No traceback on Mac OS X

[chauvjo]

I have been unable to generate a useful traceback despite using the latest compiler release. This has been discussed in other threads but I decided to move to a separate thread. Has anyone else been able to get a traceback using the latest compiler and Mac OS 10.6.2 without all unknown shown? Here is a typical example of what I am seeing using the following command:

ifort -g-save-zero-checkbounds-traceback -o a.out test.f

[cpp]mage              PC                         Routine                 Line        Source
a.out          00000001003D893C     Unknown               Unknown  Unknown
a.out          00000001003D7474     Unknown               Unknown  Unknown
Unknown    000000010101AC77     Unknown               Unknown  Unknown [/cpp]

Here is a simple division by zero case:

[cpp]      PROGRAM Main
      DOUBLE PRECISION a,b,c
      a = 1.0D0
      b = 0.0D0
      c = a/b
      STOP
      END[/cpp]

Intel compiler gives the following:

ifort -O0 -g -fpe0 -traceback -o crash ./crash.f

[cpp]forrtl: error (73): floating divide by zero
Abort trap
[/cpp]

On a SGI UNIX workstation. This is what I am expecting....

[cpp]******* TRAP TRACE FOR PID 61346 DIVZERO   1 *********
dbx version 7.3.5 (92898_Aug07 MR) Aug  7 2003 14:15:31
Process 61346 (crash) stopped at [__waitsys:24 +0x8,0xfa54468]
         Source (of /xlv86/patches/7207/work/irix/lib/libc/libc_n32_M4/proc/waitsys.s) not available for Process 61346
>  0 __waitsys(0x0, 0xef51, 0x7fff0160, 0x3, 0x11139c, 0x0, 0x10004190, 0x1000124c) ["/xlv86/patches/7207/work/irix/lib/libc/libc_n32_M4/proc/waitsys.s":24, 0xfa54468]
   1 _system(0x7fff1a70, 0xef51, 0x7fff0160, 0x3, 0x11139c, 0x0, 0x10004190, 0x1000124c) ["/xlv86/patches/7207/work/irix/lib/libc/libc_n32_M4/stdio/system.c":116, 0xfa60ba8]
   2 __fpe_trace_option(0x3, 0xef51, 0x7fff0160, 0x3, 0x11139c, 0x0, 0x10004190, 0x1000124c) ["/j7/mtibuild/v744/workarea/v7.4.4m/libfpe/fpe_trace.c":159, 0xad1c530]
   3 __catch(0x8, 0x0, 0x7fff2b70, 0x3, 0x11139c, 0x0, 0x10004190, 0x1000124c) ["/j7/mtibuild/v744/workarea/v7.4.4m/libfpe/fpe_handler.c":610, 0xad166b8]
   4 _sigtramp(0x8, 0xef51, 0x7fff0160, 0x3, 0x11139c, 0x0, 0x10004190, 0x1000124c) ["/xlv86/patches/7207/work/irix/lib/libc/libc_n32_M4/signal/sigtramp.s":71, 0xfaf06ac]
   5 Main(0x0, 0x7fff2e00, 0x7fff2e20, 0x0, 0x90, 0x0, 0x68a1, 0x0) ["/tmp_mnt/home/jchauvin/crash/crash.f":8, 0x1000124c]
   6 main(0x0, 0x7fff2e00, 0x7fff2e20, 0x0, 0x90, 0x0, 0x68a1, 0x0) ["/j7/mtibuild/v744/workarea/v7.4.4m/libF77/main.c":101, 0xaed9ef4]
   7 __start() ["/xlv55/kudzu-apr12/work/irix/lib/libc/libc_n32_M4/csu/crt1text.s":177, 0x10000fe8]
[/cpp]

[Ron_Green]

I use Mac OS X 10.6.2 Xcode 3.1.3

I can try to upgrade Xcode, I think you said you have 3.2.x

but for my configuration, it's working:
fordprefect:~ rwgreen$ source /opt/intel/Compiler/11.1/080/bin/ifortvars.sh intel64
fordprefect:~ rwgreen$ ifort -O0 -g -fpe0 -traceback -o crash ./crash.f
fordprefect:~ rwgreen$ ./crash
forrtl: error (73): floating divide by zero
Image PC Routine Line Source
crash 0000000100000A4B _MAIN__ 5 crash.f
crash 00000001000009EC Unknown Unknown Unknown

Stack trace terminated abnormally.
Abort trap
fordprefect:~ rwgreen$ more crash.f
PROGRAM Main
DOUBLE PRECISION a,b,c
a = 1.0D0
b = 0.0D0
c = a/b
STOP
END
fordprefect:~ rwgreen$

[Ron_Green]

Bug report ID: DPD200149111

Bingo - it was the Xcode upgrade to 3.2.1 that broke it:

fordprefect:~ rwgreen$ source /opt/intel/Compiler/11.1/080/bin/ifortvars.sh intel64
fordprefect:~ rwgreen$ ifort -O0 -g -fpe0 -traceback -o crash ./crash.f
fordprefect:~ rwgreen$ ./crash
forrtl: error (73): floating divide by zero
Abort trap
fordprefect:~ rwgreen$

Note the signature for the gcc libs used by the code for the new Xcode 3.2.1:
fordprefect:~ rwgreen$ otool -L crash
crash:
/usr/lib/libgcc_s.1.dylib (compatibility version 1.0.0, current version 246.0.0)
/usr/lib/libmx.A.dylib (compatibility version 1.0.0, current version 315.0.0)
/usr/lib/libSystem.B.dylib (compatibility version 1.0.0, current version 125.0.0)

versus the signature for the gcc libs used by the code for the old Xcode 3.1.2:
fordprefect:~ rwgreen$ otool -L crash
crash:
/usr/lib/libgcc_s.1.dylib (compatibility version 1.0.0, current version 1.0.0)
/usr/lib/libmx.A.dylib (compatibility version 1.0.0, current version 315.0.0)
/usr/lib/libSystem.B.dylib (compatibility version 1.0.0, current version 125.0.0)

I'll get a bug report going on this. I will then re-test your idb issue, this is probably the cause for it also.

It's difficult to keep up with Mac OS X software updates, as it is to keep up with Linux distros/gcc updates.

ron

[chauvjo]

Quoting - Ronald W. Green (Intel)

Bug report ID: DPD200149111

Bingo - it was the Xcode upgrade to 3.2.1 that broke it:

fordprefect:~ rwgreen$ source /opt/intel/Compiler/11.1/080/bin/ifortvars.sh intel64
fordprefect:~ rwgreen$ ifort -O0 -g -fpe0 -traceback -o crash ./crash.f
fordprefect:~ rwgreen$ ./crash
forrtl: error (73): floating divide by zero
Abort trap
fordprefect:~ rwgreen$

Note the signature for the gcc libs used by the code for the new Xcode 3.2.1:
fordprefect:~ rwgreen$ otool -L crash
crash:
/usr/lib/libgcc_s.1.dylib (compatibility version 1.0.0, current version 246.0.0)
/usr/lib/libmx.A.dylib (compatibility version 1.0.0, current version 315.0.0)
/usr/lib/libSystem.B.dylib (compatibility version 1.0.0, current version 125.0.0)

versus the signature for the gcc libs used by the code for the old Xcode 3.1.2:
fordprefect:~ rwgreen$ otool -L crash
crash:
/usr/lib/libgcc_s.1.dylib (compatibility version 1.0.0, current version 1.0.0)
/usr/lib/libmx.A.dylib (compatibility version 1.0.0, current version 315.0.0)
/usr/lib/libSystem.B.dylib (compatibility version 1.0.0, current version 125.0.0)

I'll get a bug report going on this. I will then re-test your idb issue, this is probably the cause for it also.

It's difficult to keep up with Mac OS X software updates, as it is to keep up with Linux distros/gcc updates.

ron

This is great news...saves me from more testing.... On the idb issue, I just tried gdb and it works perfectly on the same source.

Pass this on to your Mac developers...Apple on Wednesday began widespread testing of Mac OS X 10.6.3, the third planned maintenance and security update to its Snow Leopard operating system, early betas of which already includes bug fixes to over seven dozen system components with an emphasis on stabilization.People familiar with the matter say the first external build of the software -- labeled Mac OS X 10.6.3 build 10D522 and weighing in at 665.7MB in barebones delta form -- includes a total of 221 code corrections to 92 distinct system components.

[chauvjo]

Can you tell me what impact this issue has on the usability of Intel Fortran on the Mac? Can I still use the compiler for production work?

[Ron_Green]

Quoting - chauvjo

Can you tell me what impact this issue has on the usability of Intel Fortran on the Mac? Can I still use the compiler for production work?

Mac OS X 10.6.2 with Xcode 3.2.1 is a supported configuration. The ReleaseNotes_F.pdf file in your documentation directory is the definitive guide to supported OS and Xcode versions.

I would only caution that when 10.6.3 does come out, don't jump immediately to that update. It does take us a while to add support to new releases. Wait until an update compiler comes out with that version of the OS listed in the ReleaseNotes.

ron

[Ron_Green]

There is a fix coming in the 11.1 Update 5 compiler. This will be version 11.1.084.

I expect this compiler to be posted to https://registrationcenter.intel.com in the next couple of days, somewhere between 2/19 and 2/23/2010. Look for an email announcement when this compiler update is posted to RegistrationCenter.

ron