Skip to contentSkip to conversation

Revert "[IRDA]: Lockdep fix."

PATCH In review Samuel Ortiz 3 participants 4 messages Nov 25, 2006 Networking

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 Revert "[IRDA]: Lockdep fix."

Hi Linus,

commit 700f9672c9a61c12334651a94d17ec04620e1976 breaks IrDA as irlmp.c can no longer build.
This is due to the spin_lock_irqsave_nested() patches being in the -mm tree and not yet in yours.
I'll resend the patch once both trees are synchronized.

This reverts commit 700f9672c9a61c12334651a94d17ec04620e1976.

---
 net/irda/irlmp.c |    3 +--
 1 files changed, 1 insertions(+), 2 deletions(-)
net/irda/irlmp.c +1 -2
index fede837..5073261 100644
@@ -1678,8 +1678,7 @@ #endif /* CONFIG_IRDA_ULTRA */
 	 *  every IrLAP connection and check every LSAP associated with each
 	 *  the connection.
 	 */
-	spin_lock_irqsave_nested(&irlmp->links->hb_spinlock, flags,
-			SINGLE_DEPTH_NESTING);
+	spin_lock_irqsave(&irlmp->links->hb_spinlock, flags);
 	lap = (struct lap_cb *) hashbin_get_first(irlmp->links);
 	while (lap != NULL) {
 		IRDA_ASSERT(lap->magic == LMP_LAP_MAGIC, goto errlap;);
Signature · 1 line
1.4.2.3
maintainer
#1 lore raw

From: Samuel Ortiz <[email protected]>
Date: Sat, 25 Nov 2006 17:26:49 +0200

Quoted 10 lines

Hi Linus,

commit 700f9672c9a61c12334651a94d17ec04620e1976 breaks IrDA as irlmp.c can no longer build.
This is due to the spin_lock_irqsave_nested() patches being in the -mm tree and not yet in yours.
I'll resend the patch once both trees are synchronized.

This reverts commit 700f9672c9a61c12334651a94d17ec04620e1976.

Andrew is going to merge the necessary infrastructure to Linus, please don't revert this change.

Why is everyone so impatient about this issue? Just wait for Andrew to merge the fix and all will be well :-)

#2 lore raw
Quoted 3 lines · David Miller

Why is everyone so impatient about this issue? Just wait for Andrew to merge the fix and all will be well :-)

No, I think this was a total failure. We should have reverted it immediately rather than waiting for the fix. Especially as the damn thing wasn't even all that important.

I now (finally) have the patch from Andrew, but we should _not_ have had this thing broken for three days. It should have gotten reverted on the first report of trouble, instead of us telling people to just wait.

Broken compiles are simply not acceptable. Patches that cause them should be reverted _immediately_ unless a fix is available as quickly (which it wasn't due to turkey-day).

No excuses. People _should_ be impatient about idiotic failures like this.

Linus

maintainer
#3 lore raw

From: Linus Torvalds <[email protected]>
Date: Sat, 25 Nov 2006 13:26:51 -0800 (PST)

Quoted 9 lines

I now (finally) have the patch from Andrew, but we should _not_ have had this thing broken for three days. It should have gotten reverted on the first report of trouble, instead of us telling people to just wait.

Broken compiles are simply not acceptable. Patches that cause them should be reverted _immediately_ unless a fix is available as quickly (which it wasn't due to turkey-day).

No excuses. People _should_ be impatient about idiotic failures like this.

Ok, no problem.