On 13-05-27 12:56 PM, Paul Butler wrote:
From: David Mercado <david.merc...@windriver.com>

Modified arch/arm/mach-axxia/axxia-gic.c to to handle the slightly different
way in which a preempt_rt kernel manipulates the Axxia GIC API during boot.
Also updated arch/arm/configs/lsisim_defconfig to set the base level of
prempt_rt to CONFIG_PREEMPT_RTB.  The preempt_rt kernel must be set with

This doesn't match the patch ...

Bruce

a minimal preempt model, to enable CONFIG_GENERIC_LOCKBREAK, which in turn
allows spinlocks to work correctly across multiple clusters.

Signed-off-by: David Mercado <david.merc...@windriver.com>
Signed-off-by: Paul Butler <paul.but...@windriver.com>
---
  arch/arm/mach-axxia/axxia-gic.c | 54 +++++++++++++++++++++++++++++++++--------
  1 file changed, 44 insertions(+), 10 deletions(-)

diff --git a/arch/arm/mach-axxia/axxia-gic.c b/arch/arm/mach-axxia/axxia-gic.c
index 058b837..20d04ea 100644
--- a/arch/arm/mach-axxia/axxia-gic.c
+++ b/arch/arm/mach-axxia/axxia-gic.c
@@ -155,9 +155,11 @@ static void gic_mask_irq(struct irq_data *d)
        if (irqid >= 1020)
                return;

-       /* We should never be disabling the AXM IPI interrupts. */
-       if ((irqid >= IPI0_CPU0) && (irqid < MAX_AXM_IPI_NUM))
+       /* Deal with PPI interrupts directly. */
+       if (irqid > 16 && irqid < 32) {
+               _gic_mask_irq(d);
                return;
+       }

        /*
         * If the cpu that this interrupt is assigned to falls within
@@ -182,22 +184,54 @@ static void gic_mask_irq(struct irq_data *d)
        }
  }

-static void gic_unmask_irq(struct irq_data *d)
+static void _gic_unmask_irq(void *arg)
  {
+       struct irq_data *d = (struct irq_data *)arg;
        u32 mask = 1 << (gic_irq(d) % 32);

-       /*
-        * No need to run this at all clusters, since the kernel
-        * also calls gic_set_affinity when unmasking interrupts.
-        * In that routine we also unmask the interrupt, so doing
-        * it here would be redundant.
-        */
        raw_spin_lock(&irq_controller_lock);
        writel_relaxed(mask, gic_dist_base(d) + GIC_DIST_ENABLE_SET
-                       + (gic_irq(d) / 32) * 4);
+                               + (gic_irq(d) / 32) * 4);
        raw_spin_unlock(&irq_controller_lock);
  }

+static void gic_unmask_irq(struct irq_data *d)
+{
+       u32 pcpu = cpu_logical_map(smp_processor_id());
+       u32 irqid = gic_irq(d);
+
+       if (irqid >= 1020)
+               return;
+
+       /* Deal with PPI interrupts directly. */
+       if (irqid > 15 && irqid < 32) {
+               _gic_unmask_irq(d);
+               return;
+       }
+
+       /*
+        * If the cpu that this interrupt is assigned to falls within
+        * the same cluster as the cpu we're currently running on, do
+        * the IRQ masking directly. Otherwise, use the IPI mechanism
+        * to remotely do the masking.
+        */
+       if ((cpu_logical_map(irq_cpuid[irqid]) / 4) == (pcpu / 4)) {
+               _gic_unmask_irq(d);
+       } else {
+               /*
+                * We are running here with local interrupts
+                * disabled. Temporarily re-enable them to
+                * avoid possible deadlock when calling
+                * smp_call_function_single().
+                */
+               local_irq_enable();
+               smp_call_function_single(irq_cpuid[irqid],
+                                        _gic_unmask_irq,
+                                        d, 1);
+               local_irq_disable();
+       }
+}
+
  static void gic_eoi_irq(struct irq_data *d)
  {
        /*


_______________________________________________
linux-yocto mailing list
linux-yocto@yoctoproject.org
https://lists.yoctoproject.org/listinfo/linux-yocto

Reply via email to