On Monday 23 January 2006 12:01, Hans Reiser wrote:
> Hans Reiser wrote:
> >Vladimir V. Saveliev wrote:
> >>Hello
> >>
> >>On Fri, 2006-01-20 at 14:28 -0600, Jake Maciejewski wrote:
> >>>In addition to the disable write barrier warning reported by Louis-David
> >>>Mitterrand on January 10th, with 2.6.15-1 I'm getting:
> >>>
> >>> <4>reiser4[dd(25448)]: update_stale_dk
> >>> (fs/reiser4/search.c:1363)[nikita-38210]: WARNING: stale dk
> >>>
> >>>The process varies. I've also seen the stale dk warning triggered by
> >>> rsync and rm.
> >>
> >>This is harmless. I think we should remove this warning as well as the
> >>one about write barrier.
> >
> >Zam said he would change it
>
> by "it" I refer to the write barrier warning, sorry for my imprecision....
>
> > to a notice. Zam, what happened to the patch?
1. re-phrase "disabling write barrier" warning which confuses users and
make it KERN_NOTICE instead of KERN_WARNING.
2. remove the "[nikita-38210]: WARNING: stale dk" at fs/reiser4/search.c
which does not indicate any reiser4 tree or fs problems.
Signed-off-by: Alex Zarochentsev <[EMAIL PROTECTED]>
---
fs/reiser4/debug.h | 5 ++++-
fs/reiser4/search.c | 2 --
fs/reiser4/wander.c | 3 ++-
3 files changed, 6 insertions(+), 4 deletions(-)
--- linux-2.6.x.orig/fs/reiser4/debug.h
+++ linux-2.6.x/fs/reiser4/debug.h
@@ -205,10 +205,13 @@ extern int is_in_reiser4_context(void);
#define wrong_return_value( label, function ) \
impossible( label, "wrong return value from " function )
-/* Issue warning message to the console */
+/* Issue different types of reiser4 messages to the console */
#define warning( label, format, ... ) \
DCALL( KERN_WARNING, \
printk, 1, label, "WARNING: " format , ## __VA_ARGS__ )
+#define notice( label, format, ... ) \
+ DCALL( KERN_NOTICE, \
+ printk, 1, label, "NOTICE: " format , ## __VA_ARGS__ )
/* mark not yet implemented functionality */
#define not_yet( label, format, ... ) \
--- linux-2.6.x.orig/fs/reiser4/search.c
+++ linux-2.6.x/fs/reiser4/search.c
@@ -1359,8 +1359,6 @@ static void update_stale_dk(reiser4_tree
if (unlikely(ZF_ISSET(node, JNODE_RIGHT_CONNECTED) &&
right && ZF_ISSET(right, JNODE_DKSET) &&
!keyeq(&rd, znode_get_ld_key(right)))) {
- /* does this ever happen? */
- warning("nikita-38210", "stale dk");
assert("nikita-38211", ZF_ISSET(node, JNODE_DKSET));
read_unlock_dk(tree);
read_unlock_tree(tree);
--- linux-2.6.x.orig/fs/reiser4/wander.c
+++ linux-2.6.x/fs/reiser4/wander.c
@@ -230,7 +230,8 @@ static inline int reiser4_use_write_barr
static void disable_write_barrier(struct super_block * s)
{
- warning("zam-1055", "disabling write barrier\n");
+ notice("zam-1055", "%s does not support write barriers,"
+ " using synchronous write instead.\n", s->s_id);
set_bit((int)REISER4_NO_WRITE_BARRIER, &get_super_private(s)->fs_flags);
}
