Summary: | net-fs/openafs-1.4.2 - use of strerror/segmentation faults with error messages - .AFSSERVER file | ||
---|---|---|---|
Product: | Gentoo Linux | Reporter: | Volkmar Glauche <volkmar.glauche> |
Component: | [OLD] Server | Assignee: | Stefaan De Roeck (RETIRED) <stefaan> |
Status: | RESOLVED FIXED | ||
Severity: | normal | ||
Priority: | High | ||
Version: | 2006.1 | ||
Hardware: | AMD64 | ||
OS: | Linux | ||
Whiteboard: | |||
Package list: | Runtime testing required: | --- | |
Attachments: | strace output |
Description
Volkmar Glauche
2006-12-19 05:15:58 UTC
Which ebuild and version is this about? (In reply to comment #1) > Which ebuild and version is this about? > Sorry - here is the version info: openafs-1.4.2.ebuild Noone will notice again if you don't reopen. :) Could you give some hints on how to reproduce? Thanks Created attachment 105526 [details]
strace output
(In reply to comment #4) > Could you give some hints on how to reproduce? Thanks > Reproducible - always. The .AFSSERVER lookup was misguided (some strange documentation), but the segmentation fault happens whenever a fs command wants to report an error message. I have a new case for this (libafs.ko and 2.6.19-gentoo-r2 kernel on amd64), for which I may open a new bug, but attached is the output from strace'ing an fs call. I cannot reproduce this with gentoo-sources-2.6.19-r3 (nor with openafs-1.4.3-pre20061220 and 2.6.19-r3 or 2.6.19-r2). Could you please verify you've run "etc-update", and emerged openafs-kernel for the right kernel version? Oh, could you also look into your kernel ring buffer for messages regarding finding/failing to find the system call table? See bug #160270. Thanks! This happens always when fs wants to print an error message - perhaps its a glibc/gcc issue?? I have not tried to downgrade to a stable glibc/gcc yet, since everything else seems to be working fine. Please also look into all questions in comments #7 and #8. While guessing can sometimes give a brilliant idea out of the blue, verifying the facts can help eliminating possibilities with more certainty. (In reply to comment #10) > Please also look into all questions in comments #7 and #8. While guessing can > sometimes give a brilliant idea out of the blue, verifying the facts can help > eliminating possibilities with more certainty. > Sorry for not answering explicitly. Everything is up-to-date, and normal fs/afs behaviour is fine. I can reproduce the segfault with fs listacl /afs where /afs is the mount point for the AFS file system tree, and afsd is running with options /usr/sbin/afsd -fakestat -stat 2000 -dcache 800 -daemons 3 -volumes 70 -afsdb -dynroot Fixed in net-fs/openafs-1.4.3_pre20061220-r1, cause was missing prototype resulting in a bad pointer cast. Thanks for reporting! |