Patchwork irqdomain: Allow quiet failure mode

login
register
mail settings
Submitter Benjamin Herrenschmidt
Date May 6, 2013, 6:41 a.m.
Message ID <1367822494.15842.12.camel@pasglop>
Download mbox | patch
Permalink /patch/241558/
State Accepted
Commit 5fe0c1f2f0dca3351536284b0180a79f341b7854
Headers show

Comments

Benjamin Herrenschmidt - May 6, 2013, 6:41 a.m.
Some interrupt controllers refuse to map interrupts marked as
"protected" by firwmare. Since we try to map everyting in the
device-tree on some platforms, we end up with a lot of nasty
WARN's in the boot log for what is a normal situation on those
machines.

This defines a specific return code (-EPERM) from the host map()
callback which cause irqdomain to fail silently.

MPIC is updated to return this when hitting a protected source
printing only a single line message for diagnostic purposes.

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>


Grant, I'm happy to merge that myself provided I have your ack,
it fixes an annoying problem on Cell introduced by previous changes
to irqdomain (not recent but users started to notice).

 arch/powerpc/sysdev/mpic.c |   14 +++++++++++---
 kernel/irq/irqdomain.c     |   20 +++++++++++++++++---
 2 files changed, 28 insertions(+), 6 deletions(-)
Grant Likely - May 6, 2013, 11:07 a.m.
On Mon, May 6, 2013 at 7:41 AM, Benjamin Herrenschmidt
<benh@kernel.crashing.org> wrote:
> Some interrupt controllers refuse to map interrupts marked as
> "protected" by firwmare. Since we try to map everyting in the
> device-tree on some platforms, we end up with a lot of nasty
> WARN's in the boot log for what is a normal situation on those
> machines.
>
> This defines a specific return code (-EPERM) from the host map()
> callback which cause irqdomain to fail silently.
>
> MPIC is updated to return this when hitting a protected source
> printing only a single line message for diagnostic purposes.
>
> Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
>
>
> Grant, I'm happy to merge that myself provided I have your ack,
> it fixes an annoying problem on Cell introduced by previous changes
> to irqdomain (not recent but users started to notice).

Yes, please merge it since powerpc needs it.

Acked-by: Grant Likely <grant.likely@linaro.org>

g.

Patch

diff --git a/arch/powerpc/sysdev/mpic.c b/arch/powerpc/sysdev/mpic.c
index d30e6a6..ee21b5e 100644
--- a/arch/powerpc/sysdev/mpic.c
+++ b/arch/powerpc/sysdev/mpic.c
@@ -1001,8 +1001,12 @@  static int mpic_host_map(struct irq_domain *h, unsigned int virq,
 
 	if (hw == mpic->spurious_vec)
 		return -EINVAL;
-	if (mpic->protected && test_bit(hw, mpic->protected))
-		return -EINVAL;
+	if (mpic->protected && test_bit(hw, mpic->protected)) {
+		pr_warning("mpic: Mapping of source 0x%x failed, "
+			   "source protected by firmware !\n",\
+			   (unsigned int)hw);
+		return -EPERM;
+	}
 
 #ifdef CONFIG_SMP
 	else if (hw >= mpic->ipi_vecs[0]) {
@@ -1029,8 +1033,12 @@  static int mpic_host_map(struct irq_domain *h, unsigned int virq,
 	if (mpic_map_error_int(mpic, virq, hw))
 		return 0;
 
-	if (hw >= mpic->num_sources)
+	if (hw >= mpic->num_sources) {
+		pr_warning("mpic: Mapping of source 0x%x failed, "
+			   "source out of range !\n",\
+			   (unsigned int)hw);
 		return -EINVAL;
+	}
 
 	mpic_msi_reserve_hwirq(mpic, hw);
 
diff --git a/kernel/irq/irqdomain.c b/kernel/irq/irqdomain.c
index 96f3a1d..5a83dde 100644
--- a/kernel/irq/irqdomain.c
+++ b/kernel/irq/irqdomain.c
@@ -462,9 +462,23 @@  int irq_domain_associate_many(struct irq_domain *domain, unsigned int irq_base,
 		if (domain->ops->map) {
 			ret = domain->ops->map(domain, virq, hwirq);
 			if (ret != 0) {
-				pr_err("irq-%i==>hwirq-0x%lx mapping failed: %d\n",
-				       virq, hwirq, ret);
-				WARN_ON(1);
+				/*
+				 * If map() returns -EPERM, this interrupt is protected
+				 * by the firmware or some other service and shall not
+				 * be mapped.
+				 *
+				 * Since on some platforms we blindly try to map everything
+				 * we end up with a log full of backtraces.
+				 *
+				 * So instead, we silently fail on -EPERM, it is the
+				 * responsibility of the PIC driver to display a relevant
+				 * message if needed.
+				 */
+				if (ret != -EPERM) {
+					pr_err("irq-%i==>hwirq-0x%lx mapping failed: %d\n",
+					       virq, hwirq, ret);
+					WARN_ON(1);
+				}
 				irq_data->domain = NULL;
 				irq_data->hwirq = 0;
 				goto err_unmap;