New issue
Advanced search Search tips

Issue 918716 link

Starred by 1 user

Issue metadata

Status: WontFix
Owner:
Closed: Jan 2
Cc:
EstimatedDays: ----
NextAction: ----
OS: ----
Pri: 2
Type: Bug-Regression



Sign in to add a comment

40 kb regression in resource_sizes (MonochromePublic.apk) at 619185:619185

Project Member Reported by estevenson@chromium.org, Jan 2

Issue description

Caused by a skia roll. Ran tools/binary_size/diagnose_bloat.py --subrepo third_party/skia --reference-rev 7561dba7c145^ f3e6b90461e4 --all to find the actual CL:

Caused by "add run_pipeline_obs"

Commit: 05bf9319ec5edcf4e3041b5517e60b86839848b6

Link to size graph: https://chromeperf.appspot.com/report?sid=bb23072657e2d7ca892a1c3fa4643b1ee29b3a0a44d0732adda87168e89c0380&num_points=10&rev=619185
Link to trybot result: https://ci.chromium.org/p/chromium/builders/luci.chromium.try/android-binary-size/125808

Debugging size regressions is documented at: https://chromium.googlesource.com/chromium/src/+/master/docs/speed/apk_size_regressions.md#Debugging-Apk-Size-Increase

These two symbols cause the majority of the regression:
neon::run_pipeline_chunk
neon::lowp::run_pipeline_chunk

It's not clear to me whether or not this increase was expected.
Please have a look and either:

1. Close as “Won't Fix” with a short justification, or
2. Land a revert / fix-up.
 
All graphs for this bug:
  https://chromeperf.appspot.com/group_report?bug_id=918716

(For debugging:) Original alerts at time of bug-filing:
  https://chromeperf.appspot.com/group_report?sid=1ea085ad6ed4e5252ec23be795f331f4fb08e573a5da985b17265d67af2052cc


Bot(s) for this bug's original alert(s):

Android Builder Perf
Status: Assigned (was: Untriaged)
Assigning to chromium-autoroll@skia-public.iam.gserviceaccount.com because this is the only CL in range:
Roll src/third_party/skia 7561dba7c145..f3e6b90461e4 (8 commits)

https://skia.googlesource.com/skia.git/+log/7561dba7c145..f3e6b90461e4


git log 7561dba7c145..f3e6b90461e4 --date=short --no-merges --format='%ad %ae %s'
2018-12-28 brianosman@google.com Revert "Wide color support in the AA linearizing convex path renderer"
2018-12-28 brianosman@google.com Revert "Wide color support in AA Convex path renderer"
2018-12-28 skia-autoroll@skia-public.iam.gserviceaccount.com Roll third_party/externals/angle2 7f6b3674978d..cb9609fe58c2 (1 commits)
2018-12-28 mtklein@google.com move work out of GM ctor
2018-12-28 mtklein@google.com include_dirs = [ "tests" ] not needed
2018-12-28 brianosman@google.com Wide color support in AA Convex path renderer
2018-12-28 mtklein@google.com add run_pipeline_obs
2018-12-28 brianosman@google.com Wide color support in the AA linearizing convex path renderer


Created with:
  gclient setdep -r src/third_party/skia@f3e6b90461e4

The AutoRoll server is located here: https://autoroll.skia.org/r/skia-autoroll

Documentation for the AutoRoller is here:
https://skia.googlesource.com/buildbot/+/master/autoroll/README.md

If the roll is causing failures, please contact the current sheriff, who should
be CC'd on the roll, and stop the roller if necessary.

CQ_INCLUDE_TRYBOTS=luci.chromium.try:android_optional_gpu_tests_rel;luci.chromium.try:linux-blink-rel;luci.chromium.try:linux-chromeos-compile-dbg;luci.chromium.try:linux_optional_gpu_tests_rel;luci.chromium.try:mac_optional_gpu_tests_rel;luci.chromium.try:win_optional_gpu_tests_rel
TBR=robertphillips@chromium.org

Change-Id: I7b1fc1d342485535a9a04c2763225b0ba86db3a5
Reviewed-on: https://chromium-review.googlesource.com/c/1392044
Reviewed-by: chromium-autoroll <chromium-autoroll@skia-public.iam.gserviceaccount.com>
Commit-Queue: chromium-autoroll <chromium-autoroll@skia-public.iam.gserviceaccount.com>
Cr-Commit-Position: refs/heads/master@{#619185}
Owner: mtkl...@google.com
Cc: hua...@chromium.org
Also +huangs@ - the symbols in the diff didn't get demangled properly - I think there's already a bug filed for that but just in case. 
Status: WontFix (was: Assigned)
Yep, this is expected.   We're experimenting with an alternate implementation for a whole bunch of code.  It'll go back down soon, either by reverting this or deleting the other implementation.
Re. demangling problem: Are they symbols that look like:

  _ZN12_GLOBAL__N_124AAFlatteningConvexPathOp19onCombineIfPossibleEP4GrOpRK6GrCaps$39a714be4038aec63642597f522dad6f

? If so, then it's likely due to the presence of '$', which confuses llvm-cxxfilt (and c++filt). I don't know if a bug has been filed for this, but I'll look into it.

Looks like 766903 is the closest bug for this.

Sign in to add a comment