[ewg] Re: [ofa-general] questions about OFED 1.2 IPoIB bonding

[Roland Dreier]

 > > HCA catastrophic errors are either a hardware problem (either a
 > > transient condition like overheating, or a busted HCA), or a firmware
 > > bug.
 > 
 > Not really, since most kernel code uses the DMA MR,
 > they can easily be triggered by e.g. incorrect DMA API usage.
 > I've just seen this with the recent PPC bug.

Out of curiousity, why does this cause a catastrophic error?  I would
have thought a work request with a bogus bus address would generate an
affiliated error, since you know exactly resource what caused the bad
transaction.

 - R.