Bug 209171

Summary: Crash on big-endian arches in JSC::VM::VM()
Product: WebKit Reporter: Michael Gratton <mike>
Component: JavaScriptCoreAssignee: Nobody <webkit-unassigned>
Status: RESOLVED DUPLICATE    
Severity: Normal CC: berto, cgarcia, clopez, dkg, mcatanzaro
Priority: P3 Keywords: Gtk
Version: Other   
Hardware: PC   
OS: Linux   
Attachments:
Description Flags
gdb bt full none

Description Michael Gratton 2020-03-16 22:50:22 PDT 
Created attachment 393732 [details]
gdb bt full

Geary's JS tests are failing on Debian package builds with a runtime crash of the WebProcess, for an assortment of big-endian architectures - at least ppc64 and it has also been seen on s390x. After obtaining a stack, it looks like a WebKit/JSC bug (not sure if WebKitGTK specific), since it is failing when constructing a VM in JSC::VM::VM()

See https://gitlab.gnome.org/GNOME/geary/issues/737 for the downstream bug and details. Back trace is attached.
Comment 1 Alberto Garcia 2020-03-17 03:19:03 PDT 
From the backtrace it seems that the mprotect() call in JSC::Config::permanentlyFreeze() is failing.

I assume that this is WebKitGTK 2.28.0 ?
Comment 2 Michael Gratton 2020-03-17 03:30:25 PDT 
> I assume that this is WebKitGTK 2.28.0 ?

Looks like 2.26 from the build logs. Not sure if Debian has packaged 2.28 yet. It's an issue that has been present for a number of versions though.
Comment 3 Alberto Garcia 2020-03-17 03:40:11 PDT 
Yes, 2.28 is in Debian already (in testing/unstable and also in stable via backports)
Comment 4 Daniel Kahn Gillmor 2020-03-17 05:51:21 PDT 
i'm the person who generated the backtrace.

It was on a ppc64 machine (perotto.debian.net), and it was in a context with the debian webkit package 2.28.0-2 installed.

If anyone has any suggestions on how to isolate the problem so that there isn't so much geary involved, i'm happy to try to replicate it more narrowly.

This is also noted in the debian BTS as https://bugs.debian.org/954026
Comment 5 Alberto Garcia 2020-03-23 06:25:54 PDT 
I understand that this is a duplicate of bug 209236 , correct ?
Comment 6 Carlos Alberto Lopez Perez 2020-03-23 06:50:01 PDT 
(In reply to Alberto Garcia from comment #5)
> I understand that this is a duplicate of bug 209236 , correct ?

Yes.

*** This bug has been marked as a duplicate of bug 209236 ***
Comment 7 Michael Catanzaro 2020-03-23 07:26:41 PDT 
Note this crash is related to page size; it has nothing to do with endianness.