Sorry, I don't have concrete patches I could point to.

Best,
Dominik

On Wed, Dec 18, 2024 at 10:53 AM shiyu zhang <[email protected]>
wrote:

> Hi Dominik,
>
> Unfortunately, there is not an easy way to reproduce this crash. Even in
> our memory-heavy testing scenario, it happens from time to time with a low
> probability.
>
>  Yes, we are using an old version of V8 embedded in Chromium 114. Do you
> have any suspect fix patches that we can port and may help mitigating this
> crash?
>
> On Tuesday, December 17, 2024 at 8:08:59 PM UTC+8 [email protected]
> wrote:
>
>> Hi,
>>
>> from the stack trace alone it is hard to say what's going on here. If you
>> have a repro we would definitely be happy to take a look as we've seen
>> similar crashes occasionally. Note that from your code listing above it
>> looks like you are using some older version of V8. It could be that
>> upgrading to a newer version fixes this crash.
>>
>> Best,
>> Dominik
>>
>> On Tue, Dec 17, 2024 at 12:06 PM shiyu zhang <[email protected]>
>> wrote:
>>
>>> Hi, I am debugging a CPPGC crash happened occasionally in some
>>> memory-heavy scenarios like multiple browser windows with multiple tabs.
>>> According to the assembly code, the crash seems to be caused by an illegal
>>> memory access, whose address value is kGlobalHandleZapValue. I am new to
>>> the CPPGC, does anyone have any clue about the possible reason? Or any
>>> suggestions on where should I start for analyzing this kind of GC crash?
>>>
>>> *Crash stack:*
>>>
>>>  v8::internal::UnifiedHeapMarkingState::MarkAndPush(v8::TracedReferenceBase
>>> const&)
>>>
>>>
>>> /devcloud/ws/s9NlN/workspace/j_BU8C4IRU/src/out/../../v8/src/heap/cppgc-js/unified-heap-marking-state-inl.h:50:7
>>>
>>> v8::internal::UnifiedHeapMarkingVisitorBase::Visit(v8::TracedReferenceBase
>>> const&)
>>>
>>>
>>> /devcloud/ws/s9NlN/workspace/j_BU8C4IRU/src/out/../../v8/src/heap/cppgc-js/unified-heap-marking-visitor.cc:71:31
>>>
>>>
>>>
>>>      x0 = 0x1baffed00baffedf    x1 = 0x0000002003e529f0
>>>
>>>      x2 = 0x0000000000000001    x3 = 0x0000005ac52e19f4
>>>
>>>      x4 = 0x0000005ada7e4218    x5 = 0x0000000000000004
>>>
>>>      x6 = 0x0000000000000000    x7 = 0x0000000000000000
>>>
>>>      x8 = 0x1baffed00bac0000    x9 = 0x000000000000000f
>>>
>>>     x10 = 0x000000000000000f   x11 = 0x0000000000000000
>>>
>>>     x12 = 0x0000000000000000   x13 = 0x0000000000000001
>>>
>>>     x14 = 0x0000000000000001   x15 = 0x0000000000000000
>>>
>>>     x16 = 0x0000005ac502b800   x17 = 0x0000005a3e4b5494
>>>
>>>     x18 = 0x0000000000000000   x19 = 0x1baffed00baffedf
>>>
>>>     x20 = 0x0000005ada7e4218   x21 = 0x00000000000002ee
>>>
>>>     x22 = 0x0000005ad819f0a0   x23 = 0x0000005ada7e4110
>>>
>>>     x24 = 0x0000005ada7e4190   x25 = 0x0000005ada7e41d8
>>>
>>>     x26 = 0x0000005a3ea71f00   x27 = 0x0000000653a70612
>>>
>>>     x28 = 0x0000005ada7e4100    fp = 0x000008637bd05af5
>>>
>>>      lr = 0x0000005ac502b828    sp = 0x0000005ada7e4090
>>>
>>>      pc = 0x0000005ac502b834
>>>
>>>     Found by: given as instruction pointer in context
>>>
>>>
>>> *Assembly (crash occurred in the red line):*
>>>
>>> .text:0000000002B24AAC ;
>>> v8::internal::UnifiedHeapMarkingVisitorBase::Visit(v8::TracedReferenceBase
>>> const&)
>>>
>>> .text:0000000002B24AAC
>>> _ZN2v88internal29UnifiedHeapMarkingVisitorBase5VisitERKNS_19TracedReferenceBaseE
>>>
>>> .text:0000000002B24AAC ; __unwind {
>>>
>>> .text:0000000002B24AAC                 HINT            #0x1B
>>>
>>> .text:0000000002B24AB0                 STP             X30, X23,
>>> [SP,#-0x30]!
>>>
>>> .text:0000000002B24AB4                 STP             X22, X21,
>>> [SP,#0x10]
>>>
>>> .text:0000000002B24AB8                 STP             X20, X19,
>>> [SP,#0x20]
>>>
>>> .text:0000000002B24ABC                 LDR             X22, [X0,#0x10]
>>>
>>> .text:0000000002B24AC0                 LDR             X1, [X1]
>>>
>>> .text:0000000002B24AC4                 CBZ             X1, loc_2B24B74
>>>
>>> .text:0000000002B24AC8                 LDRB            W2, [X22,#0x19]
>>>
>>> .text:0000000002B24ACC                 LDAR            XZR, X0, [X1]
>>>
>>> .text:0000000002B24AD0                 BL
>>> _ZN2v88internal12_GLOBAL__N_110MarkObjectENS0_6ObjectERNS1_10TracedNodeENS0_13TracedHandles8MarkModeE
>>> ; v8::internal::`anonymous
>>> namespace'::MarkObject(v8::internal::Object,v8::internal::`anonymous
>>> namespace'::TracedNode &,v8::internal::TracedHandles::MarkMode)
>>>
>>> .text:0000000002B24AD4                 MOV             X19, X0
>>>
>>> .text:0000000002B24AD8                 TBZ             W19, #0,
>>> loc_2B24B74
>>>
>>> .text:0000000002B24ADC                 AND             X8, X19,
>>> #0xFFFFFFFFFFFC0000
>>>
>>> .text:0000000002B24AE0                 LDRB            W9, [X8,#8]
>>>
>>>
>>> * Value of the suspect address
>>> <https://source.chromium.org/chromium/chromium/src/+/main:v8/src/common/globals.h;drc=984bf90cca4c9577eefea5943afaea9f6e7ac45b;l=980>:*
>>>
>>>     constexpr uint64_t kClearedFreeMemoryValue = 0;
>>>
>>>     constexpr uint64_t kZapValue = uint64_t{0xdeadbeedbeadbeef};
>>>
>>>     constexpr uint64_t kHandleZapValue = uint64_t{0x1baddead0baddeaf};
>>>
>>>     constexpr uint64_t kGlobalHandleZapValue =
>>> uint64_t{0x1baffed00baffedf};
>>>
>>> --
>>> --
>>> v8-dev mailing list
>>> [email protected]
>>> http://groups.google.com/group/v8-dev
>>> ---
>>> You received this message because you are subscribed to the Google
>>> Groups "v8-dev" group.
>>> To unsubscribe from this group and stop receiving emails from it, send
>>> an email to [email protected].
>>> To view this discussion visit
>>> https://groups.google.com/d/msgid/v8-dev/44b72156-7441-444a-8cb8-27ae578f6713n%40googlegroups.com
>>> <https://groups.google.com/d/msgid/v8-dev/44b72156-7441-444a-8cb8-27ae578f6713n%40googlegroups.com?utm_medium=email&utm_source=footer>
>>> .
>>>
>> --
> --
> v8-dev mailing list
> [email protected]
> http://groups.google.com/group/v8-dev
> ---
> You received this message because you are subscribed to the Google Groups
> "v8-dev" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to [email protected].
> To view this discussion visit
> https://groups.google.com/d/msgid/v8-dev/53d95bb2-0ae3-4f12-8e62-1c163ac96902n%40googlegroups.com
> <https://groups.google.com/d/msgid/v8-dev/53d95bb2-0ae3-4f12-8e62-1c163ac96902n%40googlegroups.com?utm_medium=email&utm_source=footer>
> .
>

-- 
-- 
v8-dev mailing list
[email protected]
http://groups.google.com/group/v8-dev
--- 
You received this message because you are subscribed to the Google Groups 
"v8-dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to [email protected].
To view this discussion visit 
https://groups.google.com/d/msgid/v8-dev/CAEf_%3D5Az4A1arkm%2BfeTYSvJrR5_EhUL2q2_Vi2fhXtwmZ-pnPw%40mail.gmail.com.

Reply via email to