summaryrefslogtreecommitdiffstats
path: root/drivers/gpu
diff options
context:
space:
mode:
Diffstat (limited to 'drivers/gpu')
-rw-r--r--drivers/gpu/drm/amd/amdkfd/kfd_chardev.c315
-rw-r--r--drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c15
-rw-r--r--drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager.c2
-rw-r--r--drivers/gpu/drm/amd/amdkfd/kfd_priv.h18
-rw-r--r--drivers/gpu/drm/amd/include/kgd_kfd_interface.h2
-rw-r--r--drivers/gpu/drm/radeon/cikd.h2
-rw-r--r--drivers/gpu/drm/radeon/radeon_kfd.c11
7 files changed, 213 insertions, 152 deletions
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c b/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c
index fe5c543599b..fcfdf23e191 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_chardev.c
@@ -126,17 +126,14 @@ static int kfd_open(struct inode *inode, struct file *filep)
return 0;
}
-static long kfd_ioctl_get_version(struct file *filep, struct kfd_process *p,
- void __user *arg)
+static int kfd_ioctl_get_version(struct file *filep, struct kfd_process *p,
+ void *data)
{
- struct kfd_ioctl_get_version_args args;
+ struct kfd_ioctl_get_version_args *args = data;
int err = 0;
- args.major_version = KFD_IOCTL_MAJOR_VERSION;
- args.minor_version = KFD_IOCTL_MINOR_VERSION;
-
- if (copy_to_user(arg, &args, sizeof(args)))
- err = -EFAULT;
+ args->major_version = KFD_IOCTL_MAJOR_VERSION;
+ args->minor_version = KFD_IOCTL_MINOR_VERSION;
return err;
}
@@ -220,10 +217,10 @@ static int set_queue_properties_from_user(struct queue_properties *q_properties,
return 0;
}
-static long kfd_ioctl_create_queue(struct file *filep, struct kfd_process *p,
- void __user *arg)
+static int kfd_ioctl_create_queue(struct file *filep, struct kfd_process *p,
+ void *data)
{
- struct kfd_ioctl_create_queue_args args;
+ struct kfd_ioctl_create_queue_args *args = data;
struct kfd_dev *dev;
int err = 0;
unsigned int queue_id;
@@ -232,16 +229,13 @@ static long kfd_ioctl_create_queue(struct file *filep, struct kfd_process *p,
memset(&q_properties, 0, sizeof(struct queue_properties));
- if (copy_from_user(&args, arg, sizeof(args)))
- return -EFAULT;
-
pr_debug("kfd: creating queue ioctl\n");
- err = set_queue_properties_from_user(&q_properties, &args);
+ err = set_queue_properties_from_user(&q_properties, args);
if (err)
return err;
- dev = kfd_device_by_id(args.gpu_id);
+ dev = kfd_device_by_id(args->gpu_id);
if (dev == NULL)
return -EINVAL;
@@ -249,7 +243,7 @@ static long kfd_ioctl_create_queue(struct file *filep, struct kfd_process *p,
pdd = kfd_bind_process_to_device(dev, p);
if (IS_ERR(pdd)) {
- err = PTR_ERR(pdd);
+ err = -ESRCH;
goto err_bind_process;
}
@@ -262,33 +256,26 @@ static long kfd_ioctl_create_queue(struct file *filep, struct kfd_process *p,
if (err != 0)
goto err_create_queue;
- args.queue_id = queue_id;
+ args->queue_id = queue_id;
/* Return gpu_id as doorbell offset for mmap usage */
- args.doorbell_offset = args.gpu_id << PAGE_SHIFT;
-
- if (copy_to_user(arg, &args, sizeof(args))) {
- err = -EFAULT;
- goto err_copy_args_out;
- }
+ args->doorbell_offset = args->gpu_id << PAGE_SHIFT;
mutex_unlock(&p->mutex);
- pr_debug("kfd: queue id %d was created successfully\n", args.queue_id);
+ pr_debug("kfd: queue id %d was created successfully\n", args->queue_id);
pr_debug("ring buffer address == 0x%016llX\n",
- args.ring_base_address);
+ args->ring_base_address);
pr_debug("read ptr address == 0x%016llX\n",
- args.read_pointer_address);
+ args->read_pointer_address);
pr_debug("write ptr address == 0x%016llX\n",
- args.write_pointer_address);
+ args->write_pointer_address);
return 0;
-err_copy_args_out:
- pqm_destroy_queue(&p->pqm, queue_id);
err_create_queue:
err_bind_process:
mutex_unlock(&p->mutex);
@@ -296,99 +283,90 @@ err_bind_process:
}
static int kfd_ioctl_destroy_queue(struct file *filp, struct kfd_process *p,
- void __user *arg)
+ void *data)
{
int retval;
- struct kfd_ioctl_destroy_queue_args args;
-
- if (copy_from_user(&args, arg, sizeof(args)))
- return -EFAULT;
+ struct kfd_ioctl_destroy_queue_args *args = data;
pr_debug("kfd: destroying queue id %d for PASID %d\n",
- args.queue_id,
+ args->queue_id,
p->pasid);
mutex_lock(&p->mutex);
- retval = pqm_destroy_queue(&p->pqm, args.queue_id);
+ retval = pqm_destroy_queue(&p->pqm, args->queue_id);
mutex_unlock(&p->mutex);
return retval;
}
static int kfd_ioctl_update_queue(struct file *filp, struct kfd_process *p,
- void __user *arg)
+ void *data)
{
int retval;
- struct kfd_ioctl_update_queue_args args;
+ struct kfd_ioctl_update_queue_args *args = data;
struct queue_properties properties;
- if (copy_from_user(&args, arg, sizeof(args)))
- return -EFAULT;
-
- if (args.queue_percentage > KFD_MAX_QUEUE_PERCENTAGE) {
+ if (args->queue_percentage > KFD_MAX_QUEUE_PERCENTAGE) {
pr_err("kfd: queue percentage must be between 0 to KFD_MAX_QUEUE_PERCENTAGE\n");
return -EINVAL;
}
- if (args.queue_priority > KFD_MAX_QUEUE_PRIORITY) {
+ if (args->queue_priority > KFD_MAX_QUEUE_PRIORITY) {
pr_err("kfd: queue priority must be between 0 to KFD_MAX_QUEUE_PRIORITY\n");
return -EINVAL;
}
- if ((args.ring_base_address) &&
+ if ((args->ring_base_address) &&
(!access_ok(VERIFY_WRITE,
- (const void __user *) args.ring_base_address,
+ (const void __user *) args->ring_base_address,
sizeof(uint64_t)))) {
pr_err("kfd: can't access ring base address\n");
return -EFAULT;
}
- if (!is_power_of_2(args.ring_size) && (args.ring_size != 0)) {
+ if (!is_power_of_2(args->ring_size) && (args->ring_size != 0)) {
pr_err("kfd: ring size must be a power of 2 or 0\n");
return -EINVAL;
}
- properties.queue_address = args.ring_base_address;
- properties.queue_size = args.ring_size;
- properties.queue_percent = args.queue_percentage;
- properties.priority = args.queue_priority;
+ properties.queue_address = args->ring_base_address;
+ properties.queue_size = args->ring_size;
+ properties.queue_percent = args->queue_percentage;
+ properties.priority = args->queue_priority;
pr_debug("kfd: updating queue id %d for PASID %d\n",
- args.queue_id, p->pasid);
+ args->queue_id, p->pasid);
mutex_lock(&p->mutex);
- retval = pqm_update_queue(&p->pqm, args.queue_id, &properties);
+ retval = pqm_update_queue(&p->pqm, args->queue_id, &properties);
mutex_unlock(&p->mutex);
return retval;
}
-static long kfd_ioctl_set_memory_policy(struct file *filep,
- struct kfd_process *p, void __user *arg)
+static int kfd_ioctl_set_memory_policy(struct file *filep,
+ struct kfd_process *p, void *data)
{
- struct kfd_ioctl_set_memory_policy_args args;
+ struct kfd_ioctl_set_memory_policy_args *args = data;
struct kfd_dev *dev;
int err = 0;
struct kfd_process_device *pdd;
enum cache_policy default_policy, alternate_policy;
- if (copy_from_user(&args, arg, sizeof(args)))
- return -EFAULT;
-
- if (args.default_policy != KFD_IOC_CACHE_POLICY_COHERENT
- && args.default_policy != KFD_IOC_CACHE_POLICY_NONCOHERENT) {
+ if (args->default_policy != KFD_IOC_CACHE_POLICY_COHERENT
+ && args->default_policy != KFD_IOC_CACHE_POLICY_NONCOHERENT) {
return -EINVAL;
}
- if (args.alternate_policy != KFD_IOC_CACHE_POLICY_COHERENT
- && args.alternate_policy != KFD_IOC_CACHE_POLICY_NONCOHERENT) {
+ if (args->alternate_policy != KFD_IOC_CACHE_POLICY_COHERENT
+ && args->alternate_policy != KFD_IOC_CACHE_POLICY_NONCOHERENT) {
return -EINVAL;
}
- dev = kfd_device_by_id(args.gpu_id);
+ dev = kfd_device_by_id(args->gpu_id);
if (dev == NULL)
return -EINVAL;
@@ -396,23 +374,23 @@ static long kfd_ioctl_set_memory_policy(struct file *filep,
pdd = kfd_bind_process_to_device(dev, p);
if (IS_ERR(pdd)) {
- err = PTR_ERR(pdd);
+ err = -ESRCH;
goto out;
}
- default_policy = (args.default_policy == KFD_IOC_CACHE_POLICY_COHERENT)
+ default_policy = (args->default_policy == KFD_IOC_CACHE_POLICY_COHERENT)
? cache_policy_coherent : cache_policy_noncoherent;
alternate_policy =
- (args.alternate_policy == KFD_IOC_CACHE_POLICY_COHERENT)
+ (args->alternate_policy == KFD_IOC_CACHE_POLICY_COHERENT)
? cache_policy_coherent : cache_policy_noncoherent;
if (!dev->dqm->set_cache_memory_policy(dev->dqm,
&pdd->qpd,
default_policy,
alternate_policy,
- (void __user *)args.alternate_aperture_base,
- args.alternate_aperture_size))
+ (void __user *)args->alternate_aperture_base,
+ args->alternate_aperture_size))
err = -EINVAL;
out:
@@ -421,53 +399,44 @@ out:
return err;
}
-static long kfd_ioctl_get_clock_counters(struct file *filep,
- struct kfd_process *p, void __user *arg)
+static int kfd_ioctl_get_clock_counters(struct file *filep,
+ struct kfd_process *p, void *data)
{
- struct kfd_ioctl_get_clock_counters_args args;
+ struct kfd_ioctl_get_clock_counters_args *args = data;
struct kfd_dev *dev;
struct timespec time;
- if (copy_from_user(&args, arg, sizeof(args)))
- return -EFAULT;
-
- dev = kfd_device_by_id(args.gpu_id);
+ dev = kfd_device_by_id(args->gpu_id);
if (dev == NULL)
return -EINVAL;
/* Reading GPU clock counter from KGD */
- args.gpu_clock_counter = kfd2kgd->get_gpu_clock_counter(dev->kgd);
+ args->gpu_clock_counter = kfd2kgd->get_gpu_clock_counter(dev->kgd);
/* No access to rdtsc. Using raw monotonic time */
getrawmonotonic(&time);
- args.cpu_clock_counter = (uint64_t)timespec_to_ns(&time);
+ args->cpu_clock_counter = (uint64_t)timespec_to_ns(&time);
get_monotonic_boottime(&time);
- args.system_clock_counter = (uint64_t)timespec_to_ns(&time);
+ args->system_clock_counter = (uint64_t)timespec_to_ns(&time);
/* Since the counter is in nano-seconds we use 1GHz frequency */
- args.system_clock_freq = 1000000000;
-
- if (copy_to_user(arg, &args, sizeof(args)))
- return -EFAULT;
+ args->system_clock_freq = 1000000000;
return 0;
}
static int kfd_ioctl_get_process_apertures(struct file *filp,
- struct kfd_process *p, void __user *arg)
+ struct kfd_process *p, void *data)
{
- struct kfd_ioctl_get_process_apertures_args args;
+ struct kfd_ioctl_get_process_apertures_args *args = data;
struct kfd_process_device_apertures *pAperture;
struct kfd_process_device *pdd;
dev_dbg(kfd_device, "get apertures for PASID %d", p->pasid);
- if (copy_from_user(&args, arg, sizeof(args)))
- return -EFAULT;
-
- args.num_of_nodes = 0;
+ args->num_of_nodes = 0;
mutex_lock(&p->mutex);
@@ -476,7 +445,8 @@ static int kfd_ioctl_get_process_apertures(struct file *filp,
/* Run over all pdd of the process */
pdd = kfd_get_first_process_device_data(p);
do {
- pAperture = &args.process_apertures[args.num_of_nodes];
+ pAperture =
+ &args->process_apertures[args->num_of_nodes];
pAperture->gpu_id = pdd->dev->id;
pAperture->lds_base = pdd->lds_base;
pAperture->lds_limit = pdd->lds_limit;
@@ -486,7 +456,7 @@ static int kfd_ioctl_get_process_apertures(struct file *filp,
pAperture->scratch_limit = pdd->scratch_limit;
dev_dbg(kfd_device,
- "node id %u\n", args.num_of_nodes);
+ "node id %u\n", args->num_of_nodes);
dev_dbg(kfd_device,
"gpu id %u\n", pdd->dev->id);
dev_dbg(kfd_device,
@@ -502,80 +472,131 @@ static int kfd_ioctl_get_process_apertures(struct file *filp,
dev_dbg(kfd_device,
"scratch_limit %llX\n", pdd->scratch_limit);
- args.num_of_nodes++;
+ args->num_of_nodes++;
} while ((pdd = kfd_get_next_process_device_data(p, pdd)) != NULL &&
- (args.num_of_nodes < NUM_OF_SUPPORTED_GPUS));
+ (args->num_of_nodes < NUM_OF_SUPPORTED_GPUS));
}
mutex_unlock(&p->mutex);
- if (copy_to_user(arg, &args, sizeof(args)))
- return -EFAULT;
-
return 0;
}
+#define AMDKFD_IOCTL_DEF(ioctl, _func, _flags) \
+ [_IOC_NR(ioctl)] = {.cmd = ioctl, .func = _func, .flags = _flags, .cmd_drv = 0, .name = #ioctl}
+
+/** Ioctl table */
+static const struct amdkfd_ioctl_desc amdkfd_ioctls[] = {
+ AMDKFD_IOCTL_DEF(AMDKFD_IOC_GET_VERSION,
+ kfd_ioctl_get_version, 0),
+
+ AMDKFD_IOCTL_DEF(AMDKFD_IOC_CREATE_QUEUE,
+ kfd_ioctl_create_queue, 0),
+
+ AMDKFD_IOCTL_DEF(AMDKFD_IOC_DESTROY_QUEUE,
+ kfd_ioctl_destroy_queue, 0),
+
+ AMDKFD_IOCTL_DEF(AMDKFD_IOC_SET_MEMORY_POLICY,
+ kfd_ioctl_set_memory_policy, 0),
+
+ AMDKFD_IOCTL_DEF(AMDKFD_IOC_GET_CLOCK_COUNTERS,
+ kfd_ioctl_get_clock_counters, 0),
+
+ AMDKFD_IOCTL_DEF(AMDKFD_IOC_GET_PROCESS_APERTURES,
+ kfd_ioctl_get_process_apertures, 0),
+
+ AMDKFD_IOCTL_DEF(AMDKFD_IOC_UPDATE_QUEUE,
+ kfd_ioctl_update_queue, 0),
+};
+
+#define AMDKFD_CORE_IOCTL_COUNT ARRAY_SIZE(amdkfd_ioctls)
+
static long kfd_ioctl(struct file *filep, unsigned int cmd, unsigned long arg)
{
struct kfd_process *process;
- long err = -EINVAL;
+ amdkfd_ioctl_t *func;
+ const struct amdkfd_ioctl_desc *ioctl = NULL;
+ unsigned int nr = _IOC_NR(cmd);
+ char stack_kdata[128];
+ char *kdata = NULL;
+ unsigned int usize, asize;
+ int retcode = -EINVAL;
- dev_dbg(kfd_device,
- "ioctl cmd 0x%x (#%d), arg 0x%lx\n",
- cmd, _IOC_NR(cmd), arg);
+ if (nr >= AMDKFD_CORE_IOCTL_COUNT)
+ goto err_i1;
+
+ if ((nr >= AMDKFD_COMMAND_START) && (nr < AMDKFD_COMMAND_END)) {
+ u32 amdkfd_size;
+
+ ioctl = &amdkfd_ioctls[nr];
+
+ amdkfd_size = _IOC_SIZE(ioctl->cmd);
+ usize = asize = _IOC_SIZE(cmd);
+ if (amdkfd_size > asize)
+ asize = amdkfd_size;
+
+ cmd = ioctl->cmd;
+ } else
+ goto err_i1;
+
+ dev_dbg(kfd_device, "ioctl cmd 0x%x (#%d), arg 0x%lx\n", cmd, nr, arg);
process = kfd_get_process(current);
- if (IS_ERR(process))
- return PTR_ERR(process);
+ if (IS_ERR(process)) {
+ dev_dbg(kfd_device, "no process\n");
+ goto err_i1;
+ }
- switch (cmd) {
- case KFD_IOC_GET_VERSION:
- err = kfd_ioctl_get_version(filep, process, (void __user *)arg);
- break;
- case KFD_IOC_CREATE_QUEUE:
- err = kfd_ioctl_create_queue(filep, process,
- (void __user *)arg);
- break;
-
- case KFD_IOC_DESTROY_QUEUE:
- err = kfd_ioctl_destroy_queue(filep, process,
- (void __user *)arg);
- break;
-
- case KFD_IOC_SET_MEMORY_POLICY:
- err = kfd_ioctl_set_memory_policy(filep, process,
- (void __user *)arg);
- break;
-
- case KFD_IOC_GET_CLOCK_COUNTERS:
- err = kfd_ioctl_get_clock_counters(filep, process,
- (void __user *)arg);
- break;
-
- case KFD_IOC_GET_PROCESS_APERTURES:
- err = kfd_ioctl_get_process_apertures(filep, process,
- (void __user *)arg);
- break;
-
- case KFD_IOC_UPDATE_QUEUE:
- err = kfd_ioctl_update_queue(filep, process,
- (void __user *)arg);
- break;
-
- default:
- dev_err(kfd_device,
- "unknown ioctl cmd 0x%x, arg 0x%lx)\n",
- cmd, arg);
- err = -EINVAL;
- break;
+ /* Do not trust userspace, use our own definition */
+ func = ioctl->func;
+
+ if (unlikely(!func)) {
+ dev_dbg(kfd_device, "no function\n");
+ retcode = -EINVAL;
+ goto err_i1;
}
- if (err < 0)
- dev_err(kfd_device,
- "ioctl error %ld for ioctl cmd 0x%x (#%d)\n",
- err, cmd, _IOC_NR(cmd));
+ if (cmd & (IOC_IN | IOC_OUT)) {
+ if (asize <= sizeof(stack_kdata)) {
+ kdata = stack_kdata;
+ } else {
+ kdata = kmalloc(asize, GFP_KERNEL);
+ if (!kdata) {
+ retcode = -ENOMEM;
+ goto err_i1;
+ }
+ }
+ if (asize > usize)
+ memset(kdata + usize, 0, asize - usize);
+ }
- return err;
+ if (cmd & IOC_IN) {
+ if (copy_from_user(kdata, (void __user *)arg, usize) != 0) {
+ retcode = -EFAULT;
+ goto err_i1;
+ }
+ } else if (cmd & IOC_OUT) {
+ memset(kdata, 0, usize);
+ }
+
+ retcode = func(filep, process, kdata);
+
+ if (cmd & IOC_OUT)
+ if (copy_to_user((void __user *)arg, kdata, usize) != 0)
+ retcode = -EFAULT;
+
+err_i1:
+ if (!ioctl)
+ dev_dbg(kfd_device, "invalid ioctl: pid=%d, cmd=0x%02x, nr=0x%02x\n",
+ task_pid_nr(current), cmd, nr);
+
+ if (kdata != stack_kdata)
+ kfree(kdata);
+
+ if (retcode)
+ dev_dbg(kfd_device, "ret = %d\n", retcode);
+
+ return retcode;
}
static int kfd_mmap(struct file *filp, struct vm_area_struct *vma)
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c
index f44d6737b65..9c8961d2236 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_device_queue_manager.c
@@ -161,6 +161,9 @@ static void deallocate_vmid(struct device_queue_manager *dqm,
{
int bit = qpd->vmid - KFD_VMID_START_OFFSET;
+ /* Release the vmid mapping */
+ set_pasid_vmid_mapping(dqm, 0, qpd->vmid);
+
set_bit(bit, (unsigned long *)&dqm->vmid_bitmap);
qpd->vmid = 0;
q->properties.vmid = 0;
@@ -272,6 +275,18 @@ static int create_compute_queue_nocpsch(struct device_queue_manager *dqm,
return retval;
}
+ pr_debug("kfd: loading mqd to hqd on pipe (%d) queue (%d)\n",
+ q->pipe,
+ q->queue);
+
+ retval = mqd->load_mqd(mqd, q->mqd, q->pipe,
+ q->queue, q->properties.write_ptr);
+ if (retval != 0) {
+ deallocate_hqd(dqm, q);
+ mqd->uninit_mqd(mqd, q->mqd, q->mqd_mem_obj);
+ return retval;
+ }
+
return 0;
}
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager.c b/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager.c
index adc31474e78..4c3828cf45b 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_mqd_manager.c
@@ -184,7 +184,7 @@ static bool is_occupied(struct mqd_manager *mm, void *mqd,
uint32_t queue_id)
{
- return kfd2kgd->hqd_is_occupies(mm->dev->kgd, queue_address,
+ return kfd2kgd->hqd_is_occupied(mm->dev->kgd, queue_address,
pipe_id, queue_id);
}
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_priv.h b/drivers/gpu/drm/amd/amdkfd/kfd_priv.h
index f9fb81e3bb0..a5edb29507e 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_priv.h
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_priv.h
@@ -463,6 +463,24 @@ struct kfd_process {
bool is_32bit_user_mode;
};
+/**
+ * Ioctl function type.
+ *
+ * \param filep pointer to file structure.
+ * \param p amdkfd process pointer.
+ * \param data pointer to arg that was copied from user.
+ */
+typedef int amdkfd_ioctl_t(struct file *filep, struct kfd_process *p,
+ void *data);
+
+struct amdkfd_ioctl_desc {
+ unsigned int cmd;
+ int flags;
+ amdkfd_ioctl_t *func;
+ unsigned int cmd_drv;
+ const char *name;
+};
+
void kfd_process_create_wq(void);
void kfd_process_destroy_wq(void);
struct kfd_process *kfd_create_process(const struct task_struct *);
diff --git a/drivers/gpu/drm/amd/include/kgd_kfd_interface.h b/drivers/gpu/drm/amd/include/kgd_kfd_interface.h
index 47b551970a1..96a512208fa 100644
--- a/drivers/gpu/drm/amd/include/kgd_kfd_interface.h
+++ b/drivers/gpu/drm/amd/include/kgd_kfd_interface.h
@@ -183,7 +183,7 @@ struct kfd2kgd_calls {
int (*hqd_load)(struct kgd_dev *kgd, void *mqd, uint32_t pipe_id,
uint32_t queue_id, uint32_t __user *wptr);
- bool (*hqd_is_occupies)(struct kgd_dev *kgd, uint64_t queue_address,
+ bool (*hqd_is_occupied)(struct kgd_dev *kgd, uint64_t queue_address,
uint32_t pipe_id, uint32_t queue_id);
int (*hqd_destroy)(struct kgd_dev *kgd, uint32_t reset_type,
diff --git a/drivers/gpu/drm/radeon/cikd.h b/drivers/gpu/drm/radeon/cikd.h
index ba85986febe..03003f8a6de 100644
--- a/drivers/gpu/drm/radeon/cikd.h
+++ b/drivers/gpu/drm/radeon/cikd.h
@@ -2156,4 +2156,6 @@
#define ATC_VM_APERTURE1_HIGH_ADDR 0x330Cu
#define ATC_VM_APERTURE1_LOW_ADDR 0x3304u
+#define IH_VMID_0_LUT 0x3D40u
+
#endif
diff --git a/drivers/gpu/drm/radeon/radeon_kfd.c b/drivers/gpu/drm/radeon/radeon_kfd.c
index d3e78b4a6b7..8bf87f1203c 100644
--- a/drivers/gpu/drm/radeon/radeon_kfd.c
+++ b/drivers/gpu/drm/radeon/radeon_kfd.c
@@ -72,7 +72,7 @@ static int kgd_init_pipeline(struct kgd_dev *kgd, uint32_t pipe_id,
static int kgd_hqd_load(struct kgd_dev *kgd, void *mqd, uint32_t pipe_id,
uint32_t queue_id, uint32_t __user *wptr);
-static bool kgd_hqd_is_occupies(struct kgd_dev *kgd, uint64_t queue_address,
+static bool kgd_hqd_is_occupied(struct kgd_dev *kgd, uint64_t queue_address,
uint32_t pipe_id, uint32_t queue_id);
static int kgd_hqd_destroy(struct kgd_dev *kgd, uint32_t reset_type,
@@ -92,7 +92,7 @@ static const struct kfd2kgd_calls kfd2kgd = {
.init_memory = kgd_init_memory,
.init_pipeline = kgd_init_pipeline,
.hqd_load = kgd_hqd_load,
- .hqd_is_occupies = kgd_hqd_is_occupies,
+ .hqd_is_occupied = kgd_hqd_is_occupied,
.hqd_destroy = kgd_hqd_destroy,
.get_fw_version = get_fw_version
};
@@ -390,6 +390,10 @@ static int kgd_set_pasid_vmid_mapping(struct kgd_dev *kgd, unsigned int pasid,
cpu_relax();
write_register(kgd, ATC_VMID_PASID_MAPPING_UPDATE_STATUS, 1U << vmid);
+ /* Mapping vmid to pasid also for IH block */
+ write_register(kgd, IH_VMID_0_LUT + vmid * sizeof(uint32_t),
+ pasid_mapping);
+
return 0;
}
@@ -529,7 +533,7 @@ static int kgd_hqd_load(struct kgd_dev *kgd, void *mqd, uint32_t pipe_id,
return 0;
}
-static bool kgd_hqd_is_occupies(struct kgd_dev *kgd, uint64_t queue_address,
+static bool kgd_hqd_is_occupied(struct kgd_dev *kgd, uint64_t queue_address,
uint32_t pipe_id, uint32_t queue_id)
{
uint32_t act;
@@ -568,6 +572,7 @@ static int kgd_hqd_destroy(struct kgd_dev *kgd, uint32_t reset_type,
if (timeout == 0) {
pr_err("kfd: cp queue preemption time out (%dms)\n",
temp);
+ release_queue(kgd);
return -ETIME;
}
msleep(20);