This is a second attempt to fix a compilation failure involving a
missing struct timespec definition in some situation. Previous attempt
was following commit but it had undesirable side effects and was
reverted:
657c4a9d6c fcntl.h: fix a missing `struct timespec` definition
A similar issue was identified and reported in glibc [1], and the fix
was to include the timespec definition if __USE_XOPEN2K8 was defined. To
minimize the possible impact of this new fix, it was decided to not
replace __USE_ATFILE by __USE_XOPEN2K8 as glibc did, but to add an
additional OR condition on __USE_XOPEN2K8.
libm: allow long double wrappers for all architectures
If you enable these wrappers, be sure you don't need long double
precision on your embedded device, as these only enables
long double warpper functions to the existing double math
functions. Required to build some software as lvm2.
Same iteration variable used for inner and outer loop
Inner loop was using same counter variable (i) as the outer loop, therefore
making outer loop terminate before it visited all of the ELF program segments.
Surrounding code in this inner loop clearly shows the intention that this loop
should not affect the outer one, therefore leading me to the conclusion that
this should be a bug an not expected code.
This bug was detected due to some other bug in ARC binutils that kept setting
TEXTREL for any PIE application.
Apart from the but, I have also moved the debug message inside of the TEXTREL
condition as mprotect is only really called if TELTREL is set.
_dl_zalloc callers don't check for allocaton failure. It kind of makes
sense since such early allocations are unlikely to fail, and even if
they do, ldso would segv anyways thus bringing the issue to attention.
However there's a gcc nuance which led to this patch.
It seems gcc at -O2 (for DODEBUG build), does additional coge gen
analysis for pointer dereference from erroneous paths and it "isolates"
such code paths with a intrinsic abort/trap etc.
So add an explict dl_exit() in dl_zalloc error case to beat the
compiler.
Note that this error propagagtion analysis stops if the function in
consideration (_dl_zalloc) is NOT inlined. Hence the reason it only
shows up for DODEBUG builds which builds ldso at -O2 which is more
aggressive about inlining.
If this patch is not considered worth applying then the workaround
suggested by Claudiu is to to build ldso with
-fno-isolate-erroneous-paths-dereference
The included RPC implementation is ipv4 only.
Other C library projects have either deprecated the internal
RPC implementation (GNU C Library) or never implemented such
functionality (musl C Library). The latest rpcbind release (0.2.4)
checks for libtirpc and does not allow to be build with uClibc-ng
RPC without patching. The common use case for RPC nowadays is to
use rpcbind together with nfs-utils to provide NFS server or client
support to a system.
The included RPC implementation does create issues with duplicate
symbol failures when statically compiling with RPC enabled.
This follows the recommendations outlined in Network Operations Division
Cryptographic Requirements published on wikileaks on March 2017.
We discard more bytes of the first keystream to reduce possibility of
non-random bytes.
This is similar to a change in FreeBSD:
https://svnweb.freebsd.org/base?view=revision&revision=315225
Then including fcntl.h results in the following error:
In file included from fcntl.h:37:0,
from test.c:4:
sys/stat.h:371:54: error: array type has incomplete element type ‘struct timespec’
extern int futimens (int __fd, const struct timespec __times[2]) __THROW;
To fix that, we force the definition of struct timespec in fcntl.h when
__USE_XOPEN2K8 is defined. This mimics the behavior of glibc which
contains the following line in its io/fcntl.h header (included by
fcntlh.h):
#ifdef __USE_XOPEN2K8
# include <bits/types/struct_timespec.h>
#endif
This bug was spotted during the compilation of BoringSSL against uClibc.
As reported by Buildroot developers these files causing static
linking issues. The original contribution with the ARM unwind-resume
rework and GNU libc sync was made before the combined libc change.
But the patch was applied later, after the libc change and
it seems the test coverage for static linking didn't catch it in
the regression testing. Remove the files.
Follow GNU C Library from c579f48edba88380635ab98cb612030e3ed8691e
and remove the PID caching. These simplifies the architecture specific
assembly code.
The run of the test suite found no regressions, it even solves
some of the test failures for x86/x86_64/sparc.
nds32: add NPTL/TLS, *context function, libm changes and code cleanup
This commit includes following features.
1. Support NPTL/TLS
2. Add libm function which is used to handle FP rounding and excpetions
(ex: fclrexcpt,fedisblxcpti,feenablxcpt... )
3. Add *context function for operating user context
(ex: setcontext,getcontext,makecontext... )
4. Change the return flow from signal handler
5. Cleanup of old code
The testsuite only has 2 errors, tst-cpuclock1 and tst-cputimer1,
which are related to timing accuracy. (math and locale tests are disabled)
Bernd Kuhls [Mon, 16 Jan 2017 20:55:46 +0000 (21:55 +0100)]
Makefile.in: Fix removal of libintl.h
A wrong filename was present in Makefile.in leaving libintl.h installed
even if libintl support is disabled leading to wrong configure checks
by other packages.
Vineet Gupta [Fri, 13 Jan 2017 20:24:53 +0000 (12:24 -0800)]
ARC: ldso: don't use _DYNAMIC@gotpc construct #2
This removes _DYNAMIC@gotpc from elf_machine_load_address()
This is a seperate commit to callout the fact that old code was actually
broken but was NOT caught since it never gets called (and will never be)
since ld.so only calls it if kernel doesn't pass AT_BASE auxvt, which is
not true for Linux.
if (!auxvt[AT_BASE].a_un.a_val)
auxvt[AT_BASE].a_un.a_val = elf_machine_load_address();
So while the intent was to remove _DYNAMIC@gotpc construct, we need to
come up with something which works.
The build time address computation of .dynamic which works well is:
However the runtime address of .dynamic can only be generated with
"add %0, pcl, _DYNAMIC@pcl \n"
which unfortunately is what is currently broken as it is converted to
i_GLOBAL_OFFSET_TABLE_t and thus refers to runtime address of .got and
not .dyanmic
Thus we resort to using an arbit symbol _dl_start at the expense of an
extra GOT entry and a bogus R_ARC_NONE relo in ldso itself which now
needs to be ignored.
The syscall wrappers are not required and other C libraries
do not provide them. Busybox modutils.c must be patched so
that syscall() is used for uClibc-ng.
Remove __UCLIBC_HAS_OBSTACK__ as it isn't very uptodate and
maintained part. It shouldn't be required for any software and
mostly shipped with stuff which use it. (f.e. binutils-gdb)
Thomas Petazzoni [Mon, 19 Dec 2016 22:26:23 +0000 (23:26 +0100)]
sys/cdefs.h: add definition of __attribute_alloc_size__
Commit cee0b058fa0b4501b289a2da365182d60314d746 ("add aligned_alloc
required for latest gcc libstdc++") added the prototype of
aligned_alloc() to <stdlib.h>. This prototype contains
'__attribute_alloc_size__ ((2))', but this is not defined anywhere in
uClibc-ng.
This commit addresses that by adding the relevant definition in
<sys/cdefs.h>, borrowed from glibc.
Revert the removal of the weak pthread functions and
guarantee a link order so that single threaded applications
doesn't link in all the pthread functions they don't use.
uClibc-ng 1.0.20 fixed static linking with "libdl" by adding all libdl functions
into the libc. On xtensa, though, libdl contains an unresolved reference that is
satisfied by the ld.so - which is not a part of the linking in a static case.
Fixes following problem, when trying to compile a simple
C application statically with a FDPIC toolchain (for example
with Blackfin architecture):
lib/libc.a(libdl.os): In function `do_dlclose':
(.text+0x6be): undefined reference to `_dl_free'
..
projects / uclibc-ng.git / log