Add vector version mem* and str* functions and only build them when the
vector extension is enabled.
The original implementation comes from
https://github.com/sifive/sifive-libc, which we agree to contribute to
the Android Open Source Project.
Test: mma
Change-Id: I11b671a5bc571d7c783a657f272f282df7d16c29
Signed-off-by: Yun Hsiang <yun.hsiang@sifive.com>
Now the NDK doesn't support API levels below 21, we don't actually need
weird x86-specific annotations. In general we haven't been removing
this historical information because it might be interesting to someone,
and there's no real reason to remove it, but we've had versioner bugs
recently with these more complex cases.
Test: treehugger
Change-Id: Ia457bb338ecf55af8e319e411ec3bf48a03f3c03
Now the NDK doesn't support API levels below 21, we don't actually need
the different arm32 vs x86 annotations. In general we haven't been
removing this historical information because it might be interesting to
someone, and there's no real reason to remove it, but we've had
versioner bugs recently with these more complex cases.
Test: treehugger
Change-Id: Id9c9b8ecc01d232becd5dd8741509c104a8b6e19
C23 adds timegm(), gmtime_r(), and localtime_r(). We should remove the
"non-standard" text for timegm(), and while I'm here, let's just
document everything in this file.
Test: treehugger
Change-Id: Ia44c1bd155c939f694f6f8138b9cb7503519522c
No actual effect on the code, but misleading and wrong. (The previous
change only fixed the argument types; I didn't notice that some of the
return types were wrong too.)
Test: treehugger
Change-Id: I1ee5c48e2652fd8cbf8178d5659e57f79e61898e
Neither is great, but "gp" seems actively misleading (and setjmp.S
says x3 every time, so we should be consistent if nothing else).
Bug: https://github.com/riscv-non-isa/riscv-elf-psabi-doc/pull/379
Test: treehugger
Change-Id: Ibccda74d4794caa770b82e7ba2e31ce7b645b83f
The only remaining differences between vfprintf.cpp and vfwprintf.cpp
after this are the wide/narrow conversions for %c, %m, and %s. I've used
"chars" and "bytes" for the named constants for the directions because
(a) I find -1 and 1 pretty confusing and (b) although "narrow" is the
obvious opposite of "wide", only Windows actually moved to wide
characters, so "narrow" (aka "multibyte", and probably "utf8") is the
default/normal case. Even though C confuses bytes and characters via its
`char` type, "bytes" versus "chars" seems like the appropriate
terminology (and it's what Java/Python use).
Also improve the swprintf tests assertion so failures are readable.
Test: treehugger
Change-Id: Ife8f70f65ec28d96058a7d68df353945524835d2
wprintf doesn't need this (and already only has the iov stuff because
the non-wide printf implementation needs it), but we can further reduce
the diff between the two implementations by defining a no-op FLUSH() for
wide characters.
Test: treehugger
Change-Id: Ifefcb4b8474b086f995e2b0796f61558a19e2a42
Jens Gustedt suggested a better implementation last year on the musl
mailing list: https://www.openwall.com/lists/musl/2022/11/19/1
It means the constants are sparse, but in return it means we can add
future constants and they'll be backward compatible. (Sadly you'll need
to be on API level 35 before you can use anything but TIME_UTC.)
I doubt this will ever matter, because everyone should just stick to
clock_gettime()/clock_getres() anyway, and anyone who does have a
legitimate use for timespec_get() and timespec_getres() probably needs
to support non-Linux and so can't use any clocks that aren't in ISO C
anyway. But given that we don't _have_ to paint ourselves into a corner
here, we may as well take the opportunity to not do so.
Test: strace
Change-Id: I293d32fcbcf7f6703564dac0978ae2a10192a482
This is the one openlog() flag that toybox uses. We should probably try
to unify toybox's POSIX logger and Android-specific log at some point,
and this will help.
Also fix our behavior with an empty format string, noticed while adding
tests.
Test: treehugger
Test: adb shell logger -s foo
Change-Id: Ic027e78a460be3db83cc4c6f9946c9efa22be6e1
That lets us have idempotent #defines to reduce namespace pollution.
Bug: http://b/279405445
Test: treehugger
Change-Id: I00312cc8911025696cf0eea2d70e3219ab361613
wfN: Specifies that a following b, d, i, o, u, x, or X conversion specifier applies to a fastest minimum-width integer argument with a specific width where N is a positive decimal integer with no leading zeros (the argument will have been promoted according to the integer promotions, but its value shall be converted to the unpromoted type); or that a following n conversion specifier applies to a pointer to a fastest minimum-width integer type argument with a width of N bits. All fastest minimum-width integer types (7.22.1.3) defined in the header <stdint.h> shall be supported. Other supported values of N are implementation-defined.
Bug: b/271903607
Test: adb shell
Change-Id: Ida36d5a50af2a46fd04cb5fe039793d8872f9f3b
Although this breaks job control in several shells (including mksh),
this has been broken since the initial commit and no-one's noticed until
now.
Bug: https://github.com/android/ndk/issues/1878
Test: treehugger
Change-Id: Id7c4805965c5e5847db99b57df1af13355adcc22
57b8fc957a
changes the way overflows are tracked: now compiler builtins
are used instead of manual arithmetics. But as int_fast32_t on
64-bit Android takes 8 bytes, new logic behaves differently.
See time_test.cpp changes for more details.
Changes were applied using following commands:
1) Checkout tzcode repo
2) Prepare patches for all tzcode file using
git diff 2022a 2023a -- <file-name> > <file-name-patch>
3) Apply these patches to files in bionic using
patch -p1 <file-name> <file-name-patch>
Bug: 279742606
Test: CtsBionicTestCases
Test: CtsLibcoreTestCases
Test: CtsLibcoreOjTestCases
Test: atest toybox-tests
Change-Id: I7772a90538b8185bdd2f4be6e9d1740c95509d6c
The original wording implies that memory tagging can be disabled
per thread. That is not really possible, and doesn't happen. Instead,
this option only turns off some of the memory tagging initialization.
Leave this slightly vague since it's possible this might change
in the future.
Bug: 272383932
Test: NA
Change-Id: I6333f16175edb52a38aadfc67ce7c9ce76113e59
When annotating the netinet directory aosp/2552567, we realize the
argment s for vsnprintf family can be null only if the buffer size is 0.
So we correct them and add some tests to verify our assumption.
Bugs: b/245972273
Test: adb shell
Change-Id: I51063286272be0daee0d7c1453a374b1f5674481
Nothing to see here --- you'll want to keep using POSIX clock_gettime()
and clock_getres() instead. But portable code might use this eventually,
and it's trivial, so let's add it anyway.
(The whole "zero as an error return" precluding the direct use of
Linux's CLOCK_ constants is what really makes this a terrible API ---
we're going to have to add explicit translation any time they add a
new base.)
Test: treehugger
Change-Id: Iddb6cbe67b67b2b10fdd8b5ee654896d23deee47
Fix the uapi import script to look at the riscv/asm/ directory too, and
re-run it to add the missing SYS_riscv_* entry (there's only this one)
to glibc-syscalls.h.
Test: treehugger
Change-Id: Ie52c6ca1943c05bb615932174e49e7fb79725a7b
wN: Specifies that a following b, d, i, o, u, x, or X
conversion specifier applies to an integer argument with
a specific width where N is a positive decimal integer with
no leading zeros
Bug: b/271903607
Test: adb shell
Change-Id: I688f6cefeb2e5c8325b007a59935a46f4116ac29
This reverts commit fe2907c133.
Reason for revert: Breaks bionic-unit-tests b/278795547. Since the original mixed build change is being reverted for now (aosp/2547450), fe2907c133 is not needed immediately
Change-Id: I2deb06a38322bf8296d4721c840f06f35b757177
Contrary to the old comment, POSIX says nothing about whether or not
tmpfile() respects $TMPDIR, and it's significantly more useful on
Android if it does (because there's no shared /tmp that everyone can
write to).
Bug: https://issuetracker.google.com/36991167
Test: treehugger
Change-Id: I3cc45adff167420f100c8ed1c63cba1ea67e9f70
bp2build will generate the stub targets only if versions is not empty.
Test: b query //bionic/libc:* | grep libc_hwasan_stub_libs-current
Change-Id: Iac905497ae4955a44b7b29e2d29a2c702c86da8e
This mode instructs the linker to search for libraries in hwasan
subdirectories of all library search paths. This is set up to contain a
hwasan-enabled copy of libc, which is needed for HWASan programs to
operate. There are two ways this mode can be enabled:
* for native binaries, by using the linker_hwasan64 symlink as its
interpreter
* for apps: by setting the LD_HWASAN environment variable in wrap.sh
Bug: 276930343
Change-Id: I0f4117a50091616f26947fbe37a28ee573b97ad0
We want to give back a useful callee-saved general purpose
register (x18) that was only "chosen" because it was what llvm
allowed for historical reasons. gp is a better choice because it's
effectively unused otherwise anyway.
Unfortunately, that means we need extra space in jmp_buf (which I've
reserved in an earlier change, e7b3b8b467),
so let's rearrange the entries in jmp_buf to match their order in the
register file.
Bug: https://github.com/google/android-riscv64/issues/72
Bug: http://b/277909695
Test: treehugger
Change-Id: Ia629409a894c1a83d2052885702bbdd895c758e1
Bazel doesn't like it when modules produce files with the same name
as the module itself, and gives warnings.
Rename either the module or file in this case so that the file has
an extension and the module doesn't.
Bug: 198619163
Test: m nothing
Change-Id: Ic4592b06f575496ffd54ac75cb4d682118b29d93
If we switch from x18 to gp for shadow call stack, we're going to need
another slot in jmp_buf. We'll need this even for hardware shadow call
stacks too.
While I'm here, and because this is likely my last chance, let's just
round this up to 32 for safety. musl and glibc only have the minimum
needed (which I think means they'll need an ABI break to support SCS
unless they just use a callee-saved general purpose register), but since
we can't do ABI breaks after we ship, let's play it safe.
Bug: https://github.com/google/android-riscv64/issues/72
Test: treehugger
Change-Id: I60661fb7a308c900bfd08c9361f51919b798c005
The recent header nullability additions and the corresponding source
cleanup made me notice that we're missing a couple of actions that most
of the other implementations have. They've also been added to the _next_
revision of POSIX, unchanged except for the removal of the `_np` suffix.
They're trivial to implement, the testing is quite simple too, and
if they're going to be in POSIX soon, having them accessible in older
versions of Android via __RENAME() seems useful. (No-one else has shipped
the POSIX names yet.)
Bug: http://b/152414297
Test: treehugger
Change-Id: I0d2a1e47fbd2e826cff9c45038928aa1b6fcce59
Now <utmpx.h> isn't any more useful on Android than <utmp.h> is, but it
is POSIX, and -- importantly -- we can implement it with just a header
file, so code can use it on every existing API level.
macOS does indeed only have the <utmpx.h> functions (although it does
still have the <utmp.h> header!), so potentially portable code might
want <utmpx.h> on Android. (glibc/musl both have both headers.)
Bug: https://github.com/landley/toybox/pull/213
Test: treehugger
Change-Id: Iaa88167708182009a63e2e1a15f11186b251ed02
All the other architectures are already polluting the namespace with
`struct ucontext`, so make riscv64 match for source compatibility with
other Android code. (Code _should_ be using the POSIX `ucontext_t`, but
ART in particular had a lot of `struct ucontext`, and although I'll
clean that up separately, if there's some in our tree, there's probably
a lot more out there in the wild.)
Test: treehugger
Change-Id: Id0e4e97e660d7d60e792cd2462ddb9788d4772d7
We're going to dereference a null pointer if you pass one instead of a
pointer to a path, but at the moment (because of implementation sharing
between the different file actions) we won't do it until the last
minute, in the child itself. Let's crash as soon as you make the mistake
instead, to make debugging a lot easier.
Test: treehugger
Change-Id: I987d2700ba05b9867a936ebe770224259376633f
We don't really need <linux/compiler.h> and <linux/compiler_types.h>. We
already have a mechanism to remove unused macros, so let's do that. We
don't currently have a way to remove unused #includes, so we still need
<linux/compiler.h> and <linux/compiler_types.h> files (but I've clarified
the comments in them).
I've kept the empty definitions of `__user` and `__force` for source
compatibility. (We had one security test at least that was assuming
a kernel struct definition will "just work".)
Bug: http://b/262917450
Test: treehugger
Change-Id: Iacbbbc1aeef9a4fac52dabd7811ab875cc267d4f
Move the "is there a comparator?" check into the sole caller, to match
the "is there a filter?" check. Remove the unnecessary (and unlikely)
pre-sort "is the array empty?" check.
Test: treehugger
Change-Id: I8bd461380420dce4a8bc05ef5fe3511b26347d7c
No idea if this will be the actual API level or not, but that's all
the more reason to abstract it away early...
Bug: None
Test: None
Change-Id: I8a17bb42dbb08a6e760427514af8331e7dc9b549
Since it doesn't matter if these calls take a little longer than
before, use the more thorough but slightly longer purge mechanism.
Test: Unit tests pass.
Change-Id: Ifab7166a9682a13231746b78717d52673d13be1b
The strerror_buf is way too large, so instead of using a separate
buffer for just this string, reuse the already existing buffer.
Increase the buffer size to cover the maximum errno string.
Add a unit test to verify that none of the errno values are cut off
in the async_safe_format_buffer function when passing %m.
Bug: 274474681
Test: New unit test passes.
Test: Changing the buffer to a small value and verify that the test fails.
Change-Id: I4cb4652709582a8a6b958e12de5d923ec950e6b6
Use __memcpy_chk assembly to replace the implementation of c functions, which can reduce the use of instructions
Test: llvm-objdump
Change-Id: I5d75601626dc997626f6173d53af301183a64004
Signed-off-by: caowencheng <caowencheng@eswincomputing.com>
As of https://reviews.llvm.org/D143769, binaries (with -fsanitize=memtag-*)
have DT_AARCH64_MEMTAG_* dynamic entries, as per the AArch64 MemtagABI.
Android uses an OS-specific ELF note for MTE config, but we should
migrate to the new thing (while preserving backwards compatibility).
Without actually doing the migration right now, just handle these new
entries. Otherwise, you get a whole bunch of logspam about the
unrecognised dynamic entries.
Bug: 274032544
Test: Build android, don't get logspam.
Change-Id: I5c8b59f77a0058e5b93335e269d558a5014f2260
The buffer filled in by strerror_r needs to stay in scope while
it is pointed to by str.
Bug: 273807460
Change-Id: I494ca8b8aca2b28ec2f0f3da72d845db99633553
This is a new mallopt option that will force purge absolutely
everything no matter how long it takes to purge.
Wrote a unit test for the new mallopt, and added a test to help
verify that new mallopt parameters do not conflict with each other.
Modified some benchmarks to use this new parameter so that we can
get better RSS data.
Added a new M_PURGE_ALL benchmark.
Bug: 243851006
Test: All unit tests pass.
Test: Ran changed benchmarks.
Change-Id: I1b46a5e6253538108e052d11ee46fd513568adec
clang-r487747 added stack protector capability to check noreturn calls.
This caused the system to boot loop. Turn off the new capabilities as a
temporary workaround.
Test: build and boot
Change-Id: I62c912619dfdd2384672d504ce5d52330bf2a102
If the 'j' command is used here,it cannot always be called.
The 'tail' command is used here, let the compiler decide
which instruction to use,when the call distance is less
than 1M, it will be compiled into 'j' command, and when
the distance is greater than 1M, it will be compiled
into 'aupic' and 'jr' command.
Test: llvm-objdump -d
Change-Id: I53d8aa7f54b9c4c96fce491487dcba7b63348219
Signed-off-by: caowencheng <caowencheng@eswincomputing.com>
Don't assume that a `j` will always be in range. Let the toolchain
come up with the shortest sequence that will actually work.
Test: llvm-objdump -d
Change-Id: I497a68ecae434aad173c2b1c8024aed3756b8440
the "tail" command is used here, let the compiler decide
which instruction to use. when the call distance is less
than 1M, it will be compiled into "jr" command, and when
the distance is greater than 1M, it will be compiled
into "aupic" and "jr" command.
Test: make libc
Change-Id: I80d099d25c7cc2439a297d8afc65abeb5a7a360c
Signed-off-by: caowencheng <caowencheng@eswincomputing.com>
Guarding the definitions of registers with `#if definded(__USE_GNU)`.
This resolves a name conflict with the mesa3d library when
Virgl is enabled.
Test: launch_cvd -console=true -guest_enforce_security=false\
--gpu_mode=drm_virgl
Change-Id: I4402500eff74e26343700c099dd397b8a19a5ee3
Upstream made a change similar to our %s change, so we don't need that
difference any more. (But they didn't seem interested in our GNU
extensions, even though they're sensible ones that just ensure symmetry
between strftime() and strptime().)
Bug: http://b/167569813
Test: treehugger
Change-Id: I1d86d69c87b51719f0583341fafa7802869cd37e
Writing an assembly file can reduce several
assembly instructions, and compare it through
the disassembly of this file
Test: make libc
Change-Id: Ifdcc9c76742cc95b2ad9e3c14fac4796c36e12e6
Signed-off-by: caowencheng <caowencheng@eswincomputing.com>
Using %m is slightly less code, and means that perror() wouldn't touch
the strerror buffer in TLS. Doesn't matter, but noticed while I was
looking at this for another reason, and since it's _less_ code...
Test: treehugger
Change-Id: I99d31a5c326962958ad843ec4da2c3cc3b93b048
These are just one-liners, and the _FLOCK() macro seemed to me more
obscure than just inlining it (especially because there are only four
call sites total, so it's not like anyone's going to see that macro very
often).
Also add the missing CHECK_FP() calls. I don't expect this to break
anything, but if it does we can add a target API level check.
Test: treehugger
Change-Id: Ifa1a39d5d9eee46cca783acbe9ec3b3a1e6283d9
There's negative value to explicitly calling out Linux and ELF here,
when Android -- and especially this project -- is always both anyway.
No functional change.
Test: treehugger
Change-Id: Ic7514d74a5b5f6f06e64c46486c12773995076cd
The next NDK to take these headers only supports API 21 and later.
Note that this change leaves the _implementation_ of these functions
behind, so that any old apps calling these APIs should continue to work,
you just can't (without declaring the functions yourself) write new ones
that do (and declaring the functions yourself would only work on LP32
anyway, so that's not going to get you very far in 2023).
Test: treehugger
Change-Id: Ie03514e4215b40f6e9feaa6e4bf5df5b16dc8d59
The next NDK to take these headers only supports API 21 and later, but
even if it didn't --- there is no inline!
Test: treehugger
Change-Id: Ibb194e1cbf0a8551bff863d940d92c35ed06a1e1
The next NDK to take these headers only supports API 21 and later, so
clean up some of the trivial cruft.
Test: treehugger
Change-Id: Ib735a776d244cc82858f2ed629dd63a54dbaf650
The next NDK to take these headers only supports API 21 and later, so
clean up some of the trivial cruft.
This doesn't include the remaining "legacy inlines", since they're a bit
more complicated. I'll remove those in later changes.
Test: treehugger
Change-Id: I94c32f6393dd3ae831165917303ea591222baa0d
API level 19 is no longer supported by the NDK.
While I'm here, let's remove the duplicated structure defintion (as
we've already done for `struct stat`).
Test: treehugger
Change-Id: I9d8286f9e7ba803f3131b6dcb0486ff1b0f9d5d1
Get bsearch and other baremetal-compatible functions from NetBSD.
Bug: 249723852
Bug: 262344886
Test: -
Change-Id: Idd90188ae13c84ed35af8f1475370cec0fe8cdbb
These cause great confusion, so explicitly point out that apps can't use
one, and probably want the other.
Bug: https://github.com/android/ndk/issues/1255
Test: N/A
Change-Id: I287e820dc45a8446e3c72c9a2e4007db76828e3b
