xive: Fix IPI EOI logic in opal_xive_eoi()

We only want to directly EOI the interrupt used to emulate the MFRR,
for all the other "IPI" (aka XIVE produced interrupts), we want to
go via the normal source mechanism.

Signed-off-by: Benjamin Herrenschmidt <benh@kernel.crashing.org>
Signed-off-by: Stewart Smith <stewart@linux.vnet.ibm.com>

hw/xive.c

@@ -331,7 +331,7 @@ struct xive {
 	 * potentially handle more than one block per chip in the future.
 	 */
 	uint32_t	int_hw_bot;	/* Bottom of HW allocation */
-	uint32_t	int_ipi_top;	/* Highest IPI handed out so far */
+	uint32_t	int_ipi_top;	/* Highest IPI handed out so far + 1 */
 
 	/* Embedded source IPIs */
 	struct xive_src	ipis;
@@ -1995,25 +1995,25 @@ static int64_t opal_xive_eoi(uint32_t xirr)
 	}
 #endif
 
-	/* Perform source level EOI if it's a HW interrupt, otherwise,
-	 * EOI ourselves
+	/* Perform source level EOI if it's not our emulated MFRR IPI
+	 * otherwise EOI ourselves
 	 */
 	src_x = xive_from_isn(isn);
 	if (src_x) {
 		uint32_t idx = GIRQ_TO_IDX(isn);
 
 		/* Is it an IPI ? */
-		if (idx < src_x->int_ipi_top) {
-			xive_vdbg(src_x, "EOI of IDX %x in IPI range\n", idx);
+		if (special_ipi) {
 			xive_ipi_eoi(src_x, idx);
 
-			/* It was a special IPI, check mfrr and eventually
-			 * re-trigger. We check against the new CPPR since
-			 * we are about to update the HW.
+			/* Check mfrr and eventually re-trigger. We check
+			 * against the new CPPR since we are about to update
+			 * the HW.
 			 */
-			if (special_ipi && xs->mfrr < cppr)
+			if (xs->mfrr < cppr)
 				xive_ipi_trigger(src_x, idx);
 		} else {
+			/* Otherwise go through the source mechanism */
 			xive_vdbg(src_x, "EOI of IDX %x in EXT range\n", idx);
 			irq_source_eoi(isn);
 		}