If allocation on a non-main arena fails, the main arena is used
without checking to see if it is corrupt. Add a check that avoids the
main arena if it is corrupt.
* malloc/arena.c (arena_get_retry): Don't use main_arena if it is
corrupt.
The arena pointer in the first argument to arena_get2 was used in the
old days before per-thread arenas. They're unused now and hence can
be dropped.
ChangeLog:
* malloc/arena.c (arena_get2): Drop unused argument.
(arena_lock): Adjust.
(arena_get_retry): Likewise.
The change in 0b5395f052 replaced calls
to __get_cpu_features@plt followed by a mov from rax to rdx, with a
single macro LOAD_RTLD_GLOBAL_RO_RDX. It is pretty clear that there
was a typo in s_floorf and __nearbyint due to which the (now incorrect)
mov was not removed. This patch removes that mov.
* sysdeps/x86_64/fpu/multiarch/s_floorf.S (__floorf): Remove
unnecessary movq.
* sysdeps/x86_64/fpu/multiarch/s_nearbyint.S (__nearbyint):
Likewise.
The recently introduced TLS variables in the thread-local destructor
implementation (__cxa_thread_atexit_impl) used the default GD access
model, resulting in a call to __tls_get_addr. This causes a deadlock
with recent changes to the way TLS is initialized because DTV
allocations are delayed and hence despite knowing the offset to the
variable inside its TLS block, the thread has to take the global rtld
lock to safely update the TLS offset.
This causes deadlocks when a thread is instantiated and joined inside
a destructor of a dlopen'd DSO. The correct long term fix is to
somehow not take the lock, but that will need a lot deeper change set
to alter the way in which the big rtld lock is used.
Instead, this patch just eliminates the call to __tls_get_addr for the
thread-local variables inside libc.so, libpthread.so and rtld by
building all of their units with -mtls-model=initial-exec.
There were concerns that the static storage for TLS is limited and
hence we should not be using it. Additionally, dynamically loaded
modules may result in libc.so looking for this static storage pretty
late in static binaries. Both concerns are valid when using TLSDESC
since that is where one may attempt to allocate a TLS block from
static storage for even those variables that are not IE. They're not
very strong arguments for the traditional TLS model though, since it
assumes that the static storage would be used sparingly and definitely
not by default. Hence, for now this would only theoretically affect
ARM architectures.
The impact is hence limited to statically linked binaries that dlopen
modules that in turn load libc.so, all that on arm hardware. It seems
like a small enough impact to justify fixing the larger problem that
currently affects everything everywhere.
This still does not solve the original problem completely. That is,
it is still possible to deadlock on the big rtld lock with a small
tweak to the test case attached to this patch. That problem is
however not a regression in 2.22 and hence could be tackled as a
separate project. The test case is picked up as is from Alex's patch.
This change has been tested to verify that it does not cause any
issues on x86_64.
ChangeLog:
[BZ #18457]
* nptl/Makefile (tests): New test case tst-join7.
(modules-names): New test case module tst-join7mod.
* nptl/tst-join7.c: New file.
* nptl/tst-join7mod.c: New file.
* Makeconfig (tls-model): Pass -ftls-model=initial-exec for
all translation units in libc.so, libpthread.so and rtld.
When an TLS destructor is registered, we set the DF_1_NODELETE flag to
signal that the object should not be destroyed. We then clear the
DF_1_NODELETE flag when all destructors are called, which is wrong -
the flag could have been set by other means too.
This patch replaces this use of the flag by using l_tls_dtor_count
directly to determine whether it is safe to unload the object. This
change has the added advantage of eliminating the lock taking when
calling the destructors, which could result in a deadlock. The patch
also fixes the test case tst-tls-atexit - it was making an invalid
dlclose call, which would just return an error silently.
I have also added a detailed note on concurrency which also aims to
justify why I chose the semantics I chose for accesses to
l_tls_dtor_count. Thanks to Torvald for his help in getting me
started on this and (literally) teaching my how to approach the
problem.
Change verified on x86_64; the test suite does not show any
regressions due to the patch.
ChangeLog:
[BZ #18657]
* elf/dl-close.c (_dl_close_worker): Don't unload DSO if there
are pending TLS destructor calls.
* include/link.h (struct link_map): Add concurrency note for
L_TLS_DTOR_COUNT.
* stdlib/cxa_thread_atexit_impl.c (__cxa_thread_atexit_impl):
Don't touch the link map flag. Atomically increment
l_tls_dtor_count.
(__call_tls_dtors): Atomically decrement l_tls_dtor_count.
Avoid taking the load lock and don't touch the link map flag.
* stdlib/tst-tls-atexit-nodelete.c: New test case.
* stdlib/Makefile (tests): Use it.
* stdlib/tst-tls-atexit.c (do_test): dlopen
tst-tls-atexit-lib.so again before dlclose. Add conditionals
to allow tst-tls-atexit-nodelete test case to use it.
The tst-tls-atexit test case searches for its module in /proc/PID/maps
to verify that it is unloaded, which is a Linux-specific test. This
patch makes the test generic by looking for the library in the link
map list in the _r_debug structure.
Verified that the test continues to succeed on x86_64. There is a bug
in the test case where it calls dlclose once again, which is actually
incorrect but still manages to unload the DSO thanks to an existing
bug in __tls_call_dtors. This will be fixed in a later patch which
also fixes up the __cxa_thread_atexit_impl implementation. I have
added a FIXME comment to that call momentarily, which I will remove
when I fix the problem.
* stdlib/tst-tls-atexit-lib.c (do_foo): Rename to reg_dtor.
* stdlib/tst-tls-atexit.c: (is_loaded): New function.
(spawn_thread): New function.
(load): Rename to reg_dtor_and_close. Move dlopen to...
(do_test): ... here. Use IS_LOADED to test for its
availability.
The DF_1_NODELETE flag is set too late when opening a DSO, due to
which, if a DSO is already open, subsequently opening it with
RTLD_NODELETE fails to set the DF_1_NODELETE flag. This patch fixes
this by setting the flag immediately after bumping the opencount.
Verified on x86_64.
[BZ #18676]
* elf/tst-nodelete-opened.c: New test case.
* elf/tst-nodelete-opened-lib.c: New test case module.
* elf/Makefile (tests, modules-names): Use them.
* elf/dl-open.c (dl_open_worker): Set DF_1_NODELETE flag
early.
The locks don't synchronize with anything - they were likely
introduced initially to synchronize with some main thread code, but
that is no longer evident.
This script is a sample implementation that uses import_bench to
construct two benchmark objects and compare them. If detailed timing
information is available (when one does `make DETAILED=1 bench`), it
writes out graphs for all functions it benchmarks and prints
significant differences in timings of the two benchmark runs. If
detailed timing information is not available, it points out
significant differences in aggregate times.
Call this script as follows:
compare_bench.py schema_file.json bench1.out bench2.out
Alternatively, if one wants to set a different threshold for warnings
(default is a 10% difference):
compare_bench.py schema_file.json bench1.out bench2.out 25
The threshold in the example above is 25%. schema_file.json is the
JSON schema (which is $srcdir/benchtests/scripts/benchout.schema.json
for the benchmark output file) and bench1.out and bench2.out are the
two benchmark output files to compare.
The key functionality here is the compress_timings function which
groups together points that are close together into a single point
that is the mean of all its representative points. Any point in such
a group is at most 1.5x the smallest point in that group. The
detailed derivation is a comment in the function.
* benchtests/scripts/compare_bench.py: New file.
* benchtests/scripts/import_bench.py (mean): New function.
(split_list): Likewise.
(do_for_all_timings): Likewise.
(compress_timings): Likewise.
This is the beginning of a module to import and process benchmark
outputs. The module currently supports importing of a bench.out and
validating it against a schema file. In future this could grow a set
of routines that benchmark consumers may find useful to build their
own analysis tools. I have altered validate_bench to use this module
too.
* benchtests/scripts/import_bench.py: New file.
* benchtests/scripts/validate_benchout.py: Import import_bench
instead of jsonschema.
(validate_bench): Remove function.
(main): Use import_bench.
When the malloc subsystem detects some kind of memory corruption,
depending on the configuration it prints the error, a backtrace, a
memory map and then aborts the process. In this process, the
backtrace() call may result in a call to malloc, resulting in
various kinds of problematic behavior.
In one case, the malloc it calls may detect a corruption and call
backtrace again, and a stack overflow may result due to the infinite
recursion. In another case, the malloc it calls may deadlock on an
arena lock with the malloc (or free, realloc, etc.) that detected the
corruption. In yet another case, if the program is linked with
pthreads, backtrace may do a pthread_once initialization, which
deadlocks on itself.
In all these cases, the program exit is not as intended. This is
avoidable by marking the arena that malloc detected a corruption on,
as unusable. The following patch does that. Features of this patch
are as follows:
- A flag is added to the mstate struct of the arena to indicate if the
arena is corrupt.
- The flag is checked whenever malloc functions try to get a lock on
an arena. If the arena is unusable, a NULL is returned, causing the
malloc to use mmap or try the next arena.
- malloc_printerr sets the corrupt flag on the arena when it detects a
corruption
- free does not concern itself with the flag at all. It is not
important since the backtrace workflow does not need free. A free
in a parallel thread may cause another corruption, but that's not
new
- The flag check and set are not atomic and may race. This is fine
since we don't care about contention during the flag check. We want
to make sure that the malloc call in the backtrace does not trip on
itself and all that action happens in the same thread and not across
threads.
I verified that the test case does not show any regressions due to
this patch. I also ran the malloc benchmarks and found an
insignificant difference in timings (< 2%).
* malloc/Makefile (tests): New test case tst-malloc-backtrace.
* malloc/arena.c (arena_lock): Check if arena is corrupt.
(reused_arena): Find a non-corrupt arena.
(heap_trim): Pass arena to unlink.
* malloc/hooks.c (malloc_check_get_size): Pass arena to
malloc_printerr.
(top_check): Likewise.
(free_check): Likewise.
(realloc_check): Likewise.
* malloc/malloc.c (malloc_printerr): Add arena argument.
(unlink): Likewise.
(munmap_chunk): Adjust.
(ARENA_CORRUPTION_BIT): New macro.
(arena_is_corrupt): Likewise.
(set_arena_corrupt): Likewise.
(sysmalloc): Use mmap if there are no usable arenas.
(_int_malloc): Likewise.
(__libc_malloc): Don't fail if arena_get returns NULL.
(_mid_memalign): Likewise.
(__libc_calloc): Likewise.
(__libc_realloc): Adjust for additional argument to
malloc_printerr.
(_int_free): Likewise.
(malloc_consolidate): Likewise.
(_int_realloc): Likewise.
(_int_memalign): Don't touch corrupt arenas.
* malloc/tst-malloc-backtrace.c: New test case.
The conditional that evaluates if there are any FAILed test cases
currently always fails, since we ensure it fails if we find any
unexpected results in tests.sum and it would obviously fail if it does
not find failed results in tests.sum. This patch fixes this by simply
inverting the result of the egrep, i.e. succeed if egrep fails (to
find failed results) and fail if it succeeds.
Tested with 'make subdirs=localedata check' and 'make subdirs=locale
check' where all tests succeed and with 'make subdirs=elf check' where
a couple of tests fail for me.
* Makefile (summarize-tests): Fix return value on success.
I was told that Ma Shimao submitted a patch to add envz_remove to the
libc manual, but the patch could not be accepted since he does not
have a copyright assignment in place. I have been woefully behind on
libc-alpha recently and have not seen the patch or the discussion
thread. I have also not read the man page for envz_remove, so
Alexandre Oliva asked me if I could write this independently and post
a patch. The patch below is the result of the same - I have written
it based on the implementation in string/envz.c and Alex told me via
email that the function is AS, AC and MT-safe like envz_strip.
I assume Alex and Carlos cannot review this since they have been
tainted by the original patch (I haven't even tried to look for a link
to it since I don't want to be tainted) so someone else will have to
review this. If there are no reviewers till the end of the week, I
will commit this since I believe there is a chance that there are no
other reviewers who haven't read that thread.
* manual/string.texi (Envz Functions): Add envz_remove.
DNSSEC defines a number of response types that one me expect when the
DO bit is set. We don't process any of them, but since we do allow
setting the DO bit, skip them without logging an error since it is
only a nuisance.
Tested on x86_64.
[BZ #14841]
* resolv/gethnamaddr.c (getanswer): Skip logging if
RES_USE_DNSSEC is set.
* resolv/nss_dns/dns-host.c (getanswer_r): Likewise.
This seems to have been left behind as an artifact of some old changes
and can now be merged. Verified that the only generated code change
on x86_64 is that of line numbers in asserts, like so:
@@ -27253,7 +27253,7 @@ Disassembly of section .text:
416f09: 48 89 42 20 mov %rax,0x20(%rdx)
416f0d: e9 7e f6 ff ff jmpq 416590 <_int_free+0x230>
416f12: b9 3f 9f 4a 00 mov $0x4a9f3f,%ecx
- 416f17: ba d5 0f 00 00 mov $0xfd5,%edx
+ 416f17: ba d6 0f 00 00 mov $0xfd6,%edx
416f1c: be a8 9b 4a 00 mov $0x4a9ba8,%esi
416f21: bf 6a 9c 4a 00 mov $0x4a9c6a,%edi
416f26: e8 45 e8 ff ff callq 415770 <__malloc_assert>
The padding bytes in the statsdata struct are not initialized, due to
which valgrind throws a warning:
==11384== Memcheck, a memory error detector
==11384== Copyright (C) 2002-2012, and GNU GPL'd, by Julian Seward et al.
==11384== Using Valgrind-3.8.1 and LibVEX; rerun with -h for copyright info
==11384== Command: nscd -d
==11384==
Fri 25 Apr 2014 10:34:53 AM CEST - 11384: handle_request: request received (Version = 2) from PID 11396
Fri 25 Apr 2014 10:34:53 AM CEST - 11384: GETSTAT
==11384== Thread 6:
==11384== Syscall param socketcall.sendto(msg) points to uninitialised byte(s)
==11384== at 0x4E4ACDC: send (in /lib64/libpthread-2.12.so)
==11384== by 0x11AF6B: send_stats (in /usr/sbin/nscd)
==11384== by 0x112F75: nscd_run_worker (in /usr/sbin/nscd)
==11384== by 0x4E439D0: start_thread (in /lib64/libpthread-2.12.so)
==11384== by 0x599AB6C: clone (in /lib64/libc-2.12.so)
==11384== Address 0x15708395 is on thread 6's stack
Fix the warning by initializing the structure.
commit 050f7298e1 added an extern
declaration for __tls_get_addr that conflicts with the one in s390
dl-tls.h, based on whether __tls_get_addr is defined as a macro. The
rationale seems to be based on the assumption that __tls_get_addr is
exported for every architecture and hence an internal non-plt alias is
needed. This is not true for s390 though, since it exports
__tls_get_offset and not __tls_get_addr. This results in tst-audit9
being stuck in an infinite loop.
This patch fixes this by defining a __tls_get_addr macro to itself so
as to not use the conflicting declaration.
True multi-dimensional arrays were introduced in awk 4.0 and we
support awk versions as early as 3.12. Use a single subscript of the
form prefix_conf instead of two dimensions to work around this
limitation. We also need one additional array of just the conf names
subscripted by the prefix_conf to print the names for the
specifications.
* scripts/gen-posix-conf-vars.awk: Don't use multi-dimensional
arrays.
This patch adds support to generate the spec array in getconf from the
conf.list. The generated code is mostly unchanged. the only changes
are due to the change in layout of the spec and val arrays in the ELF.
The val array can also be auto-generated from posix-conf-vars.list
once the remaining macros are added to it.
* posix/posix-conf-vars.list (SPEC:XBS5): Add sysconf prefix.
* posix/confstr.c: Define NEED_SPEC_ARRAY to 0.
* posix/posix-envs.def: Likewise.
* sysdeps/posix/sysconf.c: Likewise.
* posix/getconf.c: Define NEED_SPEC_ARRAY to 1.
(specs): Remove array.
* scripts/gen-posix-conf-vars.awk: Support generation of specs
array.
This fixes the remaining -Wundef warnings. Tested on x86_64.
* posix/posix-conf-vars.list: Add _POSIX sysconf namespace.
* sysdeps/posix/sysconf.c: Include posix-conf-vars.h.
(__sysconf): Use CONF_IS_* macros.
This patch adds a file posix-conf-vars.list that is used to generate
macros to determine if a macro is defined as set, unset or not
defined. gen-posix-conf-vars.awk processes this file and generates a
header (posix-conf-vars-def.h) with these macros. A new header
posix-conf-vars.h includes this generated header and defines accessor
macros for the generated macros.
Tested on x86_64.
* posix/Makefile (before-compile): Add posix-conf-vars-def.h.
($(objpfx)posix-conf-vars-def.h): New target.
* posix/posix-conf-vars.list: New file.
* posix/posix-conf-vars.h: New file.
* posix/confstr.c: Include posix-conf-vars.h.
(confstr): Use CONF_IS_* macros.
* posix/posix-envs.def: Include posix-conf-vars.h. Use
CONF_IS_* macros.
* scripts/gen-posix-conf-vars.awk: New file.
I see this warning in my build on F21 x86_64, which seems to be due to
a weak check for array bounds. Fixed by making the bounds check
stronger.
This is not an actual bug since nscount is never set to anything
greater than MAXNS. The compiler however does not know this, so we
need the stronger bounds check to quieten the compiler.
POSIX allows applications to switch file handles when a read results
in an end of file. Unset the cached offset at this point so that it
is queried again.
The test case fails to truncate the file when a file is intended to be
opened in w or w+ mode. Add O_TRUNC to fix this. The test still
succeeds with this change.
Currently we seek to end of file if there are unflushed writes or the
stream is in write mode, to get the current offset for writing in
append mode, which is the end of file. The latter case (i.e. stream
is in write mode, but no unflushed writes) is unnecessary since it
will only happen when the stream has just been flushed, in which case
the recorded offset ought to be reliable.
Removing that case lets ftell give the correct offset when it follows
an ftruncate. The latter truncates the file, but does not change the
file position, due to which it is permissible to call ftell without an
intervening fseek call.
Tested on x86_64 to verify that the added test case fails without the
patch and succeeds with it, and that there are no additional
regressions due to it.
[BZ #17647]
* libio/fileops.c (do_ftell): Seek only when there are
unflushed writes.
* libio/wfileops.c (do_ftell_wide): Likewise.
* libio/tst-ftell-active-handler.c (do_ftruncate_test): New
test case.
(do_one_test): Call it.
gcc now warns when the arguments to memset may have been accidentally
transposed (i.e. length set to zero instead of the byte), so we don't
need that bit of the code in glibc headers anymore.
Tested on x86_64. Coe generated by gcc 4.8 is identical with or
without the patch. I also tested gcc master, which does not result in
any new failures. It does fail quite a few FORTIFY_SOURCE tests, but
those failures are not due to this patch.
This change is only useful for the conformance tests since the headers
changed are not installed. The conformance tests fail due to IS_IN
not being defined, so wrap it with a check to make sure that _ISOMAC
is defined.
* include/bits/stdlib-float.h: Use IS_IN only if _ISOMAC is
defined.
* include/mqueue.h: Likewise.
* include/stdlib.h: Likewise.
Ignoring IS_IN_build, which has a different purpose altogether, this
patch removes the last bit of IS_IN_* macro definitions. Now barring
NOT_IN_libc, all cases use the IN_MODULE scheme.
Generated code unchanged on x86_64.
* extra-lib.mk (CPPFLAGS-$(lib)): Remove IS_IN_$(lib).
Replace with IS_IN (ldconfig). No change in generated code.
* elf/Makefile (CFLAGS-ldconfig.c): Remove definition of
IS_IN_ldconfig.
* sysdeps/unix/sysv/linux/x86_64/dl-procinfo.c: Use IS_IN.
* sysdeps/unix/sysv/linux/x86_64/dl-procinfo.h: Likewise.
Replace it with IS_IN (libc) and remove the one place that it
is defined in. The generated code remains unchanged on x86_64.
* include/shlib-compat.h [!NOT_IN_libc]: Remove.
* nss/nss_files/files-parse.c (IS_IN_libc): Replace with
IS_IN (libc).
Make sure that all instances where NOT_IN_libc is defined also defines
IN_MODULE to facilitate removal NOT_IN_libc in future passes.
Verified that the generated code is unchanged on x86_64.
* elf/Makefile (libof-sotruss-lib): Set as extramodules.
Replace with IS_IN and IS_IN_LIB macros instead. This change results
in a change in generated code, because it fixes a subtle bug. The bug
was introduced when systemtap probes were added to lowlevellock.h,
which resulted in stap-probe.h being included in a number of places.
stap-probe.h always defines IN_LIB, which breaks a check in errno.h
and netdb.h since they rely on that macro to decide whether to
implement an internal version of a declaration or an external one.
The components that see a code change due to this are:
iconv_prog
libmemusage.so
libpcprofile.so
libSegFault.so
libutil.so.1
locale
localedef
nscd
All other built components (i.e. libc, libpthread, etc.) remain
unchanged by this on x86_64.
* elf/Makefile (CPPFLAGS-.os): Remove IN_LIB.
* elf/rtld-Rules (rtld-CPPFLAGS): Likewise.
* extra-lib.mk (CPPFLAGS-$(lib)): Likewise.
* include/libc-symbols.h (IS_IN_LIB): New macro.
* include/errno.h: Use IS_IN_LIB instead of IN_LIB.
* include/netdb.h: Likewise.
* include/stap-probe.h: Remove all uses of IN_LIB.
Define MODULE_NAME in the build command and define IN_MODULE using
MODULE_NAME. Verified that the generated code is unchanged on x86_64.
* Makeconfig (module-cppflags-real): Define MODULE_NAME
instead of IN_MODULE.
* include/libc-symbols.h (IN_MODULE): Define using
MODULE_NAME.
(PASTE_NAME, PASTE_NAME1): New macros.
* include/stap-probe.h (LIBC_PROBE_1): Use MODULE_NAME instead
of IN_LIB.
(STAP_PROBE_ASM): Likewise.
Remove libc-modules.h from the tree and auto-generate it from
soversions.i and the list of modules in the built-modules variable
defined in Makeconfig. Macros generated have increasing numbered
values, with built-modules having lower values starting from 1,
following which a separator value LIBS_BEGIN is added and then finally
the library names from soversions.i are appended to the list. This
allows us to conveniently differentiate between the versioned
libraries and other built modules, which is needed in errno.h and
netdb.h to decide whether to use an internal symbol or an external
one.
Verified that generated code remains unchanged on x86_64.
* Makeconfig (built-modules): List non-library modules to be
built.
(module-cppflags): Include libc-modules.h for
everything except shlib-versions.v.i.
(CPPFLAGS): Use it.
(before-compile): Add libc-modules.h.
($(common-objpfx)libc-modules.h,
$(common-objpfx)libc-modules.stmp): New targets.
(common-generated): Add libc-modules.h and libc-modules.stmp.
($(common-objpfx)Versions.v.i): Depend on libc-modules.h.
* include/libc-symbols.h: Don't include libc-modules.h.
* include/libc-modules.h: Remove file.
* scripts/gen-libc-modules.awk: New script to generate
libc-modules.h.
* sysdeps/unix/Makefile ($(common-objpfx)sysd-syscalls):
Depend on libc-modules.stmp.
The current scheme to identify which module a translation unit is
built in depends on defining multiple macros IS_IN_* and also defining
NOT_IN_libc if we're building a non-libc module. In addition, there
is an IN_LIB macro that does effectively the same thing, but for
different modules (notably the systemtap probes). This macro scheme
unifies both ideas to use just one macro IN_MODULE and assign it a
value depending on the module it is being built into. If the module
is not defined, it defaults to MODULE_libc.
Patches that follow will replace uses of IS_IN_* variables with the
IS_IN() macro. libc-symbols.h has been converted already to give an
example of how such a transition will look.
Verified that there are no relevant binary changes. One source change
that will crop up repeatedly is that of nscd_stat, since it uses the
build timestamp as a constant in its logic.
* Makeconfig (in-module): Get value of libof set for the
translation unit.
(CPPFLAGS): Use $(in-module).
* Makerules: Don't suffix routine names for nonlib.
* include/libc-modules.h: New file.
* include/libc-symbols.h: Include libc-modules.h
(IS_IN): New macro to replace IS_IN_* macros.
* elf/Makefile: Set libof-* for each routine.
* elf/rtld-Rules: Likewise.
* extra-modules.mk: Likewise.
* iconv/Makefile: Likewise.
* iconvdata/Makefile: Likewise.
* locale/Makefile: Likewise.
* malloc/Makefile: Likewise.
* nss/Makefile: Likewise.
* sysdeps/gnu/Makefile: Likewise.
* sysdeps/ieee754/ldbl-opt/Makefile: Likewise.
* sysdeps/unix/sysv/linux/Makefile: Likewise.
* sysdeps/s390/s390-64/Makefile: Likewise.
* nscd/Makefile: Set libof-* for each routine. Set CFLAGS and
CPPFLAGS for nscd instead of nonlib.
The s390 ABI requires the stack pointer to be aligned at 8-bytes.
When a program is invoked as an argument to the dynamic linker,
_dl_start_user adjusts the stack to remove the dynamic linker
arguments so that the program sees only its name and arguments. This
may result in the stack being misaligned since each argument shift is
only a word and not a double-word.
This is now fixed shifting argv and envp down instead of shifting argc
up and reclaiming the stack. This requires _dl_argv to be adjusted
and hence, is no longer relro.
The recvmsg could return 0 under some conditions and cause the
make_request function to be stuck in an infinite loop.
Thank you Jim King <jim.king@simplivity.com> for posting Paul's patch
on the list.
Replace it with including an auto-generated linker-runtime.h.
Build-tested on x86_64 and found that there was no change in the
generated code.
* elf/Makefile (CFLAGS-interp.c): Remove.
($(elf-objpfx)runtime-linker.h): Generate header with linker
path string.
* elf/interp.c: Include generated runtime-linker.h
Barring libc.so and libdl.so, none of the libraries have any entry
points, so it is pointless to add a .interp section for them. The
libdl.so entry point (in dlfcn/eval.c) is also defunct, so remove that
file as well.
Build tested for x86_64, ppc64 and s390x. I have not moved
CFLAGS-interp.c to CPPFLAGS-interp.c isnce I'll be removing it
completely in a follow-up patch.
Siddhesh
* Makerules (lib%.so): Don't include $(+interp) in
prerequisites.
* elf/Makefile (CFLAGS-interp.c): Don't define NOT_IN_libc.
* dlfcn/eval.c: Remove file.
The fix for BZ #17266 (884ddc5081)
removed changes that had gone into cdefs.h to make
__extern_always_inline usable with clang++. This patch adds back
support for clang to detect if GNU inlining semantics are available,
this time without breaking the gcc use case. The check put here is
based on the earlier patch and assertion[1] that checking if
__GNUC_STDC_INLINE__ or __GNUC_GNU_INLINE__ is defined is sufficient
to determine that clang++ suports GNU inlining semantics.
Tested with a simple program that builds with __extern_always_inline
with the patch and fails compilation without it.
#include <stdio.h>
#include <sys/cdefs.h>
extern void foo_alias (void) __asm ("foo");
__extern_always_inline void
foo (void)
{
puts ("hi oh world!");
return foo_alias ();
}
void
foo_alias (void)
{
puts ("hell oh world");
}
int
main ()
{
foo ();
}
[1] https://sourceware.org/ml/libc-alpha/2012-12/msg00306.html
[BZ #17266]
* misc/sys/cdefs.h: Define __extern_always_inline for clang
4.2 and newer.
The check for only __GNUC_STDC_INLINE__ and __GNUC_GNU_INLINE__ may
not be sufficient since those flags were added during initial support
for C99 inlining semantics. There is also a problem with always
defining __extern_inline and __extern_always_inline, since it enables
inline wrapper functions even when GNU inlining semantics are not
guaranteed. This, along with the possibility of such wrappers using
redirection (btowc for example) could result in compiler generating an
infinitely recusrive call to the function.
In fact it was such a recursion that led to this code being written
the way it was; see:
https://bugzilla.redhat.com/show_bug.cgi?id=186410
The initial change was to fix bugs 14530 and 13741, but they can be
resolved by checking if __fortify_function and/or
__extern_always_inline are defined, as it has been done in this patch.
In addition, I have audited uses of __extern_always_inline to make
sure that none of the uses result in compilation errors.
There is however a regression in this patch for llvm, since it reverts
the llvm expectation that __GNUC_STDC_INLINE__ or __GNUC_GNU_INLINE__
definition imply proper extern inline semantics.
2014-09-16 Siddhesh Poyarekar <siddhesh@redhat.com>
Jakub Jelinek <jakub@redhat.com>
[BZ #17266]
* libio/stdio.h: Check definition of __fortify_function
instead of __extern_always_inline to include bits/stdio2.h.
* math/bits/math-finite.h [__USE_XOPEN || __USE_ISOC99]: Also
check if __extern_always_inline is defined.
[__USE_MISC || __USE_XOPEN]: Likewise.
[__USE_ISOC99] Likewise.
* misc/sys/cdefs.h (__fortify_function): Define only if
__extern_always_inline is defined.
[!__cplusplus || __GNUC_PREREQ (4,3)]: Revert to defining
__extern_always_inline and __extern_inline only for g++-4.3
and newer or a compatible gcc.
There is no _POSIX_REGEX_VERSION, so don't check for it.
_REGEX_VERSION has been removed as well[1], so only keep the -1 return
for backward compatibility. I found this when trying to make the
getconf environment variables typo-proof.
* sysdeps/posix/sysconf.c (__sysconf): Return -1 for
_SC_REGEX_VERSION.
[1] http://pubs.opengroup.org/onlinepubs/009695399/functions/sysconf.html
getconf only recognizes IPV6 and RAW_SOCKETS, when the standard
requires it to recognize the actual configuration variable name[1]. I
have not removed the earlier names for compatibility.
* posix/getconf.c (vars): Add _POSIX_IPV6 and
_POSIX_RAW_SOCKETS.
[1] http://pubs.opengroup.org/onlinepubs/007904875/functions/sysconf.html
Spell _POSIX_DEVICE_SPECIFIC and _POSIX_DEVICE_SPECIFIC_R correctly.
Found when trying to make the getconf environment variables
typo-proof.
* sysdeps/posix/sysconf.c (__sysconf): Spell
_POSIX_DEVICE_SPECIFIC and _POSIX_DEVICE_SPECIFIC_R correctly.
2d63a517e4 added support to save and
restore zmm register in the dynamic linker, but did not enhance
test-xmmymm.sh to detect accidental usage of these registers. The
patch below adds that check.
The script has also been renamed to tst-ld-sse-use.sh. To see the
minimal changes, run `git show -M`.
[BZ #16194]
* sysdeps/x86/tst-xmmymm.sh: Rename file to...
* sysdeps/x86/tst-ld-sse-use.sh: ... this. Check for zmm
register usage.
* sysdeps/x86/Makefile: Adjust.
The netgroups lookup code fails when one of the groups in the search
tree is empty. In such a case it only returns the leaves of the tree
after the blank netgroup. This is because the line parser returns a
NOTFOUND status when the netgroup exists but is empty. The
__getnetgrent_internal implementation needs to be fixed to try
remaining groups if the current group is entry. This patch implements
this fix. Tested on x86_64.
[BZ #17363]
* inet/getnetgrent_r.c (__internal_getnetgrent_r): Try next
group if the current group is empty.
pthread_atfork is already built in an extra-libs context, which gives
it NOT_IN_libc in its CPPFLAGS. Adding the same definition to CFLAGS
is pointless.
Verified that the code is unchanged on x86_64.
These programs get the NOT_IN_libc twice, once through the 'other'
target and another explicitly. Remove the explicitly added CPFLAG.
* catgets/Makefile (CPPFLAGS-gencat): Remove.
* iconv/Makefile (CPPFLAGS-iconv_prog): Likewise.
(CPPFLAGS-iconvconfig): Likewise.
* timezone/Makefile (CPPFLAGS-zic): Likewise.
Since:
commit 409e00bd69
Author: H.J. Lu <hjl.tools@gmail.com>
Date: Wed Jan 29 07:51:41 2014 -0800
Disable x87 inline functions for SSE2 math
When i386 and x86-64 mathinline.h was merged into a single mathinline.h,
"gcc -m32" enables x87 inline functions on x86-64 even when -mfpmath=sse
and SSE2 is enabled. It is a regression on x86-64. We should check
__SSE2_MATH__ instead of __x86_64__ when disabling x87 inline functions.
gcc-3.2 is unable to correctly compile x86_64 routines for llrint
since it gets redefined. This is because gcc 3.2 does not set
__SSE2_MATH__ for x86_64, thus exposing the duplicate definition.
The correct fix ought to be to check for both __SSE2_MATH__ and
__x86_64__ and enable those bits only when neither are defined.
Tested fix with the reproducer for
409e00bd69 as well as with gcc-3.2.
The DEBUG macro is checked for its value in one place and if it is
defined in another. Make this consistent across the two cases and use
the same style that we did in mktime.c, which is to check if the macro
is defined and it is set.
Fixes -Wundef warnings for MBSTATE_T and HAVE_STRING_H. The
HAVE_STRING_H bit is a merge from the gnulib fnmatch.c and the
MBSTATE_T fix has been posted to gnulib for inclusion.
Summary of changes:
- Use of !_LIBC instead of HAVE_CONFIG_H
- Code changes in [!_LIBC] that don't affect us
- Minor formatting changes
- Use __builtin_expect in shared code
- Define some macros in [_LIBC] that are used in gnulib but never
defined in glibc
- Flip macro check for STRERROR_R_CHAR_P so that it does not throw a
warning
Here's an updated patch to fix the crash in bug-ga2 when the system
has no configured ipv6 address. I have taken a different approach of
using libc_freeres_fn instead of the libc_freeres_ptr since the former
gives better control over what is freed; we need that since cache may
or may not be allocated using malloc.
Verified that bug-ga2 works correctly in both cases and does not have
memory leaks in either of them.
While we're at fixing build warnings, here's one unnecessary warning
that can be fixed fairly easily. The SIZE variable is never actually
use uninitialized, but the compiler cannot make that out and thinks
(correctly) that there is a potential for accessing SIZE without
initializing it. Make this safe by initializing SIZE to 0.
Tested on x86_64.
There was a typo in the previous patch due to which resplen2 was
checked for non-zero instead of the value at resplen2. Fix that and
improve the condition by checking resplen2 for non-NULL (instead of
answerp2) and also adding the check in a third place.
Siddhesh Poyarekar