostly not using it.
Instead we should always create it, and then pass the general context
to those functions that accept it, so that they'll consistently use
our preferred memory allocation functions.
1. https://lore.kernel.org/git/3397e6797f872aedd18c6d795f4976e1c579514b.1565005867.git.gitgitgadget@gmail.com/
2. https://lore.kernel.org/git/CAPUEsphMh_ZqcH3M7PXC9jHTfEdQN3mhTAK2JDkdvKBp53YBoA@mail.gmail.com/
3. https://lore.kernel.org/git/20190806085014.47776-3-carenas@gmail.com/
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Make use of the pcre2_maketables_free() function to free the memory
allocated by pcre2_maketables().
At first sight it's strange that 10da030ab7 (grep: avoid leak of
chartables in PCRE2, 2019-10-16) which added the free() call here
doesn't make use of the pcre2_free() the author introduced in the
preceding commit in 513f2b0bbd (grep: make PCRE2 aware of custom
allocator, 2019-10-16).
The reason is that at the time the function didn't exist. It was first
introduced in PCREv2 version 10.34, released on 2019-11-21.
Let's make use of it behind a macro. I don't think this matters for
anything to do with custom allocators, but it makes our use of PCREv2
more discoverable.
At some distant point in the future we'll be able to drop the version
guard, as nobody will be running a version older than 10.34.
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Replace a use of pcre2_config(PCRE2_CONFIG_VERSION, ...) which I added
in 95ca1f987e (grep/pcre2: better support invalid UTF-8 haystacks,
2021-01-24) with the same test done at compile-time.
It might be cuter to do this at runtime since we don't have to do the
"major >= 11 || (major >= 10 && ...)" test. But in the next commit
we'll add another version comparison that absolutely needs to be done
at compile-time, so we're better of being consistent across the board.
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Add optional printing of PCREv2 allocations to stderr for a developer
who manually changes the GREP_PCRE2_DEBUG_MALLOC definition to "1".
You need to manually change the definition in the source file similar
to the DEBUG_MAILMAP, there's no Makefile knob for this.
This will be referenced a subsequent commit, and is generally useful
to manually see what's going on with PCREv2 allocations while working
on that code.
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Change pcre2_malloc() in a way that'll make it easier for a debugging
fprintf() to spew out the allocated pointer.
This doesn't introduce any functional change, it just makes a
subsequent commit's diff easier to read. Changes code added in
513f2b0bbd (grep: make PCRE2 aware of custom allocator, 2019-10-16).
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Correct a comment added in 513f2b0bbd (grep: make PCRE2 aware of
custom allocator, 2019-10-16). This comment was never correct in
git.git, but was consistent with an older version of the patch[1].
1. https://lore.kernel.org/git/20190806163658.66932-3-carenas@gmail.com/
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Remove a redundant assignment of pcre2_compile_context dating back to
my 94da9193a6 (grep: add support for PCRE v2, 2017-06-01).
In create_grep_pat() we xcalloc() the "grep_pat" struct, so there's no
need to NULL out individual members here.
I think this was probably something left over from an earlier
development version of mine.
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Drop an assignment added in b65abcafc7 (grep: use PCRE v2 for
optimized fixed-string search, 2019-07-01) and the overly cautious
assert() I added in 94da9193a6 (grep: add support for PCRE v2,
2017-06-01).
There was never a good reason for this, it's just a relic from when I
initially wrote the PCREv2 support. We're not going to have confusion
about compile_pcre2_pattern() being called when it shouldn't just
because we forgot to cargo-cult this opt->pcre2 option.
Furthermore the "struct grep_opt" is (mostly) used for the options the
user supplied, let's avoid the pattern of needlessly assigning to it.
With my recent removal of the PCREv1 backend in 7599730b7e (Remove
support for v1 of the PCRE library, 2021-01-24) there's even less
confusion around what we call where in these codepaths, which is one
more reason to remove this.
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Update support for invalid UTF-8 in PCRE2.
* ab/grep-pcre-invalid-utf8:
grep/pcre2: better support invalid UTF-8 haystacks
grep/pcre2 tests: don't rely on invalid UTF-8 data test
The support for deprecated PCRE1 library has been dropped.
* ab/retire-pcre1:
Remove support for v1 of the PCRE library
config.mak.uname: remove redundant NO_LIBPCRE1_JIT flag
Remove the hidden "grep --debug" and "log --grep-debug" options added
in 17bf35a3c7 (grep: teach --debug option to dump the parse tree,
2012-09-13).
At the time these options seem to have been intended to go along with
a documentation discussion and to help the author of relevant tests to
perform ad-hoc debugging on them[1].
Reasons to want this gone:
1. They were never documented, and the only (rather trivial) use of
them in our own codebase for testing is something I removed back
in e01b4dab01 (grep: change non-ASCII -i test to stop using
--debug, 2017-05-20).
2. Googling around doesn't show any in-the-wild uses I could dig up,
and on the Git ML the only mentions after the original discussion
seem to have been when they came up in unrelated diff contexts, or
that test commit of mine.
3. An exception to that is c581e4a749 (grep: under --debug, show
whether PCRE JIT is enabled, 2019-08-18) where we added the
ability to dump out when PCREv2 has the JIT in effect.
The combination of that and my earlier b65abcafc7 (grep: use PCRE
v2 for optimized fixed-string search, 2019-07-01) means Git prints
this out in its most common in-the-wild configuration:
$ git log --grep-debug --grep=foo --grep=bar --grep=baz --all-match
pcre2_jit_on=1
pcre2_jit_on=1
pcre2_jit_on=1
[all-match]
(or
pattern_body<body>foo
(or
pattern_body<body>bar
pattern_body<body>baz
)
)
$ git grep --debug \( -e foo --and -e bar \) --or -e baz
pcre2_jit_on=1
pcre2_jit_on=1
pcre2_jit_on=1
(or
(and
patternfoo
patternbar
)
patternbaz
)
I.e. for each pattern we're considering for the and/or/--all-match
etc. debugging we'll now diligently spew out another identical line
saying whether the PCREv2 JIT is on or not.
I think that nobody's complained about that rather glaringly obviously
bad output says something about how much this is used, i.e. it's
not.
The need for this debugging aid for the composed grep/log patterns
seems to have passed, and the desire to dump the JIT config seems to
have been another one-off around the time we had JIT-related issues on
the PCREv2 codepath. That the original author of this debugging
facility seemingly hasn't noticed the bad output since then[2] is
probably some indicator.
1. https://lore.kernel.org/git/cover.1347615361.git.git@drmicha.warpmail.net/
2. https://lore.kernel.org/git/xmqqk1b8x0ac.fsf@gitster-ct.c.googlers.com/
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Improve the support for invalid UTF-8 haystacks given a non-ASCII
needle when using the PCREv2 backend.
This is a more complete fix for a bug I started to fix in
870eea8166 (grep: do not enter PCRE2_UTF mode on fixed matching,
2019-07-26), now that PCREv2 has the PCRE2_MATCH_INVALID_UTF mode we
can make use of it.
This fixes the sort of case described in 8a5999838e (grep: stess test
PCRE v2 on invalid UTF-8 data, 2019-07-26), i.e.:
- The subject string is non-ASCII (e.g. "ævar")
- We're under a is_utf8_locale(), e.g. "en_US.UTF-8", not "C"
- We are using --ignore-case, or we're a non-fixed pattern
If those conditions were satisfied and we matched found non-valid
UTF-8 data PCREv2 might bark on it, in practice this only happened
under the JIT backend (turned on by default on most platforms).
Ultimately this fixes a "regression" in b65abcafc7 ("grep: use PCRE v2
for optimized fixed-string search", 2019-07-01), I'm putting that in
scare-quotes because before then we wouldn't properly support these
complex case-folding, locale etc. cases either, it just broke in
different ways.
There was a bug related to this the PCRE2_NO_START_OPTIMIZE flag fixed
in PCREv2 10.36. It can be worked around by setting the
PCRE2_NO_START_OPTIMIZE flag. Let's do that in those cases, and add
tests for the bug.
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Remove support for using version 1 of the PCRE library. Its use has
been discouraged by upstream for a long time, and it's in a
bugfix-only state.
Anyone who was relying on v1 in particular got a nudge to move to v2
in e6c531b808 (Makefile: make USE_LIBPCRE=YesPlease mean v2, not v1,
2018-03-11), which was first released as part of v2.18.0.
With this the LIBPCRE2 test prerequisites is redundant to PCRE. But
I'm keeping it for self-documentation purposes, and to avoid conflict
with other in-flight PCRE patches.
I'm also not changing all of our own "pcre2" names to "pcre", i.e. the
inverse of 6d4b5747f0 (grep: change internal *pcre* variable &
function names to be *pcre1*, 2017-05-25). I don't see the point, and
it makes the history/blame harder to read. Maybe if there's ever a
PCRE v3...
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
We have a `struct grep_opt` with our defaults which we then copy into
the caller's struct. Rather than zeroing the target struct and copying
each element one by one, just copy everything at once. This leaves the
code simpler and more maintainable.
We don't have any ownership issues with what we're copying now and can
just greedily copy the whole thing. If and when we do need to handle
such elements (`char *`?), we must and can handle it appropriately. Make
sure to leave a comment to our future selves.
Signed-off-by: Martin Ågren <martin.agren@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
In 15fabd1bbd ("builtin/grep.c: make configuration callback more
reusable", 2012-10-09), we learned to fill a `static struct grep_opt
grep_defaults` which we can use as a blueprint for other such structs.
At the time, we didn't consider designated initializers to be widely
useable, but these days, we do. (See, e.g., cbc0f81d96 ("strbuf: use
designated initializers in STRBUF_INIT", 2017-07-10).)
Use designated initializers to let the compiler set up the struct and so
that we don't need to remember to call `init_grep_defaults()`.
Signed-off-by: Martin Ågren <martin.agren@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
`init_grep_defaults()` fills a `static struct grep_opt grep_defaults`.
This struct is then used by `grep_init()` as a blueprint for other such
structs. Notably, `grep_init()` takes a `struct repo *` and assigns it
into the target struct.
As a result, it is unnecessary for us to take a `struct repo *` in
`init_grep_defaults()` as well. We assign it into the default struct and
never look at it again. And in light of how we return early if we have
already set up the default struct, it's not just unnecessary, but is
also a bit confusing: If we are called twice and with different repos,
is it a bug or a feature that we ignore the second repo?
Drop the repo parameter for `init_grep_defaults()`.
Signed-off-by: Martin Ågren <martin.agren@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This commit fixes a couple of minor spelling mistakes inside
comments.
Signed-off-by: Steve Kemp <steve@steve.org.uk>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
git-grep uses 'grep_read_mutex' to protect its calls to object reading
operations. But these have their own internal lock now, which ensures a
better performance (allowing parallel access to more regions). So, let's
remove the former and, instead, activate the latter with
enable_obj_read_lock().
Sections that are currently protected by 'grep_read_mutex' but are not
internally protected by the object reading lock should be surrounded by
obj_read_lock() and obj_read_unlock(). These guarantee mutual exclusion
with object reading operations, keeping the current behavior and
avoiding race conditions. Namely, these places are:
In grep.c:
- fill_textconv() at fill_textconv_grep().
- userdiff_get_textconv() at grep_source_1().
In builtin/grep.c:
- parse_object_or_die() and the submodule functions at
grep_submodule().
- deref_tag() and gitmodules_config_oid() at grep_objects().
If these functions become thread-safe, in the future, we might remove
the locking and probably get some speedup.
Note that some of the submodule functions will already be thread-safe
(or close to being thread-safe) with the internal object reading lock.
However, as some of them will require additional modifications to be
removed from the critical section, this will be done in its own patch.
Signed-off-by: Matheus Tavares <matheus.bernardino@usp.br>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
git-grep uses an internal grep_read_mutex to protect object reading
operations. Similarly, there's a grep_attr_mutex to protect calls to the
gitattributes machinery. However, two of the three functions protected
by the last mutex may also perform object reading, as seen below:
- userdiff_get_textconv() > notes_cache_init() >
notes_cache_match_validity() > lookup_commit_reference_gently() >
parse_object() > repo_has_object_file() >
repo_has_object_file_with_flags() > oid_object_info_extended()
- userdiff_find_by_path() > git_check_attr() > collect_some_attrs() >
prepare_attr_stack() > read_attr() > read_attr_from_index() >
read_blob_data_from_index() > read_object_file()
As these calls are not protected by grep_read_mutex, there might be race
conditions with other threads performing object reading (e.g. threads
calling fill_textconv() at grep.c:fill_textconv_grep()). To prevent
that, let's make sure to acquire the lock before both of these calls.
Note: this patch might slow down the threaded grep in worktree, for the
sake of thread-safeness. However, in the following patches, we should
regain performance by replacing grep_read_mutex for an internal object
reading lock and allowing parallel inflation during object reading.
Signed-off-by: Matheus Tavares <matheus.bernardino@usp.br>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Previously, the void pcre2_free() function in grep.c returned free().
While free() itself is void, afaict it's still an expression as per
section A.2.3, subsection 6.8.6 (jump-statement) in both C99 [1] and C11
[2]:
> return expression
Section 6.8.6.4 in C99 [1] and C11 [2] says that:
> A return statement with an expression shall not appear in a function
> whose return type is void.
The consequence of the old behavior was that developer builds with
pedantic errors enabled broke Git if PCRE2 was enabled and a
smart-enough compiler to detect these errors was used. This commit
fixes pedantic builds of Git that enables --with-libpcre.
[1] http://www.open-std.org/jtc1/sc22/wg14/www/docs/n1256.pdf
[2] http://www.open-std.org/jtc1/sc22/wg14/www/docs/n1548.pdf
Signed-off-by: Hans Jerry Illikainen <hji@dyntopia.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Leakfix.
* cb/pcre2-chartables-leakfix:
grep: avoid leak of chartables in PCRE2
grep: make PCRE2 aware of custom allocator
grep: make PCRE1 aware of custom allocator
94da9193a6 ("grep: add support for PCRE v2", 2017-06-01) introduced
a small memory leak visible with valgrind in t7813.
Complete the creation of a PCRE2 specific variable that was missing from
the original change and free the generated table just like it is done
for PCRE1.
Signed-off-by: Carlo Marcelo Arenas Belón <carenas@gmail.com>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
94da9193a6 (grep: add support for PCRE v2, 2017-06-01) didn't include
a way to override the system allocator, and so it is incompatible with
custom allocators (e.g. nedmalloc). This problem became obvious when we
tried to plug a memory leak by `free()`ing a data structure allocated by
PCRE2, triggering a segfault in Windows (where we use nedmalloc by
default).
PCRE2 requires the use of a general context to override the allocator
and therefore, there is a lot more code needed than in PCRE1, including
a couple of wrapper functions.
Extend the grep API with a "destructor" that could be called to cleanup
any objects that were created and used globally.
Update `builtin/grep.c` to use that new API, but any other future users
should make sure to have matching `grep_init()`/`grep_destroy()` calls
if they are using the pattern matching functionality.
Move some of the logic that was before done per thread (in the workers)
into an earlier phase to avoid degrading performance, but as the use of
PCRE2 with custom allocators is better understood it is expected more of
its functions will be instructed to use the custom allocator as well as
was done in the original code[1] this work was based on.
[1] https://public-inbox.org/git/3397e6797f872aedd18c6d795f4976e1c579514b.1565005867.git.gitgitgadget@gmail.com/
Reported-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Carlo Marcelo Arenas Belón <carenas@gmail.com>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
63e7e9d8b6 ("git-grep: Learn PCRE", 2011-05-09) didn't include a way
to override the system alocator, and so it is incompatible with
USE_NED_ALLOCATOR as reported by Dscho[1] (in similar code from PCRE2)
Note that nedmalloc, as well as other custom allocators like jemalloc
and mi-malloc, can be configured at runtime (via `LD_PRELOAD`),
therefore we cannot know at compile time whether a custom allocator is
used or not.
Make the minimum change possible to ensure this combination is supported
by extending `grep_init()` to set the PCRE1 specific functions to Git's
idea of `malloc()` and `free()` and therefore making sure all
allocations are done inside PCRE1 with the same allocator than the rest
of Git.
This change has negligible performance impact: PCRE needs to allocate
memory once per program run for the character table and for each pattern
compilation. These are both rare events compared to matching patterns
against lines. Actual measurements[2] show that the impact is lost in
the noise.
[1] https://public-inbox.org/git/pull.306.git.gitgitgadget@gmail.com
[2] https://public-inbox.org/git/7f42007f-911b-c570-17f6-1c6af0429586@web.de
Signed-off-by: Carlo Marcelo Arenas Belón <carenas@gmail.com>
Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
A few simplification and bugfixes to PCRE interface.
* ab/pcre-jit-fixes:
grep: under --debug, show whether PCRE JIT is enabled
grep: do not enter PCRE2_UTF mode on fixed matching
grep: stess test PCRE v2 on invalid UTF-8 data
grep: create a "is_fixed" member in "grep_pat"
grep: consistently use "p->fixed" in compile_regexp()
grep: stop using a custom JIT stack with PCRE v1
grep: stop "using" a custom JIT stack with PCRE v2
grep: remove overly paranoid BUG(...) code
grep: use PCRE v2 for optimized fixed-string search
grep: remove the kwset optimization
grep: drop support for \0 in --fixed-strings <pattern>
grep: make the behavior for NUL-byte in patterns sane
grep tests: move binary pattern tests into their own file
grep tests: move "grep binary" alongside the rest
grep: inline the return value of a function call used only once
t4210: skip more command-line encoding tests on MinGW
grep: don't use PCRE2?_UTF8 with "log --encoding=<non-utf8>"
log tests: test regex backends in "--encode=<enc>" tests
Make sure the grep machinery does not abort when seeing a payload
that is not UTF-8 even when JIT is not in use with PCRE1.
* cb/skip-utf8-check-with-pcre1:
grep: skip UTF8 checks explicitly
18547aacf5 ("grep/pcre: support utf-8", 2016-06-25) that was released
with git 2.10 added the PCRE_UTF8 flag to PCRE1 matching including a
call to has_non_ascii() to try to avoid breakage if there was non-utf8
encoded content in the haystack.
Usually PCRE is compiled with JIT support (even if is not the default),
and therefore the codepath used includes calling pcre_jit_exec, which
skips UTF-8 validation by design (which might result in crashes or hangs)
but when JIT support wasn't compiled we use pcre_exec instead with the
posibility that grep might be aborted if invalid UTF-8 is found in the
haystack.
PCRE1 provides a flag since Mar 5, 2007 that could be used to skip the
checks explicitly so use that to make both codepaths equivalent (the
flag is ignored by pcre1_jit_exec)
this fix is only implemented for PCRE1 because PCRE2 is likely to have
a better solution (without the risks) instead in the future
Helped-by: Johannes Schindelin <Johannes.Schindelin@gmx.de>
Helped-by: Eric Sunshine <sunshine@sunshineco.com>
Helped-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Suggested-by: Junio C Hamano <gitster@pobox.com>
Signed-off-by: Carlo Marcelo Arenas Belón <carenas@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The code used both a macro and a variable to keep track if JIT
support was desired and relied on the fact that a non JIT
enabled library will ignore a request for JIT compilation
(as defined by the second parameter of the call to pcre_study)
Cleanup the multiple levels of macros used and call pcre_study
with the right parameter after JIT support has been confirmed
and unless it was requested to be disabled with NO_LIBPCRE1_JIT
Signed-off-by: Carlo Marcelo Arenas Belón <carenas@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This information is useful and not visible anywhere else, so show it.
Signed-off-by: Beat Bolli <dev+git@drbeat.li>
Suggested-by: Johannes Schindelin <johannes.schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
As discussed in the last commit partially fix a bug introduced in
b65abcafc7 ("grep: use PCRE v2 for optimized fixed-string search",
2019-07-01). Because PCRE v2, unlike kwset, validates its UTF-8 input
we'd die on e.g.:
fatal: pcre2_match failed with error code -22: UTF-8 error:
isolated byte with 0x80 bit set
When grepping a non-ASCII fixed string. This is a more general problem
that's hard to fix, but we can at least fix the most common case of
grepping for a fixed string without "-i". I can't think of a reason
for why we'd turn on PCRE2_UTF when matching byte-for-byte like that.
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Since my b65abcafc7 ("grep: use PCRE v2 for optimized fixed-string
search", 2019-07-01) we've been dying on invalid UTF-8 data when
grepping for fixed strings if the following are all true:
* The subject string is non-ASCII (e.g. "ævar")
* We're under a is_utf8_locale(), e.g. "en_US.UTF-8", not "C"
* We compiled with PCRE v2
* That PCRE v2 did not have JIT support
The last of those is why this wasn't caught earlier, per pcre2jit(3):
"unless PCRE2_NO_UTF_CHECK is set, a UTF subject string is tested
for validity. In the interests of speed, these checks do not
happen on the JIT fast path, and if invalid data is passed, the
result is undefined."
I.e. the subject being matched against our pattern was invalid, but we
were lucky and getting away with it on the JIT path, but the non-JIT
one is stricter.
This patch does nothing to fix that, instead we sneak in support for
fixed patterns starting with "(*NO_JIT)", this disables the PCRE v2
jit with implicit fixed-string matching for testing, see
pcre2syntax(3) the syntax.
This is technically a change in behavior, but it's so obscure that I
figured it was OK. We'd previously consider this an invalid regular
expression as regcomp() would die on it, now we feed it to the PCRE v2
fixed-string path. I thought this was better than introducing yet
another GIT_TEST_* environment variable.
We're also relying on a behavior of PCRE v2 that technically could
change, but I think the test coverage is worth dipping our toe into
some somewhat undefined behavior.
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This change paves the way for later using this value the regex compile
functions themselves.
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
