We are investigating crashes which could be depending on bios version. Now it seems we only get the bios manufacturer in the reports? Would it be possible to also report the bios version upon crash? It would give us better information on how to handle these crashes. I'm not sure if breakpad has this feature already. If it hasn't. In bug 1314990 comment 3 I gave the possible ways to get this information on windows and linux like.

The CPU microcode that we've added in bug 1305120 could also be useful, but I often (if not always) see it null for AMD (e.g. on Beta with the AMD related signatures that we've seen in b9 and b11: https://crash-stats.mozilla.com/search/?signature=%3Dmozilla%3A%3ADisplayListClipState%3A%3ACreateInactiveScrollClip&signature=%3Djs%3A%3AAutoSetNewObjectMetadata%3A%3Atrace&signature=%3Djs%3A%3AHeapSlot%3A%3Aset&product=Firefox&version=50.0b&date=%3E%3D2016-10-28T09%3A43%3A00.000Z&date=%3C2016-11-04T09%3A43%3A00.000Z&_sort=-date&_facets=signature&_facets=cpu_info&_facets=cpu_microcode_version&_columns=date&_columns=signature&_columns=product&_columns=version&_columns=build_id&_columns=platform#facet-cpu_info).

(In reply to Marco Castelluccio [:marco] from comment #1) > The CPU microcode that we've added in bug 1305120 could also be useful, but > I often (if not always) see it null for AMD (e.g. on Beta with the AMD > related signatures that we've seen in b9 and b11: > https://crash-stats.mozilla.com/search/ > ?signature=%3Dmozilla%3A%3ADisplayListClipState%3A%3ACreateInactiveScrollClip > &signature=%3Djs%3A%3AAutoSetNewObjectMetadata%3A%3Atrace&signature=%3Djs%3A% > 3AHeapSlot%3A%3Aset&product=Firefox&version=50.0b&date=%3E%3D2016-10- > 28T09%3A43%3A00.000Z&date=%3C2016-11-04T09%3A43%3A00.000Z&_sort=- > date&_facets=signature&_facets=cpu_info&_facets=cpu_microcode_version&_column > s=date&_columns=signature&_columns=product&_columns=version&_columns=build_id > &_columns=platform#facet-cpu_info). That code has already helped us a lot! That way we could find out that a chunk of those crashes are specific to AuthenticAMD family 20 model 0-2.

(In reply to Hannes Verschore [:h4writer] from comment #2) > (In reply to Marco Castelluccio [:marco] from comment #1) > > The CPU microcode that we've added in bug 1305120 could also be useful, but > > I often (if not always) see it null for AMD (e.g. on Beta with the AMD > > related signatures that we've seen in b9 and b11: > > https://crash-stats.mozilla.com/search/ > > ?signature=%3Dmozilla%3A%3ADisplayListClipState%3A%3ACreateInactiveScrollClip > > &signature=%3Djs%3A%3AAutoSetNewObjectMetadata%3A%3Atrace&signature=%3Djs%3A% > > 3AHeapSlot%3A%3Aset&product=Firefox&version=50.0b&date=%3E%3D2016-10- > > 28T09%3A43%3A00.000Z&date=%3C2016-11-04T09%3A43%3A00.000Z&_sort=- > > date&_facets=signature&_facets=cpu_info&_facets=cpu_microcode_version&_column > > s=date&_columns=signature&_columns=product&_columns=version&_columns=build_id > > &_columns=platform#facet-cpu_info). > > That code has already helped us a lot! That way we could find out that a > chunk of those crashes are specific to AuthenticAMD family 20 model 0-2. That's the CPU info bit, but now we've also introduced the CPU microcode, which makes it even more detailed. For example, see bug 1296630 comment 82. The crash was specific to "GenuineIntel family 6 model 61 stepping 4", but only with some CPU microcode versions (0xe, 0x11, 0x12, 0x13, 0x16, 0x18, 0x19). Unfortunately, it looks like the code to detect the CPU microcode is not working with AMD: https://crash-stats.mozilla.com/search/?cpu_info=~AuthenticAMD&product=Firefox&_sort=-date&_facets=signature&_facets=cpu_info&_facets=cpu_microcode_version&_columns=date&_columns=signature&_columns=product&_columns=version&_columns=build_id&_columns=platform#facet-cpu_microcode_version.

I filed bug 1315213 to get the CPU microcode detection work for AMD CPUs.