| Age | Commit message (Collapse) | Author |
|---|
|
Add some definitions and fix some types to make gdb compile
happy.
|
|
elf-fdpic.h or elf-dsbt.h is included by link.h. When C++ program
includes <link.h>, we get following build failure.
../usr/include/bits/elf-fdpic.h: In function 'void* __reloc_pointer(void*, const elf32_fdpic_loadmap*)':
../usr/include/bits/elf-fdpic.h:95: error: invalid use of 'void'
void pointer addition and subtraction is not allowed in C++ as it has
undetermined size, however in C with language extension it is possible
because sizeof void is treated as one byte.
Instead of performing subtraction on void pointers, typecast it to char*
first.
This build failure is detected by Buildroot autobuilder.
http://autobuild.buildroot.net/results/a10/a10ed48e6eb8411a3d8372f57c05fd11130da0e0/
Signed-off-by: Rahul Bedarkar <rahul.bedarkar@imgtec.com>
|
|
A single test with targeting ARM showed that this feature
seems bit rotted. Remove DOMULTI and simplify Makefiles.
|
|
The syscall ABI includes the gcc functional calling ABI since a syscall
implies userland caller and kernel callee.
The current gcc ABI (v3) for ARCv2 ISA required 64-bit data be passed in
even-odd register pairs, (potentially punching reg holes when passing such
values as args). This was partly driven by the fact that the double-word
LDD/STD instructions in ARCv2 expect the register alignment and thus gcc
forcing this avoids extra MOV at the cost of a few unused register (which we
have plenty anyways).
This however was rejected as part of upstreaming gcc port to HS. So the new
ABI v4 doesn't enforce the even-odd reg restriction.
Do note that for ARCompact ISA builds v3 and v4 are practically the same in
terms of gcc code generation.
This change is dormant for now (gcc 4.8.x based tools) and will only kick
in with switch to gcc 6.x based tools.
Signed-off-by: Vineet Gupta <vgupta@synopsys.com>
|
|
Fix a regression introduced by commit
0550ecce0e6580c5ad34e9a9a39ff18ccf8774f9
Reported by Buildroot developers.
Embedded test must be extented to ARMv7 thumb2 builds to
find such regressions next time. It wasn't triggered by a
cortex-m4 ARM noMMU build.
|
|
CVE-2016-4429:
The call is technically in a loop, and under certain circumstances
(which are quite difficult to reproduce in a test case), alloca
can be invoked repeatedly during a single call to clntudp_call.
As a result, the available stack space can be exhausted (even
though individual alloca sizes are bounded implicitly by what
can fit into a UDP packet, as a side effect of the earlier
successful send operation).
From GNU libc:
https://sourceware.org/git/gitweb.cgi?p=glibc.git;a=commit;h=bc779a1a5b3035133024b21e2f339fe4219fb11c
|
|
crt1.S needed to use a got relative reference.
Libraries like pthreads define a _init and unless it is GOT, the
linker will fail because it will try to create a dynamic reloc on
.text section, more precisely on __start.
Signed-off-by: Cupertino Miranda <cmiranda@synopsys.com>
Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com>
Cc: Vlad Zakharov <vzakhar@synopsys.com>
Cc: Vineet Gupta <vgupta@synopsys.com>
|
|
This was found by Buildroot autobuilders for m68k.
|
|
This change was inspired by similar change in glibc:
https://sourceware.org/git/?p=glibc.git;a=commit;h=0cb313f7cb0e418b3d56f3a2ac69790522ab825d
Current Linux kernel requires CLONE_VM to be set with CLONE_THREAD
otherwise returning -EINVAL, see man clone2. This means we don't need
to check for both CLONE_THREAD and CLONE_VM instead we may simplify
code a lot and just check 1 bit (CLONE_THREAD).
Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com>
Acked-by: Vineet Gupta <vgupta@synopsys.com>
|
|
For thread group case (CLONE_THREAD), the cached PID of new process/thread
need not be reset. The old logic to decide that was flawed as it would be
true only for exact combination of CLONE_THREAD + _VM, but would fail for
CLONE_THREAD + _VM + _xyz.
More detailed tear-down of current and new code below.
Current implementation is:
--------------------->8--------------------
; r12 contains clone flags
mov_s r2, CLONE_THREAD_N_VM; r2 contains bit mask
and_s r2, r2, r12 ; r2 contains bit mask AND clone flags
; but r12 still contains the same flags
brne r2, r12, .Lgo_thread ; here we compare modified mask with
; flags as they were and skip pthread TID/PID
; setup if r2 != r12 which happens all
; the time except clone flags were
; exactly CLONE_THREAD | CLONE_VM
--------------------->8--------------------
New implementation is:
--------------------->8--------------------
; r12 contains clone flags
mov_s r2, CLONE_THREAD_N_VM; r2 contains bit mask
and_s r12, r12, r2 ; r12 contains clone flags AND bit mask
; i.e. we did mask all flags except
; CLONE_THREAD and CLONE_VM
breq r2, r12, .Lgo_thread ; here we compare masked flags with
; target mask and if they match we skip
; pthread TID/PID setup
--------------------->8--------------------
Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com>
Acked-by: Vineet Gupta <vgupta@synopsys.com>
|
|
Caught by tst-getpid1 test from uClibc's test-suite.
It looks like original implementation was not correct.
The code in question is supposed to recover PID of the new
thread. And by no means that could happen with clone() syscall
while getpid() does exactly this.
Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com>
Acked-by: Vineet Gupta <vgupta@synopsys.com>
Reported-by: Eugeniy Paltsev <paltsev@synopsys.com>
|
|
|
|
Reviewed-by: Max Filippov <jcmvbkbc@gmail.com>
|
|
As this is only implemented for a few architecture and not well
tested, just remove it.
|
|
|
|
As this is only implemented for a few architecture and not well
tested, just remove it.
Reviewed-by: Max Filippov <jcmvbkbc@gmail.com>
|
|
|
|
In uClibc-ng the syscall macros are in bits/syscalls.h.
|
|
Add support for Andes Technology NDS32 architecture.
See here http://www.andestech.com/en/index/index.htm for more
informaton. Verification of the port from an older uClibc
port was done on a sponsored AG101p board.
The testsuite only has 5 errors, three are related to
an existing bug in dlclose() with LT.old, also happening
on cris32 and m68k.
Failures to fallocate/posix_fallocate are unresolved.
Thanks to Andes Technology sponsoring the hardware and
being very helpful while doing the uClibc-ng porting.
Signed-off-by: Waldemar Brodkorb <wbx@uclibc-ng.org>
|
|
Commit 251a3c19cb "sleep: employ __USE_EXTERN_INLINES (with necessary fixes)"
introduces following problems:
1. __USE_EXTERN_INLINES forcibly enabled build fails
...
LD libuClibc-1.0.15.so
libc/libc_so.a(cmsg_nxthdr.os): In function `__GI___cmsg_nxthdr':
cmsg_nxthdr.c:(.text.__GI___cmsg_nxthdr+0x0): multiple definition of `__GI___cmsg_nxthdr'
libc/libc_so.a(close.os):close.c:(.text.__GI___cmsg_nxthdr+0x0): first defined here
libc/libc_so.a(creat.os): In function `__GI___cmsg_nxthdr':
creat.c:(.text.__GI___cmsg_nxthdr+0x0): multiple definition of `__GI___cmsg_nxthdr'
libc/libc_so.a(close.os):close.c:(.text.__GI___cmsg_nxthdr+0x0): first defined here
...
2. libuargp wrongly contains __cmsg_nxthdr/__sigismember/__sigdelset/__sigaddset
global symbols on platforms which includes signal.h from sys/procfs.h
As result, static linking will fail:
TEST_LINK argp/ bug-argp1
/home/wbx/ppc-static/target_qemu-ppc-macppc_uclibc-ng_hard/usr/lib/libc.a(sigsetops.os):
In function `__GI___sigismember':
sigsetops.c:(.text+0x0): multiple definition of `__sigismember'
/home/wbx/ppc-static/target_qemu-ppc-macppc_uclibc-ng_hard/usr/lib/libuargp.a(argp-xinl.os):argp-xinl.c:(.text+0x0):
first defined here
/home/wbx/ppc-static/target_qemu-ppc-macppc_uclibc-ng_hard/usr/lib/libc.a(sigsetops.os):
In function `__GI___sigaddset':
sigsetops.c:(.text+0x28): multiple definition of `__sigaddset'
/home/wbx/ppc-static/target_qemu-ppc-macppc_uclibc-ng_hard/usr/lib/libuargp.a(argp-xinl.os):argp-xinl.c:(.text+0x28):
first defined here
/home/wbx/ppc-static/target_qemu-ppc-macppc_uclibc-ng_hard/usr/lib/libc.a(sigsetops.os):
In function `__GI___sigdelset':
sigsetops.c:(.text+0x4c): multiple definition of `__sigdelset'
/home/wbx/ppc-static/target_qemu-ppc-macppc_uclibc-ng_hard/usr/lib/libuargp.a(argp-xinl.os):argp-xinl.c:(.text+0x4c):
first defined here
We have to partially revert 251a3c19cb to fix problems above. It is
safe to do this after commit
162cfaea20 *: inline constant __sig{add,del}set and __sigismember
since we are able to use new inlines from within libc and leave the
rest of world(__USE_EXTERN_INLINES) equal to glibc now.
Signed-off-by: Leonid Lisovskiy <lly.dev@gmail.com>
|
|
When resolving an unqualified host name, the resolver tries the original
name first before appending the domains from the search list. If a TLD
with the same name exists, the query will succeed (but yield no A record)
and the resolver will return HOST_NOT_FOUND without trying the search
domains.
This patch changes the lookup order for unqualified host names (without
dots) to try the search domains first and the original name last.
Signed-off-by: Ingo van Lil <inguin@gmx.de>
|
|
Currently crt1 takes address of functions (main,_init,_fini) directly
which doesn't generate truely position independent code, but zero based
values instead. e.g.
| __start:
| ...
| add_s r2,sp,0x4
| mov_s r0, main
generates to
| 000156ec <__start>:
| ...
| 156f4: add_s r2,sp,0x4
| 156f6: mov_s r0,0x15f7c
| ...
| 00015f7c <main>:
| 15f7c: push_s blink
This works just fine for the normal (non PIE) dynamic executables since
they are loaded at address 0. However this is not true for PIE
executables. So for Scrt1 we use a true position independent way when
taking function addresses.
Cc: uclibc@uclibc.org <uclibc@uclibc.org>
Cc: devel@uclibc-ng.org <devel@uclibc-ng.org>
Cc: Cupertino Miranda <cmiranda@synopsys.com>
Signed-off-by: Vineet Gupta <vgupta@synopsys.com>
|
|
Simplify the switch from uClibc to uClibc-ng.
Apps already built against uClibc-0.9.x.y require .so.0
libs to present on target which in case of current uClibc-ng is
not the case and those apps could not be run.
This change creates symlinks from .so.1 to .so.0 for
most of other libs in the same way as it was done by
23e96d89b6ab "ldso: install backward compatibility symlink by default"
Signed-off-by: Alexey Brodkin <abrodkin@synopsys.com>
Cc: Waldemar Brodkorb <wbx@uclibc-ng.org>
Cc: Vineet Gupta <vgupta@synopsys.com>
Cc: Anton Kolesov <akolesov@synopsys.com>
|
|
Linux Standard Base specifies section .note.ABI-tag that can be considered
as a marker for ELF files targeted to Linux systems. See
https://refspecs.linuxfoundation.org/LSB_1.2.0/gLSB/noteabitag.html
This section, for example, is used by the GDB to identify Linux ELFs as
compared to baremetal ELFs that do not have this section.
Signed-off-by: Anton Kolesov <Anton.Kolesov@synopsys.com>
Cc: Vineet Gupta <Vineet.Gupta1@synopsys.com>
|
|
When trying to resolve a hostname by getaddrinfo() using some specific
settings, it will always return -EAI_NONAME (Name or service not known).
To reproduce this behavior, you need to request an IPv6 address with the
additional AF_V4MAPPED flag set from an non IPv6 capable hostname. If
you choose a IPv4/IPv6 capable hostname like google.com, everything
works fine.
This patch is more or less a port [1][2] from the glibc and their behavior
for the AF_V4MAPPED flag. To test the bug you can use the following snippet.
---- 8< ----
int ret;
struct addrinfo* result;
struct addrinfo hints;
memset(&hints, 0, sizeof(struct addrinfo));
hints.ai_family = AF_INET6;
hints.ai_flags = AI_V4MAPPED;
ret = getaddrinfo("test.com", NULL, &hints, &result);
printf("getaddrinfo(): %i", ret);
---- 8< ----
[1] https://sourceware.org/git/?p=glibc.git;a=commit;f=sysdeps/posix/getaddrinfo.c;h=925c3c5c71596c02f7e58a0ffcdcaae44eb065c1
[2] https://sourceware.org/git/?p=glibc.git;a=commit;f=sysdeps/posix/getaddrinfo.c;h=28977c2c1acb789660ad47e0d88e42486059c916
Signed-off-by: Alexander Wenzel <alexander.wenzel@qsc.de>
|
|
The 'BLT' instruction checks for *signed* values. So if a3, length
parameter of memset, is negative, then value added to the PC will be
large.
memset(buf, 0xaa, 0xffff0000) triggers the bug.
GDB session without the patch:
"""
$ gdb ./main-buggy-memset.elf -q
Reading symbols from ./main-buggy-memset.elf...done.
(gdb) x/i memset
0x8770 <memset>: mov r3, r0
(gdb) r
Starting program: /root/memset/main-buggy-memset.elf
Program received signal SIGSEGV, Segmentation fault.
0x00048808 in ?? ()
"""
The $pc is outside of the memset function because:
"""
(gdb) x/i $pc
=> 0x87e4 <memset+116>: add pc, pc, r2, lsl #2
(gdb) info reg $r2
r2 0x10007 65543
"""
GDB session with the bug fixed (patch applied):
"""
$ gdb ./main-fixed-memset.elf -q
Reading symbols from ./main-fixed-memset.elf...done.
(gdb) x/i memset
0x8770 <memset>: mov r3, r0
(gdb) r
Starting program: /root/memset/main-fixed-memset.elf
Program received signal SIGSEGV, Segmentation fault.
memset () at libc/string/arm/memset.S:92
92 libc/string/arm/memset.S: No such file or directory.
(gdb) x/i $pc
=> 0x87b0 <memset+64>: stmia r3!, {r1, r12}
(gdb) info reg $r3
r3 0x15000 86016
(gdb) info proc mappings
process 5822
Mapped address spaces:
Start Addr End Addr Size Offset objfile
0x8000 0xb000 0x3000 0x0
/root/memset/main-fixed-memset.elf
0x12000 0x15000 0x3000 0x2000
/root/memset/main-fixed-memset.elf
0xb6fff000 0xb7000000 0x1000 0x0 [sigpage]
0xbefdf000 0xbf000000 0x21000 0x0
0xffff0000 0xffff1000 0x1000 0x0 [vectors]
(gdb) info reg $sp
sp 0x14d78 0x14d78
"""
GDB crashes inside the memset function, on the store instruction. This
time the crash is (as expected) because of a memory access imediately
after the memory region that contains the stack -- the buffer that's
being memset'd is allocated on the stack.
Signed-off-by: Lucian Cojocar <lucian.cojocar@vu.nl>
|
|
|
|
Do it by following the trail of the existing commented code, which
implemented it by calling get_phys_pages() and get_avphys_pages().
This patch implements these two functions, which are also glibc
extensions.
Some program/libraries (e.g. libuv) assumes that sysconf(_SC_PHYS_PAGES)
works on linux and never check for -1, thus they report an insane amount
of memory.
Signed-off-by: Nicolas Cavallari <nicolas.cavallari@green-communications.fr>
|
|
When a 'hard' error occurs, fwrite reports that all data was written or
buffered even if that is not the case. It should report how much data
was actually written and buffered.
Signed-off-by: Jan Vangorp <jan.vangorp_ext@softathome.com>
|
|
The man page for fopencookie prescribes that custom write functions
should return 0 on error (and should definitely not return a negative
value) [1].
However, the uClibc implementation expects a negative return value in
case of an error (libc/stdio/_WRITE.c). If the write function returns 0
on error, we drop into an infinite loop if the error persists.
This patch wraps the user supplied write function such that a 0 return
value is converted to -1. errno is first set to EAGAIN such that if the
custom write function does not set errno in case of error, this is
treated as a "soft" error.
Custom write functions that cater towards uClibc and _do_ return a
negative value are not affected.
If no custom write function is supplied, set errno to EINVAL such that
this condition is treated as a "hard" error. Previously the behaviour
depended on whether the last error before the write happened to be a "hard"
or a "soft" error.
[1] http://git.kernel.org/cgit/docs/man-pages/man-pages.git/tree/man3/fopencookie.3?id=6c0d0ef0c7a9c4bcf805c6f9e9bc1ef1c3865ea0#n164
Signed-off-by: Jan Vangorp <jan.vangorp_ext@softathome.com>
|
|
Nobody should use gcc 3.3 nowadays.
|
|
These files are not used, no regression found while
running the testsuite.
|
|
Remove ppc64 bits and bytes, as we don't support ppc64 and
it makes the code more readable without the dead code.
|
|
|
|
There exist some problem with the new memcpy/memset functions
imported from GNU libc/newlib. If you have any problem
with the new MIPS optimized assembly try to disable
prefetching support.
Thanks to Rene Nielsen and Matthew Fortune analyzing the
problem so far.
|
|
|
|
Sync mostly with GNU libc implementation of setjmp/longjmp.
Just generate PIC compatible code, so no difference for static
builds are made. Add sysdep.h with some handy assembly macros
used inside setjmp.S/__longjmp.S.
This unbreaks dynamic linker support, first test suite run
only shows 10 errors, which doesn't look so bad.
|
|
Linuxthreads.new isn't really useful with the existence
of NPTL/TLS for well supported architectures. There is no
reason to use LT.new for ARM/MIPS or other architectures
supporting NPTL/TLS. It is not available for noMMU architectures
like Blackfin or FR-V. To simplify the live of the few uClibc-ng
developers, LT.new is removed and LT.old is renamed to LT.
LINUXTHREADS_OLD -> UCLIBC_HAS_LINUXTHREADS
|
|
Ported over from glibc mostly without changes.
Lightly tested with mongrel2 in qemu-system-sparc.
|
|
|
|
License notices in most of the source files refer to an outdated
FSF address. Replace it with URL, like in the rest of the source
files.Signed-off-by: Nikola Forró <nforro@redhat.com>
|
|
Signed-off-by: Waldemar Brodkorb <wbx@openadk.org>
|
|
GDB 7.11 does not build if compiled against uclibc-ng (1.0.12 was used,
but the problematic code exists in 1.0.13 as well). The reason is that
GDB release includes its own obstack implementation, but elides the code
if <gnu-versions.h> declares a compatible obstack implementation in
libc. uclibc-ng does claim compatible obstack interface (GDB expects
version 2, but accepts version 1 if sizeof(int) == sizeof(size_t)),
however, uclibc-ng does not provide the _obstack_free symbol that is a
part of the interface (glibc does provide this symbol). This later
causes a link failure.
The attached patch makes uclibc-ng enables an alias _obstack_free ->
obstack_free.
Signed-off-by: Alexey Neyman <stilor@att.net>
|
|
On systems where uClibc doesn't provide an arch specific byteswap.h,
we fall back on bits/byteswap-common.h. However, there is a bug
in this header in the __bswap_constant_64(x) macro. If, for example,
a double is passed, we get 'invalid operands to binary &' in which we
mismatch a 'double' and 'long long unsigned int'. The newer glibc
headers fix this and so we import them. Also, since the inclusion of
byteswap.h is not appropriate for assembly code, we guard a portion
of endian.h which uses byteswap.h from inclusion in any assembly.
This is needed, for example, for f2fs-tools 1.6.0 on 32-bit big
endian PowerPC.
Signed-off-by: Anthony G. Basile <blueness@gentoo.org>
|
|
Currently, the Thumb support on ARM has three related Config.in
options, which are not trivial for users to understand, and are in
fact not needed:
- The USE_BX option is not needed: knowing whether BX is available or
not is easy. If you have an ARM > v4 or ARMv4T, then BX is
available, otherwise it's not. This is the logic used in glibc.
- The USE_LDREXSTREX option is not needed: whenever Thumb2 is
available, ldrex/strex are available, so we can simply rely on
__thumb2__ to determine whether ldrex/strex should be used, without
requiring a Config.in option.
- Once USE_BX and USE_LDREXSTREX are removed, the only thing left
that COMPILE_IN_THUMB does is to set -mthumb. This makes the option
unnecessary, as on ARM at least, the user is already supposed to
pass -march=<foo> or other compiler options tuning the library for
a specific ARM variant. There is no reason to do otherwise for
Thumb, which allows to get rid of the COMPILE_IN_THUMB option.
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@free-electrons.com>
|
|
Register window saving loop in __sigsetjmp incorrectly calculates offset
in the jmp_buf in case it was called with call12. The bug was introduced
in commit f4b1b7ade7971141 "xtensa: fix setjmp that didn't save all
registers correctly".
Signed-off-by: Max Filippov <jcmvbkbc@gmail.com>
|
|
As recently reported on the Buildroot list:
http://lists.busybox.net/pipermail/buildroot/2016-March/155325.html
DNS lookups with Node.js currently fails on uClibc-ng. The reason for this
is the way AI_V4MAPPED is handled. According to POSIX, AI_V4MAPPED should
be ignored unless ai_family is AF_INET6:
http://pubs.opengroup.org/onlinepubs/9699919799/functions/freeaddrinfo.html
If the AI_V4MAPPED flag is specified along with an ai_family of AF_INET6,
then getaddrinfo() shall return IPv4-mapped IPv6 addresses on finding no
matching IPv6 addresses (ai_addrlen shall be 16). The AI_V4MAPPED flag
shall be ignored unless ai_family equals AF_INET6.
uClibc-ng was also handling AI_V4MAPPED for AF_UNSPEC, fix that.
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
|
|
This lessens the load on upstream DNS servers (and it was just nonsensical).
Signed-off-by: Tim Hockin <thockin@google.com>
|
|
Currently a miss on a search-path entry is counted as a retry. This means that
users with more than (num_nameservers * retries) entries in their search path
list fail before trying all search paths. Concretely, a single nameserver with
4 search paths will never try the 4th search because the default retry is 3.
The code doesn't currently retry a given nameserver in case of an error, so
retries is sort of meaningless (though there are some comments indicating it
might come). This change only treats total failure of a nameserver (try next
server) as a retry.
Signed-off-by: Tim Hockin <thockin@google.com>
|
|
|
