[kernel,v4,2/2] powerpc/xive: Drop current cpu priority for orphaned interrupts
diff mbox series

Message ID 20190718050604.74233-3-aik@ozlabs.ru
State Changes Requested
Headers show
Series
  • powerpc/xive: Drop deregistered irqs
Related show

Checks

Context Check Description
snowpatch_ozlabs/checkpatch success total: 0 errors, 0 warnings, 0 checks, 30 lines checked
snowpatch_ozlabs/build-pmac32 success Build succeeded
snowpatch_ozlabs/build-ppc64e success Build succeeded
snowpatch_ozlabs/build-ppc64be success Build succeeded
snowpatch_ozlabs/build-ppc64le success Build succeeded
snowpatch_ozlabs/apply_patch success Successfully applied on branch next (f5c20693d8edcd665f1159dc941b9e7f87c17647)

Commit Message

Alexey Kardashevskiy July 18, 2019, 5:06 a.m. UTC
There is a race between releasing an irq on one cpu and fetching it
from XIVE on another cpu. When such released irq appears in a queue,
we take it from the queue but we do not change the current priority
on that cpu and since there is no handler for the irq, EOI is never
called and the cpu current priority remains elevated
(7 vs. 0xff==unmasked). If another irq is assigned to the same cpu,
then that device stops working until irq is moved to another cpu or
the device is reset.

This implements ppc_md.orphan_irq callback which is called if no irq
descriptor is found and which drops the current priority
to 0xff which effectively unmasks interrupts in a current CPU.

Signed-off-by: Alexey Kardashevskiy <aik@ozlabs.ru>
---
 arch/powerpc/sysdev/xive/common.c | 18 ++++++++++++++++++
 1 file changed, 18 insertions(+)

Patch
diff mbox series

diff --git a/arch/powerpc/sysdev/xive/common.c b/arch/powerpc/sysdev/xive/common.c
index 082c7e1c20f0..17e696b2d71b 100644
--- a/arch/powerpc/sysdev/xive/common.c
+++ b/arch/powerpc/sysdev/xive/common.c
@@ -283,6 +283,23 @@  static unsigned int xive_get_irq(void)
 	return irq;
 }
 
+/*
+ * Handles the case when a target CPU catches an interrupt which is being shut
+ * down on another CPU. generic_handle_irq() returns an error in such case
+ * and then the orphan_irq() handler restores the CPPR to reenable interrupts.
+ *
+ * Without orphan_irq() and valid irq_desc, there is no other way to restore
+ * the CPPR. This executes on a CPU which caught the interrupt.
+ */
+static void xive_orphan_irq(unsigned int irq)
+{
+	struct xive_cpu *xc = __this_cpu_read(xive_cpu);
+
+	xc->cppr = 0xff;
+	out_8(xive_tima + xive_tima_offset + TM_CPPR, 0xff);
+	DBG_VERBOSE("orphan_irq: irq %d, adjusting CPPR to 0xff\n", irq);
+}
+
 /*
  * After EOI'ing an interrupt, we need to re-check the queue
  * to see if another interrupt is pending since multiple
@@ -1419,6 +1436,7 @@  bool __init xive_core_init(const struct xive_ops *ops, void __iomem *area, u32 o
 	xive_irq_priority = max_prio;
 
 	ppc_md.get_irq = xive_get_irq;
+	ppc_md.orphan_irq = xive_orphan_irq;
 	__xive_enabled = true;
 
 	pr_devel("Initializing host..\n");