[Cbe-oss-dev] Potential issue with spu_stop in loop

[Goffredo Marocchi]

So basically the reason CellServ and the other poster did not experience the
crash was that they were not pushing for a high enough iteration number
(CellServ went for 9000+ iterations, but it still did not trigger an error).

Thank you all for the attention given to this isse, I gather, from the
comments Malc made in the dw thread, that there is a workaround to this
issue though or is an appropriate patch what we should look forward to ?

On Tue, Jun 3, 2008 at 11:06 AM, Kazunori Asayama <asayama at sm.sony.co.jp>
wrote:

> malc wrote:
>
>> I'd submit a patch - but I still can't see what's *consuming* the
>> stopinfo at the read-side of the pipe - breakpoint @
>> _event_spe_stop_info_read() doesn't get hit... and also I can't
>> explain why this isn't hit on the QS21 blades (as reported in the
>> developerworks thread)
>>
>
> The length of the pipe buffer depends on the page size. PS3's default
> page size is 4KB, while the Cell blade's one is 64KB. So if you enlarge
> the # of iterations of the test, the same problem will arise on the Cell
> blades.
>
> --
> (ASAYAMA Kazunori
>  (asayama at sm.sony.co.jp))
> t
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.ozlabs.org/pipermail/cbe-oss-dev/attachments/20080603/ecd84836/attachment.htm>