This patch renames the raw hard_irq_{enable,disable} into
__hard_irq_{enable,disable} and introduces a higher level
hard_irq_disable() function that can be used by any code
to enforce that IRQs are fully disabled, not only lazy
disabled.
The difference with the __ versions is that it will update
some per-processor fields so that the kernel keeps track and
properly re-enables them in the next local_irq_disable();
This prepares powerpc for my next patch that introduces
hard_irq_disable() generically.
Signed-off-by: Benjamin Herrenschmidt <[email protected]>
arch/powerpc/kernel/irq.c | 2 +-
arch/powerpc/kernel/swsusp.c | 4 ----
arch/powerpc/platforms/cell/pervasive.c | 6 ++----
include/asm-powerpc/hw_irq.h | 11 +++++++++--
4 files changed, 12 insertions(+), 11 deletions(-)
Index: linux-cell/arch/powerpc/kernel/irq.c
===================================================================
--- linux-cell.orig/arch/powerpc/kernel/irq.c 2007-05-10 14:58:56.000000000 +1000
+++ linux-cell/arch/powerpc/kernel/irq.c 2007-05-10 14:58:59.000000000 +1000
@@ -173,7 +173,7 @@ void local_irq_restore(unsigned long en)
lv1_get_version_info(&tmp);
}
- hard_irq_enable();
+ __hard_irq_enable();
}
#endif /* CONFIG_PPC64 */
Index: linux-cell/arch/powerpc/kernel/swsusp.c
===================================================================
--- linux-cell.orig/arch/powerpc/kernel/swsusp.c 2007-05-10 14:57:03.000000000 +1000
+++ linux-cell/arch/powerpc/kernel/swsusp.c 2007-05-10 14:57:07.000000000 +1000
@@ -36,8 +36,4 @@ void restore_processor_state(void)
#ifdef CONFIG_PPC32
set_context(current->active_mm->context.id, current->active_mm->pgd);
#endif
-
-#ifdef CONFIG_PPC64
- hard_irq_enable();
-#endif
}
Index: linux-cell/arch/powerpc/platforms/cell/pervasive.c
===================================================================
--- linux-cell.orig/arch/powerpc/platforms/cell/pervasive.c 2007-05-10 14:57:34.000000000 +1000
+++ linux-cell/arch/powerpc/platforms/cell/pervasive.c 2007-05-10 14:57:54.000000000 +1000
@@ -43,12 +43,10 @@ static void cbe_power_save(void)
unsigned long ctrl, thread_switch_control;
/*
- * We need to hard disable interrupts, but we also need to mark them
- * hard disabled in the PACA so that the local_irq_enable() done by
- * our caller upon return propertly hard enables.
+ * We need to hard disable interrupts, the local_irq_enable() done by
+ * our caller upon return will hard re-enable.
*/
hard_irq_disable();
- get_paca()->hard_enabled = 0;
ctrl = mfspr(SPRN_CTRLF);
Index: linux-cell/include/asm-powerpc/hw_irq.h
===================================================================
--- linux-cell.orig/include/asm-powerpc/hw_irq.h 2007-05-10 14:51:43.000000000 +1000
+++ linux-cell/include/asm-powerpc/hw_irq.h 2007-05-10 14:59:32.000000000 +1000
@@ -48,8 +48,15 @@ extern void iseries_handle_interrupts(vo
#define irqs_disabled() (local_get_flags() == 0)
-#define hard_irq_enable() __mtmsrd(mfmsr() | MSR_EE, 1)
-#define hard_irq_disable() __mtmsrd(mfmsr() & ~MSR_EE, 1)
+#define __hard_irq_enable() __mtmsrd(mfmsr() | MSR_EE, 1)
+#define __hard_irq_disable() __mtmsrd(mfmsr() & ~MSR_EE, 1)
+
+#define hard_irq_disable() \
+ do { \
+ __hard_irq_disable(); \
+ get_paca()->soft_enabled = 0; \
+ get_paca()->hard_enabled = 0; \
+ } while(0)
#else
On Thu, 10 May 2007, Benjamin Herrenschmidt wrote:
> This patch renames the raw hard_irq_{enable,disable} into
> __hard_irq_{enable,disable} and introduces a higher level
> hard_irq_disable() function that can be used by any code
> to enforce that IRQs are fully disabled, not only lazy
> disabled.
Why did you rename hard_irq_enable() too?
Isn't it more logical to have high-level hard_irq_disable() and
hard_irq_enable(), and a special low-level __hard_irq_disable()?
Gr{oetje,eeting}s,
Geert
--
Geert Uytterhoeven -- Sony Network and Software Technology Center Europe (NSCE)
[email protected] ------- The Corporate Village, Da Vincilaan 7-D1
Voice +32-2-7008453 Fax +32-2-7008622 ---------------- B-1935 Zaventem, Belgium
On Thu, 2007-05-10 at 09:44 +0200, Geert Uytterhoeven wrote:
> On Thu, 10 May 2007, Benjamin Herrenschmidt wrote:
> > This patch renames the raw hard_irq_{enable,disable} into
> > __hard_irq_{enable,disable} and introduces a higher level
> > hard_irq_disable() function that can be used by any code
> > to enforce that IRQs are fully disabled, not only lazy
> > disabled.
>
> Why did you rename hard_irq_enable() too?
>
> Isn't it more logical to have high-level hard_irq_disable() and
> hard_irq_enable(), and a special low-level __hard_irq_disable()?
Not really. If you see my subsequent patch, the idea is to introduce a
single generic hard_irq_disable() which is meant to be called with
irqs already disabled (that is within a local_irq_disable section) to
enforce that if the arch does lazy disabling, it gets hard disabled
at this point.
If we start adding hard_irq_enable() we end up in a can of worms:
- Do we want all the full set of save/restore etc... ?
- What if somebody does hard_enable while we are soft-disabled
-and- have been hard disabled because of a pending interrupt ?
- What's the point ? :-)
So overall, I want to keep the semantics as simple as they can be. Maybe
I can even add some WARN_ON() to make sure we are in a
local_irq_disable'd section even in the generic one instead of just a
NOP to enfore that.
Cheers,
Ben.