Attachment 149293 Details for Bug 220451 – Purposed Patch

Purposed Patch

linux-2.6.18-nfs-Tru64-hack.patch (text/x-patch), 1.41 KB, created by Steve Dickson on 2007-03-05 21:05:13 UTC

(hide)

Description:

Filename:

MIME Type:

Creator: Steve Dickson

Created: 2007-03-05 21:05:13 UTC

Size: 1.41 KB

patch

obsolete

>
>Deal with broken Tru64 servers that send different fsids
>(due to sign extensions) by sign extending the received
>fsid and then doing the comparison.
>
>Signed-off-by: Steve Dickson <steved@redhat.com>
>Acked-by: Peter Staubach <staubach@redhat.com>
>
>--- linux-2.6.18.i686/include/linux/nfs_xdr.h.orig	2007-03-05 11:13:35.000000000 -0500
>+++ linux-2.6.18.i686/include/linux/nfs_xdr.h	2007-03-05 13:23:00.000000000 -0500
>@@ -21,10 +21,26 @@ struct nfs_fsid {
> 
> /*
>  * Helper for checking equality between 2 fsids.
>+ * Hack Alert:
>+ * Due do broken Tru64 servers sending sign extend
>+ * 32bit fsids in most but _not_ all procedures, 
>+ * we need to be a bit clever as well as make an 
>+ * assumption when comparing fsids.
>+ *
>+ * If the assumption can be made that the bottom 32 bits 
>+ * of an fsids will *always* change when the fsid changes
>+ * then it should be safe to only look at the bottom 32 bits
>+ * during a 64bit comparison.
>+ *
>+ * So by sign extending the fsid we just received (i.e. b->major) 
>+ * and then do the comparison, we are still going a 64bit
>+ * comparison but only really "looking at" the bottom 32 bits.
>  */
> static inline int nfs_fsid_equal(const struct nfs_fsid *a, const struct nfs_fsid *b)
> {
>-	return a->major == b->major && a->minor == b->minor;
>+	return (a->major == b->major || 
>+	    a->major == (uint64_t)((int64_t)((int32_t)(b->major & 0xffffffff)))) &&
>+		a->minor == b->minor;
> }
> 
> struct nfs_fattr {

Actions: View

Attachments on bug 220451: 149293