From: Ingo Molnar <mingo@elte.hu>
The patch below adds a DEBUG_SMP_PROCESSOR_ID option to .config. It
defaults to y so i expect it to get good usage still, but people should
have the option to turn it off since it introduces overhead.
I have test-compiled and test-booted x86 and x64 on SMP+!PREEMPT and
everything is working fine.
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Andrew Morton <akpm@osdl.org>
---
25-akpm/include/linux/smp.h | 2 +-
25-akpm/kernel/sched.c | 3 ++-
25-akpm/lib/Kconfig.debug | 9 +++++++++
3 files changed, 12 insertions(+), 2 deletions(-)
diff -puN include/linux/smp.h~sched-add-debug_smp_processor_id include/linux/smp.h
--- 25/include/linux/smp.h~sched-add-debug_smp_processor_id 2004-10-05 02:00:53.284822288 -0700
+++ 25-akpm/include/linux/smp.h 2004-10-05 02:00:53.291821224 -0700
@@ -110,7 +110,7 @@ static inline void smp_send_reschedule(i
#endif /* !SMP */
#ifdef __smp_processor_id
-# ifdef CONFIG_PREEMPT
+# if defined(CONFIG_PREEMPT) && defined(CONFIG_DEBUG_SMP_PROCESSOR_ID)
/*
* temporary debugging check detecting places that use
* smp_processor_id() in a potentially unsafe way:
diff -puN kernel/sched.c~sched-add-debug_smp_processor_id kernel/sched.c
--- 25/kernel/sched.c~sched-add-debug_smp_processor_id 2004-10-05 02:00:53.286821984 -0700
+++ 25-akpm/kernel/sched.c 2004-10-05 02:00:53.294820768 -0700
@@ -2471,7 +2471,8 @@ static inline int dependent_sleeper(int
}
#endif
-#if defined(CONFIG_PREEMPT) && defined(__smp_processor_id)
+#if defined(CONFIG_PREEMPT) && defined(__smp_processor_id) && \
+ defined(CONFIG_DEBUG_SMP_PROCESSOR_ID)
/*
* Debugging check.
*/
diff -puN lib/Kconfig.debug~sched-add-debug_smp_processor_id lib/Kconfig.debug
--- 25/lib/Kconfig.debug~sched-add-debug_smp_processor_id 2004-10-05 02:00:53.288821680 -0700
+++ 25-akpm/lib/Kconfig.debug 2004-10-05 02:00:53.295820616 -0700
@@ -52,6 +52,15 @@ config DEBUG_SPINLOCK_SLEEP
If you say Y here, various routines which may sleep will become very
noisy if they are called with a spinlock held.
+config DEBUG_SMP_PROCESSOR_ID
+ bool "Preempt-unsafe smp_processor_id() checking"
+ depends on PREEMPT && X86
+ default y
+ help
+ If you say Y here then the kernel will use a debug variant of the
+ commonly used smp_processor_id() function and will print warnings
+ if kernel code uses it in a preemption-unsafe way.
+
config DEBUG_HIGHMEM
bool "Highmem debugging"
depends on DEBUG_KERNEL && HIGHMEM && (X86 || PPC32 || MIPS || SPARC32)
_