Upon further inspection, it looks like there are two places where it got flakier. In the one I mentioned above, the test went from being fairly flaky  (~25% flakes) to very very flaky (~90% flakes). Previously, though, between r572977 and r573188, it looks like it went from being a little flaky (~10%) to fairly flaky (~25%). I'll launch a bisect over both ranges.

https://test-results.appspot.com/dashboards/flakiness_dashboard.html#testType=performance_test_suite&showAllRuns=true&builder=chromium.perf%3Alinux-perf

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/1597f22ea40000

📍 Pinpoint job started.
https://pinpoint-dot-chromeperf.appspot.com/job/11e6ca0ea40000

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/c0e6d51ac794c73501f3a4a426d46196677ebcee

commit c0e6d51ac794c73501f3a4a426d46196677ebcee
Author: Charlie Andrews <charliea@chromium.org>
Date: Tue Jul 17 20:57:45 2018

Disable v8.browsing_desktop/flipboard on Linux

The test has recently become much flakier.

TBR=nednguyen@chromium.org
CC=mythria@chromium.org,ulan@chromium.org
NOTRY=true

Bug:  864718 
Change-Id: Ief78f4e4525bb562416bfb293770ed0b905dc32c
Reviewed-on: https://chromium-review.googlesource.com/1141067
Reviewed-by: Charlie Andrews <charliea@chromium.org>
Commit-Queue: Charlie Andrews <charliea@chromium.org>
Cr-Commit-Position: refs/heads/master@{#575765}
[modify] https://crrev.com/c0e6d51ac794c73501f3a4a426d46196677ebcee/tools/perf/expectations.config

📍 Couldn't reproduce a difference.
https://pinpoint-dot-chromeperf.appspot.com/job/1597f22ea40000

📍 Found a significant difference after 1 commit.
https://pinpoint-dot-chromeperf.appspot.com/job/11e6ca0ea40000

Roll src/third_party/catapult 8a075d7c378a..9eaedb7fa00f (1 commits) by catapult-chromium-autoroll@skia-buildbots.google.com.iam.gserviceaccount.com
https://chromium.googlesource.com/chromium/src/+/b5896b295b3238c960714f1aef991e022c0a45a0
0 → 0.25 (+0.25)

Assigning to sheriff sullivan@chromium.org because "Roll src/third_party/catapult 8a075d7c378a..9eaedb7fa00f (1 commits)" is a roll.

Understanding performance regressions:
  http://g.co/ChromePerformanceRegressions

Ned, looks like your CL resulted in a significant flakiness increase here. Any idea why?

Lemme look at this

I can reproduce the crash locally. I think my CL change some of the timing of the loading sequence which surface this crash.

Here is the log:
https://console.developers.google.com/m/cloudstorage/b/chrome-telemetry-output/o/0a0c4cb8-8a11-11e8-8505-509a4c654258



Found Minidump: True
Stack Trace:
********************************************************************************
	Operating system: Linux
	                  0.0.0 Linux 3.13.0-143-generic #192-Ubuntu SMP Tue Feb 27 10:45:36 UTC 2018 x86_64
	CPU: amd64
	     family 6 model 94 stepping 3
	     1 CPU
	
	GPU: UNKNOWN
	
	Crash reason:  SIGSEGV
	Crash address: 0x8
	Process uptime: not available
	
	Thread 0 (crashed)
	 0  chrome!blink::ImageResource::OnMemoryDump(blink::WebMemoryDumpLevelOfDetail, blink::WebProcessMemoryDump*) const [shared_buffer.h : 131 + 0x0]
	    rax = 0x0000000000000000   rdx = 0x0000000000000003
	    rcx = 0x00007f09eb821df8   rbx = 0x000035af09877bf0
	    rsi = 0x00001b99e371d3e0   rdi = 0x00007ffc90c86cd0
	    rbp = 0x00007ffc90c86d10   rsp = 0x00007ffc90c86cd0
	     r8 = 0x00001b832bee7390    r9 = 0x000000000000007f
	    r10 = 0x000021482b42f800   r11 = 0x00007f09dd8fb150
	    r12 = 0x00000ba725626ba8   r13 = 0x00000ba725753830
	    r14 = 0x00001b832be4d190   r15 = 0x00007ffc90c86cd0
	    rip = 0x00007f09ea5ecf12
	    Found by: given as instruction pointer in context
	 1  chrome!blink::MemoryCache::OnMemoryDump(blink::WebMemoryDumpLevelOfDetail, blink::WebProcessMemoryDump*) [memory_cache.cc : 450 + 0xa]
	    rbx = 0x00000ba725626b88   rbp = 0x00007ffc90c86e80
	    rsp = 0x00007ffc90c86d20   r12 = 0x00000ba725626ba8
	    r13 = 0x00000ba725753830   r14 = 0x00000ba725753d80
	    r15 = 0x00007f09eba3d8a0   rip = 0x00007f09e6eb9bdb
	    Found by: call frame info
	 2  chrome!non-virtual thunk to blink::MemoryCache::OnMemoryDump(blink::WebMemoryDumpLevelOfDetail, blink::WebProcessMemoryDump*) [memory_cache.cc : 0 + 0x5]
	    rbx = 0x0000000000000001   rbp = 0x00007ffc90c86e90
	    rsp = 0x00007ffc90c86e90   r12 = 0x00001b8327a81c60
	    r13 = 0x00001b8327a9c500   r14 = 0x00007f09ebdd1c48
	    r15 = 0x00007ffc90c86ea8   rip = 0x00007f09e6eb9c6d
	    Found by: call frame info
	 3  chrome!blink::MemoryCacheDumpProvider::OnMemoryDump(base::trace_event::MemoryDumpArgs const&, base::trace_event::ProcessMemoryDump*) [memory_cache_dump_provider.cc : 40 + 0xa]
	    rbx = 0x0000000000000001   rbp = 0x00007ffc90c86f10
	    rsp = 0x00007ffc90c86ea0   r12 = 0x00001b8327a81c60
	    r13 = 0x00001b8327a9c500   r14 = 0x00007f09ebdd1c48
	    r15 = 0x00007ffc90c86ea8   rip = 0x00007f09e6eb054c
	    Found by: call frame info
	 4  chrome!base::trace_event::MemoryDumpManager::ContinueAsyncProcessDump(base::trace_event::MemoryDumpManager::ProcessMemoryDumpAsyncState*) [memory_dump_manager.cc : 458 + 0x3]
	    rbx = 0x00007ffc90c86f90   rbp = 0x00007ffc90c86fd0
	    rsp = 0x00007ffc90c86f20   r12 = 0x00001b8327a81c60
	    r13 = 0x00001b8327a9c500   r14 = 0x00001b832ba2c190
	    r15 = 0x00001b8327a81c60   rip = 0x00007f09e741cdc5
	    Found by: call frame info
	 5  chrome!base::debug::TaskAnnotator::RunTask(char const*, base::PendingTask*) [callback.h : 99 + 0x2]


Assign this bug to lizeb@ who recently added the ImageResource::OnMemoryDump() code.

To reproduce this crash, you can run:

./tools/perf/run_benchmark --browser=default v8.browsing_desktop --story-filter=browse:news:flipboard  --pageset-repeat=10 --also-run-disabled-tests

Thanks!

Looking.

I ran a manual bisect locally.
Revision range: https://chromium.googlesource.com/chromium/src/+log/6d9e97dc832fdbcf9723c1b3c4dceaf9bcc74caf..d74b30f18efbeebf19ca6c29d6bd2478f925d021

d74b30f Do not pass source string when creating code cache. by Yang Guo · 4 months ago
4bd1eea Modernize unique_ptr usage in form utils by Vaclav Brozek · 4 months ago
ded3fb8 Print Preview Componentization: Port UMA metrics calls by rbpotter · 4 months ago
46f3835 Remove SyntheticForm::fieldsets by Vaclav Brozek · 4 months ago
67c5b71 Roll src-internal/ 130fd7666..30a2b933e (1 commit) by src-internal-chromium-autoroll@skia-buildbots.google.com.iam.gserviceaccount.com · 4 months ago
cfaac65 Deduplicate code in AdjustInlineBoxPositionForTextDirection() with templates by Xiaocheng Hu · 4 months ago
5f97a1d [Picture in Picture] Update interstitial text to match UX guidance. by Jennifer Apacible · 4 months ago
fd33ca8 chromeos: Enable mode-specific power button behavior. by Daniel Erat · 4 months ago
8cbea5e Settings > Display: Use a single div by Steven Bennetts · 4 months ago
35c8765 Move SyntheticForm into GetPasswordForm by Vaclav Brozek · 4 months ago

lizeb@ Ping, this is currently breaking several benchmarks and it repros nicely with the above instructions.

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/2f767d05dcc0d5bb7c7d01607382ae8c89126cad

commit 2f767d05dcc0d5bb7c7d01607382ae8c89126cad
Author: Benoit Lize <lizeb@chromium.org>
Date: Thu Aug 23 08:49:21 2018

blink: Fix ImageResource::OnMemoryDump() when the image is null.

This is hit in perf benchmarks, namely
v8.browsing_desktop/browse:news:flipboard (see attached bug).

Also fix trivial "git cl lint" warnings.

Bug:  864718 
Change-Id: If00839d578e04845f768d8480dc084f22980ed62
Reviewed-on: https://chromium-review.googlesource.com/1184843
Reviewed-by: Yutaka Hirano <yhirano@chromium.org>
Commit-Queue: Benoit L <lizeb@chromium.org>
Cr-Commit-Position: refs/heads/master@{#585431}
[modify] https://crrev.com/2f767d05dcc0d5bb7c7d01607382ae8c89126cad/third_party/blink/renderer/core/loader/resource/image_resource.cc

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/543421c268a01aef6de2088e15770c007e5f5e30

commit 543421c268a01aef6de2088e15770c007e5f5e30
Author: Benoit Lize <lizeb@chromium.org>
Date: Thu Aug 23 14:10:17 2018

Re-enable v8.browsing_desktop/browse:news:flipboard on Linux.

Test is no longer flaky locally.

Bug:  864718 
Change-Id: I79dcbf886b5365f6802144ef99a0a3ae024b3594
TBR: nednguyen@chromium.org
Cc: mythria@chromium.org,ulan@chromium.org
Reviewed-on: https://chromium-review.googlesource.com/1186469
Reviewed-by: Benoit L <lizeb@chromium.org>
Commit-Queue: Benoit L <lizeb@chromium.org>
Cr-Commit-Position: refs/heads/master@{#585478}
[modify] https://crrev.com/543421c268a01aef6de2088e15770c007e5f5e30/tools/perf/expectations.config