Bug 1578429 - firefox crashes at ContextToFP() on ppc64le
Summary: firefox crashes at ContextToFP() on ppc64le
Keywords:
Status: CLOSED DUPLICATE of bug 1498561
Alias: None
Product: Fedora
Classification: Fedora
Component: firefox
Version: 28
Hardware: ppc64le
OS: Linux
unspecified
unspecified
Target Milestone: ---
Assignee: Martin Stransky
QA Contact: Fedora Extras Quality Assurance
URL:
Whiteboard:
Depends On:
Blocks: PPCTracker
TreeView+ depends on / blocked
 
Reported: 2018-05-15 14:54 UTC by Menanteau Guy
Modified: 2018-05-16 11:56 UTC (History)
12 users (show)

Fixed In Version:
Doc Type: If docs needed, set a value
Doc Text:
Clone Of:
Environment:
Last Closed: 2018-05-16 11:56:14 UTC
Type: Bug
Embargoed:


Attachments (Terms of Use)
gdb backtrace (4.54 KB, text/plain)
2018-05-15 14:54 UTC, Menanteau Guy
no flags Details

Description Menanteau Guy 2018-05-15 14:54:27 UTC
Created attachment 1436812 [details]
gdb backtrace

Unable to start firefox-60.0-4 on a qemu ppc64le machine.

I installed a f27 fedora qemu ppc64le machine with last updates. When I start a
firefox-60.0-4 I get a Segmentation fault.
Same problem on f28.

on the console I have:
[363497.629997] firefox[32007]: unhandled signal 11 at 0000000000000000 nip 000000010000d514 lr 000000010000d790 code 1

when I use gdb (I run firefox thru a shh -X session), I do:
gdb -tui /usr/lib64/firefox/firefox
(gdb) run --no-remote

Program received signal SIGSEGV, Segmentation fault.
RedBlackTree<arena_chunk_map_t, ArenaAvailTreeTrait>::TreeNode::SetColor (
    aColor=Red, this=<synthetic pointer>)
    at /usr/src/debug/firefox-60.0-4.f27.ppc64le/memory/build/rb.h:203

   |192         NodeColor Color()                                              │
   │193         {                                                              │
   │194           return mNode ? Trait::GetTreeNode(mNode).Color() : NodeColor:│
   │195         }                                                              │
   │196                                                                        │
   │197         bool IsRed() { return Color() == NodeColor::Red; }             │
   │198                                                                        │
   │199         bool IsBlack() { return Color() == NodeColor::Black; }         │
   │200                                                                        │
   │201         void SetColor(NodeColor aColor)                                │
   │202         {                                                              │
  >│203           MOZ_RELEASE_ASSERT(mNode);                                   │
   │204           Trait::GetTreeNode(mNode).SetColor(aColor);                  │
   │205         }                                                              │
   │206                                                                        │
   │207         T* Get() { return mNode; }                                     │
   │208                                                                        │
   │209         MOZ_IMPLICIT operator bool() { return !!mNode; }               │
   │210                                                                        │
   │211         bool operator==(TreeNode& aOther) { return mNode == aOther.mNod│
   │212                                                                        │
   │213       private:                                                         │
   │214         T* mNode;                                                      │
   │215       };

Comment 1 Martin Stransky 2018-05-16 10:15:13 UTC
Yes, that's because of jemalloc. You can try jemalloc disabled builds:

https://koji.fedoraproject.org/koji/taskinfo?taskID=26989530

Also there's a crash at js/src/wasm/WasmSignalHandlers.cpp, ContextToPC() does not have handler for ppc64le and other arches here.

bt:
#0  0x00003fffb1f49edc in ContextToPC(ucontext_t*) (context=0x3fffffff5a60)
    at /home/komat/rpmbuild/BUILD/firefox-60.0/js/src/wasm/WasmSignalHandlers.cpp:441
#1  0x00003fffb1f4acf4 in RedirectJitCodeToInterruptCheck(JSContext*, ucontext_t*) (cx=0x1004b37f0, context=0x3fffffff5a60)
    at /home/komat/rpmbuild/BUILD/firefox-60.0/js/src/wasm/WasmSignalHandlers.cpp:1553
#2  0x00003fffb1f4aeac in JitInterruptHandler(int, siginfo_t*, void*) (signum=26, info=0x3fffffff67d8, context=0x3fffffff5a60)
    at /home/komat/rpmbuild/BUILD/firefox-60.0/js/src/wasm/WasmSignalHandlers.cpp:1601
#3  0x00003fffb7f90478 in <signal handler called> () at arch/powerpc/kernel/vdso64/sigtramp.S
#4  0x00003fffb136e7a8 in js::detail::DefineComparisonOps<js::PreBarriered<jsid> >::get(js::PreBarriered<jsid> const&) (v=...)
    at /home/komat/rpmbuild/BUILD/firefox-60.0/js/src/gc/Barrier.h:977
#5  0x00003fffb13382cc in operator==<js::PreBarriered<jsid> >(js::PreBarriered<jsid> const&, js::PreBarriered<jsid>::ElementType const&) (a=..., b=...) at /home/komat/rpmbuild/BUILD/firefox-60.0/objdir/dist/include/js/RootingAPI.h:1541
#6  0x00003fffb1278648 in js::Shape::searchLinear(jsid) (this=0x3fff5f62ba10, id=...)
    at /home/komat/rpmbuild/BUILD/firefox-60.0/js/src/vm/Shape.h:1623
#7  0x00003fffb1a1bb14 in js::Shape::searchNoHashify(js::Shape*, jsid) (start=0x3fff5f62ba10, id=...)
    at /home/komat/rpmbuild/BUILD/firefox-60.0/js/src/vm/Shape-inl.h:391
#8  0x00003fffb1a55c6c in js::NativeObject::lookupPure(jsid) (this=0x3fff99e8e120, id=...)
    at /home/komat/rpmbuild/BUILD/firefox-60.0/js/src/vm/NativeObject.cpp:289
#9  0x00003fffb0ea07c4 in js::NativeObject::lookupPure(js::PropertyName*) (this=0x3fff99e8e120, name=0x3fff99e28640)
    at /home/komat/rpmbuild/BUILD/firefox-60.0/js/src/vm/NativeObject.h:836
#10 0x00003fffb0eaa050 in js::GlobalObject::maybeGetIntrinsicValue(JSContext*, JS::Handle<js::GlobalObject*>, JS::Handle<js::PropertyName*>, JS::MutableHandle<JS::Value>, bool*) (cx=0x1004b37f0, global=..., name=..., vp=..., exists=0x3fffffff6ccf)
    at /home/komat/rpmbuild/BUILD/firefox-60.0/js/src/vm/GlobalObject.h:711
#11 0x00003fffb0eaa13c in js::GlobalObject::getIntrinsicValue(JSContext*, JS::Handle<js::GlobalObject*>, JS::Handle<js::PropertyName*>, JS::MutableHandle<JS::Value>) (cx=0x1004b37f0, global=..., name=..., value=...)
    at /home/komat/rpmbuild/BUILD/firefox-60.0/js/src/vm/GlobalObject.h:726
#12 0x00003fffb0eb4da0 in js::GetIntrinsicOperation(JSContext*, unsigned char*, JS::MutableHandle<JS::Value>) (cx=0x1004b37f0, pc=0x10065b718 "\217\001", vp=...) at /home/komat/rpmbuild/BUILD/firefox-60.0/js/src/vm/Interpreter-inl.h:293
#13 0x00003fffb0ed1028 in Interpret(JSContext*, js::RunState&) (cx=0x1004b37f0, state=...)
    at /home/komat/rpmbuild/BUILD/firefox-60.0/js/src/vm/Interpreter.cpp:3237
#14 0x00003fffb0ebcf98 in js::RunScript(JSContext*, js::RunState&) (cx=0x1004b37f0, state=...)


406	#if defined(_M_X64) || defined(__x86_64__)
407	# define PC_sig(p) RIP_sig(p)
408	# define FP_sig(p) RBP_sig(p)
409	# define SP_sig(p) RSP_sig(p)
410	#elif defined(_M_IX86) || defined(__i386__)
411	# define PC_sig(p) EIP_sig(p)
412	# define FP_sig(p) EBP_sig(p)
413	# define SP_sig(p) ESP_sig(p)
414	#elif defined(__arm__)
415	# define FP_sig(p) R11_sig(p)
416	# define SP_sig(p) R13_sig(p)
417	# define LR_sig(p) R14_sig(p)
418	# define PC_sig(p) R15_sig(p)
419	#elif defined(__aarch64__)
420	# define PC_sig(p) EPC_sig(p)
421	# define FP_sig(p) RFP_sig(p)
422	# define SP_sig(p) R31_sig(p)
423	# define LR_sig(p) RLR_sig(p)
424	#elif defined(__mips__)
425	# define PC_sig(p) EPC_sig(p)
426	# define FP_sig(p) RFP_sig(p)
427	# define SP_sig(p) RSP_sig(p)
428	# define LR_sig(p) R31_sig(p)
429	#endif

Missing other arches definitions.

430	
431	#if defined(PC_sig) && defined(FP_sig) && defined(SP_sig)
432	# define KNOWS_MACHINE_STATE
433	#endif
434	
445	static uint8_t*
446	ContextToFP(CONTEXT* context)
447	{
448	#ifdef KNOWS_MACHINE_STATE
449	    return reinterpret_cast<uint8_t*>(FP_sig(context));
450	#else
451	    MOZ_CRASH(); <<<
452	#endif
453	}

Comment 2 Menanteau Guy 2018-05-16 11:37:55 UTC
The bug describe in comment 1 is more relative to the bug #1498561. I just updated it.

Comment 3 Menanteau Guy 2018-05-16 11:40:53 UTC
I will try with patch of bug #1498561 and jemalloc disabled. Thanks for the info.

Comment 4 Martin Stransky 2018-05-16 11:56:14 UTC
Let's track it at Bug 1498561.

*** This bug has been marked as a duplicate of bug 1498561 ***


Note You need to log in before you can comment on or make changes to this bug.