1738381 - Crash in [@ OOM | large | mozalloc_abort | webrender::renderer::Renderer::render_impl]

Status: NEW

(Core :: Graphics: WebRender, defect)

Description

[Andrew McCreight [:mccr8]]

Crash report: https://crash-stats.mozilla.org/report/index/39eaf244-cfae-418d-8343-c66fb0211028

MOZ_CRASH Reason: MOZ_CRASH()

Top 10 frames of crashing thread:

0 mozglue.dll mozalloc_abort memory/mozalloc/mozalloc_abort.cpp:33
1 mozglue.dll mozalloc_handle_oom memory/mozalloc/mozalloc_oom.cpp:51
2 xul.dll mozglue_static::oom_hook::hook mozglue/static/rust/lib.rs:115
3 xul.dll std::alloc::rust_oom ../c8dfcfe046a7680554bf4eb612bad840e7631c4b//library/std/src/alloc.rs:329
4 xul.dll alloc::alloc::__alloc_error_handler::__rg_oom ../c8dfcfe046a7680554bf4eb612bad840e7631c4b//library/alloc/src/alloc.rs:398
5 xul.dll __rust_alloc_error_handler 
6 xul.dll webrender::renderer::Renderer::render_impl gfx/wr/webrender/src/renderer/mod.rs:1960
7 xul.dll webrender::renderer::Renderer::render gfx/wr/webrender/src/renderer/mod.rs:1723
8 xul.dll webrender_bindings::bindings::wr_renderer_render gfx/webrender_bindings/src/bindings.rs:622
9 xul.dll mozilla::wr::RenderThread::UpdateAndRender gfx/webrender_bindings/RenderThread.cpp:501

OOM size of around 1MB.

Comment 1

[Sebastian Hengst [:aryx] (needinfo me if it's about an intermittent or backout)]

This was topcrash #3 for 95.0b9 and has 60-180 affected installations per beta release.

Comment 2

[Glenn Watson [:gw]]

Removed callstack, pasted in incorrect bug.

Comment 3

[Jeff Muizelaar [:jrmuizel]]

A lot of the crashes are on buggy AMD CPUs: https://crash-stats.mozilla.org/report/index/a4a285a4-43a8-4f4e-ab4b-82a920211202

Comment 4

[Jim Mathies [:jimm]]

Overall crash rates are constant in 95 beta. Stalled, we're not sure what the cause is. Maybe be related to a buggy amd cpu.

Comment 5

[Chris Peterson [:cpeterson]]

This WebRender OOM spiked in Firefox 99.

Comment 6

[Wayne Mery (:wsmwk)]

(In reply to Chris Peterson [:cpeterson] from comment #5)

This WebRender OOM spiked in Firefox 99.

Roughly double - at a rate that exists through today.

Comment 7

[Jeff Muizelaar [:jrmuizel]]

Still a lot on AMD not much on Nightly

Comment 8

[Worcester12345]

https://crash-stats.mozilla.org/report/index/52d42d02-0b69-4c1f-8ea4-36f5c0230310

Comment 9

[Yannis Juglaret [:yannis]]

The crash volume here suggests that we may have successfully mitigated this crash in Release a while ago, and only ESR would now be largely impacted by this. Do we have any idea what may have caused this? Any potential uplift?

Comment 10

[Glenn Watson [:gw]]

I can't think of anything we've landed recently that would have fixed this. However, being an OOM, it might be that it's no longer occurring due to memory related optimizations of fixes outside of WR itself.

Comment 11

[Wayne Mery (:wsmwk)]

End of February, OOM | large | mozalloc_abort | webrender::renderer::Renderer::render_impl
changed to OOM | large | webrender::renderer::Renderer::render_impl ? (both Firefox only crashes)

Comment 12

[Glenn Watson [:gw]]

Is there any way we can get an about:memory report from a system experiencing this prior to OOM? It's possible there is a memory leak, though it's more likely just a general OOM issue that happens to hit in WR rather than WR-specific.

Comment 13

[Andrew McCreight [:mccr8]]

We do collect anonymized memory reports and put them in crash reports, but I didn't see any under these signatures, at least by looking to see if they had a "memory resident" or "memory explicit" field.