CrashKill/2013-11-18

General

• [lsblakk] confirm that bug 932100 will not be uplifted to Beta, that we are OK with the bitguard crash levels for 26.0 -- discuss whether this is safe enough for Aurora.

Socorro

• A new processor has been tested on parts of our incoming crashes over the weekend. Note that crashes with empty minidumps get an "EMPTY: no crashing thread identified; ERROR_NO_MINIDUMP_HEADER" signature with that instead of the previous "EMPTY: no crashing thread identified; corrupt dump" - this will enable us to tell apart empty dumps from other cases that makes us unable to find a proper signature.
• Tested JSON MDSW over the weekend. It was successful, and we're rolling the changes out to all processors later today. This will give us all sorts of freedom and power to try new things without continually having to make heavyweight schema changes.
• Shipped Release 65
  • Can now limit a search to just crashes with comments.
  • Explosive report re-enabled in web UI. Explosion detector runs daily at 1000 Pacific. It sounds the alarm only if it's 99.99% confident.
  • Infra improvements to crontabber that will eventually let us run multiple jobs in parallel
  • Lots of other stuff

• bsmedberg landed dump.largest_free_vm_block for JSON-MDSW bug 939141, filed bug 939871 to get this exposed in supersearch (hrm and maybe postgres)
• bsmedberg question about the status of bug 931907
• bsmedberg question about the status of raw crash fields in supersearch

Flash

• Flash beta is now 12.0.0.3!

Desktop

Beta

• Crashes per 100 ADI has nearly doubled in the past week: https://crash-stats.mozilla.com/daily?form_selection=by_version&p=Firefox&v=&v=26.0b&v=28.0a1&v=27.0a2&hang_type=any&os=Windows&os=Mac+OS+X&os=Linux&date_range_type=build&date_start=2013-11-10&date_end=2013-11-18&submit=Generate
  • Explosiveness report doesn't show obvious signatures that would account for the change in volume.

Aurora

• @0x0 | GetPropertyOperation made its way into top 10, but there are no reports on builds after 20131114. So no bug at this time.

Trunk

• bug 939746 - crash in mozalloc_abort(char const* const) | NS_DebugBreak | js::AtomizeChars(js::ExclusiveContext*, wchar_t const*, unsigned __int64, js::InternBehavior)

Mobile

• bug 925608 - Figure and fix how EGLSurface renewal works on Android - uplift?
• bug 931251 - crash in SaveSharedScriptData - Volume on 26 betas is half 25 betas

B2G

General

• bug 939865 - Associated crash reports are not showing in the Crash Reports tab and Signature Summary tab
• bug 939258 - When creating a bug for the B2G product it can't find the product in bugzilla
  • looks like rhelmer is on it?
• just have to get people on more recent builds for most part, don't have the symbols for others, or not enough data.
• need to look into why we aren't getting crash data from released 1.1

1.3

• bug 932046 - crash in mozilla::net::HttpChannelChild::OnRedirectVerifyCallback(unsigned int)
• bug 939259 - crash in js::Invoke
  • need STR

1.2/1.1

• people need to update
• need symbols

1.1

• need to take some time to investigate why we aren't getting 1.1 data

1.0.1

• same for the most part as before.

Notes

Upcoming PTO

lizzard - out Wed. morning Nov. 20, PST