[v2,13/43] bus/fslmc: get MC VFIO group FD directly

Message ID 20240918075056.1838654-14-vanshika.shukla@nxp.com (mailing list archive)
State New
Delegated to: Thomas Monjalon
Headers
Series DPAA2 specific patches |

Checks

Context Check Description
ci/checkpatch success coding style OK

Commit Message

vanshika.shukla@nxp.com Sept. 18, 2024, 7:50 a.m. UTC
From: Jun Yang <jun.yang@nxp.com>

Get vfio group fd directly from file system instead of
from RTE API to avoid conflicting with PCIe VFIO.
FSL MC VFIO should have it's own logic which doe NOT depend on
RTE VFIO.

Signed-off-by: Jun Yang <jun.yang@nxp.com>
---
 drivers/bus/fslmc/fslmc_vfio.c | 88 ++++++++++++++++++++++++++--------
 drivers/bus/fslmc/meson.build  |  3 +-
 2 files changed, 71 insertions(+), 20 deletions(-)
  

Patch

diff --git a/drivers/bus/fslmc/fslmc_vfio.c b/drivers/bus/fslmc/fslmc_vfio.c
index 17163333af..1cc256f849 100644
--- a/drivers/bus/fslmc/fslmc_vfio.c
+++ b/drivers/bus/fslmc/fslmc_vfio.c
@@ -1,7 +1,7 @@ 
 /* SPDX-License-Identifier: BSD-3-Clause
  *
  *   Copyright (c) 2015-2016 Freescale Semiconductor, Inc. All rights reserved.
- *   Copyright 2016-2021 NXP
+ *   Copyright 2016-2023 NXP
  *
  */
 
@@ -30,6 +30,7 @@ 
 #include <rte_kvargs.h>
 #include <dev_driver.h>
 #include <rte_eal_memconfig.h>
+#include <eal_vfio.h>
 
 #include "private.h"
 #include "fslmc_vfio.h"
@@ -440,6 +441,59 @@  int rte_fslmc_vfio_dmamap(void)
 	return 0;
 }
 
+static int
+fslmc_vfio_open_group_fd(int iommu_group_num)
+{
+	int vfio_group_fd;
+	char filename[PATH_MAX];
+	struct rte_mp_msg mp_req, *mp_rep;
+	struct rte_mp_reply mp_reply = {0};
+	struct timespec ts = {.tv_sec = 5, .tv_nsec = 0};
+	struct vfio_mp_param *p = (struct vfio_mp_param *)mp_req.param;
+
+	/* if primary, try to open the group */
+	if (rte_eal_process_type() == RTE_PROC_PRIMARY) {
+		/* try regular group format */
+		snprintf(filename, sizeof(filename),
+			VFIO_GROUP_FMT, iommu_group_num);
+		vfio_group_fd = open(filename, O_RDWR);
+		if (vfio_group_fd <= 0) {
+			DPAA2_BUS_ERR("Open VFIO group(%s) failed(%d)",
+				filename, vfio_group_fd);
+		}
+
+		return vfio_group_fd;
+	}
+	/* if we're in a secondary process, request group fd from the primary
+	 * process via mp channel.
+	 */
+	p->req = SOCKET_REQ_GROUP;
+	p->group_num = iommu_group_num;
+	strcpy(mp_req.name, EAL_VFIO_MP);
+	mp_req.len_param = sizeof(*p);
+	mp_req.num_fds = 0;
+
+	vfio_group_fd = -1;
+	if (rte_mp_request_sync(&mp_req, &mp_reply, &ts) == 0 &&
+	    mp_reply.nb_received == 1) {
+		mp_rep = &mp_reply.msgs[0];
+		p = (struct vfio_mp_param *)mp_rep->param;
+		if (p->result == SOCKET_OK && mp_rep->num_fds == 1) {
+			vfio_group_fd = mp_rep->fds[0];
+		} else if (p->result == SOCKET_NO_FD) {
+			DPAA2_BUS_ERR("Bad VFIO group fd");
+			vfio_group_fd = 0;
+		}
+	}
+
+	free(mp_reply.msgs);
+	if (vfio_group_fd < 0) {
+		DPAA2_BUS_ERR("Cannot request group fd(%d)",
+			vfio_group_fd);
+	}
+	return vfio_group_fd;
+}
+
 static int
 fslmc_vfio_setup_device(const char *sysfs_base, const char *dev_addr,
 		int *vfio_dev_fd, struct vfio_device_info *device_info)
@@ -455,7 +509,7 @@  fslmc_vfio_setup_device(const char *sysfs_base, const char *dev_addr,
 		return -1;
 
 	/* get the actual group fd */
-	vfio_group_fd = rte_vfio_get_group_fd(iommu_group_no);
+	vfio_group_fd = vfio_group.fd;
 	if (vfio_group_fd < 0 && vfio_group_fd != -ENOENT)
 		return -1;
 
@@ -891,6 +945,11 @@  fslmc_vfio_close_group(void)
 		}
 	}
 
+	if (vfio_group.fd > 0) {
+		close(vfio_group.fd);
+		vfio_group.fd = 0;
+	}
+
 	return 0;
 }
 
@@ -1081,7 +1140,6 @@  fslmc_vfio_setup_group(void)
 {
 	int groupid;
 	int ret;
-	int vfio_container_fd;
 	struct vfio_group_status status = { .argsz = sizeof(status) };
 
 	/* if already done once */
@@ -1100,16 +1158,9 @@  fslmc_vfio_setup_group(void)
 		return 0;
 	}
 
-	ret = rte_vfio_container_create();
-	if (ret < 0) {
-		DPAA2_BUS_ERR("Failed to open VFIO container");
-		return ret;
-	}
-	vfio_container_fd = ret;
-
 	/* Get the actual group fd */
-	ret = rte_vfio_container_group_bind(vfio_container_fd, groupid);
-	if (ret < 0)
+	ret = fslmc_vfio_open_group_fd(groupid);
+	if (ret <= 0)
 		return ret;
 	vfio_group.fd = ret;
 
@@ -1118,14 +1169,14 @@  fslmc_vfio_setup_group(void)
 	if (ret) {
 		DPAA2_BUS_ERR("VFIO error getting group status");
 		close(vfio_group.fd);
-		rte_vfio_clear_group(vfio_group.fd);
+		vfio_group.fd = 0;
 		return ret;
 	}
 
 	if (!(status.flags & VFIO_GROUP_FLAGS_VIABLE)) {
 		DPAA2_BUS_ERR("VFIO group not viable");
 		close(vfio_group.fd);
-		rte_vfio_clear_group(vfio_group.fd);
+		vfio_group.fd = 0;
 		return -EPERM;
 	}
 	/* Since Group is VIABLE, Store the groupid */
@@ -1136,11 +1187,10 @@  fslmc_vfio_setup_group(void)
 		/* Now connect this IOMMU group to given container */
 		ret = vfio_connect_container();
 		if (ret) {
-			DPAA2_BUS_ERR(
-				"Error connecting container with groupid %d",
-				groupid);
+			DPAA2_BUS_ERR("vfio group(%d) connect failed(%d)",
+				groupid, ret);
 			close(vfio_group.fd);
-			rte_vfio_clear_group(vfio_group.fd);
+			vfio_group.fd = 0;
 			return ret;
 		}
 	}
@@ -1151,7 +1201,7 @@  fslmc_vfio_setup_group(void)
 		DPAA2_BUS_ERR("Error getting device %s fd from group %d",
 			      fslmc_container, vfio_group.groupid);
 		close(vfio_group.fd);
-		rte_vfio_clear_group(vfio_group.fd);
+		vfio_group.fd = 0;
 		return ret;
 	}
 	container_device_fd = ret;
diff --git a/drivers/bus/fslmc/meson.build b/drivers/bus/fslmc/meson.build
index 162ca286fe..70098ad778 100644
--- a/drivers/bus/fslmc/meson.build
+++ b/drivers/bus/fslmc/meson.build
@@ -1,5 +1,5 @@ 
 # SPDX-License-Identifier: BSD-3-Clause
-# Copyright 2018,2021 NXP
+# Copyright 2018-2023 NXP
 
 if not is_linux
     build = false
@@ -27,3 +27,4 @@  sources = files(
 )
 
 includes += include_directories('mc', 'qbman/include', 'portal')
+includes += include_directories('../../../lib/eal/linux')