Open Bug 1840573 Opened 3 years ago Updated 26 days ago

Thunderbird Crash in [@ stackoverflow | style::values::animated::impl$10::to_animated_value] via style::values::generics::color.

Categories

(Thunderbird :: General, defect, P2)

Thunderbird 114
Unspecified
All
defect

Tracking

(thunderbird_esr102 unaffected, thunderbird_esr115 affected, thunderbird_esr128 affected, thunderbird_esr140 affected, thunderbird118 unaffected, thunderbird119 unaffected, thunderbird141 wontfix, thunderbird142 affected)

Tracking Status
thunderbird_esr102 --- unaffected
thunderbird_esr115 --- affected
thunderbird_esr128 --- affected
thunderbird_esr140 --- affected
thunderbird118 --- unaffected
thunderbird119 --- unaffected
thunderbird141 --- wontfix
thunderbird142 --- affected

People

(Reporter: wsmwk, Unassigned, NeedInfo)

References

(Depends on 2 open bugs, Regression)

Details

(Keywords: crash, regression, Whiteboard: [bulkoperations][snnot3p])

Crash Data

All crashes of the last 6 months are Thunderbird. All beta.

Earliest crash is this 114.0b2 with Crash report: https://crash-stats.mozilla.org/report/index/193b143b-9abe-4962-8d15-703ab0230517

Reason: EXCEPTION_STACK_OVERFLOW

Top 10 frames of crashing thread:

0  xul.dll  style::values::animated::impl$10::to_animated_value  servo/components/style/values/animated/mod.rs:290
0  xul.dll  style::values::generics::color::impl$9::to_animated_value  servo/components/style/values/generics/color.rs:16
0  xul.dll  style::values::generics::color::impl$16::to_animated_value<enum2$<style::values::generics::color::GenericColor<style::values::computed::percentage::Percentage> >, style::values::computed::percentage::Percentage>  servo/components/style/values/generics/color.rs:36
1  xul.dll  style::values::animated::impl$10::to_animated_value  servo/components/style/values/animated/mod.rs:290
1  xul.dll  style::values::generics::color::impl$9::to_animated_value  servo/components/style/values/generics/color.rs:16
1  xul.dll  style::values::generics::color::impl$16::to_animated_value<enum2$<style::values::generics::color::GenericColor<style::values::computed::percentage::Percentage> >, style::values::computed::percentage::Percentage>  servo/components/style/values/generics/color.rs:36
2  xul.dll  style::values::animated::impl$10::to_animated_value  servo/components/style/values/animated/mod.rs:290
2  xul.dll  style::values::generics::color::impl$9::to_animated_value  servo/components/style/values/generics/color.rs:16
2  xul.dll  style::values::generics::color::impl$16::to_animated_value<enum2$<style::values::generics::color::GenericColor<style::values::computed::percentage::Percentage> >, style::values::computed::percentage::Percentage>  servo/components/style/values/generics/color.rs:36
3  xul.dll  style::values::animated::impl$10::to_animated_value  servo/components/style/values/animated/mod.rs:290

A more recent crash is bp-7613f696-d398-4734-8ca4-a76780230622

Whiteboard: [snnot3p][tbird crash]

#5 crash for Thunderbird 115.0.
First crash is with 114.0b5

Severity: -- → S2
Priority: -- → P2
Whiteboard: [snnot3p][tbird crash] → [snnot3p][tbird topcrash]

#5 crash for Thunderbird 115.0.

Still #5 for 115.2.0

bp-24aae5b2-8062-4454-b2ef-472f50230821 indicates using quick filter? User comment is "Since the recent upgrade, Thunderbird crashes everytime I search/filter my inbox. I do have a LOT of emails saved, but I used to be able to search/filter with no problem."

Flags: needinfo?(mkmelin+mozilla)

idk. Bug 1839850 had some similarities, so maybe this one also will be gone after 117.0b1

Flags: needinfo?(mkmelin+mozilla)

(In reply to Magnus Melin [:mkmelin] from comment #3)

idk. Bug 1839850 had some similarities, so maybe this one also will be gone after 117.0b1

Last crash for this signature is 117.0b5 bp-1fc1070d-56bb-4cb7-9615-6ffa80230824 2023-08-24 buildid 20230818171547.

But that doesn't help us with version 115.

Crash Signature: [@ stackoverflow | style::values::animated::impl$10::to_animated_value] → [@ stackoverflow | style::values::animated::impl$10::to_animated_value] [@ stackoverflow | style::values::generics::color::impl$16::to_animated_value<T>] [@ stackoverflow | style::values::generics::color::impl$9::to_animated_value]

I believe this is related, submitted by me: bp-11559b9d-4524-4086-9b37-2e3370231103
Crash while compacting Gmail IMAP folder with >25,000 messages marked for deletion.

User comments tend to be bulk message deletions and moves. More signatures:

Crash Signature: [@ stackoverflow | style::values::animated::impl$10::to_animated_value] [@ stackoverflow | style::values::generics::color::impl$16::to_animated_value<T>] [@ stackoverflow | style::values::generics::color::impl$9::to_animated_value] → [@ stackoverflow | style::values::animated::impl$10::to_animated_value] [@ stackoverflow | style::values::generics::color::impl$16::to_animated_value<T>] [@ stackoverflow | style::values::generics::color::impl$9::to_animated_value] [@ <style::values::g…
Summary: Crash in [@ stackoverflow | style::values::animated::impl$10::to_animated_value] → Thunderbird Crash in [@ stackoverflow | style::values::animated::impl$10::to_animated_value]
Whiteboard: [snnot3p][tbird topcrash] → [bulkoperations][snnot3p][tbird topcrash]

(In reply to Rui from comment #5)

I believe this is related, submitted by me: bp-11559b9d-4524-4086-9b37-2e3370231103
Crash while compacting Gmail IMAP folder with >25,000 messages marked for deletion.

Rui, thank you for the crash report. Do you still experience crashes?

Flags: needinfo?(rmssf.pt)

(In reply to Wayne Mery (:wsmwk) from comment #7)

Rui, thank you for the crash report. Do you still experience crashes?

So far that was the only time it crashed but I still didn't reproduce the same conditions at the time of the crash. If needed I can work on reproducing those same conditions to see if it crashes again.
I'm running version 115.6.0 and have not upgraded to 115.6.1 yet. Are you asking because you think this bug may have got fixed in 115.6.1?

Depends on: 1835537

I'm running version 115.6.0 and have not upgraded to 115.6.1 yet. Are you asking because you think this bug may have got fixed in 115.6.1?

I have no specific information about whether it might be fixed for you or not.

Flags: needinfo?(rmssf.pt)
Duplicate of this bug: 1911800

Version 128 signature @ stackoverflow | style::values::animated::impl$11::to_animated_value

Crash Signature: , Percentage> as style::values::animated::ToAnimatedValue>::to_animated_value ] → , Percentage> as style::values::animated::ToAnimatedValue>::to_animated_value ] [@ stackoverflow | style::values::animated::impl$11::to_animated_value ]

bp-18e1a2ed-8734-4760-9be6-76ed50240821 and bp-8d6decf4-c694-4c57-8b81-80e6d0240820 both state "trying to rebuild inbox". But unclear what that means.

See Also: → 1930462

Guessing stackoverflow | style::values::generics::color::impl$26::to_animated_value<T> is related, which is on the rise as version 128 users grow.

Crash Signature: , Percentage> as style::values::animated::ToAnimatedValue>::to_animated_value ] [@ stackoverflow | style::values::animated::impl$11::to_animated_value ] → , Percentage> as style::values::animated::ToAnimatedValue>::to_animated_value ] [@ stackoverflow | style::values::animated::impl$11::to_animated_value ] [@ stackoverflow | style::values::generics::color::impl$26::to_animated_value<T>]
Duplicate of this bug: 1934450
Blocks: 1930462
Duplicate of this bug: 1930462
See Also: 1930462 →

From bug 1840573 comment 0 ...

We have a number of crash signatures where user mentions folder repair, the vast majority are version 128. (None mention or appear to be during shutdown)

These all have style::values::animated::impl$11 on the stack:

bp-ba7b451f-dfa1-4853-872b-21e390241108 stackoverflow | style::values::animated::impl$11::to_animated_value 128.4.2esr (bug 1840573)
bp-1d65dcb1-6697-46e9-b984-acd130241102 stackoverflow | style::values::animated::impl$11::to_animated_value 128.4.0esr (bug 1840573)
bp-0db68589-6baf-4ef8-b651-f4c310241003 stackoverflow | style::values::generics::color::impl$26::to_animated_value<T> 128.2.3esr
bp-33160720-b167-4138-9d56-90aeb0240923 stackoverflow | style::values::animated::impl$11::to_animated_value 128.2.3esr (bug 1840573)
bp-4db92c8b-3754-4ea7-a737-ae2d00240911 stackoverflow | MaybePageAlloc 128.2.0esr (linked to Bug 1930619 and Bug 1933104)
bp-f5ab85a3-1483-4642-83b1-3c47b0240915 stackoverflow | style::values::animated::impl$11::to_animated_value 128.2.0esr (bug 1840573)
bp-6008f186-472b-4984-834f-fa66a0240911 stackoverflow | MaybePageAlloc 128.2.0esr
bp-69501c85-f7b4-4e10-be2b-070220240911 stackoverflow | style::values::generics::color::impl$26::to_animated_value<T> 128.2.0esr
bp-139195bc-d5ce-4c3b-a921-2780b0240823 stackoverflow | MaybePageAlloc 128.1.0esr
These do not have style::values::animated::impl$11 on the stack:

bp-124ab641-ecae-4e48-85ca-0d2f10241106 mozilla::dom::ServiceWorkerRegistrar::GetShutdownPhase 128.4.1esr
bp-df7250c1-db98-4460-80d0-d2af60241101 Detour_VariantClear 128.4.0esr
bp-91dc302b-e8f8-4537-9bb9-5ec7e0241031 mozilla::dom::Console::ShouldProceed 128.4.0esr
bp-84d32a3b-f664-405d-9174-784dd0241028 PL_strncpy | PL_strncat | nsParseMailMessageState::GetAggregateHeader 128.3.3esr
bp-ec88fc75-c22b-485e-a274-fbf380241024 strnlen 128.3.2esr
bp-be30f0f4-1ccf-449f-978b-6f26d0241020 OOM | small 128.3.1esr
bp-dc25d6c8-bcef-456d-af27-f87fc0241026 libxul.so | libnspr4.so 128.3.1esr
bp-1d48ade6-aa0f-4ea4-9695-76c300241014 OOM | small 128.3.0esr
bp-91a3a3ee-e64a-4079-bf38-fe52e0241106 OOM | small 128.2.3esr
Note, bug 1840573 predates version 128, but the crash signatures cited here are on the rise for version 128.

Crash rate perhaps a little lower for esr140.*

Crash reports for 141 and 142 have signature stackoverflow | style::values::generics::color::impl$28::from_animated_value

Like comment 5, many user comments state they were operating on many messages, deletes, moves, etc.

Severity: S2 → S3
Crash Signature: , Percentage> as style::values::animated::ToAnimatedValue>::to_animated_value ] [@ stackoverflow | style::values::animated::impl$11::to_animated_value ] [@ stackoverflow | style::values::generics::color::impl$26::to_animated_value<T>] → , Percentage> as style::values::animated::ToAnimatedValue>::to_animated_value ] [@ stackoverflow | style::values::animated::impl$11::to_animated_value ] [@ stackoverflow | style::values::generics::color::impl$26::to_animated_value<T>] [@ stackoverflow …
OS: Windows 10 → All
Summary: Thunderbird Crash in [@ stackoverflow | style::values::animated::impl$10::to_animated_value] → Thunderbird Crash in [@ stackoverflow | style::values::animated::impl$10::to_animated_value] via style::values::generics::color.
Whiteboard: [bulkoperations][snnot3p][tbird topcrash] → [bulkoperations][snnot3p]
Depends on: 583365

Depends on: 583365

Can you explain your reasoning here? They do not seem to be related at all.

Flags: needinfo?(vseerror)

(In reply to Max Emig [:maxe] from comment #18)

Depends on: 583365

Can you explain your reasoning here? They do not seem to be related at all.

based on user's crash comments, i.e. symptoms

Flags: needinfo?(vseerror)

I'll treat this as the bug for the Thunderbird part and bug 2033796 for the Firefox/Gecko side of things making the potentially stupid CSS crash.

Keywords: regression
Regressed by: 1824454
Depends on: 2033796

Hi all,

Like comment 5, many user comments state they were operating on many messages, deletes, moves, etc.

Yeah, checks out unfortunately. It seems like the primary cause are nesting fill and stroke transitions on the message-list row <img> icons (Starred, Read, Junk).

We could either just wait for Gecko to make it no longer crash, surpress certain animations for bulk actions or make them less fancy in general.

It has confused me at first why i.e. one star is written so many times, however each <tr> is tied to a position in the list, not a concrete message that should be touched like 5 times max in a bulk action.

What do you think?

Thanks

Max

Flags: needinfo?(mkmelin+mozilla)
Flags: needinfo?(geoff)
Flags: needinfo?(alessandro)
Flags: needinfo?(mkmelin+mozilla)

Still transitions shouldn't be interrupted that often in practice? Have you investigated / found the problematic transition?

Uh this is weird, thanks for the ping.
I wouldn't remove that fill and stroke transition because then than the problem of bug 1824454 comes back.
Removing the transition timing entirely would make the experience look a lot less polished.

Is this a problem of the CSS nesting or the fact that we have a transition on the parent and child element?
Should we maybe optimize this by being a bit more specific and limit to images that are visible? (Not sure if that would change anything).
I have the same curiosity raised by Emilio.

Flags: needinfo?(alessandro)

Should we maybe optimize this by being a bit more specific and limit to images that are visible?

My educated guess is that mutating for example the star icon in a half mutated (in transition) state over and over again in in huge batch operation triggers this.

That said, the only proof of that is Socorro comments and summaries of those, I chose to reproduce in Firefox and demonstrate equivalence to the crash signatures.

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