[PATCH AUTOSEL for 4.4 05/35] KVM: x86: ioapic: Clear Remote IRR when entry is switched to edge-triggered

[Sasha Levin]

From: Nikita Leshenko <nikita.leshchenko@xxxxxxxxxx>

[ Upstream commit a8bfec2930525808c01f038825d1df3904638631 ]

Some OSes (Linux, Xen) use this behavior to clear the Remote IRR bit for
IOAPICs without an EOI register. They simulate the EOI message manually
by changing the trigger mode to edge and then back to level, with the
entry being masked during this.

QEMU implements this feature in commit ed1263c363c9
("ioapic: clear remote irr bit for edge-triggered interrupts")

As a side effect, this commit removes an incorrect behavior where Remote
IRR was cleared when the redirection table entry was rewritten. This is not
consistent with the manual and also opens an opportunity for a strange
behavior when a redirection table entry is modified from an interrupt
handler that handles the same entry: The modification will clear the
Remote IRR bit even though the interrupt handler is still running.

Signed-off-by: Nikita Leshenko <nikita.leshchenko@xxxxxxxxxx>
Reviewed-by: Liran Alon <liran.alon@xxxxxxxxxx>
Signed-off-by: Konrad Rzeszutek Wilk <konrad.wilk@xxxxxxxxxx>
Reviewed-by: Wanpeng Li <wanpeng.li@xxxxxxxxxxx>
Reviewed-by: Steve Rutherford <srutherford@xxxxxxxxxx>
Signed-off-by: Radim KrÄmÃÅ <rkrcmar@xxxxxxxxxx>
Signed-off-by: Sasha Levin <alexander.levin@xxxxxxxxxxxxx>
---
 arch/x86/kvm/ioapic.c | 11 ++++++++++-
 1 file changed, 10 insertions(+), 1 deletion(-)

diff --git a/arch/x86/kvm/ioapic.c b/arch/x86/kvm/ioapic.c
index 96ee7091becf..403dd464965c 100644
--- a/arch/x86/kvm/ioapic.c
+++ b/arch/x86/kvm/ioapic.c
@@ -296,8 +296,17 @@ static void ioapic_write_indirect(struct kvm_ioapic *ioapic, u32 val)
 		} else {
 			e->bits &= ~0xffffffffULL;
 			e->bits |= (u32) val;
-			e->fields.remote_irr = 0;
 		}
+
+		/*
+		 * Some OSes (Linux, Xen) assume that Remote IRR bit will
+		 * be cleared by IOAPIC hardware when the entry is configured
+		 * as edge-triggered. This behavior is used to simulate an
+		 * explicit EOI on IOAPICs that don't have the EOI register.
+		 */
+		if (e->fields.trig_mode == IOAPIC_EDGE_TRIG)
+			e->fields.remote_irr = 0;
+
 		mask_after = e->fields.mask;
 		if (mask_before != mask_after)
 			kvm_fire_mask_notifiers(ioapic->kvm, KVM_IRQCHIP_IOAPIC, index, mask_after);
-- 
2.11.0