| Age | Commit message (Collapse) | Author |
|---|
|
posix_fallocate implementation in uClibc relies on fallocate
system call - it just returns what fallocate returns. However
fallocate returns -1 on failure and assigns an error number
to errno variable. In the same time posix_fallocate must
return an error number but not -1.
What does this patch: if fallocate returns -1 then posix_fallocate
returns errno. Otherwise posix_fallocate returns 0 on success.
However there is a side effect - posix_fallocate sets errno on
failure because fallocate does it. But POSIX does not forbid it
thus it's not a problem.
Signed-off-by: Yuriy Kolerov <yuriy.kolerov@synopsys.com>
|
|
For common generic syscall ABI fallocate syscall handler in kernel
expects a 64-bit signed arguments for offset and len. However uClibc
has 2 wrappers for this syscall: fallocate and fallocate64.
On 32-bit machines fallocate (not fallocate64) expects 32-bit values of
offset and len. Thus in this case uClibc's fallocate must pass to the
syscall those values with sign extension. High word of 64-bit value must
be 0 or 0xFFFFFFFF depending on sign of the original 32-bit value (offset
or len). It is how sign extansion works - all high bits of the negative
value must be 1.
So on 32-bit machines uClibc's fallocate does sign extension incorrectly
when 32-bit values are passed (offset or len). It just fills the second
word of 64-bit value by zeros. E.g. fallocate works incorrectly when offset
or length is negative value - in this case kernel thinks that positive
values are passed.
Solution is to call fallocate64 from fallocate and pass 32-bit values of
offset and len to fallocate64. off_t type is automatically converted to
off64_t with an appropriate sign extension. Then fallocate64 invokes
kernel's system call properly.
This error is detected in LTP's test kernel/syscalls/fallocate02:
----------->8----------
fallocate(..., 1, -1024, 1024) failed, expected errno:22: TEST_ERRNO=0
fallocate(..., 1, 1024, -1024) failed, expected errno:22: TEST_ERRNO=0
fallocate(..., 1, 12288, -1024) failed, expected errno:22: TEST_ERRNO=0
fallocate(..., 1, -24576, 1024) failed, expected errno:22: TEST_ERRNO=0
----------->8----------
fallocate does not emit an error because negative values are passed to the
kernel without sign extension and kernel thinks that it got valid positive
values.
Signed-off-by: Yuriy Kolerov <yuriy.kolerov@synopsys.com>
|
|
fallocate system call must return 0 on success. On error, -1 is returned
and errno is set to indicate the error.
However there is an error in fallocate which is fixed by this patch - it
does not set errno and returns invalid value on error (it returns error
code instead of -1).
This error is detected in LTP's test kernel/syscalls/fallocate02:
----------->8----------
fallocate(..., 1, 0, 1024) failed, expected errno:9: TEST_ERRNO=0
fallocate(..., 1, -1024, 1024) failed, expected errno:22: TEST_ERRNO=0
fallocate(..., 1, 1024, -1024) failed, expected errno:22: TEST_ERRNO=0
fallocate(..., 1, 12288, 0) failed, expected errno:22: TEST_ERRNO=0
fallocate(..., 1, 12288, -1024) failed, expected errno:22: TEST_ERRNO=0
fallocate(..., 1, -24576, 1024) failed, expected errno:22: TEST_ERRNO=0
----------->8----------
Signed-off-by: Yuriy Kolerov <yuriy.kolerov@synopsys.com>
|
|
uClibc doesn't define signal info for the SIGSYS signal which is issued
in case of hitting a syscall prohibited by seccomp.
This is sad as it makes debugging seccomp filter policies impossible on
some architectures (at least ARM and PowerPC, maybe also others) which
do not coincidentally set si_value.sival_int as the syscall number.
To fix this, import the definitions and macros needed from glibc.
Signed-off-by: Daniel Golle <daniel@makrotopia.org>
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
|
|
This reverts commit bd3eaf83ef1b4954b6c0e7ba8bbdd29b2cd4a833.
They are now implemented.
|
|
When compiling python you get duplicate symbol problem.
Seen in the autobuilders of buildroot project.
|
|
Only the stub warnings left for now.
|
|
|
|
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
|
|
Introduce a <sys/random.h> for it.
/* FIXME: aren't there a couple of __restrict and const missing ?
*/
extern int getrandom(void *__buf, size_t count, unsigned int flags)
__nonnull ((1)) __wur;
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
|
|
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
|
|
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
|
|
https://dev.openwrt.org/browser/trunk/toolchain/uClibc/patches-0.9.33.2/616-mips_fix_stat_time.patch
When booting in Qemu you get for example failures from mdev -s otherwise.
|
|
Add __LEAF to all __THROW, introduce non-leaf __THROWNL
Adjust affected spots accordingly.
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
|
|
Sync with GNU libc, see here why r7 is not usable for
thumb mode:
https://sourceware.org/git/?p=glibc.git;a=blob;f=sysdeps/unix/sysv/linux/arm/sysdep.h;h=37eac192b1e2e7f53f112b16450b9ce57054e27f;hb=HEAD
Remove old OABI support.
Fixes build breakage for ARM noMMU builds.
Still no runtime testing.
|
|
|
|
include/bits/sigset.h:219:62: warning: right-hand operand of comma expression has no effect [-Wunused-value]
include/bits/sigset.h:210:63: warning: right-hand operand of comma expression has no effect [-Wunused-value]
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
|
|
|
|
|
|
Currently UCLIBC_SYSCALL_ALIGN_64BIT is not explicitly handled. Fix that
and make sure the special handling is done for powerpc/xtensa which use
UCLIBC_SYSCALL_ALIGN_64BIT but don't use hole punched syscall handler in
kernel.
Signed-off-by: Vineet Gupta <vgupta@synopsys.com>
Cc: Baruch Siach <baruch@tkos.co.il>
Cc: Max Filippov <jcmvbkbc@gmail.com>
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
|
|
arm/powerpc/xtensa pass @advice as 2nd arg to syscall (vs. canonical 4th)
Current code however does this for UCLIBC_SYSCALL_ALIGN_64BIT which
powerpc/xtensa also happen to define.
This is not true for ARCv2 ISA and possibly other arch of future with
64-bit even register requirement, which uses the standard syscall
handler in kernel.
Fix that by providing 2 variants of SYSCALL_ALIGN_64BIT
Signed-off-by: Vineet Gupta <vgupta@synopsys.com>
Cc: Baruch Siach <baruch@tkos.co.il>
Cc: Max Filippov <jcmvbkbc@gmail.com>
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
|
|
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
|
|
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
|
|
This reverts commit 6b6ede3d15f04fe825cfa9f697507457e3640344.
|
|
|
|
|
|
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
|
|
Fix copy'n paste error for ppc64 and other sync_file_range2 arches
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
|
|
mknod() in glibc/eglibc will check the argument, like this,
...
if (k_dev != dev) {
__set_errno (EINVAL);
return -1;
}
...
So add argument check in uclibc's mknod() too.
Signed-off-by: Wang Yufen <wangyufen@huawei.com>
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
|
|
|
|
Fixes compile errors with latest util-linux.
|
|
|
|
The GNU variant of getopt() previously had no way to turn off
getopt_long() support.
|
|
mknod() in glibc/eglibc will check the argument, like this,
...
if (k_dev != dev) {
__set_errno (EINVAL);
return -1;
}
...
So add argument check in uclibc's mknod() too.
Signed-off-by: Wang Yufen <wangyufen@huawei.com>
|
|
|
|
statfs() is a Linux-specific function. When building without
UCLIBC_LINUX_SPECIFIC set, libc_hidden_proto(statfs) in
include/sys/statfs.h is unmatched with libc_hidden_def(statfs)
in libc/sysdeps/linux/common/statfs.c, leading to a compile
error. This patch fixes this, as well as adds statfs() to the
list of Linux-specific functions in extra/Configs/Config.in.
Signed-off-by: Anthony G. Basile <blueness@gentoo.org>
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
|
|
Signed-off-by: Khem Raj <raj.khem@gmail.com>
|
|
e.g. fminf() is not implemented, only fmin(), but both are defined.
Signed-off-by: Cristian Morales Vega <cristian@samknows.com>
|
|
define NOT_IN_libc / IS_IN_libxxx appropriately
to fix pthread_once
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
|
|
We add the Linux-specific function fallocate() which allows the user to
directly manipulate allocate space for a file. fallocate() can operate
in different modes, but the default mode is equivalent to posix_fallocate()
which is specified in POSIX.1.
Recent releases of e2fsprogs 1.42.11 and above expect fallocate64() to be
available.
Signed-off-by: Anthony G. Basile <blueness@gentoo.org>
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
|
|
We cannot override individual functions on a per-arch basis otherwise
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
|
|
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
|
|
Signed-off-by: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
|
|
There was a runtime error in systems without large file support. Call
fseek(fd, 4096, SEEK_SET) has been failing with EINVAL, though it was
succeeding for offset = 4092. This has been happening because llseek system
call accepts 64-bit value as an offset argument and lseek function has been
ordering 32-bits words that form this offset value, according to the
endianness. However this ordering to match endianness is not required,
because llseek doesn't accept one 64-bit offset argument, it accepts two
32-bit offset argument, then stitches them into one following its
endianness. As a result on little endian system, order of words has been
swapped two time: in libc and in kernel. Thus call to fseek with offset 4096
(0x1000) was doing a system call to llseek with offset 0x1000_0000_0000. I'm
not entirely sure why then offset = 4092 hasn't been failing then.
This patch removes malicious swap of words when calling llseek.
Signed-off-by: Anton Kolesov <Anton.Kolesov@synopsys.com>
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
|
|
commit 00571b43df2e "libc: posix_fadvise: restore implementation for xtensa"
enabled posix_fadvise() for all arches (it was just not generated
before).
However this also unearthed an issue introduced by ee84b8b400
"linux: posix_fadvise: use new SYSCALL_ALIGN_64BIT" which is to
referencing LFS'ish code (off64_t) w/o proper checks which causes build
to break for !LFS.
Fix this by calling posix_fadvise64() only for LFS case and open-code
it's equivalent for !LFS.
Signed-off-by: Vineet Gupta <vgupta@synopsys.com>
Cc: Mike Frysinger <vapier@gentoo.org>
Cc: Baruch Siach <baruch@tkos.co.il>
Cc: Khem Raj <raj.khem@gmail.com>
Cc: markos Chandras <markos.chandras@gmail.com>
Cc: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
|
|
Signed-off-by: Mike Frysinger <vapier@gentoo.org>
|
|
Signed-off-by: Vineet Gupta <vgupta@synopsys.com>
Cc: Mike Frysinger <vapier@gentoo.org>
Cc: Denys Vlasenko <vda.linux@googlemail.com>
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
|
|
For ARC builds atleast, __NR_xxx was not trickling into in open64.c
causing open64->openat a needless detour via open
Signed-off-by: Vineet Gupta <vgupta@synopsys.com>
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
|
|
Patch "LT.old: Make __errno_location/__h_errno_location thread safe"
uncovered yet another bug with static linking and errno (hopefully this
is last of them all).
Currently, __errno_location is declared weak but is defined strong.
While this provides with the desired weak semantics in dso, it
is subtly broken in static links.
Quoting Joern Rennecke (ARC gcc expert):
| I think the issue is that you declare the function as weak in the
| header file. That is a rare instance where you want the reference
| use declaration that differs a bit from the definition.
| If the reference uses a weakly declared function, that creates a
| weakref, i.e. the linker won't bother to look for this symbol at
| all - if it gets linked in for some other reason, fine,
| otherwise, it stays zero.
So the solution to declare strong, define weak.
Supporting data
-----------------
orig code: ARM mmap wrapper (LT.old build + my prev patch for errno)
_mmap:
@ args = 8, pretend = 0, frame = 0
@ frame_needed = 0, uses_anonymous_args = 0
stmfd sp!, {r4, r5, r7, lr}
ldr r5, [sp, #20]
movs ip, r5, asl #20
beq .L2
bl __errno_location(PLT)
mov r3, #22
str r3, [r0, #0]
mvn r0, #0
...
...
.weak __errno_location
A statically linked hello world program which uses mmap too.
As we can see__errno_location is completely gone - which is
semantically wrong - we need functional errno.
00008274 <__GI_mmap>:
8274: e92d40b0 push {r4, r5, r7, lr}
8278: e59d5014 ldr r5, [sp, #20]
827c: e1b0ca05 lsls ip, r5, #20
8280: 0a000004 beq 8298
8284: e320f000 nop {0}
^^^^^^^^^^
8288: e3a03016 mov r3, #22
828c: e5803000 str r3, [r0]
8290: e3e00000 mvn r0, #0
This in turn is due to a fixup in ARM ld which transforms branch-to-null
into a nop. It is better than crashing but still wrong since errno
handling is removed.
With the patch, errno_location is restored back in test program.
00008274 <__GI_mmap>:
8274: e92d40b0 push {r4, r5, r7, lr}
8278: e59d5014 ldr r5, [sp, #20]
827c: e1b0ca05 lsls ip, r5, #20
8280: 0a000004 beq 8298 <__GI_mmap+0x24>
8284: eb000010 bl 82cc <__errno_location>
8288: e3a03016 mov r3, #22
828c: e5803000 str r3, [r0]
Cc: Christian Ruppert <christian.ruppert@abilis.com>
CC: Francois Bedard <Francois.Bedard@synopsys.com>
Cc: Anton Kolesov <Anton.Kolesov@synopsys.com>
Cc: Joern Rennecke <joern.rennecke@embecosm.com>
Cc: Jeremy Bennett <jeremy.bennett@embecosm.com>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Peter Korsgaard <peter@korsgaard.com>
Cc: Khem Raj <raj.khem@gmail.com>
Signed-off-by: Vineet Gupta <vgupta@synopsys.com>
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
|
|
WHY: errno in uClibc is not thread safe
HOW: __errno_location and it's sibling __h_errno_location need to be called
via PLT - even from within libc. That way when linked with pthread,
intra-uClibc callers will also use the thread safe version.
This is achieved by removing the GI alias for these functions,
ensuring that they get called normally (via PLT)
Verified with ARC LT.old and ARM cubieboard2 buildroot (LT.old)
NPTL is unaffected by this bug.
------------ History behind this patch ---------------------
This is a long standing bug (https://bugs.busybox.net/show_bug.cgi?id=2089)
Others have tried to fix it in past (alteast Peter Korsgaard's patch in 2010),
but somehow failed to be merged (or were backed out afterwards).
http://lists.uclibc.org/pipermail/uclibc/2010-July/044176.html
One of the causes could be side effect of atleast one more bug related to
pthreads and static link which has now been fixed.
http://lists.uclibc.org/pipermail/uclibc/2013-October/047958.html
I have solved this w/o looking at other pacthes but would like to give
credit to Peter and others for confirming that it makes sense.
Cc: Christian Ruppert <christian.ruppert@abilis.com>
CC: Francois Bedard <Francois.Bedard@synopsys.com>
Cc: Joern Rennecke <joern.rennecke@embecosm.com>
Cc: Jeremy Bennett <jeremy.bennett@embecosm.com>
Cc: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
Cc: Peter Korsgaard <peter@korsgaard.com>
Cc: Khem Raj <raj.khem@gmail.com>
Cc: buildroot@busybox.net
Signed-off-by: Vineet Gupta <vgupta@synopsys.com>
Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
|
