Roland Dreier wrote:
No, I don't think we need to worry about using up vmalloc() space really. I don't think anyone is going to consume more than a few MB of ring buffers with IPoIB, and if they do then it's almost certainly going to be on a 64-bit system.
So the patch you want to see is to make ipoib use vmalloc in those 1-3 places which are subject to kmalloc failure?
Or _______________________________________________ general mailing list [email protected] http://lists.openfabrics.org/cgi-bin/mailman/listinfo/general To unsubscribe, please visit http://openib.org/mailman/listinfo/openib-general
