New issue
Advanced search Search tips

Issue 881137 link

Starred by 1 user

Issue metadata

Status: Duplicate
Owner:
Closed: Oct 9
Cc:
Components:
EstimatedDays: ----
NextAction: 2018-09-15
OS: Mac
Pri: 1
Type: Bug



Sign in to add a comment

Follow-up on 10.12-specific crashes for 27__CFPasteboardDeallocate

Project Member Reported by tapted@chromium.org, Sep 6

Issue description

Chrome Version       : 70.0.3534.4
OS Version: Mac 10.12

Follow-up to Issue 877979 to ensure the crashes are properly dealt with.

"""
nothing appears on m71 or m70 beta/stable channels. That CL will change stack signatures, so let's use this link (27__CFPasteboardDeallocate): 

https://crash.corp.google.com/browse?q=EXISTS+%28SELECT+1+FROM+UNNEST%28CrashedStackTrace.StackFrame%29+WHERE+FunctionName%3D%27__CFPasteboardDeallocate%27%29#-propertyselector,samplereports,productname:1000,+productversion,magicsignature:50,-magicsignature2:50,-stablesignature:50,-magicsignaturesorted:50

there are actually 5 magic signatures that are hit, this bug is just one of them (at 23%)

more are actually under -[WebDragSource startDrag] (71%), and that one is *only* on 10.13, with no crashes on 10.12 :/. There's an old bug -   Issue 509652   - I guess that regressed in 10.13. Filed Issue 879412.

Restricting to 10.12:

https://crash.corp.google.com/browse?q=EXISTS+%28SELECT+1+FROM+UNNEST%28CrashedStackTrace.StackFrame%29+WHERE+FunctionName%3D%27__CFPasteboardDeallocate%27%29+AND+expanded_custom_data.ChromeCrashProto.os_family%3D%2710.12+%28Sierra%29%27#-propertyselector,productname:1000,+magicsignature:50,-magicsignature2:50,-stablesignature:50,operatingsystem,-magicsignaturesorted:50

that shows the crashes also coming from -[CocoaDragDropDataProvider .cxx_destruct], but they went away in m67.
"""
 
Components: -Blink>DataTransfer -UI>Browser>TabStrip UI>Browser>Downloads
The NextAction date has arrived: 2018-09-15
Labels: -Pri-3 Pri-1
still nothing since 70.0.3535.0 

but there are crashes on m69 stable, now with a new stack:
 - https://crash.corp.google.com/browse?q=expanded_custom_data.ChromeCrashProto.magic_signature_1.name%3D%27DragDownloadItem%27

So the merge didn't fix anything, but it has allowed us to confirm this is still a problem a lot sooner than we would otherwise have been able to.

0x00007fffb010ee89	(libdispatch.dylib + 0x00008e89 )	_dispatch_semaphore_dispose
0x00007fffb0108854	(libdispatch.dylib + 0x00002854 )	_dispatch_dispose
0x00007fff9aa71c36	(CoreFoundation + 0x0014dc36 )	____CFPasteboardDeallocate_block_invoke
0x00007fff9aa68bf6	(CoreFoundation + 0x00144bf6 )	__CFPasteboardCacheAccessSandboxExtensionState
0x00007fff9aa324f6	(CoreFoundation + 0x0010e4f6 )	__CFPasteboardDeallocate
0x00007fff9aa791a2	(CoreFoundation + 0x001551a2 )	_CFRelease
0x00007fff98b66299	(AppKit + 0x00708299 )	-[NSPasteboard dealloc]
0x00007fffaf84c3bc	(libobjc.A.dylib + 0x0000a3bc )	(anonymous namespace)::AutoreleasePoolPage::pop(void*)
0x00007fff9a96a025	(CoreFoundation + 0x00046025 )	_CFAutoreleasePoolPop
0x00007fff98743b89	(AppKit + 0x002e5b89 )	NSCoreDragTrackingProc
0x00007fff9967fbbe	(HIServices + 0x0000bbbe )	DoTrackingMessage
0x00007fff99680df7	(HIServices + 0x0000cdf7 )	SendTrackingMessage
0x00007fff99680028	(HIServices + 0x0000c028 )	DragInApplication
0x00007fff9967f0b2	(HIServices + 0x0000b0b2 )	CoreDragStartDragging
0x00007fff9874292b	(AppKit + 0x002e492b )	-[NSCoreDragManager _dragUntilMouseUp:accepted:]
0x00007fff9873f8b5	(AppKit + 0x002e18b5 )	-[NSCoreDragManager dragImage:fromWindow:at:offset:event:pasteboard:source:slideBack:]
0x00007fff9873f3fb	(AppKit + 0x002e13fb )	-[NSWindow(NSDrag) dragImage:at:offset:event:pasteboard:source:slideBack:]
0x0000000115277c96	(Google Chrome Framework -download_item_drag_mac.mm:41 )	DragDownloadItem(download::DownloadItem const*, gfx::Image*, NSView*)
0x00000001153f983d	(Google Chrome Framework -download_item_view.cc:433 )	DownloadItemView::OnMouseDragged(ui::MouseEvent const&)
0x0000000114399b52	(Google Chrome Framework -view.cc:2466 )	views::View::ProcessMouseDragged(ui::MouseEvent const&)
0x0000000114399739	(Google Chrome Framework -view.cc:1105 )	views::View::OnMouseEvent(ui::MouseEvent*)
0x000000011431fb18	(Google Chrome Framework -ink_drop_host_view.cc:279 )	views::InkDropHostView::OnMouseEvent(ui::MouseEvent*)
0x0000000113552df9	(Google Chrome Framework -event_handler.cc:32 )	ui::EventHandler::OnEvent(ui::Event*)
0x000000011355278d	(Google Chrome Framework -event_dispatcher.cc:191 )	ui::EventDispatcher::ProcessEvent(ui::EventTarget*, ui::Event*)
0x00000001135525c3	(Google Chrome Framework -event_dispatcher.cc:86 )	ui::EventDispatcherDelegate::DispatchEvent(ui::EventTarget*, ui::Event*)
0x00000001143a382e	(Google Chrome Framework -root_view.cc:426 )	views::internal::RootView::OnMouseDragged(ui::MouseEvent const&)
0x00000001143a8c32	(Google Chrome Framework -widget.cc:1242 )	views::Widget::OnMouseEvent(ui::MouseEvent*)
0x000000011432ea2c	(Google Chrome Framework -bridged_content_view.mm:703 )	-[BridgedContentView mouseEvent:]
0x00000001143354a1	(Google Chrome Framework -cocoa_mouse_capture.mm:91 )	___ZN5views17CocoaMouseCapture14ActiveEventTap4InitEv_block_invoke
0x00007fff986257f9	(AppKit + 0x001c77f9 )	_NSSendEventToObservers
0x00007fff98c1e23e	(AppKit + 0x007c023e )	-[NSApplication(NSEvent) sendEvent:]
0x000000011272f73b	(Google Chrome Framework -chrome_browser_application_mac.mm:328 )	__34-[BrowserCrApplication sendEvent:]_block_invoke
0x0000000112b0a279	(Google Chrome Framework + 0x02296279 )	base::mac::CallWithEHFrame(void () block_pointer)
0x000000011272f3e6	(Google Chrome Framework -chrome_browser_application_mac.mm:311 )	-[BrowserCrApplication sendEvent:]
0x00007fff98499426	(AppKit + 0x0003b426 )	-[NSApplication run]

Cc: a...@chromium.org
Interestingly.. that signature points to Issue 883360 in the crash reporter, which suggests a fix in 70.0.3536.0 (r586903).

So.. I think it's likely that avi has fixed the root cause of this already, but only for m70.
Mergedinto: 883360
Status: Duplicate (was: Assigned)
Last crash in 27__CFPasteboardDeallocate is still 70.0.3535.0 which was ~40 days ago. All still 10.12-only. I think we are out of the woods here.

Sign in to add a comment