Thread: [B2-devel] sigsegv.cpp `have skip check` fix for gcc 4.4.3

SourceForge Headquarters 1320 Columbia Street Suite 310 San Diego, CA 92101 +1 (858) 422-6466

I'm half sure of this fix, because check works with previous versions of
gcc (<4.4.3), so it looks like regresion on gcc side.

What exactly is this fixing?

2010/3/7 Amadeusz Sławiński <sa...@gm...>:
> I'm half sure of this fix, because check works with previous versions of
> gcc (<4.4.3), so it looks like regresion on gcc side.
>
> ------------------------------------------------------------------------------
> Download Intel&#174; Parallel Studio Eval
> Try the new software tools for yourself. Speed compiling, find bugs
> proactively, and fine-tune applications for parallel performance.
> See why Intel Parallel Studio got high marks during beta.
> http://p.sf.net/sfu/intel-sw-dev
> _______________________________________________
> basilisk-devel mailing list
> bas...@li...
> https://lists.sourceforge.net/lists/listinfo/basilisk-devel
>
>

Output of ./configure
...
checking whether we can skip instruction in SIGSEGV handler... no
...

from config.log
...
configure:9849: checking whether we can skip instruction in SIGSEGV
handler
configure:9875: g++ -o conftest -g -O2    conftest.cpp -lrt -lpthread
-lm   -lSM -lICE  -lX11 -lXext  -lXxf86vm >&5
configure:9875: $? = 0
configure:9875: ./conftest
./configure: line 2130:  4065 Segmentation fault
./conftest$ac_exeext
configure:9875: $? = 139
configure: program exited with status 139
configure: failed program was:
| /* confdefs.h */
| #define PACKAGE_NAME "SheepShaver"
| #define PACKAGE_TARNAME "SheepShaver"
| #define PACKAGE_VERSION "2.3"
| #define PACKAGE_STRING "SheepShaver 2.3"
| #define PACKAGE_BUGREPORT "Chr...@un..."
| #define PACKAGE_URL ""
| #define EMULATED_PPC 1
| #define HAVE_LIBM 1
| #define HAVE_LIBPTHREAD 1
| #define HAVE_PTHREAD_CANCEL 1
| #define HAVE_PTHREAD_COND_INIT 1
| #define HAVE_PTHREAD_TESTCANCEL 1
| #define HAVE_PTHREAD_MUTEXATTR_SETPROTOCOL 1
| #define HAVE_PTHREAD_MUTEXATTR_SETTYPE 1
| #define HAVE_PTHREAD_MUTEXATTR_SETPSHARED 1
| #define HAVE_SEM_INIT 1
| #define HAVE_PTHREADS 1
| #define ENABLE_FBDEV_DGA 1
| #define ENABLE_XF86_VIDMODE 1
| #define ENABLE_GTK 1
| #define STDC_HEADERS 1
| #define HAVE_SYS_WAIT_H 1
| #define HAVE_MALLOC_H 1
| #define HAVE_STDINT_H 1
| #define HAVE_SYS_MMAN_H 1
| #define HAVE_UNISTD_H 1
| #define HAVE_FCNTL_H 1
| #define HAVE_BYTESWAP_H 1
| #define HAVE_DIRENT_H 1
| #define HAVE_SYS_SOCKET_H 1
| #define HAVE_SYS_IOCTL_H 1
| #define HAVE_SYS_BITYPES_H 1
| #define HAVE_SYS_WAIT_H 1
| #define HAVE_SYS_TIME_H 1
| #define HAVE_SYS_POLL_H 1
| #define HAVE_SYS_SELECT_H 1
| #define HAVE_ARPA_INET_H 1
| #define HAVE_LINUX_IF_H 1
| #define HAVE_LINUX_IF_TUN_H 1
| #define HAVE_NET_IF_H 1
| #define HAVE_FENV_H 1
| #define SIZEOF_SHORT 2
| #define SIZEOF_INT 4
| #define SIZEOF_LONG 8
| #define SIZEOF_LONG_LONG 8
| #define SIZEOF_FLOAT 4
| #define SIZEOF_DOUBLE 8
| #define SIZEOF_VOID_P 8
| #define RETSIGTYPE void
| #define TIME_WITH_SYS_TIME 1
| #define HAVE_SIGNAL_SA_RESTORER 1
| #define HAVE_STRDUP 1
| #define HAVE_STRERROR 1
| #define HAVE_CFMAKERAW 1
| #define HAVE_NANOSLEEP 1
| #define HAVE_SIGACTION 1
| #define HAVE_SIGNAL 1
| #define HAVE_MMAP 1
| #define HAVE_MPROTECT 1
| #define HAVE_MUNMAP 1
| #define HAVE_EXP2F 1
| #define HAVE_LOG2F 1
| #define HAVE_EXP2 1
| #define HAVE_LOG2 1
| #define HAVE_FLOORF 1
| #define HAVE_ROUNDF 1
| #define HAVE_CEILF 1
| #define HAVE_TRUNCF 1
| #define HAVE_FLOOR 1
| #define HAVE_ROUND 1
| #define HAVE_CEIL 1
| #define HAVE_TRUNC 1
| #define HAVE_POLL 1
| #define HAVE_INET_ATON 1
| #define HAVE_CLOCK_GETTIME 1
| #define HAVE_CLOCK_NANOSLEEP 1
| #define HAVE_STRINGS_H 1
| #define HAVE_SYS_STAT_H 1
| #define HAVE_PTY_H 1
| #define HAVE_VHANGUP 1
| #define HAVE_SLIRP 1
| #define ENABLE_TUNTAP 1
| #define HAVE_MMAP_VM 1
| #define HAVE_MMAP_ANON 1
| #define HAVE_MMAP_ANONYMOUS 1
| #define HAVE_MMAP_VM 1
| #define HAVE_SIGINFO_T 1
| /* end confdefs.h.  */
| 
|     #define HAVE_SIGSEGV_SKIP_INSTRUCTION 1
|     #define CONFIGURE_TEST_SIGSEGV_RECOVERY
|     #include "vm_alloc.cpp"
|     #include "sigsegv.cpp"
| 
configure:9893: result: no

As I mentioned it works on previous versions of gcc(<4.4.3), so it is probably
compiler bug.
Also works if you do something like
volatile int label_hack = 3;                                       
switch (label_hack) {
case 3:
...
case 2:

On Mon, Mar 08, 2010 at 08:37:25PM -0500, Alexei Svitkine wrote:
> What exactly is this fixing?
> 
> 2010/3/7 Amadeusz Sławiński <sa...@gm...>:
> > I'm half sure of this fix, because check works with previous versions of
> > gcc (<4.4.3), so it looks like regresion on gcc side.

> Also works if you do something like
> volatile int label_hack = 3;
> switch (label_hack) {
> case 3:
> ...
> case 2:

I'd prefer that approach then, since it will likely also work with the
older compilers that do require the volatile.

-Alexei

On Tue, Mar 09, 2010 at 09:33:38AM -0500, Alexei Svitkine wrote:
> > Also works if you do something like
> > volatile int label_hack = 3;
> > switch (label_hack) {
> > case 3:
> > ...
> > case 2:
> 
> I'd prefer that approach then, since it will likely also work with the
> older compilers that do require the volatile.
> 
> -Alexei
> 

Should have thought that way, patch included.

Committed. Thanks.

2010/3/9 Amadeusz Sławiński <sa...@gm...>:
> On Tue, Mar 09, 2010 at 09:33:38AM -0500, Alexei Svitkine wrote:
>> > Also works if you do something like
>> > volatile int label_hack = 3;
>> > switch (label_hack) {
>> > case 3:
>> > ...
>> > case 2:
>>
>> I'd prefer that approach then, since it will likely also work with the
>> older compilers that do require the volatile.
>>
>> -Alexei
>>
>
> Should have thought that way, patch included.
>
> ------------------------------------------------------------------------------
> Download Intel&#174; Parallel Studio Eval
> Try the new software tools for yourself. Speed compiling, find bugs
> proactively, and fine-tune applications for parallel performance.
> See why Intel Parallel Studio got high marks during beta.
> http://p.sf.net/sfu/intel-sw-dev
> _______________________________________________
> basilisk-devel mailing list
> bas...@li...
> https://lists.sourceforge.net/lists/listinfo/basilisk-devel
>
>