xfs_iget_core bug.
David Eisenstein
deisenst at gtw.net
Thu Oct 26 05:25:44 UTC 2006
Karl Hudnut wrote:
> Hi,
>
> I have been working on a problem that contains a fix for a bug, not a
> vulnerability, in 2.6.x kernels. Based on kernel.org info I cannot say
> exactly when it was patched. Here is the patch:
>
> ======================================================================================
> --- 1.20/fs/xfs/xfs_iget.c Fri Jan 9 07:20:13 2004
> +++ edited/fs/xfs/xfs_iget.c Mon Feb 23 14:47:03 2004
> @@ -236,13 +236,14 @@
>
> goto again;
> }
> -/* Chances are the other vnode (the one in the inode) is being torn
> - * down right now, and we landed on top of it. Question is, what do
> - * we do? Unhook the old inode and hook up the new one?
> - */
> - cmn_err(CE_PANIC,
> - "xfs_iget_core: ambiguous vns: vp/0x%p, invp/0x%p",
> - inode_vp, vp);
> +
> + printk("%s: ambiguous vns: vp/0x%p, invp/0x%p",
> + __FUNCTION__, inode_vp, vp);
> + printk("v_vflag = 0x%x, v_type = %d\n",
> + inode_vp->v_flag, inode_vp->v_type);
> + printk("i_state = 0x%x, i_count = %d, i_nlink = %d\n",
> + inode->i_state, inode->i_count, inode->i_nlink);
> + BUG();
> }
>
> read_unlock(&ih->ih_lock);
>
> ======================================================================================
>
> I think the patch was introduced after 2.6.11, not totally sure. I can tell it was
> patched by 2.6.17. I need this patch. Can anyone verify if this is included in
> kernel-smp-2.6.12-2.3.legacy_FC3.x86_64.rpm for FC3 from Fedora Legacy?
>
> Thanks. If the answer is yes, please point me to the evidence so I can show it to the
> rest of the SysAdmin Team here at COSMIC. (Not that we would doubt it, but it would
> be better to see it for ourselves. Paranoid mode set = 1 and so on.)
>
Hi Dr. Hudnut,
I did a scan through all of the patch files in the source rpm at
<http://download.fedoralegacy.org/fedora/3/updates/SRPMS/kernel-2.6.12-2.3.legacy_FC3.src.rpm>
and could not find any patches that patch the 'fs/xfs/xfs_iget.c' source file.
The xfs_iget.c source file itself appears not to be patched with the code you
enclosed. It appears to be the original code, which calls 'cmn_err()'.
Hope this helps. -David
More information about the fedora-legacy-list
mailing list