Add --serial mode to dm, runs unit tests serially

infra/bots/recipes/swarm_test.py

 # Copyright 2016 The Chromium Authors. All rights reserved.
 # Use of this source code is governed by a BSD-style license that can be
 # found in the LICENSE file.
 
 
 # Recipe module for Skia Swarming test.
 
 
 DEPS = [
   'build/file',
@@ skipping 289 matching lines @@
     match.append('~CompressCheckerboard')
 
   if 'GalaxyS3' in bot:  # skia:1699
     match.append('~WritePixels')
 
   if 'AndroidOne' in bot:  # skia:4711
     match.append('~WritePixels')
 
   if 'NexusPlayer' in bot:
     match.append('~ResourceCache')
+    match.append('~ColorSpaceCRBug')

[mtklein, 2016/08/16 23:54:06]

Is this still going to be necessary once we've made the test run in complete
isolation?  I don't suppose we can construct a smaller test case?

[msarett, 2016/08/17 12:40:45]

I'll run the trybot, but I think we'll still need it.

Unforuntately, the bug only reproduces if we have 3 giant gamma tables - large
enough to overflow uint32_t when their sizes are added together.  And if the ICC
profile isn't large enough to contain the tables, we reject it early.  So we
need a really large ICC profile.

[mtklein, 2016/08/17 12:45:12]

So you're saying it allocates 2 giant tables, then the third tips it over?

Can we write the parsing to check the sizes of all its allocations first before
doing any?

[msarett, 2016/08/17 12:51:47]

This is actually what we do - the large memory use is from the decompressed ICC
profile.  We allocate gammas in one block of memory - so we haven't allocated
anything yet.  The bug fix prevents us from trying to allocate if we overflow
when adding sizeof(gamma1) + sizeof(gamma2) + sizeof(gamma3).

FWIW, I intend to write another CL that rejects all tables that are greater than
<some large number of entries>.  That may incidentally also fix this bug, but I
think the overflow checks are important on their own.

 
   if 'Nexus10' in bot: # skia:5509
     match.append('~CopySurface')
 
   if 'ANGLE' in bot and 'Debug' in bot:
     match.append('~GLPrograms') # skia:4717
 
   if 'MSAN' in bot:
     match.extend(['~Once', '~Shared'])  # Not sure what's up with these tests.
 
   if 'TSAN' in bot:
     match.extend(['~ReadWriteAlpha'])   # Flaky on TSAN-covered on nvidia bots.
 
+  # For memory intensive tests that we don't want to run multi-threaded.
+  serial = []
+  serial.append('ColorSpaceCRBug')
+
   if blacklist:
     args.append('--blacklist')
     args.extend(blacklist)
 
   if match:
     args.append('--match')
     args.extend(match)
 
+  if serial:
+    args.append('--serial')
+    args.extend(serial)
+
   # These bots run out of memory running RAW codec tests. Do not run them in
   # parallel
   if ('NexusPlayer' in bot or 'Nexus5' in bot or 'Nexus9' in bot
       or 'Win8-MSVC-ShuttleB' in bot):
     args.append('--noRAW_threading')
 
   return args
 
 
 def key_params(api):
@@ skipping 513 matching lines @@
         api.path['slave_build'].join('skia', 'infra', 'bots', 'assets',
                                      'skimage', 'VERSION'),
         api.path['slave_build'].join('skia', 'infra', 'bots', 'assets',
                                      'skp', 'VERSION'),
         api.path['slave_build'].join('skia', 'infra', 'bots', 'assets',
                                      'svg', 'VERSION'),
         api.path['slave_build'].join('tmp', 'uninteresting_hashes.txt')
     ) +
     api.platform('win', 64)
   )