Mozilla Status Board

mh+mozilla@glandium.org

Current Status - 2016-08-23

Done:

Reviewed: - bug 1291356, bug 1292653, bug 1293868, bug 1291424, bug 1267270, bug 1254779, bug 1294510, bug 1289194, bug 1288845, bug 1294331, bug 1240134, bug 1294731, bug 972945, bug 1294864, bug 1294803, bug 1291944, bug 1293448, bug 1295439, bug 1295516, bug 1294475, bug 1295760, bug 1295780, bug 1295784, bug 1295811, bug 1295853, bug 1287326, bug 1143421, bug 1296449, bug 1294660, bug 1295688, bug 1295695, bug 1296137

Landed bugs:

  • bug 1292463: Move MOZ_C{,XX}_SUPPORTS_WARNING to python configure
  • bug 1293579: Add a dependable function/decorator to avoid some of the depends('--help') weirdness in python configure
  • bug 1293837: Move HAVE_64_BIT_BUILD test to python configure
  • bug 1293840: Determine HAVE_64BIT_BUILD from --target
  • bug 1294585: Move --with-*-keyfile options to python configure.
  • bug 1294632: Remove unused old_configure_options
  • bug 1256004: Frequent TaskCluster build gmake[5]: * [/home/worker/workspace/build/src/obj-firefox/addon-sdk/source/test/addons/.mkdir.done] Error 1
  • bug 1294630: Move --enable-ipc-fuzzer to python configure
  • bug 1295480: Move various options to python configure
  • bug 1295885: Remove netwerk/protocol/rtsp
  • bug 1295751: Make it possible to copy configure.in to configure
  • bug 1291768: Firefox for Android Crashes when performance recording is started
  • bug 1257049: Stop spawning a separate process for config.status from configure.py
  • bug 1296502: js-config contains the python representation of a list for NSPR_CFLAGS
  • bug 1276003: Move --enable-necko-protocols to python configure
  • bug 1296503: Dump config.status with unicode or byte strings, matching what is in the config we get out of python configure

Next:

  • More python configure infrastructure (for introspection/lint)

Coordination:

Waiting for review: - bug 1287671: Simplify jscpucfg.h

Waiting for other bug(s) to be fixed: - bug 1256730: Fail configure when api keyfiles don't exist

Older Posts

Member of Projects

Team Posts

mh+mozilla@glandium.org 5 days ago

Done:

Reviewed: - bug 1291356, bug 1292653, bug 1293868, bug 1291424, bug 1267270, bug 1254779, bug 1294510, bug 1289194, bug 1288845, bug 1294331, bug 1240134, bug 1294731, bug 972945, bug 1294864, bug 1294803, bug 1291944, bug 1293448, bug 1295439, bug 1295516, bug 1294475, bug 1295760, bug 1295780, bug 1295784, bug 1295811, bug 1295853, bug 1287326, bug 1143421, bug 1296449, bug 1294660, bug 1295688, bug 1295695, bug 1296137

Landed bugs:

  • bug 1292463: Move MOZ_C{,XX}_SUPPORTS_WARNING to python configure
  • bug 1293579: Add a dependable function/decorator to avoid some of the depends('--help') weirdness in python configure
  • bug 1293837: Move HAVE_64_BIT_BUILD test to python configure
  • bug 1293840: Determine HAVE_64BIT_BUILD from --target
  • bug 1294585: Move --with-*-keyfile options to python configure.
  • bug 1294632: Remove unused old_configure_options
  • bug 1256004: Frequent TaskCluster build gmake[5]: * [/home/worker/workspace/build/src/obj-firefox/addon-sdk/source/test/addons/.mkdir.done] Error 1
  • bug 1294630: Move --enable-ipc-fuzzer to python configure
  • bug 1295480: Move various options to python configure
  • bug 1295885: Remove netwerk/protocol/rtsp
  • bug 1295751: Make it possible to copy configure.in to configure
  • bug 1291768: Firefox for Android Crashes when performance recording is started
  • bug 1257049: Stop spawning a separate process for config.status from configure.py
  • bug 1296502: js-config contains the python representation of a list for NSPR_CFLAGS
  • bug 1276003: Move --enable-necko-protocols to python configure
  • bug 1296503: Dump config.status with unicode or byte strings, matching what is in the config we get out of python configure

Next:

  • More python configure infrastructure (for introspection/lint)

Coordination:

Waiting for review: - bug 1287671: Simplify jscpucfg.h

Waiting for other bug(s) to be fixed: - bug 1256730: Fail configure when api keyfiles don't exist

nfroyd@mozilla.com 6 days ago

Done:

  • bug 1294809: remote TestPipes unit test in favor of more complete gtests.
  • bug 1295192: non-null-checked operator new usage in layout/.
  • bug 1296078: avoid format string mismatches.
  • bug 1296642: avoid debug checking function calls in release builds (size win!).

In-progress:

  • minor IPDL size reductions.

One-offs:

  • Reviews. Trying to think about how timers should work and how the GPU process should interact with xpcom startup.
  • 1:1 with Nick.
  • Monthly company meeting.
  • Investigated AWS for Windows work, didn't seem hugely better than the laptop.
  • clang-cl work, which currently involves fighting through compile errors/warnings.

Next:

  • More of the same.

nnethercote@mozilla.com 6 days ago

Done:

Landed:

  • Bug 1294645 - Don't use NS_CALLBACK for callbacks in nsI{Input,Output,UnicharInput}Stream.idl.
  • Bug 1294620 - Use infallible XPIDL attribute getters more.
  • Bug 1295053 - Remove most uses of NS_METHOD and NS_CALLBACK.
  • Bug 1293596 - DeCOMtaminate nsIWidget (partial).

jseward@acm.org 3 weeks ago

Done:

Races in Gecko:

  • Started writing a parser for TSan error messages.
  • 1274256: ffmpeg races: Measured perf loss of my proposed fixes as directed by an FFmpeg developer. Looks like they will fix this upstream at some point.

Undef value errors in Gecko:

  • 1288726: newly enabled Seccomp sandbox nukes Valgrind: landed
  • 1289098: avoid V FPs with gcc-5.4 -O2 builds: made patch

Valgrind:

  • Hacked up some simple RDTSC based code to measure JIT-vs-run costs in V.
  • Considered in some detail how to do instruction selection, register allocation and assembly in a single integrated pass in the JIT. This would probably cut 35%-40% off the JIT cost. It would be quite a big change but could be done incrementally, on a per-arch basis.
  • In the interim, implemented a small package of changes that reduces JIT cost by around 14%. Experienced some weirdness with branch prediction on Haswell that I'd never have been able to track down without Linux's "perf" package.
  • Fixed a bunch of ubsan-reported errors in Valgrind, all to do with signed left shifts.

Other:

  • 1289770: excessive CPU usage for simple web pages: investigated, but gave up.
  • Learnt more about Rust. Yay!

Next:

Continue with all the above, but with emphasis on the orange-race correlation stuff.

erahm@mozilla.com 3 months ago

Done:

Fixed:

  • Bug 1173017 - Building deadlock detector with stacks enabled is broken
  • Bug 1181142 - crash in strncpy | js::DecompileValueGenerator(JSContext*, int, JS::Handle
  • Bug 1219992 - 1,500 instances of "RasterImage::Init failed" emitted from image/ImageFactory.cpp during linux64 debug testing
  • Bug 1236923 - libexpat
  • Bug 1261304 - crash in OOM | large | mozalloc_abort | mozalloc_handle_oom | moz_xrealloc | je_free | nsTArray_Impl
  • Bug 1266256 - Don't try to create null URLs in _isMinifiedURL
  • Bug 1268215 - Make CCGraphBuilder::AddNode fallible
  • Bug 1268616 - Very large allocations in call to input_overflow_buf_.reserve()
  • Bug 1269047 - 3,800 instances of NS_ENSURE_SUCCESS(res, res), NS_ENSURE_SUCCESS(rv, rv), NS_ENSURE_TRUE(aSelection->RangeCount()) emitted during linux64 debug testing
  • Bug 1270126 - Crash in OOM | large | mozalloc_abort | mozalloc_handle_oom | moz_xmalloc | nsTArray_base
  • Bug 1271456 - [skiplist] Add huge_palloc to prefixSignatureRegEx
  • Bug 1272163 - [skiplist] Add nsAString_internal::BeginWriting to prefixSignatureRegEx
  • Bug 1273324 - Permafailing Win64 xpcshell test_dmd.js | test - [test : 113] full-sampled-live - false == true
  • Bug 1274404 - Annotate crashes for large messages in ProcessLink::SendMessage()
  • Bug 1270938 - Hang while running APZCBasicTester.OverScroll_Bug1152051a (infinite loop in BlockingResourceBase::Release())

Pending review:

  • Bug 1270310 - Crash in OOM | large | NS_ABORT_OOM | nsAString_internal::BeginWriting

Logspam:

  • Bug 1267879 - 2,100 instances of "robustness marked as unsupported" emitted from gfx/gl/GLContextFeatures.cpp during linux64 debug testing
  • Bug 1269448 - 8,400 instances of "NS_ENSURE_TRUE(aURI) failed" emitted from netwerk/cache2/CacheStorage.cpp during linux64 debug testing
  • Bug 1269011 - Lots of "SkCairoFTTypeface::onFilterRec unimplemented" spam during pdf.js debug test runs
  • Added bisection support
  • Added bug filing

Reviewed:

  • Bug 1219464 - Replace PRLogModuleInfo usage with LazyLogModule in widget/
  • Bug 1227347 - Remove whitelisting of CompositorChild leaks for XPCOM and LSan leak checkers
  • Bug 1239686 - Add button to enable logging to about:networking
  • Bug 1263197 - Consider disabling Gecko logging on Fennec
  • Bug 1267227 - gtest.GrowToMaxCapacity fails often when running in windows7 VM in AWS
  • Bug 1268772 - Use MOZ_MUST_USE more in xpcom/ds/
  • Bug 1273712 - Clean up Base64.{h,cpp}
  • Bug 1274148 - Use MOZ_MUST_USE more in xpcom/io/

Next:

OOM hunting, vacation.

Coordination:

Waiting on 2 reviews from peterv