* [PATCH] Don't use IFUNC resolver for longjmp or system in libpthread (bug 21041)
@ 2017-08-08 14:50 Andreas Schwab
2017-08-08 14:55 ` Florian Weimer
` (3 more replies)
0 siblings, 4 replies; 10+ messages in thread
From: Andreas Schwab @ 2017-08-08 14:50 UTC (permalink / raw)
To: libc-alpha
Unlike the vfork forwarder and like the fork forwarder as in bug 19861,
there won't be a problem when the compiler does not turn this into a tail
call.
Andreas.
* nptl/pt-longjmp.c (longjmp, siglongjmp): Don't use IFUNC resolver.
* nptl/pt-system.c (system): Likewise.
diff --git a/nptl/pt-longjmp.c b/nptl/pt-longjmp.c
index 2ef757e687..8f3c6b3a09 100644
--- a/nptl/pt-longjmp.c
+++ b/nptl/pt-longjmp.c
@@ -25,21 +25,14 @@
symbol in libpthread, but the historical ABI requires it. For static
linking, there is no need to provide anything here--the libc version
will be linked in. For shared library ABI compatibility, there must be
- longjmp and siglongjmp symbols in libpthread.so; so we define them using
- IFUNC to redirect to the libc function. */
+ longjmp and siglongjmp symbols in libpthread.so.
-#if SHLIB_COMPAT (libpthread, GLIBC_2_0, GLIBC_2_22)
-
-# if HAVE_IFUNC
-
-# undef INIT_ARCH
-# define INIT_ARCH()
-# define DEFINE_LONGJMP(name) libc_ifunc (name, &__libc_longjmp)
-
-extern __typeof(longjmp) longjmp_ifunc;
-extern __typeof(siglongjmp) siglongjmp_ifunc;
+ With an IFUNC resolver, it would be possible to avoid the indirection,
+ but the IFUNC resolver might run before the __libc_longjmp symbol has
+ been relocated, in which case the IFUNC resolver would not be able to
+ provide the correct address. */
-# else /* !HAVE_IFUNC */
+#if SHLIB_COMPAT (libpthread, GLIBC_2_0, GLIBC_2_22)
static void __attribute__ ((noreturn, used))
longjmp_compat (jmp_buf env, int val)
@@ -47,14 +40,10 @@ longjmp_compat (jmp_buf env, int val)
__libc_longjmp (env, val);
}
-# define DEFINE_LONGJMP(name) strong_alias (longjmp_compat, name)
-
-# endif /* HAVE_IFUNC */
-
-DEFINE_LONGJMP (longjmp_ifunc)
-compat_symbol (libpthread, longjmp_ifunc, longjmp, GLIBC_2_0);
+strong_alias (longjmp_compat, longjmp_alias)
+compat_symbol (libpthread, longjmp_alias, longjmp, GLIBC_2_0);
-strong_alias (longjmp_ifunc, siglongjmp_ifunc)
-compat_symbol (libpthread, siglongjmp_ifunc, siglongjmp, GLIBC_2_0);
+strong_alias (longjmp_alias, siglongjmp_alias)
+compat_symbol (libpthread, siglongjmp_alias, siglongjmp, GLIBC_2_0);
#endif
diff --git a/nptl/pt-system.c b/nptl/pt-system.c
index f8ca6ba0d9..b30ddf2b39 100644
--- a/nptl/pt-system.c
+++ b/nptl/pt-system.c
@@ -25,29 +25,21 @@
libpthread, but the historical ABI requires it. For static linking,
there is no need to provide anything here--the libc version will be
linked in. For shared library ABI compatibility, there must be a
- 'system' symbol in libpthread.so; so we define it using IFUNC to
- redirect to the libc function. */
+ 'system' symbol in libpthread.so.
-#if SHLIB_COMPAT (libpthread, GLIBC_2_0, GLIBC_2_22)
-
-# if HAVE_IFUNC
-
-extern __typeof(system) system_ifunc;
-# undef INIT_ARCH
-# define INIT_ARCH()
-libc_ifunc (system_ifunc, &__libc_system)
+ With an IFUNC resolver, it would be possible to avoid the indirection,
+ but the IFUNC resolver might run before the __libc_system symbol has
+ been relocated, in which case the IFUNC resolver would not be able to
+ provide the correct address. */
-# else /* !HAVE_IFUNC */
+#if SHLIB_COMPAT (libpthread, GLIBC_2_0, GLIBC_2_22)
static int __attribute__ ((used))
system_compat (const char *line)
{
return __libc_system (line);
}
-strong_alias (system_compat, system_ifunc)
-
-# endif /* HAVE_IFUNC */
-
-compat_symbol (libpthread, system_ifunc, system, GLIBC_2_0);
+strong_alias (system_compat, system_alias)
+compat_symbol (libpthread, system_alias, system, GLIBC_2_0);
#endif
--
2.14.0
--
Andreas Schwab, SUSE Labs, schwab@suse.de
GPG Key fingerprint = 0196 BAD8 1CE9 1970 F4BE 1748 E4D4 88E3 0EEA B9D7
"And now for something completely different."
^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: [PATCH] Don't use IFUNC resolver for longjmp or system in libpthread (bug 21041)
2017-08-08 14:50 [PATCH] Don't use IFUNC resolver for longjmp or system in libpthread (bug 21041) Andreas Schwab
@ 2017-08-08 14:55 ` Florian Weimer
2017-08-08 14:57 ` H.J. Lu
` (2 subsequent siblings)
3 siblings, 0 replies; 10+ messages in thread
From: Florian Weimer @ 2017-08-08 14:55 UTC (permalink / raw)
To: Andreas Schwab, libc-alpha
On 08/08/2017 04:50 PM, Andreas Schwab wrote:
> Unlike the vfork forwarder and like the fork forwarder as in bug 19861,
> there won't be a problem when the compiler does not turn this into a tail
> call.
>
> Andreas.
>
> * nptl/pt-longjmp.c (longjmp, siglongjmp): Don't use IFUNC resolver.
> * nptl/pt-system.c (system): Likewise.
Looks good to me.
Thanks,
Florian
^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: [PATCH] Don't use IFUNC resolver for longjmp or system in libpthread (bug 21041)
2017-08-08 14:50 [PATCH] Don't use IFUNC resolver for longjmp or system in libpthread (bug 21041) Andreas Schwab
2017-08-08 14:55 ` Florian Weimer
@ 2017-08-08 14:57 ` H.J. Lu
2017-08-09 8:55 ` Andreas Schwab
2017-08-08 18:09 ` Joseph Myers
2017-11-01 13:11 ` Romain Naour
3 siblings, 1 reply; 10+ messages in thread
From: H.J. Lu @ 2017-08-08 14:57 UTC (permalink / raw)
To: Andreas Schwab; +Cc: GNU C Library
On Tue, Aug 8, 2017 at 7:50 AM, Andreas Schwab <schwab@suse.de> wrote:
> Unlike the vfork forwarder and like the fork forwarder as in bug 19861,
> there won't be a problem when the compiler does not turn this into a tail
> call.
>
> Andreas.
>
> * nptl/pt-longjmp.c (longjmp, siglongjmp): Don't use IFUNC resolver.
> * nptl/pt-system.c (system): Likewise.
Missing BZ #.
Can you add a testcase from
https://sourceware.org/bugzilla/show_bug.cgi?id=21041
> diff --git a/nptl/pt-longjmp.c b/nptl/pt-longjmp.c
> index 2ef757e687..8f3c6b3a09 100644
> --- a/nptl/pt-longjmp.c
> +++ b/nptl/pt-longjmp.c
> @@ -25,21 +25,14 @@
> symbol in libpthread, but the historical ABI requires it. For static
> linking, there is no need to provide anything here--the libc version
> will be linked in. For shared library ABI compatibility, there must be
> - longjmp and siglongjmp symbols in libpthread.so; so we define them using
> - IFUNC to redirect to the libc function. */
> + longjmp and siglongjmp symbols in libpthread.so.
>
> -#if SHLIB_COMPAT (libpthread, GLIBC_2_0, GLIBC_2_22)
> -
> -# if HAVE_IFUNC
> -
> -# undef INIT_ARCH
> -# define INIT_ARCH()
> -# define DEFINE_LONGJMP(name) libc_ifunc (name, &__libc_longjmp)
> -
> -extern __typeof(longjmp) longjmp_ifunc;
> -extern __typeof(siglongjmp) siglongjmp_ifunc;
> + With an IFUNC resolver, it would be possible to avoid the indirection,
> + but the IFUNC resolver might run before the __libc_longjmp symbol has
> + been relocated, in which case the IFUNC resolver would not be able to
> + provide the correct address. */
>
> -# else /* !HAVE_IFUNC */
> +#if SHLIB_COMPAT (libpthread, GLIBC_2_0, GLIBC_2_22)
>
> static void __attribute__ ((noreturn, used))
> longjmp_compat (jmp_buf env, int val)
> @@ -47,14 +40,10 @@ longjmp_compat (jmp_buf env, int val)
> __libc_longjmp (env, val);
> }
>
> -# define DEFINE_LONGJMP(name) strong_alias (longjmp_compat, name)
> -
> -# endif /* HAVE_IFUNC */
> -
> -DEFINE_LONGJMP (longjmp_ifunc)
> -compat_symbol (libpthread, longjmp_ifunc, longjmp, GLIBC_2_0);
> +strong_alias (longjmp_compat, longjmp_alias)
> +compat_symbol (libpthread, longjmp_alias, longjmp, GLIBC_2_0);
>
> -strong_alias (longjmp_ifunc, siglongjmp_ifunc)
> -compat_symbol (libpthread, siglongjmp_ifunc, siglongjmp, GLIBC_2_0);
> +strong_alias (longjmp_alias, siglongjmp_alias)
> +compat_symbol (libpthread, siglongjmp_alias, siglongjmp, GLIBC_2_0);
>
> #endif
> diff --git a/nptl/pt-system.c b/nptl/pt-system.c
> index f8ca6ba0d9..b30ddf2b39 100644
> --- a/nptl/pt-system.c
> +++ b/nptl/pt-system.c
> @@ -25,29 +25,21 @@
> libpthread, but the historical ABI requires it. For static linking,
> there is no need to provide anything here--the libc version will be
> linked in. For shared library ABI compatibility, there must be a
> - 'system' symbol in libpthread.so; so we define it using IFUNC to
> - redirect to the libc function. */
> + 'system' symbol in libpthread.so.
>
> -#if SHLIB_COMPAT (libpthread, GLIBC_2_0, GLIBC_2_22)
> -
> -# if HAVE_IFUNC
> -
> -extern __typeof(system) system_ifunc;
> -# undef INIT_ARCH
> -# define INIT_ARCH()
> -libc_ifunc (system_ifunc, &__libc_system)
> + With an IFUNC resolver, it would be possible to avoid the indirection,
> + but the IFUNC resolver might run before the __libc_system symbol has
> + been relocated, in which case the IFUNC resolver would not be able to
> + provide the correct address. */
>
> -# else /* !HAVE_IFUNC */
> +#if SHLIB_COMPAT (libpthread, GLIBC_2_0, GLIBC_2_22)
>
> static int __attribute__ ((used))
> system_compat (const char *line)
> {
> return __libc_system (line);
> }
> -strong_alias (system_compat, system_ifunc)
> -
> -# endif /* HAVE_IFUNC */
> -
> -compat_symbol (libpthread, system_ifunc, system, GLIBC_2_0);
> +strong_alias (system_compat, system_alias)
> +compat_symbol (libpthread, system_alias, system, GLIBC_2_0);
>
> #endif
> --
> 2.14.0
>
>
> --
> Andreas Schwab, SUSE Labs, schwab@suse.de
> GPG Key fingerprint = 0196 BAD8 1CE9 1970 F4BE 1748 E4D4 88E3 0EEA B9D7
> "And now for something completely different."
--
H.J.
^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: [PATCH] Don't use IFUNC resolver for longjmp or system in libpthread (bug 21041)
2017-08-08 14:50 [PATCH] Don't use IFUNC resolver for longjmp or system in libpthread (bug 21041) Andreas Schwab
2017-08-08 14:55 ` Florian Weimer
2017-08-08 14:57 ` H.J. Lu
@ 2017-08-08 18:09 ` Joseph Myers
2017-08-09 8:55 ` Andreas Schwab
2017-11-01 13:11 ` Romain Naour
3 siblings, 1 reply; 10+ messages in thread
From: Joseph Myers @ 2017-08-08 18:09 UTC (permalink / raw)
To: Andreas Schwab; +Cc: libc-alpha
I suspect this of having caused s390 build failures.
s390x-glibc-linux-gnu-gcc -m31
../sysdeps/unix/sysv/linux/s390/pt-longjmp.c -c -std=gnu11 -fgnu89-inline
-O2 -Wall -Werror -Wundef -Wwrite-strings -fmerge-all-constants
-fno-stack-protector -frounding-math -g -Wstrict-prototypes
-Wold-style-definition -mlong-double-128 -fPIC -ftls-model=initial-exec
-I../include
-I/scratch/jmyers/glibc-bot/build/glibcs/s390-linux-gnu/glibc/nptl
-I/scratch/jmyers/glibc-bot/build/glibcs/s390-linux-gnu/glibc
-I../sysdeps/unix/sysv/linux/s390/s390-32
-I../sysdeps/unix/sysv/linux/s390/fpu -I../sysdeps/s390/fpu
-I../sysdeps/unix/sysv/linux/s390 -I../sysdeps/s390/nptl
-I../sysdeps/ieee754/ldbl-64-128 -I../sysdeps/ieee754/ldbl-opt
-I../sysdeps/unix/sysv/linux/include -I../sysdeps/unix/sysv/linux
-I../sysdeps/nptl -I../sysdeps/pthread -I../sysdeps/gnu
-I../sysdeps/unix/inet -I../sysdeps/unix/sysv -I../sysdeps/unix
-I../sysdeps/posix -I../sysdeps/s390/s390-32/multiarch
-I../sysdeps/s390/s390-32 -I../sysdeps/wordsize-32
-I../sysdeps/s390/multiarch -I../sysdeps/s390
-I../sysdeps/ieee754/ldbl-128 -I../sysdeps/ieee754/dbl-64
-I../sysdeps/ieee754/flt-32 -I../sysdeps/ieee754 -I../sysdeps/generic
-I.. -I../libio -I. -D_LIBC_REENTRANT -include
/scratch/jmyers/glibc-bot/build/glibcs/s390-linux-gnu/glibc/libc-modules.h
-DMODULE_NAME=libpthread -include ../include/libc-symbols.h -DPIC
-DSHARED -DTOP_NAMESPACE=glibc -o
/scratch/jmyers/glibc-bot/build/glibcs/s390-linux-gnu/glibc/nptl/pt-longjmp.os
-MD -MP -MF
/scratch/jmyers/glibc-bot/build/glibcs/s390-linux-gnu/glibc/nptl/pt-longjmp.os.dt
-MT
/scratch/jmyers/glibc-bot/build/glibcs/s390-linux-gnu/glibc/nptl/pt-longjmp.os
In file included from <command-line>:0:0:
../sysdeps/unix/sysv/linux/s390/pt-longjmp.c:29:15: error: 'longjmp_ifunc'
undeclared here (not in a function); did you mean 'longjmp_alias'?
strong_alias (longjmp_ifunc, __v2longjmp)
^
./../include/libc-symbols.h:127:20: note: in definition of macro
'_strong_alias'
extern __typeof (name) aliasname __attribute__ ((alias (#name)));
^~~~
../sysdeps/unix/sysv/linux/s390/pt-longjmp.c:29:1: note: in expansion of
macro 'strong_alias'
strong_alias (longjmp_ifunc, __v2longjmp)
^~~~~~~~~~~~
../sysdeps/unix/sysv/linux/s390/pt-longjmp.c:31:15: error:
'siglongjmp_ifunc' undeclared here (not in a function); did you mean
'longjmp_ifunc'?
strong_alias (siglongjmp_ifunc, __v2siglongjmp)
^
./../include/libc-symbols.h:127:20: note: in definition of macro
'_strong_alias'
extern __typeof (name) aliasname __attribute__ ((alias (#name)));
^~~~
../sysdeps/unix/sysv/linux/s390/pt-longjmp.c:31:1: note: in expansion of
macro 'strong_alias'
strong_alias (siglongjmp_ifunc, __v2siglongjmp)
^~~~~~~~~~~~
../sysdeps/unix/sysv/linux/s390/pt-longjmp.c:31:33: error:
'__v2siglongjmp' aliased to undefined symbol 'siglongjmp_ifunc'
strong_alias (siglongjmp_ifunc, __v2siglongjmp)
^
./../include/libc-symbols.h:127:20: note: in definition of macro
'_strong_alias'
extern __typeof (name) aliasname __attribute__ ((alias (#name)));
^~~~
../sysdeps/unix/sysv/linux/s390/pt-longjmp.c:31:1: note: in expansion of
macro 'strong_alias'
strong_alias (siglongjmp_ifunc, __v2siglongjmp)
^~~~~~~~~~~~
../sysdeps/unix/sysv/linux/s390/pt-longjmp.c:31:33: error:
'__v2siglongjmp' aliased to undefined symbol 'siglongjmp_ifunc'
strong_alias (siglongjmp_ifunc, __v2siglongjmp)
^
./../include/libc-symbols.h:127:26: note: in definition of macro
'_strong_alias'
extern __typeof (name) aliasname __attribute__ ((alias (#name)));
^~~~~~~~~
../sysdeps/unix/sysv/linux/s390/pt-longjmp.c:31:1: note: in expansion of
macro 'strong_alias'
strong_alias (siglongjmp_ifunc, __v2siglongjmp)
^~~~~~~~~~~~
../sysdeps/unix/sysv/linux/s390/pt-longjmp.c:29:30: error: '__v2longjmp'
aliased to undefined symbol 'longjmp_ifunc'
strong_alias (longjmp_ifunc, __v2longjmp)
^
./../include/libc-symbols.h:127:26: note: in definition of macro
'_strong_alias'
extern __typeof (name) aliasname __attribute__ ((alias (#name)));
^~~~~~~~~
../sysdeps/unix/sysv/linux/s390/pt-longjmp.c:29:1: note: in expansion of
macro 'strong_alias'
strong_alias (longjmp_ifunc, __v2longjmp)
^~~~~~~~~~~~
/scratch/jmyers/glibc-bot/build/glibcs/s390-linux-gnu/glibc/sysd-rules:151:
recipe for target
'/scratch/jmyers/glibc-bot/build/glibcs/s390-linux-gnu/glibc/nptl/pt-longjmp.os'
failed
make[3]: ***
[/scratch/jmyers/glibc-bot/build/glibcs/s390-linux-gnu/glibc/nptl/pt-longjmp.os]
Error 1
--
Joseph S. Myers
joseph@codesourcery.com
^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: [PATCH] Don't use IFUNC resolver for longjmp or system in libpthread (bug 21041)
2017-08-08 14:57 ` H.J. Lu
@ 2017-08-09 8:55 ` Andreas Schwab
0 siblings, 0 replies; 10+ messages in thread
From: Andreas Schwab @ 2017-08-09 8:55 UTC (permalink / raw)
To: H.J. Lu; +Cc: GNU C Library
I've commited this.
Andreas.
* nptl/Makefile (tests) [$(build-shared) = yes]: Add
tst-compat-forwarder.
(modules-names): Add tst-compat-forwarder-mod.
($(objpfx)tst-compat-forwarder): Depend on
$(objpfx)tst-compat-forwarder-mod.so.
* nptl/tst-compat-forwarder.c: New file.
* nptl/tst-compat-forwarder-mod.c: New file.
diff --git a/nptl/Makefile b/nptl/Makefile
index 128a52eadc..9925f0f745 100644
--- a/nptl/Makefile
+++ b/nptl/Makefile
@@ -367,7 +367,7 @@ tests += tst-cancelx2 tst-cancelx3 tst-cancelx4 tst-cancelx5 \
tst-cleanupx0 tst-cleanupx1 tst-cleanupx2 tst-cleanupx3 tst-cleanupx4 \
tst-oncex3 tst-oncex4
ifeq ($(build-shared),yes)
-tests += tst-atfork2 tst-tls4 tst-_res1 tst-fini1
+tests += tst-atfork2 tst-tls4 tst-_res1 tst-fini1 tst-compat-forwarder
tests-internal += tst-tls3 tst-tls3-malloc tst-tls5 tst-stackguard1
tests-nolibpthread += tst-fini1
ifeq ($(have-z-execstack),yes)
@@ -379,7 +379,7 @@ modules-names = tst-atfork2mod tst-tls3mod tst-tls4moda tst-tls4modb \
tst-tls5mod tst-tls5moda tst-tls5modb tst-tls5modc \
tst-tls5modd tst-tls5mode tst-tls5modf tst-stack4mod \
tst-_res1mod1 tst-_res1mod2 tst-execstack-mod tst-fini1mod \
- tst-join7mod
+ tst-join7mod tst-compat-forwarder-mod
extra-test-objs += $(addsuffix .os,$(strip $(modules-names))) \
tst-cleanup4aux.o tst-cleanupx4aux.o
test-extras += tst-cleanup4aux tst-cleanupx4aux
@@ -718,6 +718,8 @@ $(objpfx)tst-oddstacklimit.out: $(objpfx)tst-oddstacklimit $(objpfx)tst-basic1
$(evaluate-test)
endif
+$(objpfx)tst-compat-forwarder: $(objpfx)tst-compat-forwarder-mod.so
+
# The tests here better do not run in parallel
ifneq ($(filter %tests,$(MAKECMDGOALS)),)
.NOTPARALLEL:
diff --git a/nptl/tst-compat-forwarder-mod.c b/nptl/tst-compat-forwarder-mod.c
new file mode 100644
index 0000000000..823bfa22de
--- /dev/null
+++ b/nptl/tst-compat-forwarder-mod.c
@@ -0,0 +1,28 @@
+/* Copyright (C) 2017 Free Software Foundation, Inc.
+ This file is part of the GNU C Library.
+
+ The GNU C Library is free software; you can redistribute it and/or
+ modify it under the terms of the GNU Lesser General Public
+ License as published by the Free Software Foundation; either
+ version 2.1 of the License, or (at your option) any later version.
+
+ The GNU C Library is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with the GNU C Library; if not, see
+ <http://www.gnu.org/licenses/>. */
+
+/* Call the function system through a statically initialized pointer. */
+
+#include <stdlib.h>
+
+int (*system_function) (const char *) = system;
+
+void
+call_system (void)
+{
+ system_function (NULL);
+}
diff --git a/nptl/tst-compat-forwarder.c b/nptl/tst-compat-forwarder.c
new file mode 100644
index 0000000000..f96806b7fe
--- /dev/null
+++ b/nptl/tst-compat-forwarder.c
@@ -0,0 +1,35 @@
+/* Copyright (C) 2017 Free Software Foundation, Inc.
+ This file is part of the GNU C Library.
+
+ The GNU C Library is free software; you can redistribute it and/or
+ modify it under the terms of the GNU Lesser General Public
+ License as published by the Free Software Foundation; either
+ version 2.1 of the License, or (at your option) any later version.
+
+ The GNU C Library is distributed in the hope that it will be useful,
+ but WITHOUT ANY WARRANTY; without even the implied warranty of
+ MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU
+ Lesser General Public License for more details.
+
+ You should have received a copy of the GNU Lesser General Public
+ License along with the GNU C Library; if not, see
+ <http://www.gnu.org/licenses/>. */
+
+/* Test that the compat forwaders in libpthread work correctly. */
+
+#include <support/test-driver.h>
+
+extern void call_system (void);
+
+int
+do_test (void)
+{
+ /* Calling the system function from a shared library that is not linked
+ against libpthread, when the main program is linked against
+ libpthread, should not crash. */
+ call_system ();
+
+ return 0;
+}
+
+#include <support/test-driver.c>
--
2.14.0
--
Andreas Schwab, SUSE Labs, schwab@suse.de
GPG Key fingerprint = 0196 BAD8 1CE9 1970 F4BE 1748 E4D4 88E3 0EEA B9D7
"And now for something completely different."
^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: [PATCH] Don't use IFUNC resolver for longjmp or system in libpthread (bug 21041)
2017-08-08 18:09 ` Joseph Myers
@ 2017-08-09 8:55 ` Andreas Schwab
0 siblings, 0 replies; 10+ messages in thread
From: Andreas Schwab @ 2017-08-09 8:55 UTC (permalink / raw)
To: Joseph Myers; +Cc: libc-alpha
Sorry, fixed now.
Andreas.
--
Andreas Schwab, SUSE Labs, schwab@suse.de
GPG Key fingerprint = 0196 BAD8 1CE9 1970 F4BE 1748 E4D4 88E3 0EEA B9D7
"And now for something completely different."
^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: [PATCH] Don't use IFUNC resolver for longjmp or system in libpthread (bug 21041)
2017-08-08 14:50 [PATCH] Don't use IFUNC resolver for longjmp or system in libpthread (bug 21041) Andreas Schwab
` (2 preceding siblings ...)
2017-08-08 18:09 ` Joseph Myers
@ 2017-11-01 13:11 ` Romain Naour
2017-11-06 8:48 ` Andreas Schwab
3 siblings, 1 reply; 10+ messages in thread
From: Romain Naour @ 2017-11-01 13:11 UTC (permalink / raw)
To: Andreas Schwab; +Cc: libc-alpha
Hi All,
Le 08/08/2017 à 16:50, Andreas Schwab a écrit :
> Unlike the vfork forwarder and like the fork forwarder as in bug 19861,
> there won't be a problem when the compiler does not turn this into a tail
> call.
>
> Andreas.
>
> * nptl/pt-longjmp.c (longjmp, siglongjmp): Don't use IFUNC resolver.
> * nptl/pt-system.c (system): Likewise.
>
What's the status of this issue ?
This patch has been merged in master (upcoming 2.27) but the ticket is still
"NEW" [1].
The issue has been reported on the Buildroot mailing list by a user using glibc
2.25. I'm able to reproduce it using glibc 2.26-73.
It's safe to backport this patch to glibc 2.25 and 2.26 stable branches ?
Best regards,
Romain
[1] https://sourceware.org/bugzilla/show_bug.cgi?id=21041
[2] http://lists.busybox.net/pipermail/buildroot/2017-October/205719.html
^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: [PATCH] Don't use IFUNC resolver for longjmp or system in libpthread (bug 21041)
2017-11-01 13:11 ` Romain Naour
@ 2017-11-06 8:48 ` Andreas Schwab
2017-11-06 13:30 ` Bartłomiej Piotrowski
0 siblings, 1 reply; 10+ messages in thread
From: Andreas Schwab @ 2017-11-06 8:48 UTC (permalink / raw)
To: Romain Naour; +Cc: libc-alpha
On Nov 01 2017, Romain Naour <romain.naour@gmail.com> wrote:
> Hi All,
>
> Le 08/08/2017 à 16:50, Andreas Schwab a écrit :
>> Unlike the vfork forwarder and like the fork forwarder as in bug 19861,
>> there won't be a problem when the compiler does not turn this into a tail
>> call.
>>
>> Andreas.
>>
>> * nptl/pt-longjmp.c (longjmp, siglongjmp): Don't use IFUNC resolver.
>> * nptl/pt-system.c (system): Likewise.
>>
>
> What's the status of this issue ?
> This patch has been merged in master (upcoming 2.27) but the ticket is still
> "NEW" [1].
The bug is about moving a symbol between libraries, as such still
unresolved.
> It's safe to backport this patch to glibc 2.25 and 2.26 stable branches ?
You can try.
Andreas.
--
Andreas Schwab, SUSE Labs, schwab@suse.de
GPG Key fingerprint = 0196 BAD8 1CE9 1970 F4BE 1748 E4D4 88E3 0EEA B9D7
"And now for something completely different."
^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: [PATCH] Don't use IFUNC resolver for longjmp or system in libpthread (bug 21041)
2017-11-06 8:48 ` Andreas Schwab
@ 2017-11-06 13:30 ` Bartłomiej Piotrowski
2017-11-06 21:13 ` Romain Naour
0 siblings, 1 reply; 10+ messages in thread
From: Bartłomiej Piotrowski @ 2017-11-06 13:30 UTC (permalink / raw)
To: libc-alpha
On 2017-11-06 09:48, Andreas Schwab wrote:
> On Nov 01 2017, Romain Naour <romain.naour@gmail.com> wrote:
>
>> Hi All,
>>
>> Le 08/08/2017 à 16:50, Andreas Schwab a écrit :
>>> Unlike the vfork forwarder and like the fork forwarder as in bug 19861,
>>> there won't be a problem when the compiler does not turn this into a tail
>>> call.
>>>
>>> Andreas.
>>>
>>> * nptl/pt-longjmp.c (longjmp, siglongjmp): Don't use IFUNC resolver.
>>> * nptl/pt-system.c (system): Likewise.
>>>
>>
>> What's the status of this issue ?
>> This patch has been merged in master (upcoming 2.27) but the ticket is still
>> "NEW" [1].
>
> The bug is about moving a symbol between libraries, as such still
> unresolved.
>
>> It's safe to backport this patch to glibc 2.25 and 2.26 stable branches ?
>
> You can try.
>
> Andreas.
>
I backported this to 2.26 in Arch some time ago. Looks fine on my end.
BartÅomiej
^ permalink raw reply [flat|nested] 10+ messages in thread
* Re: [PATCH] Don't use IFUNC resolver for longjmp or system in libpthread (bug 21041)
2017-11-06 13:30 ` Bartłomiej Piotrowski
@ 2017-11-06 21:13 ` Romain Naour
0 siblings, 0 replies; 10+ messages in thread
From: Romain Naour @ 2017-11-06 21:13 UTC (permalink / raw)
To: Bartłomiej Piotrowski, libc-alpha, Andreas Schwab
Hi Andreas, BartÅomiej,
Le 06/11/2017 à 14:30, BartÅomiej Piotrowski a écrit :
> On 2017-11-06 09:48, Andreas Schwab wrote:
>> On Nov 01 2017, Romain Naour <romain.naour@gmail.com> wrote:
>>
>>> Hi All,
>>>
>>> Le 08/08/2017 à 16:50, Andreas Schwab a écrit :
>>>> Unlike the vfork forwarder and like the fork forwarder as in bug 19861,
>>>> there won't be a problem when the compiler does not turn this into a tail
>>>> call.
>>>>
>>>> Andreas.
>>>>
>>>> * nptl/pt-longjmp.c (longjmp, siglongjmp): Don't use IFUNC resolver.
>>>> * nptl/pt-system.c (system): Likewise.
>>>>
>>>
>>> What's the status of this issue ?
>>> This patch has been merged in master (upcoming 2.27) but the ticket is still
>>> "NEW" [1].
>>
>> The bug is about moving a symbol between libraries, as such still
>> unresolved.
>>
>>> It's safe to backport this patch to glibc 2.25 and 2.26 stable branches ?
>>
>> You can try.
>>
>> Andreas.
>>
>
> I backported this to 2.26 in Arch some time ago. Looks fine on my end.
Thank you both for the feedback.
I did a runtime test with ffmpeg [1] and the "Relink" message is gone.
No regression so far.
[1] http://lists.busybox.net/pipermail/buildroot/2017-October/205865.html
Best regards,
Romain
>
> BartÅomiej
>
^ permalink raw reply [flat|nested] 10+ messages in thread
end of thread, other threads:[~2017-11-06 21:13 UTC | newest]
Thread overview: 10+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2017-08-08 14:50 [PATCH] Don't use IFUNC resolver for longjmp or system in libpthread (bug 21041) Andreas Schwab
2017-08-08 14:55 ` Florian Weimer
2017-08-08 14:57 ` H.J. Lu
2017-08-09 8:55 ` Andreas Schwab
2017-08-08 18:09 ` Joseph Myers
2017-08-09 8:55 ` Andreas Schwab
2017-11-01 13:11 ` Romain Naour
2017-11-06 8:48 ` Andreas Schwab
2017-11-06 13:30 ` Bartłomiej Piotrowski
2017-11-06 21:13 ` Romain Naour
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).