ERROR:frame_sink_provider_impl.cc No RenderWidgetHost exists with id |
|||||||||
Issue descriptionBrowser test [SUCCESS], but is a flaky times-out once in a blue moon, very odd. Not sure if the ERROR report from frame_sink_provider_impl.cc is trying to tell us something here. 21090:21090:0428/122024.631610:INFO:CONSOLE(0)] "[SUCCESS] [rotateImageOnDrive]", source: chrome-extension://ejhcmmdhhpdhhgmifplfmjobgegbibkn/_generated_background_page.html (0) [21090:21090:0428/122026.171214:ERROR:frame_sink_provider_impl.cc(37)] No RenderWidgetHost exists with id 3 in process 9 BrowserTestBase received signal: Terminated. Backtrace: #0 0x00000096ba71 (/b/s/w/ir/out/Release/browser_tests+0x96ba70) #1 0x00000f3f82ec (/b/s/w/ir/out/Release/browser_tests+0xf3f82eb) #2 0x000010983046 (/b/s/w/ir/out/Release/browser_tests+0x10983045) #3 0x7f4855d83cb0 (/lib/x86_64-linux-gnu/libc-2.19.so+0x36caf) #4 0x00000f4224dc (/b/s/w/ir/out/Release/browser_tests+0xf4224db) #5 0x00000f30c44d (/b/s/w/ir/out/Release/browser_tests+0xf30c44c) #6 0x00000f2d5eed (/b/s/w/ir/out/Release/browser_tests+0xf2d5eec) #7 0x00000f22dfa0 (/b/s/w/ir/out/Release/browser_tests+0xf22df9f) #8 0x000015fa6ed8 (/b/s/w/ir/out/Release/browser_tests+0x15fa6ed7) #9 0x000015fa5cc8 (/b/s/w/ir/out/Release/browser_tests+0x15fa5cc7) #10 0x000015f923b5 (/b/s/w/ir/out/Release/browser_tests+0x15f923b4) #11 0x000014fc62bb (/b/s/w/ir/out/Release/browser_tests+0x14fc62ba) #12 0x000014fc88fb (/b/s/w/ir/out/Release/browser_tests+0x14fc88fa) #13 0x000014fc88fb (/b/s/w/ir/out/Release/browser_tests+0x14fc88fa) #14 0x000014fc88fb (/b/s/w/ir/out/Release/browser_tests+0x14fc88fa) #15 0x000014fc88fb (/b/s/w/ir/out/Release/browser_tests+0x14fc88fa) #16 0x000014fc88fb (/b/s/w/ir/out/Release/browser_tests+0x14fc88fa) #17 0x000014fc88fb (/b/s/w/ir/out/Release/browser_tests+0x14fc88fa) #18 0x000014fc1586 (/b/s/w/ir/out/Release/browser_tests+0x14fc1585) #19 0x00001518c47e (/b/s/w/ir/out/Release/browser_tests+0x1518c47d) #20 0x00001518aacb (/b/s/w/ir/out/Release/browser_tests+0x1518aaca) #21 0x0000151c5b8f (/b/s/w/ir/out/Release/browser_tests+0x151c5b8e) #22 0x0000151c8c10 (/b/s/w/ir/out/Release/browser_tests+0x151c8c0f) #23 0x00000f1c8ed1 (/b/s/w/ir/out/Release/browser_tests+0xf1c8ed0) #24 0x00000f46a631 (/b/s/w/ir/out/Release/browser_tests+0xf46a630) #25 0x00000f233c92 (/b/s/w/ir/out/Release/browser_tests+0xf233c91) #26 0x00000f2347b6 (/b/s/w/ir/out/Release/browser_tests+0xf2347b5) #27 0x00000f2350c7 (/b/s/w/ir/out/Release/browser_tests+0xf2350c6) #28 0x00000f4360e1 (/b/s/w/ir/out/Release/browser_tests+0xf4360e0) #29 0x00000f232765 (/b/s/w/ir/out/Release/browser_tests+0xf232764) #30 0x00000f2c8e1e (/b/s/w/ir/out/Release/browser_tests+0xf2c8e1d) #31 0x000010a1a2fb (/b/s/w/ir/out/Release/browser_tests+0x10a1a2fa) #32 0x00000f5e0f31 (/b/s/w/ir/out/Release/browser_tests+0xf5e0f30) #33 0x00000f5e0b12 (/b/s/w/ir/out/Release/browser_tests+0xf5e0b11) #34 0x000010982bd9 (/b/s/w/ir/out/Release/browser_tests+0x10982bd8) #35 0x00000f6e6d22 (/b/s/w/ir/out/Release/browser_tests+0xf6e6d21) #36 0x00000f6e3d8f (/b/s/w/ir/out/Release/browser_tests+0xf6e3d8e) #37 0x00000508b602 (/b/s/w/ir/out/Release/browser_tests+0x508b601) #38 0x0000090cf4f8 (/b/s/w/ir/out/Release/browser_tests+0x90cf4f7) #39 0x00000a179be8 (/b/s/w/ir/out/Release/browser_tests+0xa179be7) #40 0x0000090cb305 (/b/s/w/ir/out/Release/browser_tests+0x90cb304) #41 0x0000090d7555 (/b/s/w/ir/out/Release/browser_tests+0x90d7554) #42 0x0000090c4000 (/b/s/w/ir/out/Release/browser_tests+0x90c3fff) #43 0x00000edecc6b (/b/s/w/ir/out/Release/browser_tests+0xedecc6a) #44 0x00000edf05f9 (/b/s/w/ir/out/Release/browser_tests+0xedf05f8) #45 0x000016648fd4 (/b/s/w/ir/out/Release/browser_tests+0x16648fd3) #46 0x00000ede9864 (/b/s/w/ir/out/Release/browser_tests+0xede9863) #47 0x000010981411 (/b/s/w/ir/out/Release/browser_tests+0x10981410) #48 0x00000f5db524 (/b/s/w/ir/out/Release/browser_tests+0xf5db523) #49 0x0000060be8d3 (/b/s/w/ir/out/Release/browser_tests+0x60be8d2) #50 0x0000060c0b75 (/b/s/w/ir/out/Release/browser_tests+0x60c0b74) #51 0x0000060c1f27 (/b/s/w/ir/out/Release/browser_tests+0x60c1f26) #52 0x0000060e7657 (/b/s/w/ir/out/Release/browser_tests+0x60e7656) #53 0x0000060e68c7 (/b/s/w/ir/out/Release/browser_tests+0x60e68c6) #54 0x00000f62362c (/b/s/w/ir/out/Release/browser_tests+0xf62362b) #55 0x00000f19db73 (/b/s/w/ir/out/Release/browser_tests+0xf19db72) #56 0x000010a0c9f1 (/b/s/w/ir/out/Release/browser_tests+0x10a0c9f0) #57 0x00000f19e988 (/b/s/w/ir/out/Release/browser_tests+0xf19e987) #58 0x00000f19d98f (/b/s/w/ir/out/Release/browser_tests+0xf19d98e) #59 0x7f4855d6ef45 (/lib/x86_64-linux-gnu/libc-2.19.so+0x21f44) #60 0x00000091f78a (/b/s/w/ir/out/Release/browser_tests+0x91f789) [0428/192030.412005:ERROR:nacl_helper_linux.cc(310)] NaCl helper process running without a sandbox! Most likely you need to configure your SUID sandbox correctly [0428/122030.412045:ERROR:nacl_helper_linux.cc(310)] NaCl helper process running without a sandbox! Most likely you need to configure your SUID sandbox correctly [537/860] GalleryBrowserTest.RotateImageOnDrive (TIMED OUT)
,
May 1 2018
,
May 9 2018
Can you provide more information about what is going on here? What test and target is the flake in? The run you linked has a failure in mash_ash_unittests but that doesn't seem related.
,
May 11 2018
Sorry, open the log in #1, search for browser_tests, open its 'stdout', should take you here. https://logs.chromium.org/v/?s=chromium%2Fbb%2Fchromium.memory%2FLinux_Chromium_OS_ASan_LSan_Tests__1_%2F27234%2F%2B%2Frecipes%2Fsteps%2Fbrowser_tests%2F0%2Fstdout Search for the test GalleryBrowserTest.RotateImageOnDrive as an example. There are many other tests with "No RenderWidgetHost exists with ..." ERROR messages. Not sure what it means if if it's something to worry about.
,
May 11 2018
The error message means that the RenderWidgetHostImpl with that id either never existed or has already been destroyed. If it never existed that is probably indicative of a bug in the code or the test. If it has already been destroyed, well there is no ordering guarantees across mojo message pipes so maybe the test is already done with that particular RWHI. I'm not familiar with the tests in question or the code though. fsamuel@ might have more to add since he's worked on content more.
,
May 12 2018
See the log in https://chromium-review.googlesource.com/c/chromium/src/+/1055216 It exhibits the current bug (you asked for examples #3, so here's another :) but more interestingly, the log also has issue 837950 . Makes me suspicious that they are somehow related. Seems like there's something racey going on, seems it's ASAN-only so far, and ASAN slowness is maybe exposing a race condition. And since I'm just the reporter of this bug :), re-assigning.
,
May 21 2018
Passing along to Kyle. This is fixed, right?
,
May 22 2018
Not fixed? At least I didn't do anything to fix these errors and don't plan to. The log errors are indicative of a race in the test. The test is shutting down before a renderer has finished initializing itself. This might not be a problem if the test doesn't need that renderer but I have no idea. If the race is fine then the errors are WAI. The RenderWidgetHost doesn't exist by the time the CompositorFrameSinkRequest has arrived from the renderer and it harmlessly errors out with that log message.
,
May 22 2018
Just to add this appears like a race with CrOS file manager UI and browser_tests. I'm not familiar with that part of the code though.
,
May 24 2018
Assigning back to noel to investigate.
,
May 24 2018
Kyle, Files.app browser tests, and also Files.app in chrome-os, has/opens 4 extensions. The Files.app front-end is opened, and is JS/CSS/HTML code running in a chrome renderer. The tests are JS: test does its thing and ends in JS. When it ends, the browser test C++ StartTest() call returns, which shuts down the browser_tests however it is done. Being in JS, we have no control over that. Perhaps returning StartTest() closes open RenderWidgetHosts, and perhaps before CompositorFrameSinkRequest notices. They might on different Mojo interfaces and are not co-ordinating with each other, or on the same mojo pipe with no ordering guarantees. Seem the C++ code could race in that case, but since I don't work in that part of the code, what would I know :) Anyho Kyle, have a look at recent build, https://logs.chromium.org/v/?s=chromium%2Fbb%2Fchromium.memory%2FLinux_Chromium_OS_ASan_LSan_Tests__1_%2F27577%2F%2B%2Frecipes%2Fsteps%2Fbrowser_tests%2F0%2Fstdout This "ERROR" does not just happen in Files.app tests. The "ERROR" happens in a lot in other browser tests, per that log. Looks to be the normal way of things at least to me, so maybe it's not an "ERROR"? Hard for me to tell. If you have access to content::'s browser is shutting down thing, you could try using that to decide whether to LOG(ERROR) or not.
,
May 24 2018
|
|||||||||
►
Sign in to add a comment |
|||||||||
Comment 1 by noel@chromium.org
, Apr 29 2018Components: Platform>Apps>FileManager Internals>Services>Viz Tests>Flaky
Owner: kylec...@chromium.org
Status: Assigned (was: Untriaged)