Issue metadata
Sign in to add a comment
|
ATK: launching accerciser after Chromium (with a11y enabled) crashes |
||||||||||||||||||||||||
Issue descriptionUserAgent: Mozilla/5.0 (X11; Linux x86_64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/62.0.3202.75 Safari/537.36 Steps to reproduce the problem: Precondition: accerciser is not running. 1. Launch chromium with ACCESSIBILITY_ENABLED=1 set in environment 2. Launch accerciser 3. What is the expected behavior? Accerciser shows Chromium accessibility tree (with proper contents, so more than 0 children of application node). What went wrong? Chromium crashes: Received signal 11 SEGV_MAPERR 000000000020 #0 0x7f882d96bf3d base::debug::StackTrace::StackTrace() #1 0x7f882d96a36c base::debug::StackTrace::StackTrace() #2 0x7f882d96b8f5 base::debug::(anonymous namespace)::StackDumpSignalHandler() #3 0x7f882dec7150 <unknown> #4 0x7f8828cbab40 content::BrowserAccessibilityAuraLinux::GetAtkObject() #5 0x7f8828cbb7d7 content::browser_accessibility_ref_child() #6 0x7f881199a163 <unknown> #7 0x7f881199a253 <unknown> #8 0x7f8817d36775 g_type_create_instance #9 0x7f8817d175e8 <unknown> #10 0x7f8817d18d85 g_object_new_with_properties #11 0x7f8817d19801 g_object_new #12 0x7f881199bf01 <unknown> #13 0x7f881199c77f <unknown> #14 0x7f88119a3a65 <unknown> #15 0x7f88119a140d <unknown> #16 0x7f8811261fd3 <unknown> #17 0x7f8811253624 dbus_connection_dispatch #18 0x7f880d2b5525 <unknown> #19 0x7f8817a38fb7 g_main_context_dispatch #20 0x7f8817a391f0 <unknown> #21 0x7f8817a3927c g_main_context_iteration #22 0x7f882da1ce1f base::MessagePumpGlib::Run() #23 0x7f882da187cc base::MessageLoop::Run() #24 0x7f882dac58fb base::RunLoop::Run() #25 0x559fe6325add ChromeBrowserMainParts::MainMessageLoopRun() #26 0x7f8827e897f6 content::BrowserMainLoop::RunMainMessageLoopParts() #27 0x7f8827e9363f content::BrowserMainRunnerImpl::Run() #28 0x7f8827e7d1ab content::BrowserMain() #29 0x7f88299b6e08 content::RunNamedProcessTypeMain() #30 0x7f88299b99fa content::ContentMainRunnerImpl::Run() #31 0x7f88299afebd content::ContentServiceManagerMainDelegate::RunEmbedderProcess() #32 0x7f882e16089d service_manager::Main() #33 0x7f88299b5c4f content::ContentMain() #34 0x559fe4b6d22e ChromeMain #35 0x559fe4b6d142 main #36 0x7f881530e1c1 __libc_start_main #37 0x559fe4b6d029 <unknown> Crashed report ID: How much crashed? Whole browser Is it a problem with a plugin? No Did this work before? N/A Chrome version: 64.0.3261.0 Channel: canary OS Version: Ubuntu 17.04 Flash Version: It looks like, as atk bridge traverses the accessibility tree due to the request from accerciser, we find an AXNode that has no ATK object created, and fetching it we get a crash.
,
Nov 6 2017
,
Nov 6 2017
,
Nov 7 2017
,
Nov 7 2017
Problem was partial adaptation to the PlatformGetChild/PlatformChildCount API. I'll submit a fix today.
,
Nov 8 2017
The following revision refers to this bug: https://chromium.googlesource.com/chromium/src.git/+/354403c148b16c6a99e4175e5a52766a269b1da2 commit 354403c148b16c6a99e4175e5a52766a269b1da2 Author: Jose Dapena Paz <jose.dapena@lge.com> Date: Wed Nov 08 07:55:02 2017 Fix ATK object children retrieval if there is a child manager. Desktop Linux Chromium would crash with accessibility enabled if AT-SPI would retrieve the full tree after Chromium was launched (i.e. launching Chromium, then accerciser). The reason was that, while get_n_children would call PlatformChildCount, ref_child would call InternalGetChild. This way we were not taking into account the case of a children manager, that is not retrieved through InternalGetChild. Bug: 781902 Change-Id: I28a23a539fe08699551fd866f2b1aa74f5e40a98 Reviewed-on: https://chromium-review.googlesource.com/757318 Commit-Queue: José Dapena Paz <jose.dapena@lge.com> Reviewed-by: Dominic Mazzoni <dmazzoni@chromium.org> Cr-Commit-Position: refs/heads/master@{#514775} [modify] https://crrev.com/354403c148b16c6a99e4175e5a52766a269b1da2/content/browser/accessibility/browser_accessibility_auralinux.cc
,
Nov 29 2017
|
|||||||||||||||||||||||||
►
Sign in to add a comment |
|||||||||||||||||||||||||
Comment 1 by ligim...@chromium.org
, Nov 6 2017