[PATCH net-next v4 2/7] enic: Make MSI-X I/O interrupts come after the other required ones

Nelson Escobar posted 7 patches 1 week, 2 days ago
[PATCH net-next v4 2/7] enic: Make MSI-X I/O interrupts come after the other required ones
Posted by Nelson Escobar 1 week, 2 days ago
The VIC hardware has a constraint that the MSIX interrupt used for errors
be specified as a 7 bit number.  Before this patch, it was allocated after
the I/O interrupts, which would cause a problem if 128 or more I/O
interrupts are in use.

So make the required interrupts come before the I/O interrupts to
guarantee the error interrupt offset never exceeds 7 bits.

Co-developed-by: John Daley <johndale@cisco.com>
Signed-off-by: John Daley <johndale@cisco.com>
Co-developed-by: Satish Kharat <satishkh@cisco.com>
Signed-off-by: Satish Kharat <satishkh@cisco.com>
Reviewed-by: Simon Horman <horms@kernel.org>
Signed-off-by: Nelson Escobar <neescoba@cisco.com>
---
 drivers/net/ethernet/cisco/enic/enic.h     | 20 +++++++++++++++-----
 drivers/net/ethernet/cisco/enic/enic_res.c | 11 +++++++----
 2 files changed, 22 insertions(+), 9 deletions(-)

diff --git a/drivers/net/ethernet/cisco/enic/enic.h b/drivers/net/ethernet/cisco/enic/enic.h
index 07459eac2592ce5185321a619577404232cfbc2c..ec83a273d1ca40ae89f3c193207cf26814f6b277 100644
--- a/drivers/net/ethernet/cisco/enic/enic.h
+++ b/drivers/net/ethernet/cisco/enic/enic.h
@@ -280,18 +280,28 @@ static inline unsigned int enic_msix_wq_intr(struct enic *enic,
 	return enic->cq[enic_cq_wq(enic, wq)].interrupt_offset;
 }
 
-static inline unsigned int enic_msix_err_intr(struct enic *enic)
-{
-	return enic->rq_count + enic->wq_count;
-}
+/* MSIX interrupts are organized as the error interrupt, then the notify
+ * interrupt followed by all the I/O interrupts.  The error interrupt needs
+ * to fit in 7 bits due to hardware constraints
+ */
+#define ENIC_MSIX_RESERVED_INTR 2
+#define ENIC_MSIX_ERR_INTR	0
+#define ENIC_MSIX_NOTIFY_INTR	1
+#define ENIC_MSIX_IO_INTR_BASE	ENIC_MSIX_RESERVED_INTR
+#define ENIC_MSIX_MIN_INTR	(ENIC_MSIX_RESERVED_INTR + 2)
 
 #define ENIC_LEGACY_IO_INTR	0
 #define ENIC_LEGACY_ERR_INTR	1
 #define ENIC_LEGACY_NOTIFY_INTR	2
 
+static inline unsigned int enic_msix_err_intr(struct enic *enic)
+{
+	return ENIC_MSIX_ERR_INTR;
+}
+
 static inline unsigned int enic_msix_notify_intr(struct enic *enic)
 {
-	return enic->rq_count + enic->wq_count + 1;
+	return ENIC_MSIX_NOTIFY_INTR;
 }
 
 static inline bool enic_is_err_intr(struct enic *enic, int intr)
diff --git a/drivers/net/ethernet/cisco/enic/enic_res.c b/drivers/net/ethernet/cisco/enic/enic_res.c
index 60be09acb9fd56b642b7cabc77fac01f526b29a2..72b51e9d8d1a26a2cd18df9c9d702e5b11993b70 100644
--- a/drivers/net/ethernet/cisco/enic/enic_res.c
+++ b/drivers/net/ethernet/cisco/enic/enic_res.c
@@ -221,9 +221,12 @@ void enic_init_vnic_resources(struct enic *enic)
 
 	switch (intr_mode) {
 	case VNIC_DEV_INTR_MODE_INTX:
+		error_interrupt_enable = 1;
+		error_interrupt_offset = ENIC_LEGACY_ERR_INTR;
+		break;
 	case VNIC_DEV_INTR_MODE_MSIX:
 		error_interrupt_enable = 1;
-		error_interrupt_offset = enic->intr_count - 2;
+		error_interrupt_offset = enic_msix_err_intr(enic);
 		break;
 	default:
 		error_interrupt_enable = 0;
@@ -249,15 +252,15 @@ void enic_init_vnic_resources(struct enic *enic)
 
 	/* Init CQ resources
 	 *
-	 * CQ[0 - n+m-1] point to INTR[0] for INTx, MSI
-	 * CQ[0 - n+m-1] point to INTR[0 - n+m-1] for MSI-X
+	 * All CQs point to INTR[0] for INTx, MSI
+	 * CQ[i] point to INTR[ENIC_MSIX_IO_INTR_BASE + i] for MSI-X
 	 */
 
 	for (i = 0; i < enic->cq_count; i++) {
 
 		switch (intr_mode) {
 		case VNIC_DEV_INTR_MODE_MSIX:
-			interrupt_offset = i;
+			interrupt_offset = ENIC_MSIX_IO_INTR_BASE + i;
 			break;
 		default:
 			interrupt_offset = 0;

-- 
2.35.6
Re: [PATCH net-next v4 2/7] enic: Make MSI-X I/O interrupts come after the other required ones
Posted by Vadim Fedorenko 1 week, 1 day ago
On 13/11/2024 23:56, Nelson Escobar wrote:
> The VIC hardware has a constraint that the MSIX interrupt used for errors
> be specified as a 7 bit number.  Before this patch, it was allocated after
> the I/O interrupts, which would cause a problem if 128 or more I/O
> interrupts are in use.
> 
> So make the required interrupts come before the I/O interrupts to
> guarantee the error interrupt offset never exceeds 7 bits.
> 
> Co-developed-by: John Daley <johndale@cisco.com>
> Signed-off-by: John Daley <johndale@cisco.com>
> Co-developed-by: Satish Kharat <satishkh@cisco.com>
> Signed-off-by: Satish Kharat <satishkh@cisco.com>
> Reviewed-by: Simon Horman <horms@kernel.org>
> Signed-off-by: Nelson Escobar <neescoba@cisco.com>

Reviewed-by: Vadim Fedorenko <vadim.fedorenko@linux.dev>