New issue
Advanced search Search tips
Note: Color blocks (like or ) mean that a user may not be available. Tooltip shows the reason.

Issue 856601 link

Starred by 4 users

Issue metadata

Status: Assigned
Owner:
Cc:
Components:
EstimatedDays: ----
NextAction: ----
OS: Linux
Pri: 1
Type: Bug

Blocking:
issue 873045



Sign in to add a comment

Many *idlharness* WPT tests fail on chromium.webkit/WebKit Linux Trusty MSAN

Project Member Reported by sheriff-...@appspot.gserviceaccount.com, Jun 26 2018

Issue description

Filed by sheriff-o-matic@appspot.gserviceaccount.com on behalf of grt@chromium.org

webkit_layout_tests failing on chromium.webkit/WebKit Linux Trusty MSAN

Builders failed on: 
- WebKit Linux Trusty MSAN: 
  https://ci.chromium.org/buildbot/chromium.webkit/WebKit%20Linux%20Trusty%20MSAN


 

Comment 1 by grt@chromium.org, Jun 26 2018

Components: Blink
Labels: Stability-Memory-MemorySanitizer OS-Linux Type-Bug
A number of tests started timing out in https://ci.chromium.org/buildbot/chromium.webkit/WebKit%20Linux%20Trusty%20MSAN/8522:

Unexpected Failures:
* external/wpt/FileAPI/idlharness.html
* external/wpt/IndexedDB/interfaces.any.html
* external/wpt/dom/nodes/Element-classlist.html
* external/wpt/mediacapture-record/idlharness.html
* external/wpt/presentation-api/controlling-ua/idlharness.https.html
* fast/css/visited-link-hang.html
* http/tests/devtools/elements/styles-4/styles-inline-element-style-changes-should-not-force-style-recalc.js
* virtual/feature-policy-permissions/external/wpt/mediacapture-streams/MediaDevices-IDL-enumerateDevices.html
Project Member

Comment 2 by bugdroid1@chromium.org, Jun 26 2018

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/06e083bda350bb8c39dd2d54e46748db0154ce07

commit 06e083bda350bb8c39dd2d54e46748db0154ce07
Author: Greg Thompson <grt@chromium.org>
Date: Tue Jun 26 13:30:55 2018

Mark various layout tests as timing out on Linux.

BUG=856601
TBR=petermayo@chromium.org

Change-Id: I82d31d27ab3acecca5dbd345127bd0290e54f933
Reviewed-on: https://chromium-review.googlesource.com/1114841
Commit-Queue: Greg Thompson <grt@chromium.org>
Reviewed-by: Peter Mayo <petermayo@chromium.org>
Reviewed-by: Greg Thompson <grt@chromium.org>
Cr-Commit-Position: refs/heads/master@{#570393}
[modify] https://crrev.com/06e083bda350bb8c39dd2d54e46748db0154ce07/third_party/WebKit/LayoutTests/TestExpectations

Components: -Blink Blink>Infra
Project Member

Comment 4 by bugdroid1@chromium.org, Jul 9

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/e3589ef074fe15fbdd4310ac39788502ada70aa7

commit e3589ef074fe15fbdd4310ac39788502ada70aa7
Author: Takashi Sakamoto <tasak@google.com>
Date: Mon Jul 09 11:41:26 2018

[Gardening] mark various layout tests as timed out on Linux.

WebKit Linux Trusty MSAN Build:
* external/wpt/FileAPI/idlharness.worker.html
* external/wpt/bluetooth/idl/idlharness.tentative.window.html
* external/wpt/css/cssom-view/interfaces.html
* external/wpt/mediacapture-streams/MediaDevices-IDL-enumerateDevices.html
* external/wpt/page-visibility/idlharness.html
* external/wpt/touch-events/idlharness.window.html
* external/wpt/web-animations/interfaces/Animation/idlharness.html
* http/tests/devtools/network/network-filters.js

TBR=pmonette@chromium.org

Bug: 856601
Change-Id: I8f3fb802b935fa1c321a89c46b7d63a801e8c2c4
No-Try: true
Reviewed-on: https://chromium-review.googlesource.com/1128836
Reviewed-by: Takashi Sakamoto <tasak@google.com>
Commit-Queue: Takashi Sakamoto <tasak@google.com>
Cr-Commit-Position: refs/heads/master@{#573273}
[modify] https://crrev.com/e3589ef074fe15fbdd4310ac39788502ada70aa7/third_party/WebKit/LayoutTests/TestExpectations

Project Member

Comment 5 by bugdroid1@chromium.org, Jul 9

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/8a49ff6a3074257bdf00a33468d909ce0007bbff

commit 8a49ff6a3074257bdf00a33468d909ce0007bbff
Author: Vasilii Sukhanov <vasilii@chromium.org>
Date: Mon Jul 09 13:59:38 2018

Mark http/tests/event-timing/event-timing-retrievability.html as timing out

TBR=pmonette@chromium.org

Bug: 856601
Change-Id: Idac2114ffb81e7e02acf75b8b30f194cfa51c6a1
Reviewed-on: https://chromium-review.googlesource.com/1128967
Reviewed-by: Vasilii Sukhanov <vasilii@chromium.org>
Cr-Commit-Position: refs/heads/master@{#573294}
[modify] https://crrev.com/8a49ff6a3074257bdf00a33468d909ce0007bbff/third_party/WebKit/LayoutTests/TestExpectations

Cc: foolip@chromium.org
Components: -Blink>Infra Blink>Infra>Ecosystem
Summary: Many *idlharness* WPT tests fail on chromium.webkit/WebKit Linux Trusty MSAN (was: webkit_layout_tests failing on chromium.webkit/WebKit Linux Trusty MSAN)
Revising summary.

Most of the failing tests are *idlharness* or *interface* tests. If you search TestExpectations you see many entries, starting around 2018-06-26.

It's hard to tell when these started because MSAN isn't on the Flakiness Dashboard.

I don't see changes upstream to WPT that could have caused the regression.

+foolip: any ideas?
Cc: meade@chromium.org guidou@chromium.org waff...@chromium.org
 Issue 862588  has been merged into this issue.
Owner: lukebjerring@chromium.org
Status: Assigned (was: Available)
Sorry, my WPT checkout just wasn't updated. There are a lot of WPT IDL changes from Luke in recent days that align with these test failures.

Luke: Can you take a look please?
Surprised to see these timing out; starting with FileAPI, we can at least parallelize the fetches and clean up the tests a little;
https://github.com/web-platform-tests/wpt/pull/11957

We'll see if that improves things. Otherwise it might just need to be a long timeout, if `new File` and `new Blob` are just dang slow. Will look at the others soon.
Chiming in to add some more context:

The MSAN/ASAN/dbg bots are indeed much slower than release bots. If a test takes multiple seconds (very close to timeout) on normal bots, the test is likely to time out on MSAN/ASAN/dbg bots, in which case the recommended solution is to first speed up the test and perhaps give it a long timeout (by adding a meta timeout tag), which will also reduce the potential flakiness of the test on release bots.

In rare cases, a test might run very fast (<1s) on release bots but still time out on slower bots. We can then add specific expectations for the test in {ASAN,MSAN}Expectations.
Seems like a common denominator here is that these files are pulling in the `html.idl` dependency, which is a whopping 2500 lines or so. Might need to flag those tests as long timeout, or come up with a fancy ".idl" test type that generates a test with the timeout META tag, but we'll try with parallel fetches first.
The complication here is that we can't at run time increase the timeout, so it's not as simple as adding an if to idlharness.js. Besides, there are a LOT of idlharness tests that depend on dom/html.

This wasn't an issue before because many of these tests added ad-hoc (usually empty) dependent interfaces instead of pulling in the dependencies properly.

We suspect parsing dom.idl and html.idl is computationally expensive and might suffer from MSAN/ASAN overhead more than other tests, but we'd like to get the low-hanging fruits first (making fetches of IDL files parallel) and see if it helps.
Update: I've sifted through the expectations (https://cs.chromium.org/chromium/src/third_party/WebKit/LayoutTests/TestExpectations?type=cs&q=file:testexpectations&sq=package:chromium&g=0&l=1) to ensure that all of the tests tagged with Timeout have been modified to have parallel fetches already, or in a pending PR.

filtered list:
/bluetooth/idl/idlharness.tentative.window.html
/cookie-store/idlharness.tentative.https.html
/css/cssom-view/interfaces.html
/dom/nodes/Element-classlist.html
/fetch/cors-rfc1918/idlharness.tentative.any.html
/FileAPI/idlharness.html
/FileAPI/idlharness.worker.html
/gamepad/idlharness.html
/IndexedDB/interfaces.any.html
/mediacapture-record/idlharness.html
/mediacapture-streams/idlharness.https.window.html
/page-visibility/idlharness.html
/payment-request/idlharness.https.window.html
/presentation-api/controlling-ua/idlharness.https.html
/push-api/idlharness.https.any.sharedworker.html
/push-api/idlharness.https.any.worker.html
/resource-timing/idlharness.any.html
/resource-timing/idlharness.any.worker.html
/selection/interfaces.html
/service-workers/service-worker/interfaces-sw.https.html
/touch-events/idlharness.window.html
/wpt/mediacapture-streams/MediaDevices-IDL-enumerateDevices.html
Project Member

Comment 14 by bugdroid1@chromium.org, Jul 27

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/6f03c4c03992a2a99f1ef3f829bafc6ed4ea1572

commit 6f03c4c03992a2a99f1ef3f829bafc6ed4ea1572
Author: Robert Ma <robertma@chromium.org>
Date: Fri Jul 27 22:28:25 2018

Move MSAN-specific expectations to MSANExpectations

They don't need to (and shouldn't) be in the bloated TestExpectations.

Most of these are related to idlharness.js and their corresponding bugs
have been merged to crbug.com/856601, so update the bug links
accordingly.

Besides, two expectations were incorrectly removed in an import
https://crrev.com/c/1152529; they are also added back in this CL.

Bug: 856601
Change-Id: Idf3a58cd3a4b5ae9f149bbff7b5150bfda3ed645
Reviewed-on: https://chromium-review.googlesource.com/1153615
Commit-Queue: Philip Jägenstedt <foolip@chromium.org>
Reviewed-by: Philip Jägenstedt <foolip@chromium.org>
Cr-Commit-Position: refs/heads/master@{#578831}
[modify] https://crrev.com/6f03c4c03992a2a99f1ef3f829bafc6ed4ea1572/third_party/WebKit/LayoutTests/MSANExpectations
[modify] https://crrev.com/6f03c4c03992a2a99f1ef3f829bafc6ed4ea1572/third_party/WebKit/LayoutTests/TestExpectations

What is the recommended way to fix these? It looks like we need to merge more bugs/update more expectations:
 issue 866965 
issue 810963
issue 819778
Project Member

Comment 16 by bugdroid1@chromium.org, Jul 30

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/a6a03569dea7b9baeb4ea58895ca57e7e7811aa4

commit a6a03569dea7b9baeb4ea58895ca57e7e7811aa4
Author: Kenichi Ishibashi <bashi@chromium.org>
Date: Mon Jul 30 02:14:59 2018

Move some S13nServiceWorker related test expectations to MSANExpectations

These tests are timing out only on msan bots. Move these entries
to MSANExpectations so that we can make sure these tests pass
on linux bots.

Bug:  866965 , 856601
Change-Id: I48d15a00a508c989865ebba2fa88987ddcd46c80
Reviewed-on: https://chromium-review.googlesource.com/1154609
Reviewed-by: Matt Falkenhagen <falken@chromium.org>
Commit-Queue: Kenichi Ishibashi <bashi@chromium.org>
Cr-Commit-Position: refs/heads/master@{#578964}
[modify] https://crrev.com/a6a03569dea7b9baeb4ea58895ca57e7e7811aa4/third_party/WebKit/LayoutTests/MSANExpectations
[modify] https://crrev.com/a6a03569dea7b9baeb4ea58895ca57e7e7811aa4/third_party/WebKit/LayoutTests/TestExpectations

Project Member

Comment 17 by bugdroid1@chromium.org, Aug 1

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/f3d5d112714f7b12bad5dc23820d89b6cfc141e1

commit f3d5d112714f7b12bad5dc23820d89b6cfc141e1
Author: Christos Froussios <cfroussios@chromium.org>
Date: Wed Aug 01 07:34:44 2018

Disable external/wpt/IndexedDB/idlharness.any.sharedworker.html on MSAN

It is mostly failing (timeout)

Bug: 856601
No-try: true
No-Presubmit: true
Change-Id: Ic8785e8bf5237d4e252c94a48a00302ce6e0303a
Reviewed-on: https://chromium-review.googlesource.com/1158144
Commit-Queue: Christos Froussios <cfroussios@chromium.org>
Reviewed-by: Christos Froussios <cfroussios@chromium.org>
Cr-Commit-Position: refs/heads/master@{#579719}
[modify] https://crrev.com/f3d5d112714f7b12bad5dc23820d89b6cfc141e1/third_party/WebKit/LayoutTests/MSANExpectations

Project Member

Comment 18 by bugdroid1@chromium.org, Aug 1

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/d9ce625fa18367381d40a7c876fb1b7acd96ad7a

commit d9ce625fa18367381d40a7c876fb1b7acd96ad7a
Author: Christos Froussios <cfroussios@chromium.org>
Date: Wed Aug 01 08:07:38 2018

Disable several *idlharness* tests

They are flaking with timeout

Bug: 856601
No-try: true
No-Presubmit: true
Change-Id: I40e09e6b8aed3cb2014a284d012abc8122d614b7
Reviewed-on: https://chromium-review.googlesource.com/1158228
Reviewed-by: Christos Froussios <cfroussios@chromium.org>
Commit-Queue: Christos Froussios <cfroussios@chromium.org>
Cr-Commit-Position: refs/heads/master@{#579723}
[modify] https://crrev.com/d9ce625fa18367381d40a7c876fb1b7acd96ad7a/third_party/WebKit/LayoutTests/MSANExpectations

Labels: -Pri-2 Pri-1
OK we need to take this more seriously.

Does anyone know what kind of workload would be affected the most on MSAN builds? CPU-bound or IO-bound? And how do we enable MSAN locally to reproduce the flakes?
Blocking: 873045
 Issue 872731  has been merged into this issue.
I've detected 5 new tests that have timed out multiple times in the past 14 runs:

* external/wpt/css/css-typed-om/interfaces.html [x11]
* external/wpt/mediacapture-streams/idlharness.https.window.html [x5]
* external/wpt/notifications/idlharness.https.any.html [x3]
* external/wpt/push-api/idlharness.https.any.html [x2]
* external/wpt/speech-api/idlharness.window.html [x6]

Adding disables for these 5 (see  Issue 873045 ).
Project Member

Comment 23 by bugdroid1@chromium.org, Aug 10

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/67c441d2c33570fc281cef35a61169ad41a1a76b

commit 67c441d2c33570fc281cef35a61169ad41a1a76b
Author: Matt Giuca <mgiuca@chromium.org>
Date: Fri Aug 10 05:42:52 2018

WebKit Layout Tests: Disable a lot of flaky tests on MSAN.

Adds a bunch of tests with high-probability flake due to timeout or
crash, to existing lists of very similar tests that were already
disabled.

Tbr: mgiuca@chromium.org
Bug:  873045 ,  838057 , 856601, 869818
Change-Id: I66e14994ef4551d5690d3a66d51fce84a1eaf456
No-Presubmit: true
No-Tree-Checks: true
No-Try: true
Reviewed-on: https://chromium-review.googlesource.com/1170450
Reviewed-by: Matt Giuca <mgiuca@chromium.org>
Commit-Queue: Matt Giuca <mgiuca@chromium.org>
Cr-Commit-Position: refs/heads/master@{#582068}
[modify] https://crrev.com/67c441d2c33570fc281cef35a61169ad41a1a76b/third_party/WebKit/LayoutTests/MSANExpectations

Project Member

Comment 24 by bugdroid1@chromium.org, Aug 13

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/67d786b7e1c30e47ce0c0093c60709abdd6e75f2

commit 67d786b7e1c30e47ce0c0093c60709abdd6e75f2
Author: Makoto Shimazu <shimazu@chromium.org>
Date: Mon Aug 13 07:08:11 2018

Mark more idlharness.js tests on MSan as slow

TBR=robertma@chromium.org
NOTRY=true

Bug: 856601
Change-Id: I8527c9d7c1df382f08620378ac9931634cfdd5aa
Reviewed-on: https://chromium-review.googlesource.com/1172226
Reviewed-by: Makoto Shimazu <shimazu@chromium.org>
Commit-Queue: Makoto Shimazu <shimazu@chromium.org>
Cr-Commit-Position: refs/heads/master@{#582519}
[modify] https://crrev.com/67d786b7e1c30e47ce0c0093c60709abdd6e75f2/third_party/WebKit/LayoutTests/MSANExpectations

Project Member

Comment 25 by bugdroid1@chromium.org, Aug 14

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/a1a1d10c0e71ad349a47c3eeea15fe7ff603a30e

commit a1a1d10c0e71ad349a47c3eeea15fe7ff603a30e
Author: Makoto Shimazu <shimazu@chromium.org>
Date: Tue Aug 14 07:33:28 2018

Mark more idlharness tests as [ Timeout Pass ] on MSan bots

TBR=robertma@chromium.org
NOTRY=true

Bug: 856601
Change-Id: Iad0f3b236494cc6d1f14af0fc03518ebf3e0671a
Reviewed-on: https://chromium-review.googlesource.com/1174144
Reviewed-by: Makoto Shimazu <shimazu@chromium.org>
Commit-Queue: Makoto Shimazu <shimazu@chromium.org>
Cr-Commit-Position: refs/heads/master@{#582856}
[modify] https://crrev.com/a1a1d10c0e71ad349a47c3eeea15fe7ff603a30e/third_party/WebKit/LayoutTests/MSANExpectations

What's a progress on this bug?

So many tests have been timing out, although they are non-deterministic (i.e., the tests are just slow). It doesn't really look productive to mark tests as [Timeout] one by one every time we find them... :/


Cc: rhalavati@chromium.org
 Issue 886565  has been merged into this issue.
Owner: robertma@chromium.org
Pawning this off to Robert who had some thoughts about working out what's slow (in an MSAN build).
Cc: shend@chromium.org
 Issue 872728  has been merged into this issue.
Cc: -rhalavati@chromium.org
We're still seeing timeouts on several tests. Especially the first one:
* external/wpt/push-api/idlharness.https.any.serviceworker.html
* external/wpt/BackgroundSync/interfaces.https.any.html
* external/wpt/IndexedDB/idlharness.any.worker.html
* external/wpt/notifications/idlharness.https.any.serviceworker.html
* external/wpt/BackgroundSync/interfaces.https.any.html
* external/wpt/IndexedDB/idlharness.any.worker.html
* external/wpt/gamepad/idlharness.window.html

https://ci.chromium.org/buildbot/chromium.webkit/WebKit%20Linux%20Trusty%20MSAN/
Project Member

Comment 33 by bugdroid1@chromium.org, Oct 5

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/f8b83bcb698c1fbd01e784b11198e3dcc6cefa64

commit f8b83bcb698c1fbd01e784b11198e3dcc6cefa64
Author: Marijn Kruisselbrink <mek@chromium.org>
Date: Fri Oct 05 21:48:28 2018

Mark more idlharness tests as [ Timeout Pass ] on MSan bots.

Tbr: robertma@chromium.org
Bug: 856601
Change-Id: Ib6b0b9c13c89de9637222601d0830cef7bcaea8f
Reviewed-on: https://chromium-review.googlesource.com/c/1265561
Reviewed-by: Marijn Kruisselbrink <mek@chromium.org>
Commit-Queue: Marijn Kruisselbrink <mek@chromium.org>
Cr-Commit-Position: refs/heads/master@{#597320}
[modify] https://crrev.com/f8b83bcb698c1fbd01e784b11198e3dcc6cefa64/third_party/WebKit/LayoutTests/MSANExpectations

Ping robertma@chromium.org, from your friendly neighbourhood ecosystem infra rotation :-)
wpt/css/filter-effects/interfaces.any.html is also timeout on Linux Trusty MSAN

https://ci.chromium.org/buildbot/chromium.webkit/WebKit%20Linux%20Trusty%20MSAN/10984
Project Member

Comment 36 by bugdroid1@chromium.org, Nov 15

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/a4b02559ebf271f64a4d26c4e41765f657eac2d6

commit a4b02559ebf271f64a4d26c4e41765f657eac2d6
Author: Ella Ge <eirage@chromium.org>
Date: Thu Nov 15 20:08:55 2018

filter-effects/interfaces.any.html failed on MSAN

add MSAN expectation

TBR=robertma@chromium.org

Bug: 856601
Change-Id: Idb002c64071199228cb2b5881e1e773de0325fbc
Reviewed-on: https://chromium-review.googlesource.com/c/1338299
Reviewed-by: Ella Ge <eirage@chromium.org>
Reviewed-by: Robert Ma <robertma@chromium.org>
Commit-Queue: Ella Ge <eirage@chromium.org>
Cr-Commit-Position: refs/heads/master@{#608485}
[modify] https://crrev.com/a4b02559ebf271f64a4d26c4e41765f657eac2d6/third_party/WebKit/LayoutTests/MSANExpectations

Cc: robertma@chromium.org
Owner: kyleju@chromium.org
Kyle will take over the investigation.
 Issue 906211  has been merged into this issue.
Project Member

Comment 39 by bugdroid1@chromium.org, Nov 22

The following revision refers to this bug:
  https://chromium.googlesource.com/chromium/src.git/+/5a2a7a91adc9e7c97d867a098fa315d2b2bef7b1

commit 5a2a7a91adc9e7c97d867a098fa315d2b2bef7b1
Author: Anders Hartvoll Ruud <andruud@chromium.org>
Date: Thu Nov 22 11:06:48 2018

Flaky on MSAN: http/tests/devtools/elements/elements-save-to-temp-var.js.

TBR=kyleju@chromium.org

Bug: 856601
Change-Id: I4ab0da613ba282fab128206a91805593952ba0ff
Reviewed-on: https://chromium-review.googlesource.com/c/1347362
Reviewed-by: Anders Hartvoll Ruud <andruud@chromium.org>
Commit-Queue: Anders Hartvoll Ruud <andruud@chromium.org>
Cr-Commit-Position: refs/heads/master@{#610375}
[modify] https://crrev.com/5a2a7a91adc9e7c97d867a098fa315d2b2bef7b1/third_party/WebKit/LayoutTests/MSANExpectations

Cc: dgozman@chromium.org ellyjo...@chromium.org
 Issue 914900  has been merged into this issue.

Sign in to add a comment