Remove the "else" branches of the HAVE_VARIADIC_MACROS macro, which
have been unconditionally omitted since 765dc16888 (git-compat-util:
always enable variadic macros, 2021-01-28).
Since were always omitted, anyone trying to use a compiler without
variadic macro support to compile a git since version
git v2.31.0 or later would have had a compilation error. 10 months
across a few releases since then should have been enough time for
anyone who cared to run into that and report the issue.
In addition to that, for anyone unsetting HAVE_VARIADIC_MACROS we've
been emitting extremely verbose warnings since at least
ee4512ed48 (trace2: create new combined trace facility,
2019-02-22). That's because there is no such thing as a
"region_enter_printf" or "region_leave_printf" format, so at least
under GCC and Clang everything that includes trace.h (almost every
file) emits a couple of warnings about that.
There's a large benefit to being able to have a hard dependency rely
on variadic macros, the code surrounding usage.c is hard to maintain
if we need to write two implementations of everything, and by relying
on "__FILE__" and "__LINE__" along with "__VA_ARGS__" we can in the
future make error(), die() etc. log where they were called from. We've
also recently merged d67fc4bf0b (Merge branch 'bc/require-c99',
2021-12-10) which further cements our hard dependency on C99.
So let's delete the fallback code, and update our CodingGuidelines to
note that we depend on this. The added bullet-point starts with
lower-case for consistency with other bullet-points in that section.
The diff in "trace.h" is relatively hard to read, since we need to
retain the existing API docs, which were comments on the code used if
HAVE_VARIADIC_MACROS was not defined.
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Change a comment added in e208f9cc75 (make error()'s constant return
value more visible, 2012-12-15). It's not correct that this is GCC-ism
anymore, it's code that uses standard C99 features.
The comment being changed here pre-dates the HAVE_VARIADIC_MACROS
define, which we got in e05bed960d (trace: add 'file:line' to all
trace output, 2014-07-12).
The original implementation of an error() macro) in e208f9cc75 used a
GCC-ism with the paste operator (see the commit message for mention of
it), but that was dropped later by 9798f7e5f9 (Use __VA_ARGS__ for all
of error's arguments, 2013-02-08), giving us the C99-portable version
we have now.
While we could remove the __GNUC__ define here, it might cause issues
for other compilers or static analysis systems, so let's not. See
87fe5df365 (inline constant return from error() function, 2014-05-06)
for one such issue.
See also e05bed960d (trace: add 'file:line' to all trace output,
2014-07-12) for another comment about GNUC's handling of __VA_ARGS__.
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The build procedure has been taught to notice older version of zlib
and enable our replacement uncompress2() automatically.
* ab/auto-detect-zlib-compress2:
compat: auto-detect if zlib has uncompress2()
Pick a better random number generator and use it when we prepare
temporary filenames.
* bc/csprng-mktemps:
wrapper: use a CSPRNG to generate random file names
wrapper: add a helper to generate numbers from a CSPRNG
We have a copy of uncompress2() implementation in compat/ so that we
can build with an older version of zlib that lack the function, and
the build procedure selects if it is used via the NO_UNCOMPRESS2
$(MAKE) variable. This is yet another "annoying" knob the porters
need to tweak on platforms that are not common enough to have the
default set in the config.mak.uname file.
Attempt to instead ask the system header <zlib.h> to decide if we
need the compatibility implementation. This is a deviation from the
way we have been handling the "compatiblity" features so far, and if
it can be done cleanly enough, it could work as a model for features
that need compatibility definition we discover in the future. With
that goal in mind, avoid expedient but ugly hacks, like shoving the
code that is conditionally compiled into an unrelated .c file, which
may not work in future cases---instead, take an approach that uses a
file that is independently compiled and stands on its own.
Compile and link compat/zlib-uncompress2.c file unconditionally, but
conditionally hide the implementation behind #if/#endif when zlib
version is 1.2.9 or newer, and unconditionally archive the resulting
object file in the libgit.a to be picked up by the linker.
There are a few things to note in the shape of the code base after
this change:
- We no longer use NO_UNCOMPRESS2 knob; if the system header
<zlib.h> claims a version that is more cent than the library
actually is, this would break, but it is easy to add it back when
we find such a system.
- The object file compat/zlib-uncompress2.o is always compiled and
archived in libgit.a, just like a few other compat/ object files
already are.
- The inclusion of <zlib.h> is done in <git-compat-util.h>; we used
to do so from <cache.h> which includes <git-compat-util.h> as the
first thing it does, so from the *.c codes, there is no practical
change.
- Until objects in libgit.a that is already used gains a reference
to the function, the reftable code will be the only one that
wants it, so libgit.a on the linker command line needs to appear
once more at the end to satisify the mutual dependency.
- Beat found a trick used by OpenSSL to avoid making the
conditionally-compiled object truly empty (apparently because
they had to deal with compilers that do not want to see an
effectively empty input file). Our compat/zlib-uncompress2.c
file borrows the same trick for portabilty.
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Helped-by: Beat Bolli <dev+git@drbeat.li>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
There are many situations in which having access to a cryptographically
secure pseudorandom number generator (CSPRNG) is helpful. In the
future, we'll encounter one of these when dealing with temporary files.
To make this possible, let's add a function which reads from a system
CSPRNG and returns some bytes.
We know that all systems will have such an interface. A CSPRNG is
required for a secure TLS or SSH implementation and a Git implementation
which provided neither would be of little practical use. In addition,
POSIX is set to standardize getentropy(2) in the next version, so in the
(potentially distant) future we can rely on that.
For systems which lack one of the other interfaces, we provide the
ability to use OpenSSL's CSPRNG. OpenSSL is highly portable and
functions on practically every known OS, and we know it will have access
to some source of cryptographically secure randomness. We also provide
support for the arc4random in libbsd for folks who would prefer to use
that.
Because this is a security sensitive interface, we take some
precautions. We either succeed by filling the buffer completely as we
requested, or we fail. We don't return partial data because the caller
will almost never find that to be a useful behavior.
Specify a makefile knob which users can use to specify one or more
suitable CSPRNGs, and turn the multiple string options into a set of
defines, since we cannot match on strings in the preprocessor. We allow
multiple options to make the job of handling this in autoconf easier.
The order of options is important here. On systems with arc4random,
which is most of the BSDs, we use that, since, except on MirBSD and
macOS, it uses ChaCha20, which is extremely fast, and sits entirely in
userspace, avoiding a system call. We then prefer getrandom over
getentropy, because the former has been available longer on Linux, and
then OpenSSL. Finally, if none of those are available, we use
/dev/urandom, because most Unix-like operating systems provide that API.
We prefer options that don't involve device files when possible because
those work in some restricted environments where device files may not be
available.
Set the configuration variables appropriately for Linux and the BSDs,
including macOS, as well as Windows and NonStop. We specifically only
consider versions which receive publicly available security support
here. For the same reason, we don't specify getrandom(2) on Linux,
because CentOS 7 doesn't support it in glibc (although its kernel does)
and we don't want to resort to making syscalls.
Finally, add a test helper to allow this to be tested by hand and in
tests. We don't add any tests, since invoking the CSPRNG is not likely
to produce interesting, reproducible results.
Signed-off-by: brian m. carlson <sandals@crustytoothpaste.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Code clean-up to hide vreportf() from public API.
* ab/usage-die-message:
config API: use get_error_routine(), not vreportf()
usage.c + gc: add and use a die_message_errno()
gc: return from cmd_gc(), don't call exit()
usage.c API users: use die_message() for error() + exit 128
usage.c API users: use die_message() for "fatal :" + exit 128
usage.c: add a die_message() routine
The conditions to choose different definitions of the FLEX_ARRAY
macro for vendor compilers has been simplified to make it easier to
maintain.
* jc/flex-array-definition:
flex-array: simplify compiler-specific workaround
We use "type array[];" syntax for the flex-array member at the end
of a struct under C99 or later, except when we are building with
older SUNPRO_C compilers. As we find more vendor compilers that
claim to grok C99 but not understand the flex-array syntax, the
existing "If we are using C99, but not with these compilers..."
conditional will keep growing.
Make it more manageable by listing vendor-specific exceptions
earlier, with the expectation that new exceptions will not be
combined into existing ones to make the condition longer, and
instead will be implemented as a new "#elif" in the cascade of
similar to old SUNPRO_C, we can just add a single line
#elif defined(_MSC_VER)
immediately before "#elif defined(__GNUC__)" to cause us to fallback
to the safer but a bit wasteful version.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Change the git_die_config() function added in 5a80e97c82 (config: add
`git_die_config()` to the config-set API, 2014-08-07) to use the
public callbacks in the usage.[ch] API instead of the the underlying
vreportf() function.
In preceding commits the rest of the vreportf() users outside of
usage.c was migrated to die_message(), so we can now make it "static".
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Change the "error: " output when we exit with 128 due to gc.log errors
to use a "fatal: " prefix instead. To do this add a
die_message_errno() a sibling function to the die_errno() added in a
preceding commit.
Before this we'd expect report_last_gc_error() to return -1 from
error_errno() in this case. It already treated a status of 0 and 1
specially. Let's just document that anything that's not 0 or 1 should
be returned.
We could also retain the "ret < 0" behavior here without hardcoding
128 by returning -128, and having the caller do a "return -ret", but I
think this makes more sense, and preserves the path from
die_message*()'s return value to the "return" without hardcoding
"128".
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
We have code in various places that would like to call die(), but
wants to defer the exit(128) it would invoke, e.g. to print an
additional message, or adjust the exit code. Add a die_message()
helper routine to bridge this gap in the API.
Functionally this behaves just like the error() routine, except it'll
print a "fatal: " prefix, and it will return with 128 instead of -1,
this is so that caller can pass the return value to "exit()", instead
of having to hardcode "exit(128)".
Note that as with the other routines the "die_message_builtin" needs
to return "void" and otherwise conform to the "report_fn"
signature.
As we'll see in a subsequent commit callers will want to replace
e.g. their default "die_routine" with a "die_message_routine".
For now we're just adding the routine and making die_builtin() in
usage.c itself use it. In order to do that we need to add a
get_die_message_routine() function, which works like the other
get_*_routine() functions in usage.c. There is no
set_die_message_rotine(), as it hasn't been needed yet. We can add it
if we ever need it.
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The C99 standard was released in January 1999, now 22 years ago. It
provides a variety of useful features, including variadic arguments for
macros, declarations after statements, designated initializers, and a
wide variety of other useful features, many of which we already use.
We'd like to take advantage of these features, but we want to be
cautious. As far as we know, all major compilers now support C99 or a
later C standard, such as C11 or C17. POSIX has required C99 support as
a requirement for the 2001 revision, so we can safely assume any POSIX
system which we are interested in supporting has C99.
Even MSVC, long a holdout against modern C, now supports both C11 and
C17 with an appropriate update. Moreover, even if people are using an
older version of MSVC on these systems, they will generally need some
implementation of the standard Unix utilities for the testsuite, and GNU
coreutils, the most common option, has required C99 since 2009.
Therefore, we can safely assume that a suitable version of GCC or clang
is available to users even if their version of MSVC is not sufficiently
capable.
Let's add a test balloon to git-compat-util.h to see if anyone is using
an older compiler. We'll add a comment telling people how to enable
this functionality on GCC and Clang, even though modern versions of both
will automatically do the right thing, and ask people still experiencing
a problem to report that to us on the list.
Note that C89 compilers don't provide the __STDC_VERSION__ macro, so we
use a well-known hack of using "- 0". On compilers with this macro, it
doesn't change the value, and on C89 compilers, the macro will be
replaced with nothing, and our value will be 0.
For sparse, we explicitly request the gnu99 style because we've
traditionally taken advantage of some GCC- and clang-specific extensions
when available and we'd like to retain the ability to do that. sparse
also defaults to C89 without it, so things will fail for us if we don't.
Update the cmake configuration to require C11 for MSVC. We do this
because this will make MSVC to use C11, since it does not explicitly
support C99. We do this with a compiler options because setting the
C_STANDARD option does not work in our CI on MSVC and at the moment, we
don't want to require C11 for Unix compilers.
In the Makefile, don't set any compiler flags for the compiler itself,
since on some systems, such as FreeBSD, we actually need C11, and asking
for C99 causes things to fail to compile. The error message should make
it obvious what's going wrong and allow a user to set the appropriate
option when building in the event they're using a Unix compiler that
doesn't support it by default.
Signed-off-by: brian m. carlson <sandals@crustytoothpaste.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The clean/smudge conversion code path has been prepared to better
work on platforms where ulong is narrower than size_t.
* mc/clean-smudge-with-llp64:
clean/smudge: allow clean filters to process extremely large files
odb: guard against data loss checking out a huge file
git-compat-util: introduce more size_t helpers
odb: teach read_blob_entry to use size_t
t1051: introduce a smudge filter test for extremely large files
test-lib: add prerequisite for 64-bit platforms
test-tool genzeros: generate large amounts of data more efficiently
test-genzeros: allow more than 2G zeros in Windows
The compatibility implementation for unsetenv(3) were written to
mimic ancient, non-POSIX, variant seen in an old glibc; it has been
changed to return an integer to match the more modern era.
* jc/unsetenv-returns-an-int:
unsetenv(3) returns int, not void
mingw-w64's pthread_unistd.h had a bug that mistakenly (because there is
no support for the *lockfile() functions required[1]) defined
_POSIX_THREAD_SAFE_FUNCTIONS and that was being worked around since
3ecd153a3b (compat/mingw: support MSys2-based MinGW build, 2016-01-14).
The bug was fixed in winphtreads, but as a side effect, leaves the
reentrant functions from time.h no longer visible and therefore breaks
the build.
Since the intention all along was to avoid using the fallback functions,
formalize the use of POSIX by setting the corresponding feature flag and
compile out the implementation for the fallback functions.
[1] https://unix.org/whitepapers/reentrant.html
Signed-off-by: Carlo Marcelo Arenas Belón <carenas@gmail.com>
Acked-by: Johannes Schindelin <Johannes.Schindelin@gmx.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
This compatilibity implementation has been returning a wrong type,
ever since 731043fd (Add compat/unsetenv.c ., 2006-01-25) added to
the system, yet nobody noticed it in the past 16 years, presumably
because no code checks failures in their unsetenv() calls. Sigh.
For now, make it always succeed.
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Remove the unused wrapper function.
Reported-by: Randall S. Becker <rsbecker@nexbridge.com>
Signed-off-by: Carlo Marcelo Arenas Belón <carenas@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Code cleanup.
* ab/repo-settings-cleanup:
repository.h: don't use a mix of int and bitfields
repo-settings.c: simplify the setup
read-cache & fetch-negotiator: check "enum" values in switch()
environment.c: remove test-specific "ignore_untracked..." variable
wrapper.c: add x{un,}setenv(), and use xsetenv() in environment.c
Adjust credential-cache helper to Windows.
* cb/unix-sockets-with-windows:
git-compat-util: include declaration for unix sockets in windows
credential-cache: check for windows specific errors
t0301: fixes for windows compatibility
Add fatal wrappers for setenv() and unsetenv(). In d7ac12b25d (Add
set_git_dir() function, 2007-08-01) we started checking its return
value, and since 48988c4d0c (set_git_dir: die when setenv() fails,
2018-03-30) we've had set_git_dir_1() die if we couldn't set it.
Let's provide a wrapper for both, this will be useful in many other
places, a subsequent patch will make another use of xsetenv().
The checking of the return value here is over-eager according to
setenv(3) and POSIX. It's documented as returning just -1 or 0, so
perhaps we should be checking -1 explicitly.
Let's just instead die on any non-zero, if our C library is so broken
as to return something else than -1 on error (and perhaps not set
errno?) the worst we'll do is die with a nonsensical errno value, but
we'll want to die in either case.
Let's make these return "void" instead of "int". As far as I can tell
there's no other x*() wrappers that needed to make the decision of
deviating from the signature in the C library, but since their return
value is only used to indicate errors (so we'd die here), we can catch
unreachable code such as
if (xsetenv(...) < 0)
[...];
I think it would be OK skip the NULL check of the "name" here for the
calls to die_errno(). Almost all of our setenv() callers are taking a
constant string hardcoded in the source as the first argument, and for
the rest we can probably assume they've done the NULL check
themselves. Even if they didn't, modern C libraries are forgiving
about it (e.g. glibc formatting it as "(null)"), on those that aren't,
well, we were about to die anyway. But let's include the check anyway
for good measure.
1. https://pubs.opengroup.org/onlinepubs/009604499/functions/setenv.html
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Available since Windows 10 release 1803 and Windows Server 2019.
NO_UNIX_SOCKETS is still the default for Windows builds, as they need
to keep backward compatibility with releases up to Windows 7, but allow
including the header otherwise.
Signed-off-by: Carlo Marcelo Arenas Belón <carenas@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Remove the USE_PARENS_AROUND_GETTEXT_N compile-time option which was
meant to catch an inadvertent mistake which is too obscure to
maintain this facility.
The backstory of how USE_PARENS_AROUND_GETTEXT_N came about is: When I
added the N_() macro in 6578483036 (i18n: add no-op _() and N_()
wrappers, 2011-02-22) it was defined as:
#define N_(msgid) (msgid)
This is non-standard C, as was noticed and fixed in 642f85faab (i18n:
avoid parenthesized string as array initializer, 2011-04-07).
I.e. this needed to be defined as:
#define N_(msgid) msgid
Then in e62cd35a3e (i18n: log: mark parseopt strings for translation,
2012-08-20) when "builtin_log_usage" was marked for translation the
string concatenation for passing to usage() added in 1c370ea4e5
(Show usage string for 'git log -h', 'git show -h' and 'git diff -h',
2009-08-06) was faithfully preserved:
- "git log [<options>] [<since>..<until>] [[--] <path>...]\n"
- " or: git show [options] <object>...",
+ N_("git log [<options>] [<since>..<until>] [[--] <path>...]\n")
+ N_(" or: git show [options] <object>..."),
This was then fixed to be the expected array of usage strings in
e66dc0cc4b (log.c: fix translation markings, 2015-01-06) rather than
a string with multiple "\n"-delimited usage strings, and finally in
290c8e7a3f (gettext.h: add parentheses around N_ expansion if
supported, 2015-01-11) USE_PARENS_AROUND_GETTEXT_N was added to ensure
this mistake didn't happen again.
I think that even if this was a N_()-specific issue this
USE_PARENS_AROUND_GETTEXT_N facility wouldn't be worth it, the issue
would be too rare to worry about.
But I also think that 290c8e7a3f which introduced
USE_PARENS_AROUND_GETTEXT_N misattributed the problem. The issue
wasn't with the N_() macro added in e62cd35a3e, but that before the
N_() macro existed in the codebase the initial migration to
parse_options() in 1c370ea4e5 continued passsing in a "\n"-delimited
string, when the new API it was migrating to supported and expected
the passing of an array.
Helped-by: Eric Sunshine <sunshine@sunshineco.com>
Signed-off-by: Ævar Arnfjörð Bjarmason <avarab@gmail.com>
Signed-off-by: Carlo Marcelo Arenas Belón <carenas@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Linux users may benefit from additional information on how to
avoid ENOMEM from mmap despite the system having enough RAM to
accomodate them. We can't reliably unmap pack windows to work
around the issue since malloc and other library routines may
mmap without our knowledge.
Signed-off-by: Eric Wong <e@80x24.org>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Fix typos in documentation, code comments, and RelNotes which repeat
various words. In trivial cases, just delete the duplicated word and
rewrap text, if needed. Reword the affected sentence in
Documentation/RelNotes/1.8.4.txt for it to make sense.
Signed-off-by: Andrei Rybak <rybak.a.v@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
The xsize_t helper aims to safely convert an off_t to a size_t,
erroring out when a file offset is too large to fit into a memory
address. It does this by using two casts:
size_t size = (size_t) len;
if (len != (off_t) size)
... error out ...
On a platform with sizeof(size_t) < sizeof(off_t), this check is safe
and correct. The first cast truncates to a size_t by finding the
remainder modulo SIZE_MAX+1 (see C99 section 6.3.1.3 Signed and
unsigned integers) and the second promotes to an off_t, meaning the
result is true if and only if len is representable as a size_t.
On other platforms, this two-casts strategy still works well (always
succeeds) for len >= 0. But for len < 0, when the first cast succeeds
and produces SIZE_MAX + 1 + len, the resulting value is too large to
be represented as an off_t, so the second cast produces implementation
defined behavior. In practice, it is likely to produce a result of
true despite len not being representable as size_t.
Simplify by replacing with a more straightforward check: compare len
to the relevant bounds and then cast it. (To avoid a -Wsign-compare
warning, after checking that len >= 0, we explicitly convert to a
sufficiently-large unsigned type before comparing to SIZE_MAX.)
In practice, this is not likely to come up since typical callers use
nonnegative len. Still, it's helpful to handle this case to make the
behavior easy to reason about.
Historical note: the original bounds-checking in 46be82dfd0 (xsize_t:
check whether we lose bits, 2010-07-28) did not produce this
implementation-defined behavior, though it still did not handle
negative offsets. It was not until 73560c793a (git-compat-util.h:
xsize_t() - avoid -Wsign-compare warnings, 2017-09-21) introduced the
double cast that the implementation-defined behavior was triggered.
Signed-off-by: Jonathan Nieder <jrnieder@gmail.com>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Streamline the codepath to fix the UTF-8 encoding issues in the
argv[] and the prefix on macOS.
* tb/precompose-prefix-simplify:
macOS: precompose startup_info->prefix
precompose_utf8: make precompose_string_if_needed() public
commit 5c327502 (MacOS: precompose_argv_prefix(), 2021-02-03) uses
the function precompose_string_if_needed() internally. It is only
used from precompose_argv_prefix() and therefore static in
compat/precompose_utf8.c
Expose this function, it will be used in the next commit.
While there, allow passing a NULL pointer, which will return NULL.
Signed-off-by: Torsten Bögershausen <tboegi@web.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
It does not make sense to make ".gitattributes", ".gitignore" and
".mailmap" symlinks, as they are supposed to be usable from the
object store (think: bare repositories where HEAD:.mailmap etc. are
used). When these files are symbolic links, we used to read the
contents of the files pointed by them by mistake, which has been
corrected.
* jk/open-dotgitx-with-nofollow:
mailmap: do not respect symlinks for in-tree .mailmap
exclude: do not respect symlinks for in-tree .gitignore
attr: do not respect symlinks for in-tree .gitattributes
exclude: add flags parameter to add_patterns()
attr: convert "macro_ok" into a flags field
add open_nofollow() helper
Make CALLOC_ARRAY usable like a function by requiring callers to supply
the trailing semicolon, which all of the current ones already do. With
the extra semicolon e.g. the following code wouldn't compile because it
disconnects the "else" from the "if":
if (condition)
CALLOC_ARRAY(ptr, n);
else
whatever();
Signed-off-by: René Scharfe <l.s.r@web.de>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Work around platforms whose open() is reported to return EINTR (it
shouldn't, as we do our signals with SA_RESTART).
* jk/open-returns-eintr:
config.mak.uname: enable OPEN_RETURNS_EINTR for macOS Big Sur
Makefile: add OPEN_RETURNS_EINTR knob
On some platforms, open() reportedly returns EINTR when opening regular
files and we receive a signal (usually SIGALRM from our progress meter).
This shouldn't happen, as open() should be a restartable syscall, and we
specify SA_RESTART when setting up the alarm handler. So it may actually
be a kernel or libc bug for this to happen. But it has been reported on
at least one version of Linux (on a network filesystem):
https://lore.kernel.org/git/c8061cce-71e4-17bd-a56a-a5fed93804da@neanderfunk.de/
as well as on macOS starting with Big Sur even on a regular filesystem.
We can work around it by retrying open() calls that get EINTR, just as
we do for read(), etc. Since we don't ever _want_ to interrupt an open()
call, we can get away with just redefining open, rather than insisting
all callsites use xopen().
We actually do have an xopen() wrapper already (and it even does this
retry, though there's no indication of it being an observed problem back
then; it seems simply to have been lifted from xread(), etc). But it is
used hardly anywhere, and isn't suitable for general use because it will
die() on error. In theory we could combine the two, but it's awkward to
do so because of the variable-args interface of open().
This patch adds a Makefile knob for enabling the workaround. It's not
enabled by default for any platforms in config.mak.uname yet, as we
don't have enough data to decide how common this is (I have not been
able to reproduce on either Linux or Big Sur myself). It may be worth
enabling preemptively anyway, since the cost is pretty low (if we don't
see an EINTR, it's just an extra conditional).
However, note that we must not enable this on Windows. It doesn't do
anything there, and the macro overrides the existing mingw_open()
redirection. I've added a preemptive #undef here in the mingw header
(which is processed first) to just quietly disable it (we could also
make it an #error, but there is little point in being so aggressive).
Reported-by: Aleksey Kliger <alklig@microsoft.com>
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
Some callers of open() would like to use O_NOFOLLOW, but it is not
available on all platforms. Let's abstract this into a helper function
so we can provide system-specific implementations.
Some light web-searching reveals that we might be able to get something
similar on Windows using FILE_FLAG_OPEN_REPARSE_POINT. I didn't dig into
this further.
For other systems without O_NOFOLLOW or any equivalent, we have two
options for fallback:
- we can just open anyway, following symlinks; this may have security
implications (e.g., following untrusted in-tree symlinks)
- we can determine whether the path is a symlink with lstat().
This is slower (two syscalls instead of one), but that may be
acceptable for infrequent uses like looking up .gitattributes files
(especially because we can get away with a single syscall for the
common case of ENOENT).
It's also racy, but should be sufficient for our needs (we are
worried about in-tree symlinks that we ourselves would have
previously created). We could make it non-racy at the cost of making
it even slower, by doing an fstat() on the opened descriptor and
comparing the dev/ino fields to the original lstat().
This patch implements the lstat() option in its slightly-faster racy
form.
Signed-off-by: Jeff King <peff@peff.net>
Signed-off-by: Junio C Hamano <gitster@pobox.com>
When commands are started from a subdirectory, they may have to
compare the path to the subdirectory (called prefix and found out
from $(pwd)) with the tracked paths. On macOS, $(pwd) and
readdir() yield decomposed path, while the tracked paths are
usually normalized to the precomposed form, causing mismatch. This
has been fixed by taking the same approach used to normalize the
command line arguments.
* tb/precompose-prefix-too:
MacOS: precompose_argv_prefix()
* maint-2.22:
Git 2.22.5
Git 2.21.4
Git 2.20.5
Git 2.19.6
Git 2.18.5
Git 2.17.6
unpack_trees(): start with a fresh lstat cache
run-command: invalidate lstat cache after a command finished
checkout: fix bug that makes checkout follow symlinks in leading path
* maint-2.21:
Git 2.21.4
Git 2.20.5
Git 2.19.6
Git 2.18.5
Git 2.17.6
unpack_trees(): start with a fresh lstat cache
run-command: invalidate lstat cache after a command finished
checkout: fix bug that makes checkout follow symlinks in leading path
* maint-2.20:
Git 2.20.5
Git 2.19.6
Git 2.18.5
Git 2.17.6
unpack_trees(): start with a fresh lstat cache
run-command: invalidate lstat cache after a command finished
checkout: fix bug that makes checkout follow symlinks in leading path
