Re: [PATCH v2 63/93] tcg/tci: Use ffi for calls

[Richard Henderson]

On 2/8/21 11:04 AM, Stefan Weil wrote:
> 
> Am 08.02.21 um 18:39 schrieb Richard Henderson:
>> On 2/8/21 5:07 AM, Stefan Weil wrote:
>>> Richard, this commit is also the one which breaks qemu-system-i386 on 
>>> sparc64
>>> for me:
>> You'll have to give me more details than that, because qemu-system-i386 works
>> for me on a niagara5 w/ debian sid.
> 
> 
> I am testing on a similar Debian system (debian-ports unstable), but with a
> Niagara3 cpu:
> 
> Linux gcc102.fsffrance.org 5.10.0-3-sparc64-smp #1 SMP Debian 5.10.12-1
> (2021-01-30) sparc64 GNU/Linux
> 
> gcc (Debian 10.2.1-6) 10.2.1 20210110
> 
> $ cat /proc/cpuinfo
> cpu        : UltraSparc T3 (Niagara3)
> fpu        : UltraSparc T3 integrated FPU
> pmu        : niagara3
> prom        : OBP 4.34.6.c 2017/03/22 13:55
> type        : sun4v
> ncpus probed    : 256
> ncpus active    : 256
> D$ parity tl1    : 0
> I$ parity tl1    : 0
> cpucaps        :
> flush,stbar,swap,muldiv,v9,blkinit,n2,mul32,div32,v8plus,popc,vis,vis2,ASIBlkInit,fmaf,vis3,hpc
> 

Ok, I've reproduced something on a T3 (gcc102.fsffrance.org).
Running the same code side-by-side vs the T5, I get different results.

I'll see if I can track down the difference, since they're both running the
same base os.


r~