On Thu, 2005-06-23 at 02:42, Grant Grundler wrote: > On Wed, Jun 22, 2005 at 12:00:10PM -0700, Roland Dreier wrote: > > Roland> I just checked in a change that converts from using > > Roland> pci_alloc_consistent(), which is always GFP_ATOMIC, to > > Roland> dma_alloc_coherent(..., GFP_KERNEL). > > This arch specific. > > On ia64 I expect both functions to land in the same bit of code. > For ZX1 platforms, that is sba_alloc_coherent(). > See arch/ia64/hp/common/sba_iommu.c since the behavior is > different depending on CONFIG_NUMA setting. > > Sorry, I missed/forgot what the original problem was.
The original problem (for me) was that there was a page allocation failure with kdapl/mthca on a create QP (and the driver changed but kdapl didn't in terms of this). I'm not sure which change introduced the problem but it was fixed with r2666. -- Hal _______________________________________________ openib-general mailing list [email protected] http://openib.org/mailman/listinfo/openib-general To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
