[PATCH v3 7/8] x86/irq: WARN_ONCE() if irq_move_cleanup is called on a pending interrupt

From: H. Peter Anvin
Date: Wed May 19 2021 - 17:22:31 EST


From: "H. Peter Anvin (Intel)" <hpa@xxxxxxxxx>

The current IRQ vector allocation code should be "clean" and never
issue a IRQ_MOVE_CLEANUP_VECTOR IPI for an interrupt that could still
be pending. This should make it possible to move it to the "normal"
system IRQ vector range. This should probably be a three-step process:

1. Introduce this WARN_ONCE() on this event ever occurring.
2. Remove the self-IPI hack.
3. Move the IRQ_MOVE_CLEANUP_VECTOR to the sysvec range.

This implements step 1.

[ Previous versions of this patch had steps 2 and 3 reversed. ]

Suggested-by: Thomas Gleixner <tglx@xxxxxxxxxxxxx>
Signed-off-by: H. Peter Anvin (Intel) <hpa@xxxxxxxxx>
---
arch/x86/kernel/apic/vector.c | 5 +++++
1 file changed, 5 insertions(+)

diff --git a/arch/x86/kernel/apic/vector.c b/arch/x86/kernel/apic/vector.c
index 6dbdc7c22bb7..7ba2982a3585 100644
--- a/arch/x86/kernel/apic/vector.c
+++ b/arch/x86/kernel/apic/vector.c
@@ -939,9 +939,14 @@ DEFINE_IDTENTRY_SYSVEC(sysvec_irq_move_cleanup)
* to this CPU. IRQ_MOVE_CLEANUP_VECTOR is the lowest
* priority external vector, so on return from this
* interrupt the device interrupt will happen first.
+ *
+ * *** This should never happen with the current IRQ
+ * cleanup code, so WARN_ONCE() for now, and
+ * eventually get rid of this hack.
*/
irr = apic_read(APIC_IRR + (vector / 32 * 0x10));
if (irr & (1U << (vector % 32))) {
+ WARN_ONCE(1, "irq_move_cleanup called on still pending interrupt\n");
apic->send_IPI_self(IRQ_MOVE_CLEANUP_VECTOR);
continue;
}
--
2.31.1