drm/amdgpu: Add H/W agnostic amdgpu <--> amdkfd interface
This patch adds an interface file between amdgpu and amdkfd. This
interface file is H/W agnostic, thus containing functions that
operate the same for any AMD APU/GPU H/W generation.
The functions in this interface mirror (some) of the functions in
radeon_kfd.c (the radeon<-->amdkfd interface file). The main functions
are:
- amdgpu_amdkfd_init - initialize the amdkfd module
- amdgpu_amdkfd_load_interface - load the H/W interface according to the
currently probed device
- amdgpu_amdkfd_device_probe - probe the device in amdkfd
- amdgpu_amdkfd_device_init - initialize the device in amdkfd
- amdgpu_amdkfd_interrupt - call the ISR of amdkfd
- amdgpu_amdkfd_suspend - suspend callback from amdgpu
- amdgpu_amdkfd_resume - resume callback from amdgpu
This patch also modifies the relevant amdgpu files, to use this new
interface.
Signed-off-by: Oded Gabbay <oded.gabbay@gmail.com>
diff --git a/drivers/gpu/drm/amd/amdgpu/Makefile b/drivers/gpu/drm/amd/amdgpu/Makefile
index 616dfd4..891f02c 100644
--- a/drivers/gpu/drm/amd/amdgpu/Makefile
+++ b/drivers/gpu/drm/amd/amdgpu/Makefile
@@ -71,6 +71,10 @@
amdgpu_vce.o \
vce_v3_0.o
+# add amdkfd interfaces
+amdgpu-y += \
+ amdgpu_amdkfd.o
+
amdgpu-$(CONFIG_COMPAT) += amdgpu_ioc32.o
amdgpu-$(CONFIG_VGA_SWITCHEROO) += amdgpu_atpx_handler.o
amdgpu-$(CONFIG_ACPI) += amdgpu_acpi.o
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
index 0165783..f3791e0d 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
@@ -2011,6 +2011,9 @@
/* tracking pinned memory */
u64 vram_pin_size;
u64 gart_pin_size;
+
+ /* amdkfd interface */
+ struct kfd_dev *kfd;
};
bool amdgpu_device_is_px(struct drm_device *dev);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c
new file mode 100644
index 0000000..2071a9b
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c
@@ -0,0 +1,262 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+#include "amdgpu_amdkfd.h"
+#include "amdgpu_family.h"
+#include <drm/drmP.h>
+#include "amdgpu.h"
+#include <linux/module.h>
+
+const struct kfd2kgd_calls *kfd2kgd;
+const struct kgd2kfd_calls *kgd2kfd;
+bool (*kgd2kfd_init_p)(unsigned, const struct kgd2kfd_calls**);
+
+bool amdgpu_amdkfd_init(void)
+{
+#if defined(CONFIG_HSA_AMD_MODULE)
+ bool (*kgd2kfd_init_p)(unsigned, const struct kgd2kfd_calls**);
+
+ kgd2kfd_init_p = symbol_request(kgd2kfd_init);
+
+ if (kgd2kfd_init_p == NULL)
+ return false;
+#endif
+ return true;
+}
+
+bool amdgpu_amdkfd_load_interface(struct amdgpu_device *rdev)
+{
+#if defined(CONFIG_HSA_AMD_MODULE)
+ bool (*kgd2kfd_init_p)(unsigned, const struct kgd2kfd_calls**);
+#endif
+
+ switch (rdev->asic_type) {
+ case CHIP_KAVERI:
+ default:
+ return false;
+ }
+
+#if defined(CONFIG_HSA_AMD_MODULE)
+ kgd2kfd_init_p = symbol_request(kgd2kfd_init);
+
+ if (kgd2kfd_init_p == NULL) {
+ kfd2kgd = NULL;
+ return false;
+ }
+
+ if (!kgd2kfd_init_p(KFD_INTERFACE_VERSION, &kgd2kfd)) {
+ symbol_put(kgd2kfd_init);
+ kfd2kgd = NULL;
+ kgd2kfd = NULL;
+
+ return false;
+ }
+
+ return true;
+#elif defined(CONFIG_HSA_AMD)
+ if (!kgd2kfd_init(KFD_INTERFACE_VERSION, &kgd2kfd)) {
+ kfd2kgd = NULL;
+ kgd2kfd = NULL;
+ return false;
+ }
+
+ return true;
+#else
+ kfd2kgd = NULL;
+ return false;
+#endif
+}
+
+void amdgpu_amdkfd_fini(void)
+{
+ if (kgd2kfd) {
+ kgd2kfd->exit();
+ symbol_put(kgd2kfd_init);
+ }
+}
+
+void amdgpu_amdkfd_device_probe(struct amdgpu_device *rdev)
+{
+ if (kgd2kfd)
+ rdev->kfd = kgd2kfd->probe((struct kgd_dev *)rdev,
+ rdev->pdev, kfd2kgd);
+}
+
+void amdgpu_amdkfd_device_init(struct amdgpu_device *rdev)
+{
+ if (rdev->kfd) {
+ struct kgd2kfd_shared_resources gpu_resources = {
+ .compute_vmid_bitmap = 0xFF00,
+
+ .first_compute_pipe = 1,
+ .compute_pipe_count = 4 - 1,
+ };
+
+ amdgpu_doorbell_get_kfd_info(rdev,
+ &gpu_resources.doorbell_physical_address,
+ &gpu_resources.doorbell_aperture_size,
+ &gpu_resources.doorbell_start_offset);
+
+ kgd2kfd->device_init(rdev->kfd, &gpu_resources);
+ }
+}
+
+void amdgpu_amdkfd_device_fini(struct amdgpu_device *rdev)
+{
+ if (rdev->kfd) {
+ kgd2kfd->device_exit(rdev->kfd);
+ rdev->kfd = NULL;
+ }
+}
+
+void amdgpu_amdkfd_interrupt(struct amdgpu_device *rdev,
+ const void *ih_ring_entry)
+{
+ if (rdev->kfd)
+ kgd2kfd->interrupt(rdev->kfd, ih_ring_entry);
+}
+
+void amdgpu_amdkfd_suspend(struct amdgpu_device *rdev)
+{
+ if (rdev->kfd)
+ kgd2kfd->suspend(rdev->kfd);
+}
+
+int amdgpu_amdkfd_resume(struct amdgpu_device *rdev)
+{
+ int r = 0;
+
+ if (rdev->kfd)
+ r = kgd2kfd->resume(rdev->kfd);
+
+ return r;
+}
+
+u32 pool_to_domain(enum kgd_memory_pool p)
+{
+ switch (p) {
+ case KGD_POOL_FRAMEBUFFER: return AMDGPU_GEM_DOMAIN_VRAM;
+ default: return AMDGPU_GEM_DOMAIN_GTT;
+ }
+}
+
+int alloc_gtt_mem(struct kgd_dev *kgd, size_t size,
+ void **mem_obj, uint64_t *gpu_addr,
+ void **cpu_ptr)
+{
+ struct amdgpu_device *rdev = (struct amdgpu_device *)kgd;
+ struct kgd_mem **mem = (struct kgd_mem **) mem_obj;
+ int r;
+
+ BUG_ON(kgd == NULL);
+ BUG_ON(gpu_addr == NULL);
+ BUG_ON(cpu_ptr == NULL);
+
+ *mem = kmalloc(sizeof(struct kgd_mem), GFP_KERNEL);
+ if ((*mem) == NULL)
+ return -ENOMEM;
+
+ r = amdgpu_bo_create(rdev, size, PAGE_SIZE, true, AMDGPU_GEM_DOMAIN_GTT,
+ AMDGPU_GEM_CREATE_CPU_GTT_USWC, NULL, &(*mem)->bo);
+ if (r) {
+ dev_err(rdev->dev,
+ "failed to allocate BO for amdkfd (%d)\n", r);
+ return r;
+ }
+
+ /* map the buffer */
+ r = amdgpu_bo_reserve((*mem)->bo, true);
+ if (r) {
+ dev_err(rdev->dev, "(%d) failed to reserve bo for amdkfd\n", r);
+ goto allocate_mem_reserve_bo_failed;
+ }
+
+ r = amdgpu_bo_pin((*mem)->bo, AMDGPU_GEM_DOMAIN_GTT,
+ &(*mem)->gpu_addr);
+ if (r) {
+ dev_err(rdev->dev, "(%d) failed to pin bo for amdkfd\n", r);
+ goto allocate_mem_pin_bo_failed;
+ }
+ *gpu_addr = (*mem)->gpu_addr;
+
+ r = amdgpu_bo_kmap((*mem)->bo, &(*mem)->cpu_ptr);
+ if (r) {
+ dev_err(rdev->dev,
+ "(%d) failed to map bo to kernel for amdkfd\n", r);
+ goto allocate_mem_kmap_bo_failed;
+ }
+ *cpu_ptr = (*mem)->cpu_ptr;
+
+ amdgpu_bo_unreserve((*mem)->bo);
+
+ return 0;
+
+allocate_mem_kmap_bo_failed:
+ amdgpu_bo_unpin((*mem)->bo);
+allocate_mem_pin_bo_failed:
+ amdgpu_bo_unreserve((*mem)->bo);
+allocate_mem_reserve_bo_failed:
+ amdgpu_bo_unref(&(*mem)->bo);
+
+ return r;
+}
+
+void free_gtt_mem(struct kgd_dev *kgd, void *mem_obj)
+{
+ struct kgd_mem *mem = (struct kgd_mem *) mem_obj;
+
+ BUG_ON(mem == NULL);
+
+ amdgpu_bo_reserve(mem->bo, true);
+ amdgpu_bo_kunmap(mem->bo);
+ amdgpu_bo_unpin(mem->bo);
+ amdgpu_bo_unreserve(mem->bo);
+ amdgpu_bo_unref(&(mem->bo));
+ kfree(mem);
+}
+
+uint64_t get_vmem_size(struct kgd_dev *kgd)
+{
+ struct amdgpu_device *rdev =
+ (struct amdgpu_device *)kgd;
+
+ BUG_ON(kgd == NULL);
+
+ return rdev->mc.real_vram_size;
+}
+
+uint64_t get_gpu_clock_counter(struct kgd_dev *kgd)
+{
+ struct amdgpu_device *rdev = (struct amdgpu_device *)kgd;
+
+ if (rdev->asic_funcs->get_gpu_clock_counter)
+ return rdev->asic_funcs->get_gpu_clock_counter(rdev);
+ return 0;
+}
+
+uint32_t get_max_engine_clock_in_mhz(struct kgd_dev *kgd)
+{
+ struct amdgpu_device *rdev = (struct amdgpu_device *)kgd;
+
+ /* The sclk is in quantas of 10kHz */
+ return rdev->pm.dpm.dyn_state.max_clock_voltage_on_ac.sclk / 100;
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h
new file mode 100644
index 0000000..c81242e
--- /dev/null
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h
@@ -0,0 +1,64 @@
+/*
+ * Copyright 2014 Advanced Micro Devices, Inc.
+ *
+ * Permission is hereby granted, free of charge, to any person obtaining a
+ * copy of this software and associated documentation files (the "Software"),
+ * to deal in the Software without restriction, including without limitation
+ * the rights to use, copy, modify, merge, publish, distribute, sublicense,
+ * and/or sell copies of the Software, and to permit persons to whom the
+ * Software is furnished to do so, subject to the following conditions:
+ *
+ * The above copyright notice and this permission notice shall be included in
+ * all copies or substantial portions of the Software.
+ *
+ * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+ * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+ * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL
+ * THE COPYRIGHT HOLDER(S) OR AUTHOR(S) BE LIABLE FOR ANY CLAIM, DAMAGES OR
+ * OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE,
+ * ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR
+ * OTHER DEALINGS IN THE SOFTWARE.
+ */
+
+/* amdgpu_amdkfd.h defines the private interface between amdgpu and amdkfd. */
+
+#ifndef AMDGPU_AMDKFD_H_INCLUDED
+#define AMDGPU_AMDKFD_H_INCLUDED
+
+#include <linux/types.h>
+#include <kgd_kfd_interface.h>
+
+struct amdgpu_device;
+
+struct kgd_mem {
+ struct amdgpu_bo *bo;
+ uint64_t gpu_addr;
+ void *cpu_ptr;
+};
+
+bool amdgpu_amdkfd_init(void);
+void amdgpu_amdkfd_fini(void);
+
+bool amdgpu_amdkfd_load_interface(struct amdgpu_device *rdev);
+
+void amdgpu_amdkfd_suspend(struct amdgpu_device *rdev);
+int amdgpu_amdkfd_resume(struct amdgpu_device *rdev);
+void amdgpu_amdkfd_interrupt(struct amdgpu_device *rdev,
+ const void *ih_ring_entry);
+void amdgpu_amdkfd_device_probe(struct amdgpu_device *rdev);
+void amdgpu_amdkfd_device_init(struct amdgpu_device *rdev);
+void amdgpu_amdkfd_device_fini(struct amdgpu_device *rdev);
+
+struct kfd2kgd_calls *amdgpu_amdkfd_gfx_7_get_functions(void);
+
+/* Shared API */
+int alloc_gtt_mem(struct kgd_dev *kgd, size_t size,
+ void **mem_obj, uint64_t *gpu_addr,
+ void **cpu_ptr);
+void free_gtt_mem(struct kgd_dev *kgd, void *mem_obj);
+uint64_t get_vmem_size(struct kgd_dev *kgd);
+uint64_t get_gpu_clock_counter(struct kgd_dev *kgd);
+
+uint32_t get_max_engine_clock_in_mhz(struct kgd_dev *kgd);
+
+#endif /* AMDGPU_AMDKFD_H_INCLUDED */
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
index 56da962..115906f 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_drv.c
@@ -44,6 +44,8 @@
#include "amdgpu.h"
#include "amdgpu_irq.h"
+#include "amdgpu_amdkfd.h"
+
/*
* KMS wrapper.
* - 3.0.0 - initial driver
@@ -527,12 +529,15 @@
driver->num_ioctls = amdgpu_max_kms_ioctl;
amdgpu_register_atpx_handler();
+ amdgpu_amdkfd_init();
+
/* let modprobe override vga console setting */
return drm_pci_init(driver, pdriver);
}
static void __exit amdgpu_exit(void)
{
+ amdgpu_amdkfd_fini();
drm_pci_exit(driver, pdriver);
amdgpu_unregister_atpx_handler();
}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ih.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ih.c
index db5422e..fb44dd2 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ih.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ih.c
@@ -24,6 +24,7 @@
#include <drm/drmP.h>
#include "amdgpu.h"
#include "amdgpu_ih.h"
+#include "amdgpu_amdkfd.h"
/**
* amdgpu_ih_ring_alloc - allocate memory for the IH ring
@@ -199,6 +200,12 @@
rmb();
while (adev->irq.ih.rptr != wptr) {
+ u32 ring_index = adev->irq.ih.rptr >> 2;
+
+ /* Before dispatching irq to IP blocks, send it to amdkfd */
+ amdgpu_amdkfd_interrupt(adev,
+ (const void *) &adev->irq.ih.ring[ring_index]);
+
amdgpu_ih_decode_iv(adev, &entry);
adev->irq.ih.rptr &= adev->irq.ih.ptr_mask;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c
index 5533434..8c40a96 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_kms.c
@@ -34,6 +34,7 @@
#include <linux/vga_switcheroo.h>
#include <linux/slab.h>
#include <linux/pm_runtime.h>
+#include "amdgpu_amdkfd.h"
#if defined(CONFIG_VGA_SWITCHEROO)
bool amdgpu_has_atpx(void);
@@ -61,6 +62,8 @@
pm_runtime_get_sync(dev->dev);
+ amdgpu_amdkfd_device_fini(adev);
+
amdgpu_acpi_fini(adev);
amdgpu_device_fini(adev);
@@ -118,6 +121,10 @@
"Error during ACPI methods call\n");
}
+ amdgpu_amdkfd_load_interface(adev);
+ amdgpu_amdkfd_device_probe(adev);
+ amdgpu_amdkfd_device_init(adev);
+
if (amdgpu_device_is_px(dev)) {
pm_runtime_use_autosuspend(dev->dev);
pm_runtime_set_autosuspend_delay(dev->dev, 5000);
diff --git a/drivers/gpu/drm/amd/amdgpu/cik.c b/drivers/gpu/drm/amd/amdgpu/cik.c
index 341c566..b3b66a0 100644
--- a/drivers/gpu/drm/amd/amdgpu/cik.c
+++ b/drivers/gpu/drm/amd/amdgpu/cik.c
@@ -64,6 +64,8 @@
#include "oss/oss_2_0_d.h"
#include "oss/oss_2_0_sh_mask.h"
+#include "amdgpu_amdkfd.h"
+
/*
* Indirect registers accessor
*/
@@ -2448,14 +2450,21 @@
{
struct amdgpu_device *adev = (struct amdgpu_device *)handle;
+ amdgpu_amdkfd_suspend(adev);
+
return cik_common_hw_fini(adev);
}
static int cik_common_resume(void *handle)
{
+ int r;
struct amdgpu_device *adev = (struct amdgpu_device *)handle;
- return cik_common_hw_init(adev);
+ r = cik_common_hw_init(adev);
+ if (r)
+ return r;
+
+ return amdgpu_amdkfd_resume(adev);
}
static bool cik_common_is_idle(void *handle)