Skip to contentSkip to conversation

Linux 2.5.4-dj1

DISCUSSION Dave Jones 2 participants 3 messages Feb 12, 2002

lore
Share & tools
lore

Dave Jones

author
#0 lore raw

Mostly compile fixes, and a point-release resync. Diffsize hasn't come down as much as I expected, so I'll push a few of the larger chunks to Linus tomorrow.

Patch against 2.5.4 vanilla is available from:
ftp://ftp.kernel.org/pub/linux/kernel/people/davej/patches/2.5/

-- Davej.

2.5.4-dj1

o   Merge 2.5.4
o   Fix inverted parameters in NCR5380			(Rasmus Andersen)
o   NSC Geode Companion chip workaround.		(Hiroshi MIURA)
    | TODO: Nuke the CONFIG option, replace with
    | PCI detection.
o   Improve kiobuf_init performance.			(Various Intel folks)
o   uidhash cleanup.					(William Lee Irwin III)
o   Fix /proc 'read past end of buffer' bug.		(Thomas Hood)
o   PnPBIOS updates (ESCD support).			(Thomas Hood)
o   signal.c missing binfmt include compile fix.	(Udo A. Steinberg)
o   thread_saved_pc compile fix.			(Andrew Morton)
o   Various reiserfs updates.				(Oleg Drokin, Namesys)
o   Fix UP Preempt compilation.				(Mikael Pettersson)
o   Kill sleep_on in Olympic TR driver.			(Mike Phillips)
o   Drop 64bit DRM fixes.
o   Fix zftape compile.					(Me)
o   Hack around synclink non-compile.			(Me)

2.5.3-dj5

o   Merge 2.5.4pre5
o   Add some missing MODULE_LICENSE tags	(Hubert Mantel)
o   Fix ptrace PEEKUSR oops.			(Manfred Spraul, others)
o   Drop some bogus bits from USB & netdrivers.	(Me)
o   sbpcd bio fixes.				(Paul Gortmaker)
o   pci id trigraph warning fixes.		(Steven J. Hill)
o   Tridentfb resource management fixes.	(Geert Uytterhoeven)
o   53c700 locking cleanup.			(James Bottomley)
o   Workaround ext2 trying to free block -1	(Andreas Dilger)
o   Fix up deviceio Docbook generation.		(Jason Ferguson)
o   removal of isa_read/writes from ibmtr.	(Mike Phillips)
o   kthread abstraction.			(Christoph Hellwig)
Signature · 2 lines
Dave Jones.                    http://www.codemonkey.org.uk
SuSE Labs.
#1 lore raw
Quoted 1 line · Dave Jones

o Fix UP Preempt compilation. (Mikael Pettersson)

I ended up sending the following patch to Linus instead.

Would you merge this into your next release, so we can keep the trees in sync (and not inadvertently push your fix over Linus's later on). This approach removes the conditional altogether in the UP+preempt case, so it is optimal.

Thanks,

Robert Love

include/asm-i386/smplock.h +7 -3
diff -urN linux-2.5.4-dj1/include/asm-i386/smplock.h linux/include/asm-i386/smplock.h
@@ -12,10 +12,15 @@
 
 #ifdef CONFIG_SMP
 #define kernel_locked()		spin_is_locked(&kernel_flag)
+#define check_irq_holder(cpu) \
+do { \
+	if (global_irq_holder == (cpu)) \
+		BUG(); \
+} while(0)
 #else
 #ifdef CONFIG_PREEMPT
 #define kernel_locked()		preempt_get_count()
-#define global_irq_holder      0xFF    /* NO_PROC_ID */
+#define check_irq_holder(cpu)	do { } while(0)
 #else
 #define kernel_locked()		1
 #endif
@@ -28,8 +33,7 @@
 do {						\
 	if (unlikely(task->lock_depth >= 0)) {	\
 		spin_unlock(&kernel_flag);	\
-		if (global_irq_holder == (cpu))	\
-			BUG();			\
+		check_irq_holder(cpu);		\
 	}					\
 } while (0)
 
author
#2 lore raw
In reply to Dave Jones
Quoted 5 lines · Robert Love
Quoted 1 line

o Fix UP Preempt compilation. (Mikael Pettersson)

I ended up sending the following patch to Linus instead.

Would you merge this into your next release, so we can keep the trees in sync (and not inadvertently push your fix over Linus's later on).

Done.

Signature · 2 lines
| Dave Jones.        http://www.codemonkey.org.uk
| SuSE Labs