[openib-general] Re: spinlock wrong CPU on CPU#1, ib_addr

[Or Gerlitz]

Sean Hefty wrote:

 > Can you describe more what iSER is doing in the callback or post the
 > code?

As of Monday, the code is posted under ulp/iser at the trunk.

When iser_cma_handler gets ADDR_RESOLVED event, it calls
iser_adaptor_find_by_device which if needed, allocates all the required 
IB resources associated with it (PD, CQ, DMA MR, FMR pool) and then 
calls to cma_resolve_route.

 > I haven't seen this bug before, so I'm not sure what to make of it.
 > The ib_addr code doesn't even acquire spinlocks.

I see, good chances the issue is within iser and the kernel debug print 
is somehow missleading, any idea what "spinlock wrong CPU" means?

Or.