On s390x, gcc inlines more aggresive compared to other architectures.
This occaisionally leads to build warnings / errors.
Therefore this patch adds a s390x glibc variant with optimization.
There is the ccopts field which contain ABI options which are passed
to configure as CC / CXX. Now there is also the cflags field which
contains non-ABI options like -g or -O. Those are passed to configure
as CFLAGS / CXXFLAGS.
Currently CC is passed to conformtest.py or linknamespace.py but not
the CFLAGS.
If build with optimization, stdio.h and sys/stat.h are defining some inlining
functions. This leads to test fails if glibc is build with the following
commands. (Note that the conformtests usually builds without optimization or
other CFLAGS):
<glibc>/configure CC="gcc -O3" --prefix=/usr
make
make subdirs=conform check
- FAIL: conform/XPG4/stdio.h/conform
- FAIL: conform/XPG42/stdio.h/conform
out-files:
...
PASSCOMBINED: Availability of variable optopt
PASSCOMBINED: Type of variable optopt
Namespace violation: "getc_unlocked"
Namespace violation: "getchar_unlocked"
Namespace violation: "putc_unlocked"
Namespace violation: "putchar_unlocked"
FAIL: Namespace of <stdio.h>
----------------------------------------------------------------------------
Total number of tests : 168
Number of failed tests : 1
Number of xfailed tests : 0
Number of skipped tests : 0
- FAIL: conform/POSIX2008/sys/stat.h/conform
out-file:
...
PASSCOMBINED: Availability of function utimensat
PASSCOMBINED: Type of function utimensat
Namespace violation: "mknodat"
FAIL: Namespace of <sys/stat.h>
----------------------------------------------------------------------------
Total number of tests : 97
Number of failed tests : 1
Number of xfailed tests : 0
Number of skipped tests : 0
For getc_unlocked, getchar_unlocked, putc_unlocked, putchar_unlocked in stdio.h,
those are defined "# ifdef __USE_POSIX" instead of "#ifdef __USE_POSIX199506"
for the non-inlining declaration. See also
"Bug 20014 - stdio.h namespace for pre-threads POSIX"
(https://sourceware.org/bugzilla/show_bug.cgi?id=20014).
For mknodat in sys/stat.h, those are defined "# ifdef __USE_ATFILE" instead of
the additional guard "# if defined __USE_MISC || defined __USE_XOPEN_EXTENDED".
C2x adds BOOL_MAX and BOOL_WIDTH macros to <limits.h>. This patch
adds them to glibc's <limits.h> for the case when they aren't defined
by GCC's <limits.h>.
Tested for x86_64.
This patch adds the new STATX_MNT_ID, STATX_ATTR_MOUNT_ROOT and
STATX_ATTR_DAX macros from Linux 5.8 to glibc's bits/statx-generic.h.
(As with previous such changes, this only does anything if glibc is
being used with old kernel headers.)
A comment in the Linux kernel headers indicates that STATX_ALL is
deliberately not being changed.
Tested for x86_64.
Properly serialize the access to the global state shared between the
syslog functions, to avoid races in multithreaded processes. Protect a
local allocation in the __vsyslog_internal function from leaking during
cancellation.
Since binutils has obsoleted ia64 support, use --enable-obsolete for
now when configuring binutils for ia64 in build-many-glibcs.py (which
requires adding support for architecture-specific binutils configure
options there). As with other obsoletions, the removal of support for
ia64 in any of (binutils, GCC, Linux kernel) should imply its removal
from glibc.
Tested with build-many-glibcs.py for ia64-linux-gnu (compilers and
glibcs build).
This patch updates the kernel version in the test tst-mman-consts.py
to 5.8. (There are no new MAP_* constants covered by this test in 5.8
that need any other header changes.)
Tested with build-many-glibcs.py.
The pthread_clockjoin_np and pthread_timedjoin_np have been converted to
support 64 bit time.
This change introduces new futex_timed_wait_cancel64 function in
./sysdeps/nptl/futex-internal.h, which uses futex_time64 where possible
and tries to replace low-level preprocessor macros from
lowlevellock-futex.h
The pthread_{timed|clock}join_np only accept absolute time. Moreover,
there is no need to check for NULL passed as *abstime pointer as
clockwait_tid() always passes struct __timespec64.
For systems with __TIMESIZE != 64 && __WORDSIZE == 32:
- Conversions between 64 bit time to 32 bit are necessary
- Redirection to __pthread_{clock|timed}join_np64 will provide support
for 64 bit time
Build tests:
./src/scripts/build-many-glibcs.py glibcs
Run-time tests:
- Run specific tests on ARM/x86 32bit systems (qemu):
https://github.com/lmajewski/meta-y2038 and run tests:
https://github.com/lmajewski/y2038-tests/commits/master
Above tests were performed with Y2038 redirection applied as well as without
to test the proper usage of both __pthread_{timed|clock}join_np64 and
__pthread_{timed|clock}join_np.
Reviewed-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>
Reviewed-by: Alistair Francis <alistair.francis@wdc.com>
This provides correct AT_EACCESS handling and also takes
Linux security modules into account.
Reviewed-by: Adhemerval Zanella <adhemerval.zanella@linaro.org>
Linux 5.8 has one new syscall, faccessat2. Update syscall-names.list
and regenerate the arch-syscall.h headers with build-many-glibcs.py
update-syscalls.
Tested with build-many-glibcs.py.
Making the brk start exactly at the end of the main application binary was
requiring to get it through the _end symbol, which does not work any more
with recent toolchains, and actually produces in libc.so a confusing
external _end symbol that produces odd results, see
https://sourceware.org/bugzilla/show_bug.cgi?id=23499
Trying to do so is quite outdated anyway with the tendency for address
randomization.
Using _end was also allowing to include the main binary data within
the RLIMIT_DATA, but this also seems outdated with dynamic library
loading, and nowadays' memory consumption via malloc and mmap rather than
statically-allocated data.
This adds a BRK_START macro in <vm_param.h> that just tells where we
want to start the brk, and thus removes the _end symbol.
* sysdeps/mach/hurd/i386/vm_param.h: New file.
* sysdeps/mach/hurd/brk.c: Use BRK_START as brk start instead of _end.
Also ignore __data_start.
* hurd/Versions: Remove _end symbol.
* sysdeps/mach/hurd/i386/libc.abilist: Remove _end symbol.
Intel64 and IA-32 Architectures Software Developer’s Manual has changed
the following CPU feature names:
1. The CPU feature of Enhanced Intel SpeedStep Technology is renamed
from EST to EIST.
2. The CPU feature which supports Platform Quality of Service Monitoring
(PQM) capability is changed to Intel Resource Director Technology
(Intel RDT) Monitoring capability, i.e. PQM is renamed to RDT_M.
3. The CPU feature which supports Platform Quality of Service
Enforcement (PQE) capability is changed to Intel Resource Director
Technology (Intel RDT) Allocation capability, i.e. PQE is renamed to
RDT_A.
Sync this file from Gnulib, thus incorporating the following
fix for a bug with regexps with 16 or more subexpressions:
* posix/regex_internal.h (struct re_backref_cache_entry):
Use bitset_word_t as the type of eps_reachable_subexps_map,
instead of unsigned short int. This fixes a bug I introduced
to glibc in 2005-09-28T17:33:18Z!drepper@redhat.com (glibc commit
2c05d33f90, BZ #1302).
Remove unused member 'unused'.
* posix/regex.c (__STDC_WANT_IEC_60559_BFP_EXT__):
Define, for ULONG_WIDTH. This syncs regex.c from Gnujlib.
* posix/regex_internal.h (ULONG_WIDTH):
Use a more-portable fallback, from Gnulib.
(BITSET_WORD_BITS): Now defined in terms of ULONG_WIDTH.
* include/intprops.h: Sync from Gnulib. This improves
performance of INT_MULTIPLY_WRAPV on recent GCC, which affects
glibc only in the support library.
After some discussions it seems the original news was not clear
and that it is valid to manually pass the branch protection flags
iff GCC target libs are built with them too. The main difference
between manually passing the flags and using the configure
option is that the latter also makes branch protection the
default in GCC which may not be desirable in some cases.
Reviewed-by: Carlos O'Donell <carlos@redhat.com>
__GLRO loaded the word after the requested variable on big-endian
PowerPC, where LOWORD is 4. This can cause the memset implement
go wrong because the masking with the cache line size produces
wrong results, particularly if the loaded value happens to be 1.
The __GLRO macro is not used in any place where loading the lower
32-bit word of a 64-bit value is desired, so the +4 offset is always
wrong.
Fixes commit 18363b4f01
("powerpc: Move cache line size to rtld_global_ro") and bug 26332.
Reviewed-by: Carlos O'Donell <carlos@redhat.com>
It was fixed in commit d937694059
("Fix array overflow in backtrace on PowerPC (bug 25423)"), which
went into glibc 2.31.
Reviewed-by: Carlos O'Donell <carlos@redhat.com>
Storing user databases in DNS, without client-side DNSSEC validation,
is problematic from a security point of view.
Reviewed-by: Carlos O'Donell <carlos@redhat.com>
nptl has
/* Opcodes and data types for communication with the signal handler to
change user/group IDs. */
struct xid_command
{
int syscall_no;
long int id[3];
volatile int cntr;
volatile int error;
};
/* This must be last, otherwise the current thread might not have
permissions to send SIGSETXID syscall to the other threads. */
result = INTERNAL_SYSCALL_NCS (cmdp->syscall_no, 3,
cmdp->id[0], cmdp->id[1], cmdp->id[2]);
But the second argument of setgroups syscal is a pointer:
int setgroups (size_t size, const gid_t *list);
But on x32, pointers passed to syscall must have pointer type so that
they will be zero-extended. The kernel XID arguments are unsigned and
do not require sign extension. Change xid_command to
struct xid_command
{
int syscall_no;
unsigned long int id[3];
volatile int cntr;
volatile int error;
};
so that all arguments are zero-extended. A testcase is added for x32 and
setgroups returned with EFAULT when running as root without the fix.
