[Gems-users] crashing with OpenMP program

[Marcelo Cintra <mc@xxxxxxxxxxxx>]

Hello,

I'm running an OpenMP program on ruby+opal and it crashes with the 
following messages:

1) On the simics console:

simics> opal0.sim-step 10000000000
patch  NPC: 0xffffffffff3c2a5c 0xff3c2a5c
patch  NPC: 0xffffffffff3c2a5c 0xff3c2a5c
patch  NPC: 0xffffffffff3c2a5c 0xff3c2a5c
patch  NPC: 0xffffffffff3c2a5c 0xff3c2a5c
patch  NPC: 0xffffffffff3c2a5c 0xff3c2a5c
patch  NPC: 0xffffffffff3c2a5c 0xff3c2a5c
patch  NPC: 0xffffffffff3c2a5c 0xff3c2a5c
patch  PC: 0x1000d00 0xf000d294
patch  NPC: 0x1000d04 0xf000d298
control-C: ending simulation...
Exception error message: control C
pstate_t::simcontinue: caught exception (#16).
patch  PC: 0x100c90c 0x100c908
patch  NPC: 0x100c908 0x100c90c
simulate: completed 16144286 instructions, cycle: 15609294
simics>

2) On the target console:

# time ./tree
send mondo timeout [18454 NACK 0 BUSY]
IDSR 0x8  aids: 2
panic: failed to stop cpu2

panic[cpu1]/thread=2a100097d40: send_mondo_set: timeout

000002a100096f40 SUNW,UltraSPARC-III+:send_mondo_set+208 (3, 8, 0, 
89dfb965, 2a, 89dfb38b)
  %l0-3: 0000000089dfb965 0000000000000000 000002a1000970f0 
0000000000000001
  %l4-7: 0000000000000008 0000000000000020 0000000000000003 
0000000000000000
000002a100097040 unix:xt_some+1b8 (0, 2, 2a100097140, 0, 1467400, 1467400)


The program runs perfectly on simics without ruby+opal.

Any ideas, anyone? Thanks a lot,

		Marcelo