----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: http://reviews.gem5.org/r/3767/#review9241 -----------------------------------------------------------
Thanks for this patch Eric. Can you elaborate a bit more on the case that fails? I was under the impression that the current conditions that trigger this panic cannot be satisfied by cacheable requests. The panic here https://www.mail-archive.com/[email protected]/msg13105.html seems to be due to a WriteReq which in the form of a snoop can only be seen as an uncacheable request if I am not mistaken. - Nikos Nikoleris On Dec. 19, 2016, 7:39 p.m., Eric Clark wrote: > > ----------------------------------------------------------- > This is an automatically generated e-mail. To reply, visit: > http://reviews.gem5.org/r/3767/ > ----------------------------------------------------------- > > (Updated Dec. 19, 2016, 7:39 p.m.) > > > Review request for Default. > > > Repository: gem5 > > > Description > ------- > > Changeset 11766:3a8eb6336012 > --------------------------- > mem: Allow non-invalidating uncacheable snoops > > This is a follow up to changeset 11285. Uncacheable snoops do > not invalidate so this panics if we are responding to an uncacheable > write snoop. > > This panic happens under ARM Linux 4.3. It affects myself and at > least one other user: > > https://www.mail-archive.com/[email protected]/msg13105.html > > > Diffs > ----- > > src/mem/cache/cache.cc 29d401db3746 > > Diff: http://reviews.gem5.org/r/3767/diff/ > > > Testing > ------- > > Ran {ARM,X86} debug/quick/se regressions but I don't have the spec binaries. > Used with ARM full system simulation to complete my course project. > > > Thanks, > > Eric Clark > > _______________________________________________ gem5-dev mailing list [email protected] http://m5sim.org/mailman/listinfo/gem5-dev
