git/linux-cryptodev-2.6
1
0
Fork 0
mirror of https://git.kernel.org/pub/scm/linux/kernel/git/herbert/cryptodev-2.6.git synced 2026-04-23 05:56:14 -04:00
Code Issues Packages Projects Releases Wiki Activity

drm/amdkfd: Add initial VI support for KQ

Browse Source 
This patch starts to add support for the VI APU in the KQ (kernel queue)
module.

Because most (more than 90%) of the KQ code is shared among AMD's APUs, we
chose a design that performs most/all the code in the shared KQ file
(kfd_kernel_queue.c). If there is H/W specific code to be executed,
than it is written in an asic-specific extension function for that H/W.

That asic-specific extension function is called from the shared function at the
appropriate time. This requires that for every asic-specific extension function
that is implemented in a specific ASIC, there will be an equivalent
implementation in ALL ASICs, even if those implementations are just stubs.

That way we achieve:

- Maintainability: by having one copy of most of the code, we only need to
  fix bugs at one locations

- Readability: very clear what is the shared code and what is done per ASIC

- Extensibility: very easy to add new H/W specific files/functions

Signed-off-by: Ben Goz <ben.goz@amd.com>
Signed-off-by: Oded Gabbay <oded.gabbay@amd.com>
Reviewed-by: Alex Deucher <alexander.deucher@amd.com>
This commit is contained in:
Ben Goz
2014-12-02 16:38:57 +02:00
committed by Oded Gabbay
parent 443fbd5f11
commit 6898f0a568
5 changed files with 127 additions and 4 deletions
Download Patch File Download Diff File Expand all files Collapse all files

21
drivers/gpu/drm/amd/amdkfd/kfd_kernel_queue.c
View File

@@ -73,13 +73,16 @@ static bool initialize(struct kernel_queue *kq, struct kfd_dev *dev,
goto err_get_kernel_doorbell;
retval = kfd_gtt_sa_allocate(dev, queue_size, &kq->pq);
if (retval != 0)
goto err_pq_allocate_vidmem;
kq->pq_kernel_addr = kq->pq->cpu_ptr;
kq->pq_gpu_addr = kq->pq->gpu_addr;
retval = kq->ops_asic_specific.initialize(kq, dev, type, queue_size);
if (retval == false)
goto err_eop_allocate_vidmem;
retval = kfd_gtt_sa_allocate(dev, sizeof(*kq->rptr_kernel),
&kq->rptr_mem);
@@ -111,6 +114,8 @@ static bool initialize(struct kernel_queue *kq, struct kfd_dev *dev,
prop.queue_address = kq->pq_gpu_addr;
prop.read_ptr = (uint32_t *) kq->rptr_gpu_addr;
prop.write_ptr = (uint32_t *) kq->wptr_gpu_addr;
prop.eop_ring_buffer_address = kq->eop_gpu_addr;
prop.eop_ring_buffer_size = PAGE_SIZE;
if (init_queue(&kq->queue, prop) != 0)
goto err_init_queue;
@@ -156,6 +161,8 @@ err_init_queue:
err_wptr_allocate_vidmem:
kfd_gtt_sa_free(dev, kq->rptr_mem);
err_rptr_allocate_vidmem:
kfd_gtt_sa_free(dev, kq->eop_mem);
err_eop_allocate_vidmem:
kfd_gtt_sa_free(dev, kq->pq);
err_pq_allocate_vidmem:
pr_err("kfd: error init pq\n");
@@ -182,6 +189,7 @@ static void uninitialize(struct kernel_queue *kq)
kfd_gtt_sa_free(kq->dev, kq->rptr_mem);
kfd_gtt_sa_free(kq->dev, kq->wptr_mem);
kq->ops_asic_specific.uninitialize(kq);
kfd_gtt_sa_free(kq->dev, kq->pq);
kfd_release_kernel_doorbell(kq->dev,
kq->queue->properties.doorbell_ptr);
@@ -300,6 +308,13 @@ struct kernel_queue *kernel_queue_init(struct kfd_dev *dev,
kq->ops.sync_with_hw = sync_with_hw;
kq->ops.rollback_packet = rollback_packet;
switch (dev->device_info->asic_family) {
case CHIP_CARRIZO:
kernel_queue_init_vi(&kq->ops_asic_specific);
case CHIP_KAVERI:
kernel_queue_init_cik(&kq->ops_asic_specific);
}
if (kq->ops.initialize(kq, dev, type, KFD_KERNEL_QUEUE_SIZE) == false) {
pr_err("kfd: failed to init kernel queue\n");
kfree(kq);
@@ -324,7 +339,7 @@ static __attribute__((unused)) void test_kq(struct kfd_dev *dev)
BUG_ON(!dev);
pr_debug("kfd: starting kernel queue test\n");
pr_err("kfd: starting kernel queue test\n");
kq = kernel_queue_init(dev, KFD_QUEUE_TYPE_HIQ);
BUG_ON(!kq);
@@ -336,7 +351,7 @@ static __attribute__((unused)) void test_kq(struct kfd_dev *dev)
kq->ops.submit_packet(kq);
kq->ops.sync_with_hw(kq, 1000);
pr_debug("kfd: ending kernel queue test\n");
pr_err("kfd: ending kernel queue test\n");
}