diff mbox series

[iwl-next,07/12] idpf: compile singleq code only under default-n CONFIG_IDPF_SINGLEQ

Message ID 20240528134846.148890-8-aleksander.lobakin@intel.com
State Changes Requested
Delegated to: Anthony Nguyen
Headers show
Series idpf: XDP chapter I: convert Rx to libeth | expand

Commit Message

Alexander Lobakin May 28, 2024, 1:48 p.m. UTC
Currently, all HW supporting idpf supports the singleq model, but none
of it advertises it by default, as splitq is supported and preferred
for multiple reasons. Still, this almost dead code often times adds
hotpath branches and redundant cacheline accesses.
While it can't currently be removed, add CONFIG_IDPF_SINGLEQ and build
the singleq code only when it's enabled manually. This corresponds to
-10 Kb of object code size and a good bunch of hotpath checks.
idpf_is_queue_model_split() works as a gate and compiles out to `true`
when the config option is disabled.

Reviewed-by: Przemek Kitszel <przemyslaw.kitszel@intel.com>
Signed-off-by: Alexander Lobakin <aleksander.lobakin@intel.com>
---
 drivers/net/ethernet/intel/Kconfig            | 13 +--------
 drivers/net/ethernet/intel/idpf/Kconfig       | 27 +++++++++++++++++++
 drivers/net/ethernet/intel/idpf/Makefile      |  3 ++-
 drivers/net/ethernet/intel/idpf/idpf.h        |  3 ++-
 drivers/net/ethernet/intel/idpf/idpf_txrx.c   |  2 +-
 .../net/ethernet/intel/idpf/idpf_virtchnl.c   | 15 ++++++++---
 6 files changed, 44 insertions(+), 19 deletions(-)
 create mode 100644 drivers/net/ethernet/intel/idpf/Kconfig

Comments

Jacob Keller May 29, 2024, 12:47 a.m. UTC | #1
On 5/28/2024 6:48 AM, Alexander Lobakin wrote:
> Currently, all HW supporting idpf supports the singleq model, but none
> of it advertises it by default, as splitq is supported and preferred
> for multiple reasons. Still, this almost dead code often times adds
> hotpath branches and redundant cacheline accesses.
> While it can't currently be removed, add CONFIG_IDPF_SINGLEQ and build
> the singleq code only when it's enabled manually. This corresponds to
> -10 Kb of object code size and a good bunch of hotpath checks.
> idpf_is_queue_model_split() works as a gate and compiles out to `true`
> when the config option is disabled.
> 

Could you clarify why we can't remove it? Presumably there exist some
users out there who depend on this (and can thus set the CONFIG option)?
Or is it because we intend to support some hardware which only supports
singleq sometime soon?

This is a huge improvement over leaving it as-is, but I do wonder why
not just completely remove it. Its also a surprisingly small patch, but
I guess thats because the singleq_txrx is a separate file.
Alexander Lobakin June 12, 2024, 1:15 p.m. UTC | #2
From: Jacob Keller <jacob.e.keller@intel.com>
Date: Tue, 28 May 2024 17:47:32 -0700

> 
> 
> On 5/28/2024 6:48 AM, Alexander Lobakin wrote:
>> Currently, all HW supporting idpf supports the singleq model, but none
>> of it advertises it by default, as splitq is supported and preferred
>> for multiple reasons. Still, this almost dead code often times adds
>> hotpath branches and redundant cacheline accesses.
>> While it can't currently be removed, add CONFIG_IDPF_SINGLEQ and build
>> the singleq code only when it's enabled manually. This corresponds to
>> -10 Kb of object code size and a good bunch of hotpath checks.
>> idpf_is_queue_model_split() works as a gate and compiles out to `true`
>> when the config option is disabled.
>>
> 
> Could you clarify why we can't remove it? Presumably there exist some
> users out there who depend on this (and can thus set the CONFIG option)?
> Or is it because we intend to support some hardware which only supports
> singleq sometime soon?
> 
> This is a huge improvement over leaving it as-is, but I do wonder why
> not just completely remove it. Its also a surprisingly small patch, but
> I guess thats because the singleq_txrx is a separate file.

I'd be happy to remove it completely, as it has bugs, makes some
optimizations more difficult to do etc. But I was told there *might* be
real users for it, although with no particular examples, so it can't be
removed. Maybe Emil or someone else from the idpf team will finally give
some ¯\_(ツ)_/¯
In general, I have no idea why one might want to use singleq when splitq
is available: it's way faster (one queue is strictly RO, the other one
is strictly WO -> better CL usage, less descriptor reads and writes
etc.) and more flexible. XDP will be available in the splitq mode only
-- it would be hell otherwise with no practical benefit.

Thanks,
Olek
Jacob Keller June 12, 2024, 10:43 p.m. UTC | #3
On 6/12/2024 6:15 AM, Alexander Lobakin wrote:
> From: Jacob Keller <jacob.e.keller@intel.com>
> Date: Tue, 28 May 2024 17:47:32 -0700
> 
>>
>>
>> On 5/28/2024 6:48 AM, Alexander Lobakin wrote:
>>> Currently, all HW supporting idpf supports the singleq model, but none
>>> of it advertises it by default, as splitq is supported and preferred
>>> for multiple reasons. Still, this almost dead code often times adds
>>> hotpath branches and redundant cacheline accesses.
>>> While it can't currently be removed, add CONFIG_IDPF_SINGLEQ and build
>>> the singleq code only when it's enabled manually. This corresponds to
>>> -10 Kb of object code size and a good bunch of hotpath checks.
>>> idpf_is_queue_model_split() works as a gate and compiles out to `true`
>>> when the config option is disabled.
>>>
>>
>> Could you clarify why we can't remove it? Presumably there exist some
>> users out there who depend on this (and can thus set the CONFIG option)?
>> Or is it because we intend to support some hardware which only supports
>> singleq sometime soon?
>>
>> This is a huge improvement over leaving it as-is, but I do wonder why
>> not just completely remove it. Its also a surprisingly small patch, but
>> I guess thats because the singleq_txrx is a separate file.
> 
> I'd be happy to remove it completely, as it has bugs, makes some
> optimizations more difficult to do etc. But I was told there *might* be
> real users for it, although with no particular examples, so it can't be
> removed. Maybe Emil or someone else from the idpf team will finally give
> some ¯\_(ツ)_/¯
> In general, I have no idea why one might want to use singleq when splitq
> is available: it's way faster (one queue is strictly RO, the other one
> is strictly WO -> better CL usage, less descriptor reads and writes
> etc.) and more flexible. XDP will be available in the splitq mode only
> -- it would be hell otherwise with no practical benefit.
> 
> Thanks,
> Olek

Makes sense, thanks for the explanation.
diff mbox series

Patch

diff --git a/drivers/net/ethernet/intel/Kconfig b/drivers/net/ethernet/intel/Kconfig
index e0287fbd501d..0375c7448a57 100644
--- a/drivers/net/ethernet/intel/Kconfig
+++ b/drivers/net/ethernet/intel/Kconfig
@@ -384,17 +384,6 @@  config IGC_LEDS
 	  Optional support for controlling the NIC LED's with the netdev
 	  LED trigger.
 
-config IDPF
-	tristate "Intel(R) Infrastructure Data Path Function Support"
-	depends on PCI_MSI
-	select DIMLIB
-	select PAGE_POOL
-	select PAGE_POOL_STATS
-	help
-	  This driver supports Intel(R) Infrastructure Data Path Function
-	  devices.
-
-	  To compile this driver as a module, choose M here. The module
-	  will be called idpf.
+source "drivers/net/ethernet/intel/idpf/Kconfig"
 
 endif # NET_VENDOR_INTEL
diff --git a/drivers/net/ethernet/intel/idpf/Kconfig b/drivers/net/ethernet/intel/idpf/Kconfig
new file mode 100644
index 000000000000..9082c16edb7e
--- /dev/null
+++ b/drivers/net/ethernet/intel/idpf/Kconfig
@@ -0,0 +1,27 @@ 
+# SPDX-License-Identifier: GPL-2.0-only
+# Copyright (C) 2024 Intel Corporation
+
+config IDPF
+	tristate "Intel(R) Infrastructure Data Path Function Support"
+	depends on PCI_MSI
+	select DIMLIB
+	select PAGE_POOL
+	select PAGE_POOL_STATS
+	help
+	  This driver supports Intel(R) Infrastructure Data Path Function
+	  devices.
+
+	  To compile this driver as a module, choose M here. The module
+	  will be called idpf.
+
+if IDPF
+
+config IDPF_SINGLEQ
+	bool "idpf singleq support"
+	help
+	  This option enables support for legacy single Rx/Tx queues w/no
+	  completion and fill queues. Only enable if you have hardware which
+	  wants to work in this mode as it increases the driver size and adds
+	  runtme checks on hotpath.
+
+endif # IDPF
diff --git a/drivers/net/ethernet/intel/idpf/Makefile b/drivers/net/ethernet/intel/idpf/Makefile
index 6844ead2f3ac..2ce01a0b5898 100644
--- a/drivers/net/ethernet/intel/idpf/Makefile
+++ b/drivers/net/ethernet/intel/idpf/Makefile
@@ -12,7 +12,8 @@  idpf-y := \
 	idpf_ethtool.o		\
 	idpf_lib.o		\
 	idpf_main.o		\
-	idpf_singleq_txrx.o	\
 	idpf_txrx.o		\
 	idpf_virtchnl.o 	\
 	idpf_vf_dev.o
+
+idpf-$(CONFIG_IDPF_SINGLEQ)	+= idpf_singleq_txrx.o
diff --git a/drivers/net/ethernet/intel/idpf/idpf.h b/drivers/net/ethernet/intel/idpf/idpf.h
index f9e43d171f17..5d9529f5b41b 100644
--- a/drivers/net/ethernet/intel/idpf/idpf.h
+++ b/drivers/net/ethernet/intel/idpf/idpf.h
@@ -599,7 +599,8 @@  struct idpf_adapter {
  */
 static inline int idpf_is_queue_model_split(u16 q_model)
 {
-	return q_model == VIRTCHNL2_QUEUE_MODEL_SPLIT;
+	return !IS_ENABLED(CONFIG_IDPF_SINGLEQ) ||
+	       q_model == VIRTCHNL2_QUEUE_MODEL_SPLIT;
 }
 
 #define idpf_is_cap_ena(adapter, field, flag) \
diff --git a/drivers/net/ethernet/intel/idpf/idpf_txrx.c b/drivers/net/ethernet/intel/idpf/idpf_txrx.c
index 5dd1b1a9e624..e79a8f9dfc40 100644
--- a/drivers/net/ethernet/intel/idpf/idpf_txrx.c
+++ b/drivers/net/ethernet/intel/idpf/idpf_txrx.c
@@ -1309,7 +1309,7 @@  static void idpf_vport_calc_numq_per_grp(struct idpf_vport *vport,
 static void idpf_rxq_set_descids(const struct idpf_vport *vport,
 				 struct idpf_rx_queue *q)
 {
-	if (vport->rxq_model == VIRTCHNL2_QUEUE_MODEL_SPLIT) {
+	if (idpf_is_queue_model_split(vport->rxq_model)) {
 		q->rxdids = VIRTCHNL2_RXDID_2_FLEX_SPLITQ_M;
 	} else {
 		if (vport->base_rxd)
diff --git a/drivers/net/ethernet/intel/idpf/idpf_virtchnl.c b/drivers/net/ethernet/intel/idpf/idpf_virtchnl.c
index 44602b87cd41..1aa4770dfe18 100644
--- a/drivers/net/ethernet/intel/idpf/idpf_virtchnl.c
+++ b/drivers/net/ethernet/intel/idpf/idpf_virtchnl.c
@@ -1256,12 +1256,12 @@  int idpf_send_create_vport_msg(struct idpf_adapter *adapter,
 	vport_msg->vport_type = cpu_to_le16(VIRTCHNL2_VPORT_TYPE_DEFAULT);
 	vport_msg->vport_index = cpu_to_le16(idx);
 
-	if (adapter->req_tx_splitq)
+	if (adapter->req_tx_splitq || !IS_ENABLED(CONFIG_IDPF_SINGLEQ))
 		vport_msg->txq_model = cpu_to_le16(VIRTCHNL2_QUEUE_MODEL_SPLIT);
 	else
 		vport_msg->txq_model = cpu_to_le16(VIRTCHNL2_QUEUE_MODEL_SINGLE);
 
-	if (adapter->req_rx_splitq)
+	if (adapter->req_rx_splitq || !IS_ENABLED(CONFIG_IDPF_SINGLEQ))
 		vport_msg->rxq_model = cpu_to_le16(VIRTCHNL2_QUEUE_MODEL_SPLIT);
 	else
 		vport_msg->rxq_model = cpu_to_le16(VIRTCHNL2_QUEUE_MODEL_SINGLE);
@@ -1323,10 +1323,17 @@  int idpf_check_supported_desc_ids(struct idpf_vport *vport)
 
 	vport_msg = adapter->vport_params_recvd[vport->idx];
 
+	if (!IS_ENABLED(CONFIG_IDPF_SINGLEQ) &&
+	    (vport_msg->rxq_model == VIRTCHNL2_QUEUE_MODEL_SINGLE ||
+	     vport_msg->txq_model == VIRTCHNL2_QUEUE_MODEL_SINGLE)) {
+		pci_err(adapter->pdev, "singleq mode requested, but not compiled-in\n");
+		return -EOPNOTSUPP;
+	}
+
 	rx_desc_ids = le64_to_cpu(vport_msg->rx_desc_ids);
 	tx_desc_ids = le64_to_cpu(vport_msg->tx_desc_ids);
 
-	if (vport->rxq_model == VIRTCHNL2_QUEUE_MODEL_SPLIT) {
+	if (idpf_is_queue_model_split(vport->rxq_model)) {
 		if (!(rx_desc_ids & VIRTCHNL2_RXDID_2_FLEX_SPLITQ_M)) {
 			dev_info(&adapter->pdev->dev, "Minimum RX descriptor support not provided, using the default\n");
 			vport_msg->rx_desc_ids = cpu_to_le64(VIRTCHNL2_RXDID_2_FLEX_SPLITQ_M);
@@ -1336,7 +1343,7 @@  int idpf_check_supported_desc_ids(struct idpf_vport *vport)
 			vport->base_rxd = true;
 	}
 
-	if (vport->txq_model != VIRTCHNL2_QUEUE_MODEL_SPLIT)
+	if (!idpf_is_queue_model_split(vport->txq_model))
 		return 0;
 
 	if ((tx_desc_ids & MIN_SUPPORT_TXDID) != MIN_SUPPORT_TXDID) {