[ofa-general] Bogus Receive Completions

[Roman Kononov]

On 2008-01-23 16:12, Eli Cohen wrote:
> if your software creates QPs whose receive queue size is not a power of two
> then you might experience weired problems as the patches I sent have a bug

The RQ size is 64. The SQ size is 64. The CQ size is 128.

> I am sending a patch to be applied on top of the previous libmthca patch so
> you can try it (the same fix goes for the kernel code too). Tomorow I will
> send the fixed patches again. I apologize if the patch is badly formed.

Thanks for the patches.

I have another "simple" program (700 lines) which fails with both tavor 
(4.8.200) and memfree (5.3.000) FW. Unfortunately, it takes an hour or more 
to fail; the failure is not obvious, because when it happens, very little is 
printed; and when I try to modify the code to make it more obvious the 
problem goes away. The failures are related to ordering. Either completions 
or data are out of order. If anybody is interested let me know, I will post it.

Roman