Linux CXL
 help / color / mirror / Atom feed
From: <alejandro.lucero-palau@amd.com>
To: <linux-cxl@vger.kernel.org>, <djbw@kernel.org>,
	<edward.cree@amd.com>, <davem@davemloft.net>, <kuba@kernel.org>,
	<pabeni@redhat.com>, <edumazet@google.com>,
	<dave.jiang@intel.com>
Cc: Alejandro Lucero <alucerop@amd.com>
Subject: [PATCH v26 8/8] sfc: support pio mapping based on cxl
Date: Thu, 23 Apr 2026 19:05:28 +0100	[thread overview]
Message-ID: <20260423180528.17166-9-alejandro.lucero-palau@amd.com> (raw)
In-Reply-To: <20260423180528.17166-1-alejandro.lucero-palau@amd.com>

From: Alejandro Lucero <alucerop@amd.com>

A PIO buffer is a region of device memory to which the driver can write a
packet for TX, with the device handling the transmit doorbell without
requiring a DMA for getting the packet data, which helps reducing latency
in certain exchanges. With CXL mem protocol this latency can be lowered
further.

With a device supporting CXL and successfully initialised, use the cxl
region to map the memory range and use this mapping for PIO buffers.

Add the disabling of those CXL-based PIO buffers if the callback for
potential cxl endpoint removal by the CXL core happens.

Signed-off-by: Alejandro Lucero <alucerop@amd.com>
---
 drivers/net/ethernet/sfc/ef10.c       | 72 ++++++++++++++++++++++++---
 drivers/net/ethernet/sfc/efx.h        |  1 +
 drivers/net/ethernet/sfc/efx_cxl.c    |  2 +
 drivers/net/ethernet/sfc/net_driver.h |  2 +
 drivers/net/ethernet/sfc/nic.h        |  3 ++
 5 files changed, 73 insertions(+), 7 deletions(-)

diff --git a/drivers/net/ethernet/sfc/ef10.c b/drivers/net/ethernet/sfc/ef10.c
index 7e04f115bbaa..be9bd8d5aec9 100644
--- a/drivers/net/ethernet/sfc/ef10.c
+++ b/drivers/net/ethernet/sfc/ef10.c
@@ -24,6 +24,7 @@
 #include <linux/wait.h>
 #include <linux/workqueue.h>
 #include <net/udp_tunnel.h>
+#include "efx_cxl.h"
 
 /* Hardware control for EF10 architecture including 'Huntington'. */
 
@@ -106,7 +107,7 @@ static int efx_ef10_get_vf_index(struct efx_nic *efx)
 
 static int efx_ef10_init_datapath_caps(struct efx_nic *efx)
 {
-	MCDI_DECLARE_BUF(outbuf, MC_CMD_GET_CAPABILITIES_V4_OUT_LEN);
+	MCDI_DECLARE_BUF(outbuf, MC_CMD_GET_CAPABILITIES_V7_OUT_LEN);
 	struct efx_ef10_nic_data *nic_data = efx->nic_data;
 	size_t outlen;
 	int rc;
@@ -177,6 +178,12 @@ static int efx_ef10_init_datapath_caps(struct efx_nic *efx)
 			  efx->num_mac_stats);
 	}
 
+	if (outlen < MC_CMD_GET_CAPABILITIES_V7_OUT_LEN)
+		nic_data->datapath_caps3 = 0;
+	else
+		nic_data->datapath_caps3 = MCDI_DWORD(outbuf,
+						      GET_CAPABILITIES_V7_OUT_FLAGS3);
+
 	return 0;
 }
 
@@ -771,6 +778,35 @@ static int efx_ef10_alloc_piobufs(struct efx_nic *efx, unsigned int n)
 	return rc;
 }
 
+#ifdef CONFIG_SFC_CXL
+/* Invoked from cxl core when a cxl region is removed. This is expected at
+ * driver exit linked to cxl core devm releases which does not require the
+ * below sync.
+ *
+ * However, it is required when user space actions triggger such a cxl region
+ * removal forcing any cxl piobuf usage to stop. Setting per tx queue piobuf
+ * to NULL is safe if such a tx queue is not currently in use inside
+ * efx_hard_start_xmit() implying tx_queue locked.
+ *
+ * After this the cxl region physical range can be safely unmap.
+ */
+void efx_ef10_disable_piobufs(struct efx_nic *efx)
+{
+	struct efx_tx_queue *tx_queue;
+	struct efx_channel *channel;
+
+	local_bh_disable();
+	efx_for_each_channel(channel, efx)
+		efx_for_each_channel_tx_queue(tx_queue, channel) {
+			HARD_TX_LOCK(efx->net_dev, tx_queue->core_txq,
+				     smp_processor_id());
+			tx_queue->piobuf = NULL;
+			HARD_TX_UNLOCK(efx->net_dev, tx_queue->core_txq);
+		}
+	local_bh_enable();
+}
+#endif
+
 static int efx_ef10_link_piobufs(struct efx_nic *efx)
 {
 	struct efx_ef10_nic_data *nic_data = efx->nic_data;
@@ -1140,6 +1176,9 @@ static int efx_ef10_dimension_resources(struct efx_nic *efx)
 	unsigned int channel_vis, pio_write_vi_base, max_vis;
 	struct efx_ef10_nic_data *nic_data = efx->nic_data;
 	unsigned int uc_mem_map_size, wc_mem_map_size;
+#ifdef CONFIG_SFC_CXL
+	struct efx_probe_data *probe_data;
+#endif
 	void __iomem *membase;
 	int rc;
 
@@ -1263,8 +1302,25 @@ static int efx_ef10_dimension_resources(struct efx_nic *efx)
 	iounmap(efx->membase);
 	efx->membase = membase;
 
-	/* Set up the WC mapping if needed */
-	if (wc_mem_map_size) {
+	if (!wc_mem_map_size)
+		goto skip_pio;
+
+	/* Set up the WC mapping */
+
+#ifdef CONFIG_SFC_CXL
+	probe_data = container_of(efx, struct efx_probe_data, efx);
+	if ((nic_data->datapath_caps3 &
+	    (1 << MC_CMD_GET_CAPABILITIES_V7_OUT_CXL_CONFIG_ENABLE_LBN)) &&
+	    probe_data->cxl_pio_initialised) {
+		/* Using PIO through CXL mapping */
+		nic_data->pio_write_base = probe_data->cxl->ctpio_cxl;
+		nic_data->pio_write_vi_base = pio_write_vi_base;
+
+		probe_data->cxl_pio_in_use = true;
+	} else
+#endif
+	{
+		/* Using legacy PIO BAR mapping */
 		nic_data->wc_membase = ioremap_wc(efx->membase_phys +
 						  uc_mem_map_size,
 						  wc_mem_map_size);
@@ -1279,12 +1335,14 @@ static int efx_ef10_dimension_resources(struct efx_nic *efx)
 			nic_data->wc_membase +
 			(pio_write_vi_base * efx->vi_stride + ER_DZ_TX_PIOBUF -
 			 uc_mem_map_size);
-
-		rc = efx_ef10_link_piobufs(efx);
-		if (rc)
-			efx_ef10_free_piobufs(efx);
 	}
 
+	rc = efx_ef10_link_piobufs(efx);
+	if (rc)
+		efx_ef10_free_piobufs(efx);
+
+skip_pio:
+
 	netif_dbg(efx, probe, efx->net_dev,
 		  "memory BAR at %pa (virtual %p+%x UC, %p+%x WC)\n",
 		  &efx->membase_phys, efx->membase, uc_mem_map_size,
diff --git a/drivers/net/ethernet/sfc/efx.h b/drivers/net/ethernet/sfc/efx.h
index 45e191686625..37fd1cf96582 100644
--- a/drivers/net/ethernet/sfc/efx.h
+++ b/drivers/net/ethernet/sfc/efx.h
@@ -237,4 +237,5 @@ static inline bool efx_rwsem_assert_write_locked(struct rw_semaphore *sem)
 int efx_xdp_tx_buffers(struct efx_nic *efx, int n, struct xdp_frame **xdpfs,
 		       bool flush);
 
+void efx_ef10_disable_piobufs(struct efx_nic *efx);
 #endif /* EFX_EFX_H */
diff --git a/drivers/net/ethernet/sfc/efx_cxl.c b/drivers/net/ethernet/sfc/efx_cxl.c
index 9ca16b051d62..8ba1645cfed8 100644
--- a/drivers/net/ethernet/sfc/efx_cxl.c
+++ b/drivers/net/ethernet/sfc/efx_cxl.c
@@ -10,6 +10,7 @@
 #include <cxl/cxl.h>
 #include <cxl/pci.h>
 #include "net_driver.h"
+#include "efx.h"
 #include "efx_cxl.h"
 
 #define EFX_CTPIO_BUFFER_SIZE	SZ_256M
@@ -40,6 +41,7 @@ static int efx_cxl_map_region(void *data) {
 static void efx_cxl_unmap_region(void *data) {
 	struct efx_probe_data *probe_data = data;
 
+	efx_ef10_disable_piobufs(&probe_data->efx);
 	probe_data->cxl_pio_initialised = false;
 	iounmap(probe_data->cxl->ctpio_cxl);
 }
diff --git a/drivers/net/ethernet/sfc/net_driver.h b/drivers/net/ethernet/sfc/net_driver.h
index 3964b2c56609..bea4eecdf842 100644
--- a/drivers/net/ethernet/sfc/net_driver.h
+++ b/drivers/net/ethernet/sfc/net_driver.h
@@ -1207,6 +1207,7 @@ struct efx_cxl;
  * @efx: Efx NIC details
  * @cxl: details of related cxl objects
  * @cxl_pio_initialised: cxl initialization outcome.
+ * @cxl_pio_in_use: PIO using CXL mapping
  */
 struct efx_probe_data {
 	struct pci_dev *pci_dev;
@@ -1214,6 +1215,7 @@ struct efx_probe_data {
 #ifdef CONFIG_SFC_CXL
 	struct efx_cxl *cxl;
 	bool cxl_pio_initialised;
+	bool cxl_pio_in_use;
 #endif
 };
 
diff --git a/drivers/net/ethernet/sfc/nic.h b/drivers/net/ethernet/sfc/nic.h
index ec3b2df43b68..7480f9995dfb 100644
--- a/drivers/net/ethernet/sfc/nic.h
+++ b/drivers/net/ethernet/sfc/nic.h
@@ -152,6 +152,8 @@ enum {
  *	%MC_CMD_GET_CAPABILITIES response)
  * @datapath_caps2: Further Capabilities of datapath firmware (FLAGS2 field of
  * %MC_CMD_GET_CAPABILITIES response)
+ * @datapath_caps3: Further Capabilities of datapath firmware (FLAGS3 field of
+ * %MC_CMD_GET_CAPABILITIES response)
  * @rx_dpcpu_fw_id: Firmware ID of the RxDPCPU
  * @tx_dpcpu_fw_id: Firmware ID of the TxDPCPU
  * @must_probe_vswitching: Flag: vswitching has yet to be setup after MC reboot
@@ -187,6 +189,7 @@ struct efx_ef10_nic_data {
 	bool must_check_datapath_caps;
 	u32 datapath_caps;
 	u32 datapath_caps2;
+	u32 datapath_caps3;
 	unsigned int rx_dpcpu_fw_id;
 	unsigned int tx_dpcpu_fw_id;
 	bool must_probe_vswitching;
-- 
2.34.1


  parent reply	other threads:[~2026-04-23 17:26 UTC|newest]

Thread overview: 20+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-04-23 18:05 [PATCH v26 0/8] Type2 device basic support alejandro.lucero-palau
2026-04-23 18:05 ` [PATCH v26 1/8] sfc: add cxl support alejandro.lucero-palau
2026-04-29 21:14   ` Cheatham, Benjamin
2026-05-01 10:07     ` Alejandro Lucero Palau
2026-04-23 18:05 ` [PATCH v26 2/8] cxl/sfc: Map cxl regs alejandro.lucero-palau
2026-04-23 18:05 ` [PATCH v26 3/8] cxl/sfc: Initialize dpa without a mailbox alejandro.lucero-palau
2026-04-23 18:05 ` [PATCH v26 4/8] cxl: Prepare memdev creation for type2 alejandro.lucero-palau
2026-04-30 23:23   ` Dan Williams (nvidia)
2026-04-23 18:05 ` [PATCH v26 5/8] sfc: create type2 cxl memdev alejandro.lucero-palau
2026-04-23 18:05 ` [PATCH v26 6/8] cxl: attach region to an accelerator/type2 memdev alejandro.lucero-palau
2026-04-29 21:14   ` Cheatham, Benjamin
2026-05-01 10:35     ` Alejandro Lucero Palau
2026-05-01  2:00   ` Dan Williams (nvidia)
2026-05-01 10:59     ` Alejandro Lucero Palau
2026-05-02  0:46       ` Dan Williams (nvidia)
2026-05-05 20:51         ` Alejandro Lucero Palau
2026-04-23 18:05 ` [PATCH v26 7/8] cxl: Avoid dax creation for accelerators alejandro.lucero-palau
2026-04-29 21:14   ` Cheatham, Benjamin
2026-04-23 18:05 ` alejandro.lucero-palau [this message]
2026-04-23 22:07 ` [PATCH v26 0/8] Type2 device basic support Dave Jiang

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20260423180528.17166-9-alejandro.lucero-palau@amd.com \
    --to=alejandro.lucero-palau@amd.com \
    --cc=alucerop@amd.com \
    --cc=dave.jiang@intel.com \
    --cc=davem@davemloft.net \
    --cc=djbw@kernel.org \
    --cc=edumazet@google.com \
    --cc=edward.cree@amd.com \
    --cc=kuba@kernel.org \
    --cc=linux-cxl@vger.kernel.org \
    --cc=pabeni@redhat.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox