ContentBrowserTest stack symbolication tests failing on CrWinAsan, likely after https://codereview.chromium.org/2857643002 |
|||
Issue descriptionStarted here: https://luci-milo.appspot.com/buildbot/chromium.fyi/CrWinAsan%20tester/1739 https://codereview.chromium.org/2857643002 assumes we externally symbolize. ContentBrowserTest.BrowserCrashCallStack ContentBrowserTest.RendererCrashCallStack
,
Nov 6 2017
,
Nov 8 2017
The following revision refers to this bug: https://chromium.googlesource.com/chromium/src.git/+/5d357ed23c6f0c4b34971935752877519e451191 commit 5d357ed23c6f0c4b34971935752877519e451191 Author: Reid Kleckner <rnk@google.com> Date: Wed Nov 08 23:40:51 2017 Fix crash content_browsertests callstack tests under WinASan First, there is a use-after-return in the browser process of MANUAL_RendererCrash. It registers a crash observer which gets called back after the renderer process exits. Second, RendererProcessExited checks the termination status. When ASan catches the crash access violation, it ultimately exits with the default exit code of 1. Code in base/process/kill_win.cc classifies an exit code of 1 as TERMINATION_STATUS_PROCESS_WAS_KILLED. Exit code 1 is pretty common, so that classification is suspicious to me, but it is what happens today. Third, BrowserCrashCallstack crashes with DCHECK(false), which crashes with __debugbreak. ASan isn't currently set up to handle EXCEPTION_BREAKPOINT, although it can be configured to do this. As a result, Chromium does the stack dumping, and it does not contain the string "#0 ". This is *not* the case in the renderer test case, which crashes with an access violation, and gets an ASan report. However, Windows ASan knows how to symbolize stack traces even inside a sandbox, so it's still correct to look for the symbolized frame. Last but not least, there seems to be some conflict between TSan, the Linux sandbox, and dumping the stack from a renderer process. Disabling the sandbox makes the test dump its stack as expected, so I went ahead and did that. R=wfh@chromium.org Bug: 752168 Change-Id: I639e7947db3c1d47fbefb48f3c94b11832b0765b Reviewed-on: https://chromium-review.googlesource.com/755760 Commit-Queue: Reid Kleckner <rnk@chromium.org> Reviewed-by: Alex Moshchuk <alexmos@chromium.org> Reviewed-by: Will Harris <wfh@chromium.org> Cr-Commit-Position: refs/heads/master@{#515007} [modify] https://crrev.com/5d357ed23c6f0c4b34971935752877519e451191/content/test/content_browser_test_test.cc
,
Feb 28 2018
|
|||
►
Sign in to add a comment |
|||
Comment 1 by r...@chromium.org
, Aug 8 2017