[Pkg-octave-devel] Bug#699300: octave matrix multiply grief

[Rafael Laboissiere]

Bill,

As Jordi pointed out in another message, going down the strace road is 
not going to give us enough information.  Please follow the given 
advice, in particular that one from Sebastien regarding libatlas3gf-base.

Rafael

* William Ludescher <whludescher at verizon.net> [2013-02-01 08:30]:

> Rafael, 
> 	Apologies for my ham-handed carelessness both on my test and email cc. 
> I will attempt to reform. Well I retyped your test and here are the 
> results:
>
> What I typed: echo 'a=[1 2; 3 4]; b=[5 6;7 8]; a*b' | strace octave
>
> Relevant strace output:
>
> mmap2(NULL, 4096, PROT_READ|PROT_WRITE, MAP_PRIVATE|MAP_ANONYMOUS, -1, 
> 0) = 0xb77a5000 
> read(0, "a=[1 2; 3 4]; b=[5 6;7 8]; a*b\n", 4096) = 31 
> time(NULL)                              = 1359724093 
> --- SIGILL (Illegal instruction) @ 0 (0) --- 
> futex(0xb5d1e8c0, FUTEX_WAKE_PRIVATE, 2147483647) = 0 
> write(2, "panic: ", 7panic: )                  = 7 
> write(2, "Illegal instruction", 19Illegal instruction)     = 19 
> write(2, " -- stopping myself...\n", 23 -- stopping myself... 
> ) = 23 
> write(2, "attempting to save variables to "..., 49attempting to save 
> variables to `octave-core'... 
> ) = 49 
> open("octave-core", O_WRONLY|O_CREAT|O_TRUNC|O_LARGEFILE, 0666) = 3 
> write(2, "save to `octave-core' complete\n", 31save to `octave-core' 
> complete 
> ) = 31 
> write(3, "Octave-1-L\0\10\0\0\0.nargin.\0\0\0\0\0\377\6\0\0"..., 244) = 
> 244 
> close(3)                                = 0 
> rt_sigaction(SIGILL, {SIG_DFL, [], SA_RESTART}, {0xb6ff61e0, [], 
> SA_RESTART}, 8) = 0 
> tgkill(2678, 2678, SIGILL)              = 0 
> sigreturn()                             = ? (mask now []) 
> --- SIGILL (Illegal instruction) @ 0 (0) --- 
> +++ killed by SIGILL +++ 
> Illegal instruction
>
>
> I presume that the \n came from the echo command. However, why does 
> octave think that this is an illegal instruction?
>
> Bill
>
>