2024-03-28 19:53 UTC

View Issue Details Jump to Notes ]
IDProjectCategoryView StatusLast Update
0002682NetSurfABENDpublic2020-05-23 14:30
ReporterDave Higton 
Assigned To 
SeveritycrashReproducibilityrandom 
StatusacknowledgedResolutionopen 
PlatformBeagleBoard xMOSRISC OSOS Version5.27 (21 Jul 19)
Product Version 
Target VersionFixed in Version 
Summary0002682: Occasional crash
DescriptionFor some months now, NetSurf has occasionally crashed. I can't discern any pattern to it, and it's not specific to any web site or page.
Steps To ReproduceKeep using it. It will crash, somewhere, sooner or later.
TagsNo tags attached.
Fixed in CI build #
Reported in CI build #4720
URL of problem page
Attached Files

-Relationships
has duplicate 0002716acknowledged NetSurf fatal error on the down ROOL site 
has duplicate 0002717acknowledged Segfault on GWR site 
+Relationships

-Notes
Dave Higton

~0002061

Dave Higton (developer)

Just had another one, apparently at the same location.
Dave Higton

~0002083

Dave Higton (developer)

And another, this time with CI #4835.
Dave Higton

~0002132

Dave Higton (developer)

Last edited: 2019-11-26 21:54

View 2 revisions

And another, this time with CI #4900.

If I understand the logs correctly, they show that the segfault always occurs in function sh_delentry(), by attempting to store R6 at [R5,#36] when R5 is 0. Isn't this just a classic null pointer problem?

Daniel Silverstone

~0002145

Daniel Silverstone (administrator)

Certainly that sounds possible Dave. The new libcurl has changed shape slightly so we need to wait for something new to show up before I can look again. I think it's inside the CURL code associated with cleaning up socket filehandles from its hash table.
Dave Higton

~0002146

Dave Higton (developer)

https://github.com/curl/curl/issues/3995

The lowest 14 levels of the call tree are identical. So other people are seeing the same problem in completely different contexts. But, from your comments, I see you're ahead of me!
Dave Higton

~0002198

Dave Higton (developer)

I haven't seen this crash in a long time now. Can we assume that it's been fixed by some change in cURL?
+Notes

-Issue History
Date Modified Username Field Change
2019-07-25 09:29 Dave Higton New Issue
2019-07-25 09:29 Dave Higton File Added: CrashDump
2019-08-01 09:22 Vincent Sanders Status new => acknowledged
2019-08-03 21:12 Dave Higton File Added: CrashDump-2
2019-08-03 21:12 Dave Higton Note Added: 0002061
2019-08-20 18:46 Dave Higton File Added: CrashDump-3
2019-08-20 18:46 Dave Higton Note Added: 0002083
2019-11-25 21:33 Vincent Sanders Relationship added has duplicate 0002716
2019-11-25 21:33 Vincent Sanders Relationship added has duplicate 0002717
2019-11-26 21:54 Dave Higton File Added: CrashDump-4
2019-11-26 21:54 Dave Higton Note Added: 0002132
2019-11-26 21:54 Dave Higton Note Edited: 0002132 View Revisions
2019-11-30 20:55 Daniel Silverstone Note Added: 0002145
2019-11-30 22:46 Dave Higton Note Added: 0002146
2020-05-23 14:30 Dave Higton Note Added: 0002198
+Issue History