summaryrefslogtreecommitdiff
AgeCommit message (Collapse)Author
2010-10-31libm: fix false positives in double testsDenys Vlasenko
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
2010-10-31libm: fix remaining failures in test-floatDenys Vlasenko
These two failures are fixed: Failure: Test: ilogb (0.0) == FP_ILOGB0 plus exceptions allowed Failure: Test: ilogb (NaN) == FP_ILOGBNAN plus exceptions allowed Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
2010-10-30libm: fix rint/scalb testcase failuresDenys Vlasenko
These failures no longer happen: Failure: Test: scalb (2.0, 0.5) == NaN plus invalid exception Failure: Test: scalb (3.0, -2.5) == NaN plus invalid exception Failure: Test: rint (0.5) == 0.0 Failure: Test: rint (1.5) == 2.0 Failure: Test: rint (2.5) == 2.0 Failure: Test: rint (3.5) == 4.0 Failure: Test: rint (4.5) == 4.0 Failure: Test: rint (-0.5) == -0.0 Failure: Test: rint (-1.5) == -2.0 Failure: Test: rint (-2.5) == -2.0 Failure: Test: rint (-3.5) == -4.0 Failure: Test: rint (-4.5) == -4.0 Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
2010-10-30libm: fix powf testcase failuresDenys Vlasenko
Fixed failures: Failure: Test: pow (1, NaN) == 1 Failure: Test: pow (1, inf) == 1 Failure: Test: pow (-1, inf) == 1 Failure: Test: pow (1, -inf) == 1 Failure: Test: pow (-1, -inf) == 1 Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
2010-10-30libm: fix scalb testsuite failuresDenys Vlasenko
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
2010-10-30libm: fix testuite failure in modfDenys Vlasenko
This error is gone: Failure: Test: modf (NaN, &x) == NaN Result: is: -0.00000000000000000000e+00 -0 should be: nan nan Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
2010-10-30test/math/*: unbreakDenys Vlasenko
test/math was not running at all, presume it has bit rotted: It was trying to run libm-test.c as if it is a binary. It was looking for libm-test-ulps (file with allowed errors in lower bits) in arch-specific dirs which do not exist in uclibc, as a result wrong file (libm-test.inc) was used instead. Test failure was not showing the error result, user had to fish it out from some .out files. I added libm-test-ulps-ARCH files from recent glibc, this filtered out a lot of false positives. For example, cosf(M_PI_6l * 4.0) ideally should be -0.5, we are getting -0.50000005047356477217, and this isn't a failure (the difference is one lowest bit of mantissa). "make check UCLIBC_ONLY=1 VERBOSE=1" still fails, but not as catastrophically as before. For the record, the failure occurs on the stage where we check 32-bit float functions, these tests fail: Failure: Test: modf (NaN, &x) == NaN Failure: Test: ilogb (NaN) == FP_ILOGBNAN plus exceptions allowed Failure: Test: scalb (2.0, 0.5) == NaN plus invalid exception Failure: Test: scalb (3.0, -2.5) == NaN plus invalid exception Failure: Test: scalb (0, NaN) == NaN Failure: Test: scalb (1, NaN) == NaN Failure: Test: scalb (0, inf) == NaN plus invalid exception Failure: Test: scalb (-0, inf) == NaN plus invalid exception Failure: Test: scalb (1, inf) == inf Failure: Test: scalb (-1, inf) == -inf Failure: Test: scalb (inf, -inf) == NaN plus invalid exception Failure: Test: scalb (-inf, -inf) == NaN plus invalid exception Failure: Test: scalb (1, NaN) == NaN Failure: Test: scalb (0, NaN) == NaN Failure: Test: scalb (inf, NaN) == NaN Failure: Test: pow (1, NaN) == 1 Failure: Test: pow (1, inf) == 1 Failure: Test: pow (-1, inf) == 1 Failure: Test: pow (1, -inf) == 1 Failure: Test: pow (-1, -inf) == 1 Failure: Test: rint (0.5) == 0.0 Failure: Test: rint (1.5) == 2.0 Failure: Test: rint (2.5) == 2.0 Failure: Test: rint (3.5) == 4.0 Failure: Test: rint (4.5) == 4.0 Failure: Test: rint (-0.5) == -0.0 Failure: Test: rint (-1.5) == -2.0 Failure: Test: rint (-2.5) == -2.0 Failure: Test: rint (-3.5) == -4.0 Failure: Test: rint (-4.5) == -4.0 Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
2010-10-30sleep: include Linus' email in the commentDenys Vlasenko
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
2010-10-29sleep: document testing result on 2.4.x kernelsDenys Vlasenko
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
2010-10-29sleep: add comment with test program for SIG_IGNed SIGCHLDDenys Vlasenko
Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>
2010-10-27Fix msync() redefinition for noMMU buildBernhard Reutner-Fischer
An architecture that _may_ have a MMU (and thus implements the msync syscall), but for which the kernel and uClibc are built as noMMU, we have a stub for msync. Thanks to Steven J. Magnani for noticing. Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
2010-10-27mips: fix errno setting after syscallAlexander Gordeev
If there was an error during syscall then after it's completion a3 register holds a non-zero value and v0 holds an actual error code which should be saved in errno. This can be achieved by calling __syscall_error with the value from v0 as a parameter. So this value should be stored in a0, but the appropriate assembly instructions are missing. Fixed this now by adding "move a0, v0". I think it was once fixed by 58c5f8ba4cdf62342d05a546d15404cbbb3c4e07 but then something went wrong. Again... Signed-off-by: Alexander Gordeev <lasaine@lvk.cs.msu.su> Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
2010-10-22*: inline constant __sig{add,del}set and __sigismemberDenys Vlasenko
text data bss dec hex filename - 318 4 0 322 142 libc/pwd_grp/lckpwdf.o + 312 4 0 316 13c libc/pwd_grp/lckpwdf.o - 166 0 1 167 a7 libc/stdlib/abort.o + 157 0 1 158 9e libc/stdlib/abort.o - 42 0 0 42 2a libc/sysdeps/linux/common/pause.o + 27 0 0 27 1b libc/sysdeps/linux/common/pause.o Signed-off-by: Denys Vlasenko <dvlasenk@redhat.com>
2010-10-22remove superfluous libc_hidden_proto(memcpy) and #includeDenys Vlasenko
Signed-off-by: Denys Vlasenko <dvlasenk@redhat.com>
2010-10-22sleep: tiny code shrinkDenys Vlasenko
...or rather, it WILL BE code shrink when gcc become clever enough to not emit a second, useless XORing of ebx: 31 db xor %ebx,%ebx 85 c0 test %eax,%eax 74 11 je 73 <__GI_sleep+0x73> 31 db xor %ebx,%ebx <=== ?! Signed-off-by: Denys Vlasenko <dvlasenk@redhat.com>
2010-10-22sleep: employ __USE_EXTERN_INLINES (with necessary fixes)Denys Vlasenko
__USE_EXTERN_INLINES was unused and had bit-rotted, had to fix it when it didn't work as intended at first. text data bss dec hex filename - 168 0 0 168 a8 libc/unistd/sleep.o + 146 0 0 146 92 libc/unistd/sleep.o Signed-off-by: Denys Vlasenko <dvlasenk@redhat.com>
2010-10-22sleep: check "SIGCHLD is SIG_IGN'ed" first. Saves two syscalls in common caseDenys Vlasenko
text data bss dec hex filename - 197 0 0 197 c5 libc/unistd/sleep.o + 168 0 0 168 a8 libc/unistd/sleep.o Signed-off-by: Denys Vlasenko <dvlasenk@redhat.com>
2010-10-22libm/x86: use call instead of jump for wrappersTimo Teräs
GCC can emit prologue/epilogue code for the functions in various different cases: - frame pointers - PIC build (to load ebx for indirect calls/jumps) - forced stack smashing protection If we used jump in such cases, we'd corrupt the call stack and crash. Signed-off-by: Timo Teräs <timo.teras@iki.fi> Signed-off-by: Denys Vlasenko <dvlasenk@redhat.com>
2010-10-22sleep: remove commented-out code. no code changesDenys Vlasenko
It can be easily reconstructed, since it's obvious Signed-off-by: Denys Vlasenko <dvlasenk@redhat.com>
2010-10-21sleep: code shrinkDenys Vlasenko
Use less stack by using same "sigset_t set" object for new and saved signal set, remove redundant clearing of set, and do not save/restore errno around sigprocmask(SIG_SETMASK) - it never changes it. While at it, improve comments and make code style consistent across sleep.c file. text data bss dec hex filename - 242 0 0 242 f2 libc/unistd/sleep.o + 197 0 0 197 c5 libc/unistd/sleep.o Signed-off-by: Denys Vlasenko <dvlasenk@redhat.com>
2010-10-19vfprintf.c: reduce a chunk of #ifdef forest and remove one goto inside itDenys Vlasenko
No code changes according to objdump. Signed-off-by: Denys Vlasenko <dvlasenk@redhat.com>
2010-10-19_vfprintf.c: de-obfuscate badly twisted fragment. no code changes.Denys Vlasenko
objdump confirms that I did not mess it up. Signed-off-by: Denys Vlasenko <dvlasenk@redhat.com>
2010-10-19vfprintf.c: remove endif comments which clog up the source. no code changesDenys Vlasenko
Example: --ppfs->maxposarg; Verified with objdump that no code is changed Signed-off-by: Denys Vlasenko <dvlasenk@redhat.com>
2010-10-19vfprintf.c: de-obfuscate if(with nested assignments). no logic changesDenys Vlasenko
God knows this file is hard to read as-is, some readability improvement is in order. Signed-off-by: Denys Vlasenko <dvlasenk@redhat.com>
2010-10-14getproto: increase line buffer size, simlify and fix alias handlingNatanael Copa
We increase line buffer size, reduce MAXALIASES and make sure we don't segfault when there are too manuy aliases in /etc/protocols. Signed-off-by: Natanael Copa <natanael.copa@gmail.com> Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
2010-10-14getnet: simplify alias handling and reduce MAXALIASESNatanael Copa
Reduce MAXALIASES to something lower. There will probably never be need for more than 1 alias but we allow a few extra. While here we alos fix segfault when there are too many aliases. Signed-off-by: Natanael Copa <natanael.copa@gmail.com> Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
2010-10-14getservice: fix handling of long linesNatanael Copa
Don't try to be smart by dynamically realloc buffersize as it doesn't work. Instead, be simple and allocate a buffer big enough. This fixes a memory leak when calling getserv{ent,byname,byport} multiple times. To save memory we reduce number of max aliases. We seldomly will need more than 1 anyways. While here, fix segfault that happened if there were too many aliases. Signed-off-by: Natanael Copa <natanael.copa@gmail.com> Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
2010-10-14parse_config: discard rest of incomplete lineNatanael Copa
If line is longer then size of given buffer and buffer is not allocated by the config parser itself, then discard rest of line. Signed-off-by: Natanael Copa <natanael.copa@gmail.com> Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
2010-10-14config parser: always initialize line pointerNatanael Copa
We must always initialize line pointer since data pointer might have changed due to a realloc (in getserv.c for example). Signed-off-by: Natanael Copa <natanael.copa@gmail.com> Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
2010-10-14getservice: getservent_r must return ERANGE when buffer is too smallNatanael Copa
This fixes issue introduced by 72e1a1ce186c39f07282398e2af9eb0253e60f15 This should also fix the following testcase to exit with error rather than cause an endless loop. int main(void) { if (getservbyname("non-existing", "udp") == NULL) err(1, "getservbyname"); return 0; } Reported by Pirmin Walthert http://lists.uclibc.org/pipermail/uclibc/2010-August/044277.html Signed-off-by: Natanael Copa <natanael.copa@gmail.com> Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
2010-10-14config parser: do not assume that realloc return same pointerNatanael Copa
We need to update the parser->line pointer on realloc and do not initialize the token array til after the potensial realloc in bb_get_chunk_with_continuation(). While here, also replace a realloc() with malloc() where pointer always is NULL. Signed-off-by: Natanael Copa <natanael.copa@gmail.com> Signed-off-by: Bernhard Reutner-Fischer <rep.dot.nop@gmail.com>
2010-10-01sparc: pipe.SAustin Foxley
return value of pipe on success is supposed to be 0 Signed-off-by: Austin Foxley <austinf@cetoncorp.com>
2010-09-24ldso_sh: add support for protected symbols to SHSalvatore Cro
Protected symbols should not be overridden by symbols from other modules. Such symbols are exported i.e. globally visible, but references from whithin defining modules are satisfied locally. Signed-off-by: Salvatore Cro <salvatore.cro@st.com> Signed-off-by: Carmelo Amoroso <carmelo.amoroso@st.com>
2010-09-22libc: Sort HEADERS_RM alphabeticallyCarmelo Amoroso
No functional changes, just a code tidy-up. Please keep them sorted in future changes. Signed-off-by: Filippo Arcidiacono <filippo.arcidiacono@st.com> Signed-off-by: Carmelo Amoroso <carmelo.amoroso@st.com>
2010-09-20nptl: Fix libpthread build when UCLIBC_LINUX_SPECIFIC is disabledCarmelo Amoroso
NPTL library needs both madvise and statfs symbols, that are guarded by UCLIBC_LINUX_SPECIFIC option. This fix provides these symbols too when NPTL is used, indipendently by UCLIBC_LINUX_SPECIFIC choice. Otherwise libpthread link fails as below: LD libpthread-0.9.32-git.so libpthread/nptl/libpthread_so.a(pthread_create.oS): In function `__free_tcb': pthread_create.c:(.text+0x1184): undefined reference to `madvise' libpthread/nptl/libpthread_so.a(sem_open.oS): In function `__where_is_shmfs': sem_open.c:(.text+0x764): undefined reference to `statfs' collect2: ld returned 1 exit status make: *** [lib/libpthread.so] Error 1 Signed-off-by: Carmelo Amoroso <carmelo.amoroso@st.com>
2010-09-17libc: Handle cancellation in non multiplexed socket callsCarmelo Amoroso
For those archs that provide non multiplexed socket calls it possible to implement the lib C wrappers without calling the multi-purpose __socketcall. For a subset of these functions that are cancellation point, it needs to correctly handle cancellation. Signed-off-by: Francesco Rundo <francesco.rundo@st.com> Signed-off-by: Carmelo Amoroso <carmelo.amoroso@st.com>
2010-09-16nptl: Remove nptl/sysdeps/generic directory from include listCarmelo Amoroso
All archs have now moved their generic sysdep.h header from sysdeps/generic to sysdeps/common directory, so it is possible to remove the sysdeps/generic directory from the include list. Signed-off-by: Carmelo Amoroso <carmelo.amoroso@st.com>
2010-09-16nptl_sh: Fix compiler warning due to shadowed variableCarmelo Amoroso
Signed-off-by: Carmelo Amoroso <carmelo.amoroso@st.com>
2010-09-16libc: Add canonicalize_file_name functionCarmelo Amoroso
Add canonicalize_file_name function and its related tests. Required by elfutils and coreutils (readlink). Signed-off-by: Salvatore Cro <salvatore.cro@st.com> Signed-off-by: Carmelo Amoroso <carmelo.amoroso@st.com>
2010-09-16test: Update gitignore to take into account new tests added.Carmelo Amoroso
Signed-off-by: Carmelo Amoroso <carmelo.amoroso@st.com>
2010-09-15ldso: Do not adjust dynamic section entries unnecessarilyFilippo Arcidiacono
Signed-off-by: Filippo Arcidiacono <filippo.arcidiacono@st.com> Signed-off-by: Carmelo Amoroso <carmelo.amoroso@st.com>
2010-09-15tests: Added new nptl testsSalvatore Cro
tests: Added several nptl tests from glibc 2.10.1 Signed-off-by: Salvatore Cro <salvatore.cro@st.com> Signed-off-by: Carmelo Amoroso <carmelo.amoroso@st.com>
2010-09-15sh: update the memcpy adding a new loop with aggressive prefetchingSalvatore Cro
After exploring different prefetch distance-degree combinations in this new update of the memcpy function, a new loop has been added for moving many cache lines with an aggressive prefetching schema. Prefetch has been removed when move few cache line aligned blocks. As final result, this memcpy gives us the same performances for small sizes (we already had!) and better numbers for big copies. In case of SH4-300 CPU Series, benchmarks show a gain of ~20% for sizes from 4KiB to 256KiB. In case of the SH4-200, there is a gain of ~40% for sizes bigger than 32KiB. Signed-off-by: Giuseppe Cavallaro <peppe.cavallaro@st.com> Signed-off-by: Carmelo Amoroso <carmelo.amoroso@st.com>
2010-09-15sh: move data without fetching cache block within the memsetSalvatore Cro
With this patch the movca.l instruction is used within the memset. The current memset implementation only uses the FPU and there is an real gain for all the sizes. Adding the movca.l instruction numbers always are better than the generic code. There is a big gain for size greater than 64 KiB but number are worst for 4-32KiB sizes compared with the implementation without movca.l. Time Memory Bandwidth (Mbytes) ------------------------------------------------- Generic SH4 SH4 (FPU) (FPU+movca.l) ------------------------------------------------- 512 1143 1998 1596 1 KiB 1273 2567 1915 2 KiB 1350 2993 2128 4-32KiB 1391 3262 2252 64KiB-16MiB 170 186 *830* Signed-off-by: Giuseppe Cavallaro <peppe.cavallaro@st.com> Signed-off-by: Carmelo Amoroso <carmelo.amoroso@st.com>
2010-09-15libc: Fix cancellation handling in some C functionsSalvatore Cro
According to POSIX.1-2008 standard, the following syscalls shall be cancellation points : waitid, sleep, fdatasync, ppoll. Further, if generic syscall is not available and stubs are configured, provide the stub implementation for function. Signed-off-by: Salvatore Cro <salvatore.cro@st.com> Signed-off-by: Carmelo Amoroso <carmelo.amoroso@st.com>
2010-09-15libubacktrace: Provide uClibc with backtrace functionsSalvatore Cro
A new shared object, libubacktrace.so.0 is added to uClibc to provide backtrace functions to support application self-debugging. This set of functions requires to dynamically load libgcc_s.so so they need to call dlopen/dlsym that are provided by libdl. For this reason they cannot be included into libc.so.0 but are provided by a new library. User application that wants to use backtrace needs to be compiled with -fexceptions option and -rdynamic to get full symbols printed and must be linked against libubacktrace.so Signed-off-by: Carmelo Amoroso <carmelo.amoroso@st.com>
2010-09-14misc: utmpx based logging supportSalvatore Cro
misc: Added support for accessing user accounting database based on utmpx structure. Signed-off-by: Salvatore Cro <salvatore.cro@st.com> Signed-off-by: Carmelo Amoroso <carmelo.amoroso@st.com>
2010-09-09nptl: fix static linkingNatanael Copa
We need dl-tls.c for static libc or we will get missing symbols when linking thinkgs like openssl static. Signed-off-by: Natanael Copa <natanael.copa@gmail.com> Signed-off-by: Carmelo Amoroso <carmelo.amoroso@st.com>
2010-09-09tests: nptl tests dependencies cleanupSalvatore Cro
tests: some cleanings on nptl tests' dependencies. Signed-off-by: Salvatore Cro <salvatore.cro@st.com> Signed-off-by: Carmelo Amoroso <carmelo.amoroso@st.com>
2010-09-07ldso: Fix compilation for x86_64 without TLS support.Will Newton
Add an #ifdef to avoid using TLS structures when TLS is not enabled. Signed-off-by: Will Newton <will.newton@imgtec.com> Signed-off-by: Carmelo Amoroso <carmelo.amoroso@st.com>