On 7/24/18 4:57 AM, Marc Glisse wrote:
On Tue, 24 Jul 2018, Andrew Haley wrote:
On 07/24/2018 09:44 AM, Andrew Haley wrote:
On 07/24/2018 09:43 AM, Andrew Janke wrote:
warning: unhandled dyld version (15)
Oops.
Your next trick:
Run "gcc -v foo.c" to see what command is used to launch cc1
Then run cc1 on its own in gdb.
There is the -wrapper option for that:
gcc ... -wrapper gdb,--args
Great; that got me going. Doing this stopped the program when SIGILL was
raised.
Am I reading this right that a MULX was the offending instruction? That
sounds right, since it was introduced in Haswell, and this is an older
Ivy Bridge machine.
Unfortunately, the backtrace isn't giving me file names. I guess that's
because these files weren't compiled with debug symbol info (hence "no
debugging symbols found")?
$ sudo
/Applications/Octave-4.4.0.app/Contents/Resources/usr/bin/gfortran
hello_world.f90 -wrapper gdb,-q,--args
Reading symbols from
/Applications/Octave-4.4.0.app/Contents/Resources/usr/Cellar/gcc_8.1.0/8.1.0/libexec/gcc/x86_64-apple-darwin17.6.0/8.1.0/f951...(no
debugging symbols found)...done.
(gdb) set disassemble-next-line on
(gdb) r
Starting program:
/Applications/Octave-4.4.0.app/Contents/Resources/usr/Cellar/gcc_8.1.0/8.1.0/libexec/gcc/x86_64-apple-darwin17.6.0/8.1.0/f951
hello_world.f90 -fPIC -quiet -dumpbase hello_world.f90
-mmacosx-version-min=10.13.0 -mtune=core2 -auxbase hello_world
-fintrinsic-modules-path
/Applications/Octave-4.4.0.app/Contents/Resources/usr/Cellar/gcc_8.1.0/8.1.0/lib/gcc/8/gcc/x86_64-apple-darwin17.6.0/8.1.0/finclude
-o /var/tmp//ccTDUpMv.s
[New Thread 0xf03 of process 38443]
warning: unhandled dyld version (15)
Thread 2 received signal SIGILL, Illegal instruction.
0x000000014187b4b1 in ?? ()
=> 0x000000014187b4b1: c4 62 a3 f6 16 mulx (%rsi),%r11,%r10
(gdb) bt
#0 0x000000014187b4b1 in ?? ()
#1 0xcccccccccccccccc in ?? ()
#2 0x0000000000000004 in ?? ()
#3 0x00007ffeefbfee70 in ?? ()
#4 0x0000000000000000 in ?? ()
(gdb)
