1089210 - Crash without breakpad on windows93.net

Status: RESOLVED INCOMPLETE

(Core :: Graphics: CanvasWebGL, defect)

Description

[u279076]

Firefox for Android crashes for me without showing the crash reporter when I load windows93.net. It seems to be crashing during the text-mode boot process. I tried loading the same site in Chrome and it loads to a functional GUI.

Comment 1

[Richard Newman [:rnewman]]

Cannot reproduce on current Nightly, HTC One m7. Site loads fine to 'desktop' with audio and all.

What device? What Fennec version? What Android version?

Comment 2

[Richard Newman [:rnewman]]

What does the adb log say?

(An adb log is pretty much mandatory for Android crash bugs, for future reference.)

Comment 3

[u279076]

Device Info:
> Device: Moto G (2nd Generation)
> Android: 4.4.4
> Kernel: 3.4.42-g6c9aef2-00110-gd543d01 hudsoncm@ilclbld53 #1 Thu Jul 24 16:10:12 CDT 2014

Firefox Info:
> Firefox 36.0a1 crashes
> Firefox 35.0a2 crashes
> Firefox 34.0b2 loads the page

Sorry I didn't get you an adb log. I experienced this crash while away from my computer so getting a log was impossible. I'll try to get a log for you asap.

Comment 4

[u279076]

Attachment: adb-motog-20141027.log

Here is the log from the silent crash I'm seeing.

Comment 5

[Aaron Train [:aaronmt]]

I crashed too. I'll try and get a useful stack.

Comment 6

[Aaron Train [:aaronmt]]

Here's where I crash in particular in WebGL land. Also, that site is amazing.

Program received signal SIGSEGV, Segmentation fault.
0x7c39f448 in mozilla::AssertUintParamCorrect (gl=<optimized out>, pname=pname@entry=2968, shadow=4294967295) at /home/aaronmt/Mozilla/Fennec/dom/canvas/WebGLContextUtils.cpp:709
709	      MOZ_ASSERT(false, "Bad cached value.");
(gdb) bt
#0  0x7c39f448 in mozilla::AssertUintParamCorrect (gl=<optimized out>, pname=pname@entry=2968, shadow=4294967295) at /home/aaronmt/Mozilla/Fennec/dom/canvas/WebGLContextUtils.cpp:709
#1  0x7c39fa20 in mozilla::WebGLContext::AssertCachedState (this=0x8c3e8c00) at /home/aaronmt/Mozilla/Fennec/dom/canvas/WebGLContextUtils.cpp:835
#2  0x7c399984 in mozilla::WebGLContext::SetDimensions (this=0x8c3e8c00, sWidth=<optimized out>, sHeight=<optimized out>) at /home/aaronmt/Mozilla/Fennec/dom/canvas/WebGLContext.cpp:945
#3  0x7c69e4d0 in mozilla::dom::HTMLCanvasElement::UpdateContext (this=this@entry=0x8ccf9800, aCx=aCx@entry=0x8ba048a0, aNewContextOptions=..., aNewContextOptions@entry=...)
    at /home/aaronmt/Mozilla/Fennec/content/html/content/src/HTMLCanvasElement.cpp:845
#4  0x7c69ea34 in mozilla::dom::HTMLCanvasElement::GetContext (this=0x8ccf9800, aCx=0x8ba048a0, aContextId=..., aContextOptions=..., rv=...)
    at /home/aaronmt/Mozilla/Fennec/content/html/content/src/HTMLCanvasElement.cpp:757
#5  0x7c13d8e6 in mozilla::dom::HTMLCanvasElementBinding::getContext (cx=0x8ba048a0, obj=..., self=0x8ccf9800, args=...)
    at /home/aaronmt/Mozilla/Fennec/fennec-debug/dom/bindings/HTMLCanvasElementBinding.cpp:215
#6  0x7c3746dc in mozilla::dom::GenericBindingMethod (cx=0x8ba048a0, argc=<optimized out>, vp=<optimized out>) at /home/aaronmt/Mozilla/Fennec/dom/bindings/BindingUtils.cpp:2424
#7  0x7d3c1d84 in js::CallJSNative (cx=0x8ba048a0, native=0x7c37461d <mozilla::dom::GenericBindingMethod(JSContext*, unsigned int, JS::Value*)>, args=...)
    at /home/aaronmt/Mozilla/Fennec/js/src/jscntxtinlines.h:231
#8  0x7d3baa88 in js::Invoke (cx=0x8ba048a0, args=..., construct=js::NO_CONSTRUCT) at /home/aaronmt/Mozilla/Fennec/js/src/vm/Interpreter.cpp:482
#9  0x7d3b22d8 in Interpret (cx=0x8ba048a0, state=...) at /home/aaronmt/Mozilla/Fennec/js/src/vm/Interpreter.cpp:2537
#10 0x7d3ba4ca in js::RunScript (cx=cx@entry=0x8ba048a0, state=...) at /home/aaronmt/Mozilla/Fennec/js/src/vm/Interpreter.cpp:432
#11 0x7d3baaba in js::Invoke (cx=cx@entry=0x8ba048a0, args=..., construct=construct@entry=js::CONSTRUCT) at /home/aaronmt/Mozilla/Fennec/js/src/vm/Interpreter.cpp:501
#12 0x7d3bae30 in js::InvokeConstructor (cx=cx@entry=0x8ba048a0, args=...) at /home/aaronmt/Mozilla/Fennec/js/src/vm/Interpreter.cpp:568
#13 0x7d3baf38 in js::InvokeConstructor (cx=0x8ba048a0, fval=..., argc=1, argv=0x75ef8630, rval=...) at /home/aaronmt/Mozilla/Fennec/js/src/vm/Interpreter.cpp:594
#14 0x7d163166 in js::jit::DoCallFallback (cx=0x8ba048a0, frame=0x75ef86a0, stub_=<optimized out>, argc=1, vp=0x75ef8620, res=...) at /home/aaronmt/Mozilla/Fennec/js/src/jit/BaselineIC.cpp:8703
#15 0x852ce2c8 in ?? ()
#16 0x852ce2c8 in ?? ()

Comment 7

[Kevin Brosnan [Ex-Mozilla]]

[Tracking Requested - why for this release]:Regression

Comment 8

[Richard Newman [:rnewman]]

Anthony: a logcat without the grepping would be helpful. Most of our logs only include "Gecko", and most pernicious crashes don't even have that.

Comment 9

[u279076]

(In reply to Richard Newman [:rnewman] from comment #8)
> Anthony: a logcat without the grepping would be helpful. Most of our logs
> only include "Gecko", and most pernicious crashes don't even have that.

I'm not able to crash anymore. I think they changed something in their code because the boot sequence looks different and I'm not crashing anymore in builds that were easy to crash with earlier.

Comment 10

[Lukas Blakk [:lsblakk] use ?needinfo]

Fwiw, I tried this out on a Z3 Xperia and crashed but did get the crash reporter: https://crash-stats.mozilla.com/report/index/b7d7ab17-b3bc-4629-b49d-83ee22141105

Comment 11

[James Willcox (:snorp) ([e-mail]) (he/him)]

Jeff, the stack in comment 6 looks actionable, even if we can't repro anymore.

Comment 12

[Lukas Blakk [:lsblakk] use ?needinfo]

Untracking since this isn't easily reproducible.

Comment 13

[u279076]

Jeff, I'm guessing you didn't make any progress on this. If not I nominate we just close this bug as incomplete.

Comment 14

[u279076]

I can't reproduce this anymore. Please reopen if you're still able to reproduce.