If you look at the OOM handler of Windows version
<https://source.chromium.org/chromium/chromium/src/+/main:base/allocator/partition_allocator/src/partition_alloc/oom.cc;l=30;drc=c1b9831c8c232ab9470645977a18527cfa7bb993;bpv=1;bpt=1>
in PartitionAlloc we do not debug alias the size of allocation. So you may
not see the size that was failing in the minidump.  Have you had any
success in getting a histogram of the sizes of the allocations that are
failing from those minidumps? Could it be a bug in the TextEncoding that is
asking for a very large allocation size?
Also there is CrashMemoryMetricsCollector
<https://source.chromium.org/chromium/chromium/src/+/main:components/crash/content/browser/crash_memory_metrics_collector_android.h;l=19>
on
Android that you could expose on desktop to collect any additional
histograms via shared memory (ie. I don't know if a histogram of the size
of allocation that failed might be useful or not).

Perhaps the partition alloc folks have any ideas that additional reasons
could be logged in crash keys or in the shared memory crash metrics.

dave.

On Fri, Aug 2, 2024 at 6:22 AM Yoav Weiss (@Shopify) <yoavwe...@chromium.org>
wrote:

>
>
> On Fri, Aug 2, 2024 at 11:40 AM Adam Rice <ri...@chromium.org> wrote:
>
>> Usually specs don't cover what happens when you run out of memory, as
>> implied by https://infra.spec.whatwg.org/#algorithm-limits. I think
>> this is fine. I'm interested in what other browsers do, but it's hard to
>> test unless you have a VM handy.
>>
>
> This section indicates that algorithms shouldn't impose specific limits,
> but I don't think that implies that there shouldn't be standard behavior
> when those limits are (naturally) met. An example of that is the
> QuotaExceededError
> <https://webidl.spec.whatwg.org/#dom-domexception-quota_exceeded_err>
> exception. We expect developers to be able to deal with different quota
> levels in different browsers, but the exception they get when they hit the
> limit is the same everywhere.
>
> IMO, it may make sense to define an exception that developers can expect
> in those cases. (maybe depending on what other browsers are doing in that
> case)
>
>
>> On Fri, 2 Aug 2024 at 01:17, Mike Taylor <miketa...@chromium.org> wrote:
>>
>>> Hi there,
>>>
>>> Have we done any sort of web compatibility analysis of what this change
>>> implies? A broken page might be a better choice than a crashed tab, but
>>> it's hard to know without any sense of the potential impact of this change.
>>> Also, is there a plan to specify this behavior? What's the interop
>>> situation?
>>>
>>> thanks,
>>> Mike
>>> On 8/1/24 4:38 AM, 'xu ms' via blink-dev wrote:
>>>
>>> *Contact emails*: xuzha...@microsoft.com
>>>
>>> *Summary:*
>>>
>>> We are currently observing many renderer crashes occurring in text
>>> encode.Encoding Standard (whatwg.org)
>>> <https://encoding.spec.whatwg.org/>
>>> This is because DOMArrayBuffer::Create is currently used to create a
>>> buffer, and when memory allocation fails, renderer process crashes. The
>>> reasons for memory allocation failure are unclear and not solely due to
>>> allocating excessively large memory.
>>>
>>> Therefore, we want to change the logic here so that when memory
>>> allocation fails, a DOMException is thrown from text encode instead of
>>> crashing.
>>>
>>> *Blink component*: Blink>TextEncoding
>>> <https://issues.chromium.org/issues?q=customfield1222907:%22Blink%3ETextEncoding%22>
>>>
>>> *Tracking bug*:[OOM] Is it a real OOM for
>>> blink::DOMArrayBuffer::Create? [355018938] - Chromium
>>> <https://issues.chromium.org/issues/355018938>
>>>
>>>
>>> --
>>> You received this message because you are subscribed to the Google
>>> Groups "blink-dev" group.
>>> To unsubscribe from this group and stop receiving emails from it, send
>>> an email to blink-dev+unsubscr...@chromium.org.
>>> To view this discussion on the web visit
>>> https://groups.google.com/a/chromium.org/d/msgid/blink-dev/f4cfdc62-707f-4d21-80d5-43ed37ce52fan%40chromium.org
>>> <https://groups.google.com/a/chromium.org/d/msgid/blink-dev/f4cfdc62-707f-4d21-80d5-43ed37ce52fan%40chromium.org?utm_medium=email&utm_source=footer>
>>> .
>>>
>>> --
>>> You received this message because you are subscribed to the Google
>>> Groups "blink-dev" group.
>>> To unsubscribe from this group and stop receiving emails from it, send
>>> an email to blink-dev+unsubscr...@chromium.org.
>>> To view this discussion on the web visit
>>> https://groups.google.com/a/chromium.org/d/msgid/blink-dev/c6b00d44-599f-47bf-be3d-9e977681b827%40chromium.org
>>> <https://groups.google.com/a/chromium.org/d/msgid/blink-dev/c6b00d44-599f-47bf-be3d-9e977681b827%40chromium.org?utm_medium=email&utm_source=footer>
>>> .
>>>
>> --
>> You received this message because you are subscribed to the Google Groups
>> "blink-dev" group.
>> To unsubscribe from this group and stop receiving emails from it, send an
>> email to blink-dev+unsubscr...@chromium.org.
>> To view this discussion on the web visit
>> https://groups.google.com/a/chromium.org/d/msgid/blink-dev/CAC_ixdwDfGZQgUNP7HSkU03heC8VG2Zy8fqhJJWzxDerV1i8zA%40mail.gmail.com
>> <https://groups.google.com/a/chromium.org/d/msgid/blink-dev/CAC_ixdwDfGZQgUNP7HSkU03heC8VG2Zy8fqhJJWzxDerV1i8zA%40mail.gmail.com?utm_medium=email&utm_source=footer>
>> .
>>
> --
> You received this message because you are subscribed to the Google Groups
> "blink-dev" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to blink-dev+unsubscr...@chromium.org.
> To view this discussion on the web visit
> https://groups.google.com/a/chromium.org/d/msgid/blink-dev/CAOmohS%2Bcb1M4gjXTbJ-Hyv3DDpBbxFU6-U4gyZZZnxmOffvqaA%40mail.gmail.com
> <https://groups.google.com/a/chromium.org/d/msgid/blink-dev/CAOmohS%2Bcb1M4gjXTbJ-Hyv3DDpBbxFU6-U4gyZZZnxmOffvqaA%40mail.gmail.com?utm_medium=email&utm_source=footer>
> .
>

-- 
You received this message because you are subscribed to the Google Groups 
"blink-dev" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to blink-dev+unsubscr...@chromium.org.
To view this discussion on the web visit 
https://groups.google.com/a/chromium.org/d/msgid/blink-dev/CAHgVhZVw3HMDCx%3D43DP5TT4%3DmA_49EJhUd4c9%3DcJ0i-tEE22UA%40mail.gmail.com.

Reply via email to