Bug #1705

shmat on x86_64 causes SIGSEGV in x11/gtk2

Added by davshao over 4 years ago. Updated over 4 years ago.

Status:ClosedStart date:
Priority:NormalDue date:
Assignee:-% Done:

0%

Category:-
Target version:-

Description

To reproduce, on an x86_64 machine (tested on a Shuttle SG45H7 and an Intel Core
2 Duo Macbook 2.1), from a current DragonFly v2.5.1 1061.ge9fe2 X86_64_GENERIC
build, install modular xorg and gtk+-2.18.9 from current pkgsrc, and run
git-demo which SIGSEVs. No such SIGSEGV occurs on an i386 machine.

Attached is a gdb backtrace. Of interest are gtk2 GdkImage, in this case,
parameter image, a struct that has a field mem that is if shm is enabled on
x86_64 produces an

invalid address 0x8400000: Cannot access memory at address 0x8400000

Actually image_info also has a field that points to a list static_image of
GdkImage's.

gtk-demo_debug.txt Magnifier (5.77 KB) davshao, 03/26/2010 06:15 PM

History

#1 Updated by Anonymous over 4 years ago

This simple test case succeeds in x86 but fails (segfaults) in amd64:

http://gitweb.dragonflybsd.org/~beket/pcca-tests.git/blob_plain/master:/sys/shm.h/t_shmat.c

Cheers

#2 Updated by dillon over 4 years ago

:Stathis Kamperis <> added the comment:
:
:This simple test case succeeds in x86 but fails (segfaults) in amd64:
:
:http://gitweb.dragonflybsd.org/~beket/pcca-tests.git/blob_plain/master:/sys/shm.h/t_shmat.c
:
:Cheers
:
:----------
:status: unread -> chatting

It just looks like a return value problem in kern/syscalls.master
and in how the sysmg_result is set. I'll get it fixed.

-Matt
Matthew Dillon
<>

#3 Updated by alexh over 4 years ago

Commited in 65cac0a7e0f5b540e524e8872d616c4aa03ca6a8

Also available in: Atom PDF