Re: 2.6.10-rc1-mm5 - badness in enable_irg, BUG

[Ingo Molnar]

* Andrew Morton <akpm@xxxxxxxx> wrote:

> Well can we at least stick a comment in there explaining to the
> long-suffering reader what the difference is between
> smp_processor_id(), _smp_processor_id() and __smp_processor_id()?  And
> what the architecture's options are?

doc-patch against -rc2-mm1 attached.

> Or should we go through every arch and rename their smp_processor_id()
> to __smp_processor_id()?  That would make sense, and would simplify
> that piece of code.

initially the rate of false positives is like 90%, so there would only
be annoyance coming out of this patch. Arch maintainers should have the
ability to enable this on their own pace i think. Once the majority of 
arches have enabled this we can force it on for all architectures.

	Ingo

Signed-off-by: Ingo Molnar <mingo@xxxxxxx>

--- linux/include/linux/smp.h.orig
+++ linux/include/linux/smp.h
@@ -109,12 +109,24 @@ static inline void smp_send_reschedule(i
 
 #endif /* !SMP */
 
+/*
+ * DEBUG_PREEMPT support: check whether smp_processor_id() is being
+ * used in a preemption-safe way.
+ *
+ * An architecture has to enable this debugging code explicitly.
+ * It can do so by renaming the smp_processor_id() macro to
+ * __smp_processor_id().  This should only be done after some minimal
+ * testingy, because usually there are a number of false positives
+ * that an architecture will trigger.
+ *
+ * To fix a false positives (i.e. smp_processor_id() use that the
+ * debugging code reports but which use for some reason is legal),
+ * change the smp_processor_id() reference to _smp_processor_id(),
+ * which is the nondebug variant.  NOTE: dont use this to hack around
+ * real bugs.
+ */
 #ifdef __smp_processor_id
 # if defined(CONFIG_PREEMPT) && defined(CONFIG_DEBUG_PREEMPT)
-  /*
-   * temporary debugging check detecting places that use
-   * smp_processor_id() in a potentially unsafe way:
-   */
    extern unsigned int smp_processor_id(void);
 # else
 #  define smp_processor_id() __smp_processor_id()
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/