vfio: move global config out of PCI files

The vfio_cfg is a module-global variable and so together with this
variable, it is necessary to move functions:

* pci_vfio_get_group_fd
  - renamed to vfio_get_group_fd
  - pci_* version removed (no other call in EAL)

* pci_vfio_setup_device
  - renamed as vfio_setup_device

* pci_vfio_enable
  - renamed as vfio_enable
  - generalized to check for a specific vfio driver presence
  - pci_* specialization preserved as a wrapper

* pci_vfio_is_enabled
  - renamed as vfio_is_enabled
  - generalized to check for a specific vfio driver presence
    to preserve the semantics of VFIO + PCI
  - pci_* specialization preserved as a wrapper

* clear_current_group
  - private function, just moved

To stop GCC complaining about "defined but not used", the private
function pci_vfio_get_group_no has been removed entirely.

Signed-off-by: Jan Viktorin <viktorin@rehivetech.com>
This commit is contained in:
Jan Viktorin 2016-07-04 17:16:46 +02:00 committed by Thomas Monjalon
parent 0747689b56
commit 221f7c220d
5 changed files with 309 additions and 295 deletions

View File

@ -94,7 +94,6 @@ int pci_vfio_ioport_unmap(struct rte_pci_ioport *p);
/* map VFIO resource prototype */ /* map VFIO resource prototype */
int pci_vfio_map_resource(struct rte_pci_device *dev); int pci_vfio_map_resource(struct rte_pci_device *dev);
int pci_vfio_get_group_fd(int iommu_group_fd);
#endif #endif

View File

@ -69,9 +69,6 @@ static struct rte_tailq_elem rte_vfio_tailq = {
}; };
EAL_REGISTER_TAILQ(rte_vfio_tailq) EAL_REGISTER_TAILQ(rte_vfio_tailq)
/* per-process VFIO config */
static struct vfio_config vfio_cfg;
int int
pci_vfio_read_config(const struct rte_intr_handle *intr_handle, pci_vfio_read_config(const struct rte_intr_handle *intr_handle,
void *buf, size_t len, off_t offs) void *buf, size_t len, off_t offs)
@ -299,255 +296,6 @@ pci_vfio_setup_interrupts(struct rte_pci_device *dev, int vfio_dev_fd)
return -1; return -1;
} }
/* open group fd or get an existing one */
int
pci_vfio_get_group_fd(int iommu_group_no)
{
int i;
int vfio_group_fd;
char filename[PATH_MAX];
/* check if we already have the group descriptor open */
for (i = 0; i < vfio_cfg.vfio_group_idx; i++)
if (vfio_cfg.vfio_groups[i].group_no == iommu_group_no)
return vfio_cfg.vfio_groups[i].fd;
/* if primary, try to open the group */
if (internal_config.process_type == RTE_PROC_PRIMARY) {
/* try regular group format */
snprintf(filename, sizeof(filename),
VFIO_GROUP_FMT, iommu_group_no);
vfio_group_fd = open(filename, O_RDWR);
if (vfio_group_fd < 0) {
/* if file not found, it's not an error */
if (errno != ENOENT) {
RTE_LOG(ERR, EAL, "Cannot open %s: %s\n", filename,
strerror(errno));
return -1;
}
/* special case: try no-IOMMU path as well */
snprintf(filename, sizeof(filename),
VFIO_NOIOMMU_GROUP_FMT, iommu_group_no);
vfio_group_fd = open(filename, O_RDWR);
if (vfio_group_fd < 0) {
if (errno != ENOENT) {
RTE_LOG(ERR, EAL, "Cannot open %s: %s\n", filename,
strerror(errno));
return -1;
}
return 0;
}
/* noiommu group found */
}
/* if the fd is valid, create a new group for it */
if (vfio_cfg.vfio_group_idx == VFIO_MAX_GROUPS) {
RTE_LOG(ERR, EAL, "Maximum number of VFIO groups reached!\n");
close(vfio_group_fd);
return -1;
}
vfio_cfg.vfio_groups[vfio_cfg.vfio_group_idx].group_no = iommu_group_no;
vfio_cfg.vfio_groups[vfio_cfg.vfio_group_idx].fd = vfio_group_fd;
return vfio_group_fd;
}
/* if we're in a secondary process, request group fd from the primary
* process via our socket
*/
else {
int socket_fd, ret;
socket_fd = vfio_mp_sync_connect_to_primary();
if (socket_fd < 0) {
RTE_LOG(ERR, EAL, " cannot connect to primary process!\n");
return -1;
}
if (vfio_mp_sync_send_request(socket_fd, SOCKET_REQ_GROUP) < 0) {
RTE_LOG(ERR, EAL, " cannot request container fd!\n");
close(socket_fd);
return -1;
}
if (vfio_mp_sync_send_request(socket_fd, iommu_group_no) < 0) {
RTE_LOG(ERR, EAL, " cannot send group number!\n");
close(socket_fd);
return -1;
}
ret = vfio_mp_sync_receive_request(socket_fd);
switch (ret) {
case SOCKET_NO_FD:
close(socket_fd);
return 0;
case SOCKET_OK:
vfio_group_fd = vfio_mp_sync_receive_fd(socket_fd);
/* if we got the fd, return it */
if (vfio_group_fd > 0) {
close(socket_fd);
return vfio_group_fd;
}
/* fall-through on error */
default:
RTE_LOG(ERR, EAL, " cannot get container fd!\n");
close(socket_fd);
return -1;
}
}
return -1;
}
/* parse IOMMU group number for a PCI device
* returns 1 on success, -1 for errors, 0 for non-existent group
*/
static int
pci_vfio_get_group_no(const char *pci_addr, int *iommu_group_no)
{
return vfio_get_group_no(pci_get_sysfs_path(), pci_addr, iommu_group_no);
}
static void
clear_current_group(void)
{
vfio_cfg.vfio_groups[vfio_cfg.vfio_group_idx].group_no = 0;
vfio_cfg.vfio_groups[vfio_cfg.vfio_group_idx].fd = -1;
}
/**
* Setup vfio_cfg for the device indentified by its address. It discovers
* the configured I/O MMU groups or sets a new one for the device. If a new
* groups is assigned, the DMA mapping is performed.
* Returns 0 on success, a negative value on failure and a positive value in
* case the given device cannot be managed this way.
*/
static int pci_vfio_setup_device(const char *pci_addr, int *vfio_dev_fd,
struct vfio_device_info *device_info)
{
struct vfio_group_status group_status = {
.argsz = sizeof(group_status)
};
int vfio_group_fd;
int iommu_group_no;
int ret;
/* get group number */
ret = pci_vfio_get_group_no(pci_addr, &iommu_group_no);
if (ret == 0) {
RTE_LOG(WARNING, EAL, " %s not managed by VFIO driver, skipping\n",
pci_addr);
return 1;
}
/* if negative, something failed */
if (ret < 0)
return -1;
/* get the actual group fd */
vfio_group_fd = pci_vfio_get_group_fd(iommu_group_no);
if (vfio_group_fd < 0)
return -1;
/* store group fd */
vfio_cfg.vfio_groups[vfio_cfg.vfio_group_idx].group_no = iommu_group_no;
vfio_cfg.vfio_groups[vfio_cfg.vfio_group_idx].fd = vfio_group_fd;
/* if group_fd == 0, that means the device isn't managed by VFIO */
if (vfio_group_fd == 0) {
RTE_LOG(WARNING, EAL, " %s not managed by VFIO driver, skipping\n",
pci_addr);
/* we store 0 as group fd to distinguish between existing but
* unbound VFIO groups, and groups that don't exist at all.
*/
vfio_cfg.vfio_group_idx++;
return 1;
}
/*
* at this point, we know that this group is viable (meaning, all devices
* are either bound to VFIO or not bound to anything)
*/
/* check if the group is viable */
ret = ioctl(vfio_group_fd, VFIO_GROUP_GET_STATUS, &group_status);
if (ret) {
RTE_LOG(ERR, EAL, " %s cannot get group status, "
"error %i (%s)\n", pci_addr, errno, strerror(errno));
close(vfio_group_fd);
clear_current_group();
return -1;
} else if (!(group_status.flags & VFIO_GROUP_FLAGS_VIABLE)) {
RTE_LOG(ERR, EAL, " %s VFIO group is not viable!\n", pci_addr);
close(vfio_group_fd);
clear_current_group();
return -1;
}
/* check if group does not have a container yet */
if (!(group_status.flags & VFIO_GROUP_FLAGS_CONTAINER_SET)) {
/* add group to a container */
ret = ioctl(vfio_group_fd, VFIO_GROUP_SET_CONTAINER,
&vfio_cfg.vfio_container_fd);
if (ret) {
RTE_LOG(ERR, EAL, " %s cannot add VFIO group to container, "
"error %i (%s)\n", pci_addr, errno, strerror(errno));
close(vfio_group_fd);
clear_current_group();
return -1;
}
/*
* at this point we know that this group has been successfully
* initialized, so we increment vfio_group_idx to indicate that we can
* add new groups.
*/
vfio_cfg.vfio_group_idx++;
}
/*
* pick an IOMMU type and set up DMA mappings for container
*
* needs to be done only once, only when at least one group is assigned to
* a container and only in primary process
*/
if (internal_config.process_type == RTE_PROC_PRIMARY &&
vfio_cfg.vfio_container_has_dma == 0) {
/* select an IOMMU type which we will be using */
const struct vfio_iommu_type *t =
vfio_set_iommu_type(vfio_cfg.vfio_container_fd);
if (!t) {
RTE_LOG(ERR, EAL, " %s failed to select IOMMU type\n", pci_addr);
return -1;
}
ret = t->dma_map_func(vfio_cfg.vfio_container_fd);
if (ret) {
RTE_LOG(ERR, EAL, " %s DMA remapping failed, "
"error %i (%s)\n", pci_addr, errno, strerror(errno));
return -1;
}
vfio_cfg.vfio_container_has_dma = 1;
}
/* get a file descriptor for the device */
*vfio_dev_fd = ioctl(vfio_group_fd, VFIO_GROUP_GET_DEVICE_FD, pci_addr);
if (*vfio_dev_fd < 0) {
/* if we cannot get a device fd, this simply means that this
* particular port is not bound to VFIO
*/
RTE_LOG(WARNING, EAL, " %s not managed by VFIO driver, skipping\n",
pci_addr);
return 1;
}
/* test and setup the device */
ret = ioctl(*vfio_dev_fd, VFIO_DEVICE_GET_INFO, device_info);
if (ret) {
RTE_LOG(ERR, EAL, " %s cannot get device info, "
"error %i (%s)\n", pci_addr, errno, strerror(errno));
close(*vfio_dev_fd);
return -1;
}
return 0;
}
/* /*
* map the PCI resources of a PCI device in virtual memory (VFIO version). * map the PCI resources of a PCI device in virtual memory (VFIO version).
* primary and secondary processes follow almost exactly the same path * primary and secondary processes follow almost exactly the same path
@ -575,7 +323,8 @@ pci_vfio_map_resource(struct rte_pci_device *dev)
snprintf(pci_addr, sizeof(pci_addr), PCI_PRI_FMT, snprintf(pci_addr, sizeof(pci_addr), PCI_PRI_FMT,
loc->domain, loc->bus, loc->devid, loc->function); loc->domain, loc->bus, loc->devid, loc->function);
if ((ret = pci_vfio_setup_device(pci_addr, &vfio_dev_fd, &device_info))) if ((ret = vfio_setup_device(pci_get_sysfs_path(), pci_addr,
&vfio_dev_fd, &device_info)))
return ret; return ret;
/* get MSI-X BAR, if any (we have to know where it is because we can't /* get MSI-X BAR, if any (we have to know where it is because we can't
@ -834,50 +583,12 @@ pci_vfio_ioport_unmap(struct rte_pci_ioport *p)
int int
pci_vfio_enable(void) pci_vfio_enable(void)
{ {
/* initialize group list */ return vfio_enable("vfio_pci");
int i;
int vfio_available;
for (i = 0; i < VFIO_MAX_GROUPS; i++) {
vfio_cfg.vfio_groups[i].fd = -1;
vfio_cfg.vfio_groups[i].group_no = -1;
}
/* inform the user that we are probing for VFIO */
RTE_LOG(INFO, EAL, "Probing VFIO support...\n");
/* check if vfio-pci module is loaded */
vfio_available = rte_eal_check_module("vfio_pci");
/* return error directly */
if (vfio_available == -1) {
RTE_LOG(INFO, EAL, "Could not get loaded module details!\n");
return -1;
}
/* return 0 if VFIO modules not loaded */
if (vfio_available == 0) {
RTE_LOG(DEBUG, EAL, "VFIO modules not loaded, "
"skipping VFIO support...\n");
return 0;
}
vfio_cfg.vfio_container_fd = vfio_get_container_fd();
/* check if we have VFIO driver enabled */
if (vfio_cfg.vfio_container_fd != -1) {
RTE_LOG(NOTICE, EAL, "VFIO support initialized\n");
vfio_cfg.vfio_enabled = 1;
} else {
RTE_LOG(NOTICE, EAL, "VFIO support could not be initialized\n");
}
return 0;
} }
int int
pci_vfio_is_enabled(void) pci_vfio_is_enabled(void)
{ {
return vfio_cfg.vfio_enabled; return vfio_is_enabled("vfio_pci");
} }
#endif #endif

View File

@ -310,7 +310,7 @@ pci_vfio_mp_sync_thread(void __rte_unused * arg)
continue; continue;
} }
fd = pci_vfio_get_group_fd(vfio_group_no); fd = vfio_get_group_fd(vfio_group_no);
if (fd < 0) if (fd < 0)
vfio_mp_sync_send_request(conn_sock, SOCKET_ERR); vfio_mp_sync_send_request(conn_sock, SOCKET_ERR);

View File

@ -42,9 +42,296 @@
#include "eal_filesystem.h" #include "eal_filesystem.h"
#include "eal_vfio.h" #include "eal_vfio.h"
#include "eal_private.h"
#ifdef VFIO_PRESENT #ifdef VFIO_PRESENT
/* per-process VFIO config */
static struct vfio_config vfio_cfg;
int
vfio_get_group_fd(int iommu_group_no)
{
int i;
int vfio_group_fd;
char filename[PATH_MAX];
/* check if we already have the group descriptor open */
for (i = 0; i < vfio_cfg.vfio_group_idx; i++)
if (vfio_cfg.vfio_groups[i].group_no == iommu_group_no)
return vfio_cfg.vfio_groups[i].fd;
/* if primary, try to open the group */
if (internal_config.process_type == RTE_PROC_PRIMARY) {
/* try regular group format */
snprintf(filename, sizeof(filename),
VFIO_GROUP_FMT, iommu_group_no);
vfio_group_fd = open(filename, O_RDWR);
if (vfio_group_fd < 0) {
/* if file not found, it's not an error */
if (errno != ENOENT) {
RTE_LOG(ERR, EAL, "Cannot open %s: %s\n", filename,
strerror(errno));
return -1;
}
/* special case: try no-IOMMU path as well */
snprintf(filename, sizeof(filename),
VFIO_NOIOMMU_GROUP_FMT, iommu_group_no);
vfio_group_fd = open(filename, O_RDWR);
if (vfio_group_fd < 0) {
if (errno != ENOENT) {
RTE_LOG(ERR, EAL, "Cannot open %s: %s\n", filename,
strerror(errno));
return -1;
}
return 0;
}
/* noiommu group found */
}
/* if the fd is valid, create a new group for it */
if (vfio_cfg.vfio_group_idx == VFIO_MAX_GROUPS) {
RTE_LOG(ERR, EAL, "Maximum number of VFIO groups reached!\n");
close(vfio_group_fd);
return -1;
}
vfio_cfg.vfio_groups[vfio_cfg.vfio_group_idx].group_no = iommu_group_no;
vfio_cfg.vfio_groups[vfio_cfg.vfio_group_idx].fd = vfio_group_fd;
return vfio_group_fd;
}
/* if we're in a secondary process, request group fd from the primary
* process via our socket
*/
else {
int socket_fd, ret;
socket_fd = vfio_mp_sync_connect_to_primary();
if (socket_fd < 0) {
RTE_LOG(ERR, EAL, " cannot connect to primary process!\n");
return -1;
}
if (vfio_mp_sync_send_request(socket_fd, SOCKET_REQ_GROUP) < 0) {
RTE_LOG(ERR, EAL, " cannot request container fd!\n");
close(socket_fd);
return -1;
}
if (vfio_mp_sync_send_request(socket_fd, iommu_group_no) < 0) {
RTE_LOG(ERR, EAL, " cannot send group number!\n");
close(socket_fd);
return -1;
}
ret = vfio_mp_sync_receive_request(socket_fd);
switch (ret) {
case SOCKET_NO_FD:
close(socket_fd);
return 0;
case SOCKET_OK:
vfio_group_fd = vfio_mp_sync_receive_fd(socket_fd);
/* if we got the fd, return it */
if (vfio_group_fd > 0) {
close(socket_fd);
return vfio_group_fd;
}
/* fall-through on error */
default:
RTE_LOG(ERR, EAL, " cannot get container fd!\n");
close(socket_fd);
return -1;
}
}
return -1;
}
static void
clear_current_group(void)
{
vfio_cfg.vfio_groups[vfio_cfg.vfio_group_idx].group_no = 0;
vfio_cfg.vfio_groups[vfio_cfg.vfio_group_idx].fd = -1;
}
int vfio_setup_device(const char *sysfs_base, const char *dev_addr,
int *vfio_dev_fd, struct vfio_device_info *device_info)
{
struct vfio_group_status group_status = {
.argsz = sizeof(group_status)
};
int vfio_group_fd;
int iommu_group_no;
int ret;
/* get group number */
ret = vfio_get_group_no(sysfs_base, dev_addr, &iommu_group_no);
if (ret == 0) {
RTE_LOG(WARNING, EAL, " %s not managed by VFIO driver, skipping\n",
dev_addr);
return 1;
}
/* if negative, something failed */
if (ret < 0)
return -1;
/* get the actual group fd */
vfio_group_fd = vfio_get_group_fd(iommu_group_no);
if (vfio_group_fd < 0)
return -1;
/* store group fd */
vfio_cfg.vfio_groups[vfio_cfg.vfio_group_idx].group_no = iommu_group_no;
vfio_cfg.vfio_groups[vfio_cfg.vfio_group_idx].fd = vfio_group_fd;
/* if group_fd == 0, that means the device isn't managed by VFIO */
if (vfio_group_fd == 0) {
RTE_LOG(WARNING, EAL, " %s not managed by VFIO driver, skipping\n",
dev_addr);
/* we store 0 as group fd to distinguish between existing but
* unbound VFIO groups, and groups that don't exist at all.
*/
vfio_cfg.vfio_group_idx++;
return 1;
}
/*
* at this point, we know that this group is viable (meaning, all devices
* are either bound to VFIO or not bound to anything)
*/
/* check if the group is viable */
ret = ioctl(vfio_group_fd, VFIO_GROUP_GET_STATUS, &group_status);
if (ret) {
RTE_LOG(ERR, EAL, " %s cannot get group status, "
"error %i (%s)\n", dev_addr, errno, strerror(errno));
close(vfio_group_fd);
clear_current_group();
return -1;
} else if (!(group_status.flags & VFIO_GROUP_FLAGS_VIABLE)) {
RTE_LOG(ERR, EAL, " %s VFIO group is not viable!\n", dev_addr);
close(vfio_group_fd);
clear_current_group();
return -1;
}
/* check if group does not have a container yet */
if (!(group_status.flags & VFIO_GROUP_FLAGS_CONTAINER_SET)) {
/* add group to a container */
ret = ioctl(vfio_group_fd, VFIO_GROUP_SET_CONTAINER,
&vfio_cfg.vfio_container_fd);
if (ret) {
RTE_LOG(ERR, EAL, " %s cannot add VFIO group to container, "
"error %i (%s)\n", dev_addr, errno, strerror(errno));
close(vfio_group_fd);
clear_current_group();
return -1;
}
/*
* at this point we know that this group has been successfully
* initialized, so we increment vfio_group_idx to indicate that we can
* add new groups.
*/
vfio_cfg.vfio_group_idx++;
}
/*
* pick an IOMMU type and set up DMA mappings for container
*
* needs to be done only once, only when at least one group is assigned to
* a container and only in primary process
*/
if (internal_config.process_type == RTE_PROC_PRIMARY &&
vfio_cfg.vfio_container_has_dma == 0) {
/* select an IOMMU type which we will be using */
const struct vfio_iommu_type *t =
vfio_set_iommu_type(vfio_cfg.vfio_container_fd);
if (!t) {
RTE_LOG(ERR, EAL, " %s failed to select IOMMU type\n", dev_addr);
return -1;
}
ret = t->dma_map_func(vfio_cfg.vfio_container_fd);
if (ret) {
RTE_LOG(ERR, EAL, " %s DMA remapping failed, "
"error %i (%s)\n", dev_addr, errno, strerror(errno));
return -1;
}
vfio_cfg.vfio_container_has_dma = 1;
}
/* get a file descriptor for the device */
*vfio_dev_fd = ioctl(vfio_group_fd, VFIO_GROUP_GET_DEVICE_FD, dev_addr);
if (*vfio_dev_fd < 0) {
/* if we cannot get a device fd, this simply means that this
* particular port is not bound to VFIO
*/
RTE_LOG(WARNING, EAL, " %s not managed by VFIO driver, skipping\n",
dev_addr);
return 1;
}
/* test and setup the device */
ret = ioctl(*vfio_dev_fd, VFIO_DEVICE_GET_INFO, device_info);
if (ret) {
RTE_LOG(ERR, EAL, " %s cannot get device info, "
"error %i (%s)\n", dev_addr, errno, strerror(errno));
close(*vfio_dev_fd);
return -1;
}
return 0;
}
int
vfio_enable(const char *modname)
{
/* initialize group list */
int i;
int vfio_available;
for (i = 0; i < VFIO_MAX_GROUPS; i++) {
vfio_cfg.vfio_groups[i].fd = -1;
vfio_cfg.vfio_groups[i].group_no = -1;
}
/* inform the user that we are probing for VFIO */
RTE_LOG(INFO, EAL, "Probing VFIO support...\n");
/* check if vfio-pci module is loaded */
vfio_available = rte_eal_check_module(modname);
/* return error directly */
if (vfio_available == -1) {
RTE_LOG(INFO, EAL, "Could not get loaded module details!\n");
return -1;
}
/* return 0 if VFIO modules not loaded */
if (vfio_available == 0) {
RTE_LOG(DEBUG, EAL, "VFIO modules not loaded, "
"skipping VFIO support...\n");
return 0;
}
vfio_cfg.vfio_container_fd = vfio_get_container_fd();
/* check if we have VFIO driver enabled */
if (vfio_cfg.vfio_container_fd != -1) {
RTE_LOG(NOTICE, EAL, "VFIO support initialized\n");
vfio_cfg.vfio_enabled = 1;
} else {
RTE_LOG(NOTICE, EAL, "VFIO support could not be initialized\n");
}
return 0;
}
int
vfio_is_enabled(const char *modname)
{
const int mod_available = rte_eal_check_module(modname);
return vfio_cfg.vfio_enabled && mod_available;
}
const struct vfio_iommu_type * const struct vfio_iommu_type *
vfio_set_iommu_type(int vfio_container_fd) { vfio_set_iommu_type(int vfio_container_fd) {
unsigned idx; unsigned idx;

View File

@ -126,6 +126,23 @@ int
vfio_get_group_no(const char *sysfs_base, vfio_get_group_no(const char *sysfs_base,
const char *dev_addr, int *iommu_group_no); const char *dev_addr, int *iommu_group_no);
/* open group fd or get an existing one */
int
vfio_get_group_fd(int iommu_group_no);
/**
* Setup vfio_cfg for the device indentified by its address. It discovers
* the configured I/O MMU groups or sets a new one for the device. If a new
* groups is assigned, the DMA mapping is performed.
* Returns 0 on success, a negative value on failure and a positive value in
* case the given device cannot be managed this way.
*/
int vfio_setup_device(const char *sysfs_base, const char *dev_addr,
int *vfio_dev_fd, struct vfio_device_info *device_info);
int vfio_enable(const char *modname);
int vfio_is_enabled(const char *modname);
int vfio_type1_dma_map(int); int vfio_type1_dma_map(int);
int vfio_noiommu_dma_map(int); int vfio_noiommu_dma_map(int);