[rds-devel] Re: trying to reproduce the crash

[Olaf Kirch]

I noticed in the FW release note that there's a known issues with MTTs -
WRITE_MTT may fail of first MTT_INDEX is not even.

One thing I changed recently was to up the limit of pages per FMR from
256 to 257 (in order to allow unaligned RDMAs of 1Meg). I've backed out
that change and am currently testing. It's been running for ~20 minutes
now without crashing - I'll keep my fingers crossed for a little longer

I *think* the driver should print error messages to syslog when WRITE_MTT
fails, but maybe there's some other side effect.

I'll try to get the firmware upgraded ASAP - I can't do that myself, but
I'll ask the guy who reigns over these boxes whether he has time to do it
today.

Olaf
-- 
Olaf Kirch  |  --- o --- Nous sommes du soleil we love when we play
okir at lst.de |    / | \   sol.dhoop.naytheet.ah kin.ir.samse.qurax