Description of problem: When I tried to turn on jdata with gfs_tool on a ppc cluster I got the following error: gfs_tool: /mnt/gfs_vs0 is not a GFS file/filesystem and the following message on the console: ioctl32(gfs_tool:6952): Unknown cmd fd(3) cmd(00004723){00} arg(ff98f8bc) on /mnt/gfs_vs0 Version-Release number of selected component (if applicable): gfs-utils-0.1.12-1.el5 kmod-gfs-0.1.19-7.el5_1.1 kernel-2.6.18-53.1.6.el5 How reproducible: Every time Steps to Reproduce: 1. mount GFS file system on a ppc system 2. gfs_tool getsb /mnt/gfs Actual results: Every command I've tried (getsb, stat, and inherit_jdata) doesn't recognize the file system as GFS. Expected results: Additional info:
Created attachment 292677 [details] Proposed patch This patch seems to work on both ppc and x86_64 architecture. It's a bit big and a bit of a kludge. However, hopefully it will only affect systems that need compatibility ioctls, such as the failing case. Also, it doesn't affect userland.
So this looks to be needed for 5.1z too.
Created attachment 292853 [details] Revised patch This version has some small cleanups. For one thing, it only allocates as much memory as required rather than NAME_MAX for the copies parameters. Also, it fixes some formatting mistakes.
Fix was committed to the HEAD and RHEL5 branches of CVS (plus a crosswrite to the RHEL51 branch for bug #430154). Changing status to Modified. The predecessor patch was tested on the failing ppc/ppc64 system "basic" and on x86_64 system "roth-01". This patch was tested on roth-01, but "basic" was reassigned to another purpose, so I don't currently have a ppc system to test it on.
An advisory has been issued which should help the problem described in this bug report. This report is therefore being closed with a resolution of ERRATA. For more information on the solution and/or where to find the updated files, please follow the link below. You may reopen this bug report if the solution does not work for you. http://rhn.redhat.com/errata/RHBA-2008-0347.html