summaryrefslogtreecommitdiff
path: root/drivers/gpu/drm/amd/amdgpu
diff options
context:
space:
mode:
Diffstat (limited to 'drivers/gpu/drm/amd/amdgpu')
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu.h10
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c3
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c111
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h20
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v7.c1
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v8.c1
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.c1
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c180
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.c1
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c21
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_csa.c2
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c2
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_device.c52
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c7
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_gem.h2
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h1
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_ih.h2
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c42
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_object.c49
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_object.h5
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_prime.c4
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c22
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h1
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c2
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c52
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h5
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c26
-rw-r--r--drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.h1
-rw-r--r--drivers/gpu/drm/amd/amdgpu/cik_ih.c29
-rw-r--r--drivers/gpu/drm/amd/amdgpu/cz_ih.c29
-rw-r--r--drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c11
-rw-r--r--drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c16
-rw-r--r--drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c64
-rw-r--r--drivers/gpu/drm/amd/amdgpu/iceland_ih.c29
-rw-r--r--drivers/gpu/drm/amd/amdgpu/psp_gfx_if.h5
-rw-r--r--drivers/gpu/drm/amd/amdgpu/psp_v11_0.c98
-rw-r--r--drivers/gpu/drm/amd/amdgpu/psp_v3_1.c2
-rw-r--r--drivers/gpu/drm/amd/amdgpu/si_ih.c14
-rw-r--r--drivers/gpu/drm/amd/amdgpu/soc15.c3
-rw-r--r--drivers/gpu/drm/amd/amdgpu/tonga_ih.c29
-rw-r--r--drivers/gpu/drm/amd/amdgpu/vega10_ih.c85
41 files changed, 563 insertions, 477 deletions
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu.h b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
index c8ad6bf6618a..88db3c263e5f 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu.h
@@ -82,6 +82,7 @@
#include "amdgpu_bo_list.h"
#include "amdgpu_gem.h"
#include "amdgpu_doorbell.h"
+#include "amdgpu_amdkfd.h"
#define MAX_GPU_INSTANCE 16
@@ -163,6 +164,7 @@ extern int amdgpu_si_support;
extern int amdgpu_cik_support;
#endif
+#define AMDGPU_VM_MAX_NUM_CTX 4096
#define AMDGPU_SG_THRESHOLD (256*1024*1024)
#define AMDGPU_DEFAULT_GTT_SIZE_MB 3072ULL /* 3GB by default */
#define AMDGPU_WAIT_IDLE_TIMEOUT_IN_MS 3000
@@ -862,6 +864,9 @@ struct amdgpu_device {
/* GDS */
struct amdgpu_gds gds;
+ /* KFD */
+ struct amdgpu_kfd_dev kfd;
+
/* display related functionality */
struct amdgpu_display_manager dm;
@@ -875,9 +880,6 @@ struct amdgpu_device {
atomic64_t visible_pin_size;
atomic64_t gart_pin_size;
- /* amdkfd interface */
- struct kfd_dev *kfd;
-
/* soc15 register offset based on ip, instance and segment */
uint32_t *reg_offset[MAX_HWIP][HWIP_MAX_INSTANCE];
@@ -910,7 +912,9 @@ struct amdgpu_device {
bool in_gpu_reset;
struct mutex lock_reset;
struct amdgpu_doorbell_index doorbell_index;
+
int asic_reset_res;
+ struct work_struct xgmi_reset_work;
};
static inline struct amdgpu_device *amdgpu_ttm_adev(struct ttm_bo_device *bdev)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c
index 47db65926d71..4376b17ca594 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_acpi.c
@@ -886,6 +886,5 @@ void amdgpu_acpi_get_backlight_caps(struct amdgpu_device *adev,
void amdgpu_acpi_fini(struct amdgpu_device *adev)
{
unregister_acpi_notifier(&adev->acpi_nb);
- if (adev->atif)
- kfree(adev->atif);
+ kfree(adev->atif);
}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c
index d693b8047653..2dfaf158ef07 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.c
@@ -26,15 +26,26 @@
#include "amdgpu.h"
#include "amdgpu_gfx.h"
#include <linux/module.h>
+#include <linux/dma-buf.h>
const struct kgd2kfd_calls *kgd2kfd;
static const unsigned int compute_vmid_bitmap = 0xFF00;
+/* Total memory size in system memory and all GPU VRAM. Used to
+ * estimate worst case amount of memory to reserve for page tables
+ */
+uint64_t amdgpu_amdkfd_total_mem_size;
+
int amdgpu_amdkfd_init(void)
{
+ struct sysinfo si;
int ret;
+ si_meminfo(&si);
+ amdgpu_amdkfd_total_mem_size = si.totalram - si.totalhigh;
+ amdgpu_amdkfd_total_mem_size *= si.mem_unit;
+
#ifdef CONFIG_HSA_AMD
ret = kgd2kfd_init(KFD_INTERFACE_VERSION, &kgd2kfd);
if (ret)
@@ -87,8 +98,11 @@ void amdgpu_amdkfd_device_probe(struct amdgpu_device *adev)
return;
}
- adev->kfd = kgd2kfd->probe((struct kgd_dev *)adev,
- adev->pdev, kfd2kgd);
+ adev->kfd.dev = kgd2kfd->probe((struct kgd_dev *)adev,
+ adev->pdev, kfd2kgd);
+
+ if (adev->kfd.dev)
+ amdgpu_amdkfd_total_mem_size += adev->gmc.real_vram_size;
}
/**
@@ -128,7 +142,8 @@ void amdgpu_amdkfd_device_init(struct amdgpu_device *adev)
{
int i, n;
int last_valid_bit;
- if (adev->kfd) {
+
+ if (adev->kfd.dev) {
struct kgd2kfd_shared_resources gpu_resources = {
.compute_vmid_bitmap = compute_vmid_bitmap,
.num_pipe_per_mec = adev->gfx.mec.num_pipe_per_mec,
@@ -167,7 +182,7 @@ void amdgpu_amdkfd_device_init(struct amdgpu_device *adev)
&gpu_resources.doorbell_start_offset);
if (adev->asic_type < CHIP_VEGA10) {
- kgd2kfd->device_init(adev->kfd, &gpu_resources);
+ kgd2kfd->device_init(adev->kfd.dev, &gpu_resources);
return;
}
@@ -196,37 +211,37 @@ void amdgpu_amdkfd_device_init(struct amdgpu_device *adev)
gpu_resources.reserved_doorbell_mask = 0x1e0;
gpu_resources.reserved_doorbell_val = 0x0e0;
- kgd2kfd->device_init(adev->kfd, &gpu_resources);
+ kgd2kfd->device_init(adev->kfd.dev, &gpu_resources);
}
}
void amdgpu_amdkfd_device_fini(struct amdgpu_device *adev)
{
- if (adev->kfd) {
- kgd2kfd->device_exit(adev->kfd);
- adev->kfd = NULL;
+ if (adev->kfd.dev) {
+ kgd2kfd->device_exit(adev->kfd.dev);
+ adev->kfd.dev = NULL;
}
}
void amdgpu_amdkfd_interrupt(struct amdgpu_device *adev,
const void *ih_ring_entry)
{
- if (adev->kfd)
- kgd2kfd->interrupt(adev->kfd, ih_ring_entry);
+ if (adev->kfd.dev)
+ kgd2kfd->interrupt(adev->kfd.dev, ih_ring_entry);
}
void amdgpu_amdkfd_suspend(struct amdgpu_device *adev)
{
- if (adev->kfd)
- kgd2kfd->suspend(adev->kfd);
+ if (adev->kfd.dev)
+ kgd2kfd->suspend(adev->kfd.dev);
}
int amdgpu_amdkfd_resume(struct amdgpu_device *adev)
{
int r = 0;
- if (adev->kfd)
- r = kgd2kfd->resume(adev->kfd);
+ if (adev->kfd.dev)
+ r = kgd2kfd->resume(adev->kfd.dev);
return r;
}
@@ -235,8 +250,8 @@ int amdgpu_amdkfd_pre_reset(struct amdgpu_device *adev)
{
int r = 0;
- if (adev->kfd)
- r = kgd2kfd->pre_reset(adev->kfd);
+ if (adev->kfd.dev)
+ r = kgd2kfd->pre_reset(adev->kfd.dev);
return r;
}
@@ -245,8 +260,8 @@ int amdgpu_amdkfd_post_reset(struct amdgpu_device *adev)
{
int r = 0;
- if (adev->kfd)
- r = kgd2kfd->post_reset(adev->kfd);
+ if (adev->kfd.dev)
+ r = kgd2kfd->post_reset(adev->kfd.dev);
return r;
}
@@ -419,6 +434,62 @@ void amdgpu_amdkfd_get_cu_info(struct kgd_dev *kgd, struct kfd_cu_info *cu_info)
cu_info->lds_size = acu_info.lds_size;
}
+int amdgpu_amdkfd_get_dmabuf_info(struct kgd_dev *kgd, int dma_buf_fd,
+ struct kgd_dev **dma_buf_kgd,
+ uint64_t *bo_size, void *metadata_buffer,
+ size_t buffer_size, uint32_t *metadata_size,
+ uint32_t *flags)
+{
+ struct amdgpu_device *adev = (struct amdgpu_device *)kgd;
+ struct dma_buf *dma_buf;
+ struct drm_gem_object *obj;
+ struct amdgpu_bo *bo;
+ uint64_t metadata_flags;
+ int r = -EINVAL;
+
+ dma_buf = dma_buf_get(dma_buf_fd);
+ if (IS_ERR(dma_buf))
+ return PTR_ERR(dma_buf);
+
+ if (dma_buf->ops != &amdgpu_dmabuf_ops)
+ /* Can't handle non-graphics buffers */
+ goto out_put;
+
+ obj = dma_buf->priv;
+ if (obj->dev->driver != adev->ddev->driver)
+ /* Can't handle buffers from different drivers */
+ goto out_put;
+
+ adev = obj->dev->dev_private;
+ bo = gem_to_amdgpu_bo(obj);
+ if (!(bo->preferred_domains & (AMDGPU_GEM_DOMAIN_VRAM |
+ AMDGPU_GEM_DOMAIN_GTT)))
+ /* Only VRAM and GTT BOs are supported */
+ goto out_put;
+
+ r = 0;
+ if (dma_buf_kgd)
+ *dma_buf_kgd = (struct kgd_dev *)adev;
+ if (bo_size)
+ *bo_size = amdgpu_bo_size(bo);
+ if (metadata_size)
+ *metadata_size = bo->metadata_size;
+ if (metadata_buffer)
+ r = amdgpu_bo_get_metadata(bo, metadata_buffer, buffer_size,
+ metadata_size, &metadata_flags);
+ if (flags) {
+ *flags = (bo->preferred_domains & AMDGPU_GEM_DOMAIN_VRAM) ?
+ ALLOC_MEM_FLAGS_VRAM : ALLOC_MEM_FLAGS_GTT;
+
+ if (bo->flags & AMDGPU_GEM_CREATE_CPU_ACCESS_REQUIRED)
+ *flags |= ALLOC_MEM_FLAGS_PUBLIC;
+ }
+
+out_put:
+ dma_buf_put(dma_buf);
+ return r;
+}
+
uint64_t amdgpu_amdkfd_get_vram_usage(struct kgd_dev *kgd)
{
struct amdgpu_device *adev = (struct amdgpu_device *)kgd;
@@ -501,7 +572,7 @@ void amdgpu_amdkfd_set_compute_idle(struct kgd_dev *kgd, bool idle)
bool amdgpu_amdkfd_is_kfd_vmid(struct amdgpu_device *adev, u32 vmid)
{
- if (adev->kfd) {
+ if (adev->kfd.dev) {
if ((1 << vmid) & compute_vmid_bitmap)
return true;
}
@@ -515,7 +586,7 @@ bool amdkfd_fence_check_mm(struct dma_fence *f, struct mm_struct *mm)
return false;
}
-void amdgpu_amdkfd_unreserve_system_memory_limit(struct amdgpu_bo *bo)
+void amdgpu_amdkfd_unreserve_memory_limit(struct amdgpu_bo *bo)
{
}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h
index bcf587b4ba98..70429f7aa9a8 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd.h
@@ -27,7 +27,6 @@
#include <linux/types.h>
#include <linux/mm.h>
-#include <linux/mmu_context.h>
#include <linux/workqueue.h>
#include <kgd_kfd_interface.h>
#include <drm/ttm/ttm_execbuf_util.h>
@@ -35,6 +34,7 @@
#include "amdgpu_vm.h"
extern const struct kgd2kfd_calls *kgd2kfd;
+extern uint64_t amdgpu_amdkfd_total_mem_size;
struct amdgpu_device;
@@ -77,6 +77,11 @@ struct amdgpu_amdkfd_fence {
char timeline_name[TASK_COMM_LEN];
};
+struct amdgpu_kfd_dev {
+ struct kfd_dev *dev;
+ uint64_t vram_used;
+};
+
struct amdgpu_amdkfd_fence *amdgpu_amdkfd_fence_create(u64 context,
struct mm_struct *mm);
bool amdkfd_fence_check_mm(struct dma_fence *f, struct mm_struct *mm);
@@ -144,6 +149,11 @@ uint64_t amdgpu_amdkfd_get_gpu_clock_counter(struct kgd_dev *kgd);
uint32_t amdgpu_amdkfd_get_max_engine_clock_in_mhz(struct kgd_dev *kgd);
void amdgpu_amdkfd_get_cu_info(struct kgd_dev *kgd, struct kfd_cu_info *cu_info);
+int amdgpu_amdkfd_get_dmabuf_info(struct kgd_dev *kgd, int dma_buf_fd,
+ struct kgd_dev **dmabuf_kgd,
+ uint64_t *bo_size, void *metadata_buffer,
+ size_t buffer_size, uint32_t *metadata_size,
+ uint32_t *flags);
uint64_t amdgpu_amdkfd_get_vram_usage(struct kgd_dev *kgd);
uint64_t amdgpu_amdkfd_get_hive_id(struct kgd_dev *kgd);
@@ -195,7 +205,13 @@ int amdgpu_amdkfd_gpuvm_restore_process_bos(void *process_info,
int amdgpu_amdkfd_gpuvm_get_vm_fault_info(struct kgd_dev *kgd,
struct kfd_vm_fault_info *info);
+int amdgpu_amdkfd_gpuvm_import_dmabuf(struct kgd_dev *kgd,
+ struct dma_buf *dmabuf,
+ uint64_t va, void *vm,
+ struct kgd_mem **mem, uint64_t *size,
+ uint64_t *mmap_offset);
+
void amdgpu_amdkfd_gpuvm_init_mem_limits(void);
-void amdgpu_amdkfd_unreserve_system_memory_limit(struct amdgpu_bo *bo);
+void amdgpu_amdkfd_unreserve_memory_limit(struct amdgpu_bo *bo);
#endif /* AMDGPU_AMDKFD_H_INCLUDED */
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v7.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v7.c
index 72a357dae070..ff7fac7df34b 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v7.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v7.c
@@ -23,6 +23,7 @@
#include <linux/fdtable.h>
#include <linux/uaccess.h>
#include <linux/firmware.h>
+#include <linux/mmu_context.h>
#include <drm/drmP.h>
#include "amdgpu.h"
#include "amdgpu_amdkfd.h"
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v8.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v8.c
index 0e2a56b6a9b6..56ea929f524b 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v8.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v8.c
@@ -24,6 +24,7 @@
#include <linux/fdtable.h>
#include <linux/uaccess.h>
#include <linux/firmware.h>
+#include <linux/mmu_context.h>
#include <drm/drmP.h>
#include "amdgpu.h"
#include "amdgpu_amdkfd.h"
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.c
index 03b604c96d94..5c51d4910650 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gfx_v9.c
@@ -26,6 +26,7 @@
#include <linux/fdtable.h>
#include <linux/uaccess.h>
#include <linux/firmware.h>
+#include <linux/mmu_context.h>
#include <drm/drmP.h>
#include "amdgpu.h"
#include "amdgpu_amdkfd.h"
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c
index f3129b912714..be1ab43473c6 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_amdkfd_gpuvm.c
@@ -25,6 +25,7 @@
#include <linux/list.h>
#include <linux/pagemap.h>
#include <linux/sched/mm.h>
+#include <linux/dma-buf.h>
#include <drm/drmP.h>
#include "amdgpu_object.h"
#include "amdgpu_vm.h"
@@ -110,17 +111,17 @@ void amdgpu_amdkfd_gpuvm_init_mem_limits(void)
(kfd_mem_limit.max_ttm_mem_limit >> 20));
}
-static int amdgpu_amdkfd_reserve_system_mem_limit(struct amdgpu_device *adev,
+static int amdgpu_amdkfd_reserve_mem_limit(struct amdgpu_device *adev,
uint64_t size, u32 domain, bool sg)
{
- size_t acc_size, system_mem_needed, ttm_mem_needed;
+ size_t acc_size, system_mem_needed, ttm_mem_needed, vram_needed;
+ uint64_t reserved_for_pt = amdgpu_amdkfd_total_mem_size >> 9;
int ret = 0;
acc_size = ttm_bo_dma_acc_size(&adev->mman.bdev, size,
sizeof(struct amdgpu_bo));
- spin_lock(&kfd_mem_limit.mem_limit_lock);
-
+ vram_needed = 0;
if (domain == AMDGPU_GEM_DOMAIN_GTT) {
/* TTM GTT memory */
system_mem_needed = acc_size + size;
@@ -133,23 +134,30 @@ static int amdgpu_amdkfd_reserve_system_mem_limit(struct amdgpu_device *adev,
/* VRAM and SG */
system_mem_needed = acc_size;
ttm_mem_needed = acc_size;
+ if (domain == AMDGPU_GEM_DOMAIN_VRAM)
+ vram_needed = size;
}
+ spin_lock(&kfd_mem_limit.mem_limit_lock);
+
if ((kfd_mem_limit.system_mem_used + system_mem_needed >
- kfd_mem_limit.max_system_mem_limit) ||
- (kfd_mem_limit.ttm_mem_used + ttm_mem_needed >
- kfd_mem_limit.max_ttm_mem_limit))
+ kfd_mem_limit.max_system_mem_limit) ||
+ (kfd_mem_limit.ttm_mem_used + ttm_mem_needed >
+ kfd_mem_limit.max_ttm_mem_limit) ||
+ (adev->kfd.vram_used + vram_needed >
+ adev->gmc.real_vram_size - reserved_for_pt)) {
ret = -ENOMEM;
- else {
+ } else {
kfd_mem_limit.system_mem_used += system_mem_needed;
kfd_mem_limit.ttm_mem_used += ttm_mem_needed;
+ adev->kfd.vram_used += vram_needed;
}
spin_unlock(&kfd_mem_limit.mem_limit_lock);
return ret;
}
-static void unreserve_system_mem_limit(struct amdgpu_device *adev,
+static void unreserve_mem_limit(struct amdgpu_device *adev,
uint64_t size, u32 domain, bool sg)
{
size_t acc_size;
@@ -167,6 +175,11 @@ static void unreserve_system_mem_limit(struct amdgpu_device *adev,
} else {
kfd_mem_limit.system_mem_used -= acc_size;
kfd_mem_limit.ttm_mem_used -= acc_size;
+ if (domain == AMDGPU_GEM_DOMAIN_VRAM) {
+ adev->kfd.vram_used -= size;
+ WARN_ONCE(adev->kfd.vram_used < 0,
+ "kfd VRAM memory accounting unbalanced");
+ }
}
WARN_ONCE(kfd_mem_limit.system_mem_used < 0,
"kfd system memory accounting unbalanced");
@@ -176,29 +189,18 @@ static void unreserve_system_mem_limit(struct amdgpu_device *adev,
spin_unlock(&kfd_mem_limit.mem_limit_lock);
}
-void amdgpu_amdkfd_unreserve_system_memory_limit(struct amdgpu_bo *bo)
+void amdgpu_amdkfd_unreserve_memory_limit(struct amdgpu_bo *bo)
{
- spin_lock(&kfd_mem_limit.mem_limit_lock);
+ struct amdgpu_device *adev = amdgpu_ttm_adev(bo->tbo.bdev);
+ u32 domain = bo->preferred_domains;
+ bool sg = (bo->preferred_domains == AMDGPU_GEM_DOMAIN_CPU);
if (bo->flags & AMDGPU_AMDKFD_USERPTR_BO) {
- kfd_mem_limit.system_mem_used -=
- (bo->tbo.acc_size + amdgpu_bo_size(bo));
- kfd_mem_limit.ttm_mem_used -= bo->tbo.acc_size;
- } else if (bo->preferred_domains == AMDGPU_GEM_DOMAIN_GTT) {
- kfd_mem_limit.system_mem_used -=
- (bo->tbo.acc_size + amdgpu_bo_size(bo));
- kfd_mem_limit.ttm_mem_used -=
- (bo->tbo.acc_size + amdgpu_bo_size(bo));
- } else {
- kfd_mem_limit.system_mem_used -= bo->tbo.acc_size;
- kfd_mem_limit.ttm_mem_used -= bo->tbo.acc_size;
+ domain = AMDGPU_GEM_DOMAIN_CPU;
+ sg = false;
}
- WARN_ONCE(kfd_mem_limit.system_mem_used < 0,
- "kfd system memory accounting unbalanced");
- WARN_ONCE(kfd_mem_limit.ttm_mem_used < 0,
- "kfd TTM memory accounting unbalanced");
- spin_unlock(&kfd_mem_limit.mem_limit_lock);
+ unreserve_mem_limit(adev, amdgpu_bo_size(bo), domain, sg);
}
@@ -535,7 +537,7 @@ static void add_kgd_mem_to_kfd_bo_list(struct kgd_mem *mem,
struct amdgpu_bo *bo = mem->bo;
INIT_LIST_HEAD(&entry->head);
- entry->shared = true;
+ entry->num_shared = 1;
entry->bo = &bo->tbo;
mutex_lock(&process_info->lock);
if (userptr)
@@ -676,7 +678,7 @@ static int reserve_bo_and_vm(struct kgd_mem *mem,
ctx->kfd_bo.priority = 0;
ctx->kfd_bo.tv.bo = &bo->tbo;
- ctx->kfd_bo.tv.shared = true;
+ ctx->kfd_bo.tv.num_shared = 1;
ctx->kfd_bo.user_pages = NULL;
list_add(&ctx->kfd_bo.tv.head, &ctx->list);
@@ -740,7 +742,7 @@ static int reserve_bo_and_cond_vms(struct kgd_mem *mem,
ctx->kfd_bo.priority = 0;
ctx->kfd_bo.tv.bo = &bo->tbo;
- ctx->kfd_bo.tv.shared = true;
+ ctx->kfd_bo.tv.num_shared = 1;
ctx->kfd_bo.user_pages = NULL;
list_add(&ctx->kfd_bo.tv.head, &ctx->list);
@@ -885,6 +887,24 @@ update_gpuvm_pte_failed:
return ret;
}
+static struct sg_table *create_doorbell_sg(uint64_t addr, uint32_t size)
+{
+ struct sg_table *sg = kmalloc(sizeof(*sg), GFP_KERNEL);
+
+ if (!sg)
+ return NULL;
+ if (sg_alloc_table(sg, 1, GFP_KERNEL)) {
+ kfree(sg);
+ return NULL;
+ }
+ sg->sgl->dma_address = addr;
+ sg->sgl->length = size;
+#ifdef CONFIG_NEED_SG_DMA_LENGTH
+ sg->sgl->dma_length = size;
+#endif
+ return sg;
+}
+
static int process_validate_vms(struct amdkfd_process_info *process_info)
{
struct amdgpu_vm *peer_vm;
@@ -1168,6 +1188,8 @@ int amdgpu_amdkfd_gpuvm_alloc_memory_of_gpu(
{
struct amdgpu_device *adev = get_amdgpu_device(kgd);
struct amdgpu_vm *avm = (struct amdgpu_vm *)vm;
+ enum ttm_bo_type bo_type = ttm_bo_type_device;
+ struct sg_table *sg = NULL;
uint64_t user_addr = 0;
struct amdgpu_bo *bo;
struct amdgpu_bo_param bp;
@@ -1196,13 +1218,25 @@ int amdgpu_amdkfd_gpuvm_alloc_memory_of_gpu(
if (!offset || !*offset)
return -EINVAL;
user_addr = *offset;
+ } else if (flags & ALLOC_MEM_FLAGS_DOORBELL) {
+ domain = AMDGPU_GEM_DOMAIN_GTT;
+ alloc_domain = AMDGPU_GEM_DOMAIN_CPU;
+ bo_type = ttm_bo_type_sg;
+ alloc_flags = 0;
+ if (size > UINT_MAX)
+ return -EINVAL;
+ sg = create_doorbell_sg(*offset, size);
+ if (!sg)
+ return -ENOMEM;
} else {
return -EINVAL;
}
*mem = kzalloc(sizeof(struct kgd_mem), GFP_KERNEL);
- if (!*mem)
- return -ENOMEM;
+ if (!*mem) {
+ ret = -ENOMEM;
+ goto err;
+ }
INIT_LIST_HEAD(&(*mem)->bo_va_list);
mutex_init(&(*mem)->lock);
(*mem)->aql_queue = !!(flags & ALLOC_MEM_FLAGS_AQL_QUEUE_MEM);
@@ -1235,8 +1269,7 @@ int amdgpu_amdkfd_gpuvm_alloc_memory_of_gpu(
amdgpu_sync_create(&(*mem)->sync);
- ret = amdgpu_amdkfd_reserve_system_mem_limit(adev, size,
- alloc_domain, false);
+ ret = amdgpu_amdkfd_reserve_mem_limit(adev, size, alloc_domain, !!sg);
if (ret) {
pr_debug("Insufficient system memory\n");
goto err_reserve_limit;
@@ -1250,7 +1283,7 @@ int amdgpu_amdkfd_gpuvm_alloc_memory_of_gpu(
bp.byte_align = byte_align;
bp.domain = alloc_domain;
bp.flags = alloc_flags;
- bp.type = ttm_bo_type_device;
+ bp.type = bo_type;
bp.resv = NULL;
ret = amdgpu_bo_create(adev, &bp, &bo);
if (ret) {
@@ -1258,6 +1291,10 @@ int amdgpu_amdkfd_gpuvm_alloc_memory_of_gpu(
domain_string(alloc_domain), ret);
goto err_bo_create;
}
+ if (bo_type == ttm_bo_type_sg) {
+ bo->tbo.sg = sg;
+ bo->tbo.ttm->sg = sg;
+ }
bo->kfd_bo = *mem;
(*mem)->bo = bo;
if (user_addr)
@@ -1289,10 +1326,15 @@ allocate_init_user_pages_failed:
/* Don't unreserve system mem limit twice */
goto err_reserve_limit;
err_bo_create:
- unreserve_system_mem_limit(adev, size, alloc_domain, false);
+ unreserve_mem_limit(adev, size, alloc_domain, !!sg);
err_reserve_limit:
mutex_destroy(&(*mem)->lock);
kfree(*mem);
+err:
+ if (sg) {
+ sg_free_table(sg);
+ kfree(sg);
+ }
return ret;
}
@@ -1362,6 +1404,14 @@ int amdgpu_amdkfd_gpuvm_free_memory_of_gpu(
/* Free the sync object */
amdgpu_sync_free(&mem->sync);
+ /* If the SG is not NULL, it's one we created for a doorbell
+ * BO. We need to free it.
+ */
+ if (mem->bo->tbo.sg) {
+ sg_free_table(mem->bo->tbo.sg);
+ kfree(mem->bo->tbo.sg);
+ }
+
/* Free the BO*/
amdgpu_bo_unref(&mem->bo);
mutex_destroy(&mem->lock);
@@ -1664,6 +1714,60 @@ int amdgpu_amdkfd_gpuvm_get_vm_fault_info(struct kgd_dev *kgd,
return 0;
}
+int amdgpu_amdkfd_gpuvm_import_dmabuf(struct kgd_dev *kgd,
+ struct dma_buf *dma_buf,
+ uint64_t va, void *vm,
+ struct kgd_mem **mem, uint64_t *size,
+ uint64_t *mmap_offset)
+{
+ struct amdgpu_device *adev = (struct amdgpu_device *)kgd;
+ struct drm_gem_object *obj;
+ struct amdgpu_bo *bo;
+ struct amdgpu_vm *avm = (struct amdgpu_vm *)vm;
+
+ if (dma_buf->ops != &amdgpu_dmabuf_ops)
+ /* Can't handle non-graphics buffers */
+ return -EINVAL;
+
+ obj = dma_buf->priv;
+ if (obj->dev->dev_private != adev)
+ /* Can't handle buffers from other devices */
+ return -EINVAL;
+
+ bo = gem_to_amdgpu_bo(obj);
+ if (!(bo->preferred_domains & (AMDGPU_GEM_DOMAIN_VRAM |
+ AMDGPU_GEM_DOMAIN_GTT)))
+ /* Only VRAM and GTT BOs are supported */
+ return -EINVAL;
+
+ *mem = kzalloc(sizeof(struct kgd_mem), GFP_KERNEL);
+ if (!*mem)
+ return -ENOMEM;
+
+ if (size)
+ *size = amdgpu_bo_size(bo);
+
+ if (mmap_offset)
+ *mmap_offset = amdgpu_bo_mmap_offset(bo);
+
+ INIT_LIST_HEAD(&(*mem)->bo_va_list);
+ mutex_init(&(*mem)->lock);
+ (*mem)->mapping_flags =
+ AMDGPU_VM_PAGE_READABLE | AMDGPU_VM_PAGE_WRITEABLE |
+ AMDGPU_VM_PAGE_EXECUTABLE | AMDGPU_VM_MTYPE_NC;
+
+ (*mem)->bo = amdgpu_bo_ref(bo);
+ (*mem)->va = va;
+ (*mem)->domain = (bo->preferred_domains & AMDGPU_GEM_DOMAIN_VRAM) ?
+ AMDGPU_GEM_DOMAIN_VRAM : AMDGPU_GEM_DOMAIN_GTT;
+ (*mem)->mapped_to_gpu_memory = 0;
+ (*mem)->process_info = avm->process_info;
+ add_kgd_mem_to_kfd_bo_list(*mem, avm->process_info, false);
+ amdgpu_sync_create(&(*mem)->sync);
+
+ return 0;
+}
+
/* Evict a userptr BO by stopping the queues if necessary
*
* Runs in MMU notifier, may be in RECLAIM_FS context. This means it
@@ -1830,7 +1934,7 @@ static int validate_invalid_user_pages(struct amdkfd_process_info *process_info)
validate_list.head) {
list_add_tail(&mem->resv_list.head, &resv_list);
mem->resv_list.bo = mem->validate_list.bo;
- mem->resv_list.shared = mem->validate_list.shared;
+ mem->resv_list.num_shared = mem->validate_list.num_shared;
}
/* Reserve all BOs and page tables for validation */
@@ -2049,7 +2153,7 @@ int amdgpu_amdkfd_gpuvm_restore_process_bos(void *info, struct dma_fence **ef)
list_add_tail(&mem->resv_list.head, &ctx.list);
mem->resv_list.bo = mem->validate_list.bo;
- mem->resv_list.shared = mem->validate_list.shared;
+ mem->resv_list.num_shared = mem->validate_list.num_shared;
}
ret = ttm_eu_reserve_buffers(&ctx.ticket, &ctx.list,
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.c
index 14d2982a47cc..5c79da8e1150 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_bo_list.c
@@ -118,7 +118,6 @@ int amdgpu_bo_list_create(struct amdgpu_device *adev, struct drm_file *filp,
entry->priority = min(info[i].bo_priority,
AMDGPU_BO_LIST_MAX_PRIORITY);
entry->tv.bo = &bo->tbo;
- entry->tv.shared = !bo->prime_shared_count;
if (bo->preferred_domains == AMDGPU_GEM_DOMAIN_GDS)
list->gds_obj = bo;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
index dc54e9efd910..5dc3ee372e2f 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_cs.c
@@ -50,7 +50,8 @@ static int amdgpu_cs_user_fence_chunk(struct amdgpu_cs_parser *p,
bo = amdgpu_bo_ref(gem_to_amdgpu_bo(gobj));
p->uf_entry.priority = 0;
p->uf_entry.tv.bo = &bo->tbo;
- p->uf_entry.tv.shared = true;
+ /* One for TTM and one for the CS job */
+ p->uf_entry.tv.num_shared = 2;
p->uf_entry.user_pages = NULL;
drm_gem_object_put_unlocked(gobj);
@@ -598,6 +599,10 @@ static int amdgpu_cs_parser_bos(struct amdgpu_cs_parser *p,
return r;
}
+ /* One for TTM and one for the CS job */
+ amdgpu_bo_list_for_each_entry(e, p->bo_list)
+ e->tv.num_shared = 2;
+
amdgpu_bo_list_get_list(p->bo_list, &p->validated);
if (p->bo_list->first_userptr != p->bo_list->num_entries)
p->mn = amdgpu_mn_get(p->adev, AMDGPU_MN_TYPE_GFX);
@@ -717,8 +722,14 @@ static int amdgpu_cs_parser_bos(struct amdgpu_cs_parser *p,
gws = p->bo_list->gws_obj;
oa = p->bo_list->oa_obj;
- amdgpu_bo_list_for_each_entry(e, p->bo_list)
- e->bo_va = amdgpu_vm_bo_find(vm, ttm_to_amdgpu_bo(e->tv.bo));
+ amdgpu_bo_list_for_each_entry(e, p->bo_list) {
+ struct amdgpu_bo *bo = ttm_to_amdgpu_bo(e->tv.bo);
+
+ /* Make sure we use the exclusive slot for shared BOs */
+ if (bo->prime_shared_count)
+ e->tv.num_shared = 0;
+ e->bo_va = amdgpu_vm_bo_find(vm, bo);
+ }
if (gds) {
p->job->gds_base = amdgpu_bo_gpu_offset(gds) >> PAGE_SHIFT;
@@ -955,10 +966,6 @@ static int amdgpu_cs_vm_handling(struct amdgpu_cs_parser *p)
if (r)
return r;
- r = reservation_object_reserve_shared(vm->root.base.bo->tbo.resv, 1);
- if (r)
- return r;
-
p->job->vm_pd_addr = amdgpu_gmc_pd_addr(vm->root.base.bo);
if (amdgpu_vm_debug) {
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_csa.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_csa.c
index 5b550706ee76..7e22be7ca68a 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_csa.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_csa.c
@@ -74,7 +74,7 @@ int amdgpu_map_static_csa(struct amdgpu_device *adev, struct amdgpu_vm *vm,
INIT_LIST_HEAD(&list);
INIT_LIST_HEAD(&csa_tv.head);
csa_tv.bo = &bo->tbo;
- csa_tv.shared = true;
+ csa_tv.num_shared = 1;
list_add(&csa_tv.head, &list);
amdgpu_vm_get_pd_bo(vm, &list, &pd);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c
index 95f4c4139fc6..d85184b5b35c 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ctx.c
@@ -248,7 +248,7 @@ static int amdgpu_ctx_alloc(struct amdgpu_device *adev,
return -ENOMEM;
mutex_lock(&mgr->lock);
- r = idr_alloc(&mgr->ctx_handles, ctx, 1, 0, GFP_KERNEL);
+ r = idr_alloc(&mgr->ctx_handles, ctx, 1, AMDGPU_VM_MAX_NUM_CTX, GFP_KERNEL);
if (r < 0) {
mutex_unlock(&mgr->lock);
kfree(ctx);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
index c75badfa5c4c..b60afeade50a 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_device.c
@@ -515,7 +515,6 @@ void amdgpu_device_pci_config_reset(struct amdgpu_device *adev)
*/
static int amdgpu_device_doorbell_init(struct amdgpu_device *adev)
{
- amdgpu_asic_init_doorbell_index(adev);
/* No doorbell on SI hardware generation */
if (adev->asic_type < CHIP_BONAIRE) {
@@ -529,6 +528,8 @@ static int amdgpu_device_doorbell_init(struct amdgpu_device *adev)
if (pci_resource_flags(adev->pdev, 2) & IORESOURCE_UNSET)
return -EINVAL;
+ amdgpu_asic_init_doorbell_index(adev);
+
/* doorbell bar mapping */
adev->doorbell.base = pci_resource_start(adev->pdev, 2);
adev->doorbell.size = pci_resource_len(adev->pdev, 2);
@@ -1864,6 +1865,9 @@ static int amdgpu_device_ip_fini(struct amdgpu_device *adev)
{
int i, r;
+ if (adev->gmc.xgmi.num_physical_nodes > 1)
+ amdgpu_xgmi_remove_device(adev);
+
amdgpu_amdkfd_device_fini(adev);
amdgpu_device_set_pg_state(adev, AMD_PG_STATE_UNGATE);
@@ -2353,6 +2357,19 @@ bool amdgpu_device_has_dc_support(struct amdgpu_device *adev)
return amdgpu_device_asic_has_dc_support(adev->asic_type);
}
+
+static void amdgpu_device_xgmi_reset_func(struct work_struct *__work)
+{
+ struct amdgpu_device *adev =
+ container_of(__work, struct amdgpu_device, xgmi_reset_work);
+
+ adev->asic_reset_res = amdgpu_asic_reset(adev);
+ if (adev->asic_reset_res)
+ DRM_WARN("ASIC reset failed with err r, %d for drm dev, %s",
+ adev->asic_reset_res, adev->ddev->unique);
+}
+
+
/**
* amdgpu_device_init - initialize the driver
*
@@ -2451,6 +2468,8 @@ int amdgpu_device_init(struct amdgpu_device *adev,
INIT_DELAYED_WORK(&adev->gfx.gfx_off_delay_work,
amdgpu_device_delay_enable_gfx_off);
+ INIT_WORK(&adev->xgmi_reset_work, amdgpu_device_xgmi_reset_func);
+
adev->gfx.gfx_off_req_count = 1;
adev->pm.ac_power = power_supply_is_system_supplied() > 0 ? true : false;
@@ -3239,6 +3258,8 @@ bool amdgpu_device_should_recover_gpu(struct amdgpu_device *adev)
if (amdgpu_gpu_recovery == -1) {
switch (adev->asic_type) {
+ case CHIP_BONAIRE:
+ case CHIP_HAWAII:
case CHIP_TOPAZ:
case CHIP_TONGA:
case CHIP_FIJI:
@@ -3328,10 +3349,31 @@ static int amdgpu_do_asic_reset(struct amdgpu_hive_info *hive,
*/
if (need_full_reset) {
list_for_each_entry(tmp_adev, device_list_handle, gmc.xgmi.head) {
- r = amdgpu_asic_reset(tmp_adev);
- if (r)
- DRM_WARN("ASIC reset failed with err r, %d for drm dev, %s",
+ /* For XGMI run all resets in parallel to speed up the process */
+ if (tmp_adev->gmc.xgmi.num_physical_nodes > 1) {
+ if (!queue_work(system_highpri_wq, &tmp_adev->xgmi_reset_work))
+ r = -EALREADY;
+ } else
+ r = amdgpu_asic_reset(tmp_adev);
+
+ if (r) {
+ DRM_ERROR("ASIC reset failed with err r, %d for drm dev, %s",
r, tmp_adev->ddev->unique);
+ break;
+ }
+ }
+
+ /* For XGMI wait for all PSP resets to complete before proceed */
+ if (!r) {
+ list_for_each_entry(tmp_adev, device_list_handle,
+ gmc.xgmi.head) {
+ if (tmp_adev->gmc.xgmi.num_physical_nodes > 1) {
+ flush_work(&tmp_adev->xgmi_reset_work);
+ r = tmp_adev->asic_reset_res;
+ if (r)
+ break;
+ }
+ }
}
}
@@ -3518,8 +3560,6 @@ retry: /* Rest of adevs pre asic reset from XGMI hive. */
if (tmp_adev == adev)
continue;
- dev_info(tmp_adev->dev, "GPU reset begin for drm dev %s!\n", adev->ddev->unique);
-
amdgpu_device_lock_adev(tmp_adev);
r = amdgpu_device_pre_asic_reset(tmp_adev,
NULL,
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
index 7b3d1ebda9df..f4f00217546e 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.c
@@ -169,7 +169,7 @@ void amdgpu_gem_object_close(struct drm_gem_object *obj,
INIT_LIST_HEAD(&duplicates);
tv.bo = &bo->tbo;
- tv.shared = true;
+ tv.num_shared = 1;
list_add(&tv.head, &list);
amdgpu_vm_get_pd_bo(vm, &list, &vm_pd);
@@ -604,7 +604,10 @@ int amdgpu_gem_va_ioctl(struct drm_device *dev, void *data,
return -ENOENT;
abo = gem_to_amdgpu_bo(gobj);
tv.bo = &abo->tbo;
- tv.shared = !!(abo->flags & AMDGPU_GEM_CREATE_VM_ALWAYS_VALID);
+ if (abo->flags & AMDGPU_GEM_CREATE_VM_ALWAYS_VALID)
+ tv.num_shared = 1;
+ else
+ tv.num_shared = 0;
list_add(&tv.head, &list);
} else {
gobj = NULL;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.h
index d63daba9b17c..f1ddfc50bcc7 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gem.h
@@ -54,6 +54,8 @@ void *amdgpu_gem_prime_vmap(struct drm_gem_object *obj);
void amdgpu_gem_prime_vunmap(struct drm_gem_object *obj, void *vaddr);
int amdgpu_gem_prime_mmap(struct drm_gem_object *obj, struct vm_area_struct *vma);
+extern const struct dma_buf_ops amdgpu_dmabuf_ops;
+
/*
* GEM objects.
*/
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h
index 8c57924c075f..81e6070d255b 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_gmc.h
@@ -99,6 +99,7 @@ struct amdgpu_xgmi {
unsigned num_physical_nodes;
/* gpu list in the same hive */
struct list_head head;
+ bool supported;
};
struct amdgpu_gmc {
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ih.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_ih.h
index 9ce8c93ec19b..f877bb78d10a 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ih.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ih.h
@@ -51,14 +51,12 @@ struct amdgpu_ih_ring {
struct amdgpu_ih_funcs {
/* ring read/write ptr handling, called from interrupt context */
u32 (*get_wptr)(struct amdgpu_device *adev);
- bool (*prescreen_iv)(struct amdgpu_device *adev);
void (*decode_iv)(struct amdgpu_device *adev,
struct amdgpu_iv_entry *entry);
void (*set_rptr)(struct amdgpu_device *adev);
};
#define amdgpu_ih_get_wptr(adev) (adev)->irq.ih_funcs->get_wptr((adev))
-#define amdgpu_ih_prescreen_iv(adev) (adev)->irq.ih_funcs->prescreen_iv((adev))
#define amdgpu_ih_decode_iv(adev, iv) (adev)->irq.ih_funcs->decode_iv((adev), (iv))
#define amdgpu_ih_set_rptr(adev) (adev)->irq.ih_funcs->set_rptr((adev))
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c
index 6b6524f04ce0..b7968f426862 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_irq.c
@@ -145,13 +145,6 @@ static void amdgpu_irq_callback(struct amdgpu_device *adev,
u32 ring_index = ih->rptr >> 2;
struct amdgpu_iv_entry entry;
- /* Prescreening of high-frequency interrupts */
- if (!amdgpu_ih_prescreen_iv(adev))
- return;
-
- /* Before dispatching irq to IP blocks, send it to amdkfd */
- amdgpu_amdkfd_interrupt(adev, (const void *) &ih->ring[ring_index]);
-
entry.iv_entry = (const uint32_t *)&ih->ring[ring_index];
amdgpu_ih_decode_iv(adev, &entry);
@@ -371,39 +364,38 @@ void amdgpu_irq_dispatch(struct amdgpu_device *adev,
unsigned client_id = entry->client_id;
unsigned src_id = entry->src_id;
struct amdgpu_irq_src *src;
+ bool handled = false;
int r;
trace_amdgpu_iv(entry);
if (client_id >= AMDGPU_IRQ_CLIENTID_MAX) {
DRM_DEBUG("Invalid client_id in IV: %d\n", client_id);
- return;
- }
- if (src_id >= AMDGPU_MAX_IRQ_SRC_ID) {
+ } else if (src_id >= AMDGPU_MAX_IRQ_SRC_ID) {
DRM_DEBUG("Invalid src_id in IV: %d\n", src_id);
- return;
- }
- if (adev->irq.virq[src_id]) {
+ } else if (adev->irq.virq[src_id]) {
generic_handle_irq(irq_find_mapping(adev->irq.domain, src_id));
- } else {
- if (!adev->irq.client[client_id].sources) {
- DRM_DEBUG("Unregistered interrupt client_id: %d src_id: %d\n",
- client_id, src_id);
- return;
- }
- src = adev->irq.client[client_id].sources[src_id];
- if (!src) {
- DRM_DEBUG("Unhandled interrupt src_id: %d\n", src_id);
- return;
- }
+ } else if (!adev->irq.client[client_id].sources) {
+ DRM_DEBUG("Unregistered interrupt client_id: %d src_id: %d\n",
+ client_id, src_id);
+ } else if ((src = adev->irq.client[client_id].sources[src_id])) {
r = src->funcs->process(adev, src, entry);
- if (r)
+ if (r < 0)
DRM_ERROR("error processing interrupt (%d)\n", r);
+ else if (r)
+ handled = true;
+
+ } else {
+ DRM_DEBUG("Unhandled interrupt src_id: %d\n", src_id);
}
+
+ /* Send it to amdkfd as well if it isn't already handled */
+ if (!handled)
+ amdgpu_amdkfd_interrupt(adev, entry->iv_entry);
}
/**
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
index cf768acb51dc..fd271f9746a2 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.c
@@ -81,7 +81,7 @@ static void amdgpu_bo_destroy(struct ttm_buffer_object *tbo)
amdgpu_bo_subtract_pin_size(bo);
if (bo->kfd_bo)
- amdgpu_amdkfd_unreserve_system_memory_limit(bo);
+ amdgpu_amdkfd_unreserve_memory_limit(bo);
amdgpu_bo_kunmap(bo);
@@ -608,53 +608,6 @@ int amdgpu_bo_create(struct amdgpu_device *adev,
}
/**
- * amdgpu_bo_backup_to_shadow - Backs up an &amdgpu_bo buffer object
- * @adev: amdgpu device object
- * @ring: amdgpu_ring for the engine handling the buffer operations
- * @bo: &amdgpu_bo buffer to be backed up
- * @resv: reservation object with embedded fence
- * @fence: dma_fence associated with the operation
- * @direct: whether to submit the job directly
- *
- * Copies an &amdgpu_bo buffer object to its shadow object.
- * Not used for now.
- *
- * Returns:
- * 0 for success or a negative error code on failure.
- */
-int amdgpu_bo_backup_to_shadow(struct amdgpu_device *adev,
- struct amdgpu_ring *ring,
- struct amdgpu_bo *bo,
- struct reservation_object *resv,
- struct dma_fence **fence,
- bool direct)
-
-{
- struct amdgpu_bo *shadow = bo->shadow;
- uint64_t bo_addr, shadow_addr;
- int r;
-
- if (!shadow)
- return -EINVAL;
-
- bo_addr = amdgpu_bo_gpu_offset(bo);
- shadow_addr = amdgpu_bo_gpu_offset(bo->shadow);
-
- r = reservation_object_reserve_shared(bo->tbo.resv, 1);
- if (r)
- goto err;
-
- r = amdgpu_copy_buffer(ring, bo_addr, shadow_addr,
- amdgpu_bo_size(bo), resv, fence,
- direct, false);
- if (!r)
- amdgpu_bo_fence(bo, *fence, true);
-
-err:
- return r;
-}
-
-/**
* amdgpu_bo_validate - validate an &amdgpu_bo buffer object
* @bo: pointer to the buffer object
*
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h
index 7d3312d0da11..9291c2f837e9 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_object.h
@@ -267,11 +267,6 @@ int amdgpu_bo_fault_reserve_notify(struct ttm_buffer_object *bo);
void amdgpu_bo_fence(struct amdgpu_bo *bo, struct dma_fence *fence,
bool shared);
u64 amdgpu_bo_gpu_offset(struct amdgpu_bo *bo);
-int amdgpu_bo_backup_to_shadow(struct amdgpu_device *adev,
- struct amdgpu_ring *ring,
- struct amdgpu_bo *bo,
- struct reservation_object *resv,
- struct dma_fence **fence, bool direct);
int amdgpu_bo_validate(struct amdgpu_bo *bo);
int amdgpu_bo_restore_shadow(struct amdgpu_bo *shadow,
struct dma_fence **fence);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_prime.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_prime.c
index 3e44d889f7af..71913a18d142 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_prime.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_prime.c
@@ -39,8 +39,6 @@
#include <drm/amdgpu_drm.h>
#include <linux/dma-buf.h>
-static const struct dma_buf_ops amdgpu_dmabuf_ops;
-
/**
* amdgpu_gem_prime_get_sg_table - &drm_driver.gem_prime_get_sg_table
* implementation
@@ -332,7 +330,7 @@ static int amdgpu_gem_begin_cpu_access(struct dma_buf *dma_buf,
return ret;
}
-static const struct dma_buf_ops amdgpu_dmabuf_ops = {
+const struct dma_buf_ops amdgpu_dmabuf_ops = {
.attach = amdgpu_gem_map_attach,
.detach = amdgpu_gem_map_detach,
.map_dma_buf = drm_gem_map_dma_buf,
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c
index e05dc66b1090..6759d898b3ab 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.c
@@ -155,10 +155,22 @@ psp_cmd_submit_buf(struct psp_context *psp,
return ret;
}
-static void psp_prep_tmr_cmd_buf(struct psp_gfx_cmd_resp *cmd,
+bool psp_support_vmr_ring(struct psp_context *psp)
+{
+ if (amdgpu_sriov_vf(psp->adev) && psp->sos_fw_version > 0x80045)
+ return true;
+ else
+ return false;
+}
+
+static void psp_prep_tmr_cmd_buf(struct psp_context *psp,
+ struct psp_gfx_cmd_resp *cmd,
uint64_t tmr_mc, uint32_t size)
{
- cmd->cmd_id = GFX_CMD_ID_SETUP_TMR;
+ if (psp_support_vmr_ring(psp))
+ cmd->cmd_id = GFX_CMD_ID_SETUP_VMR;
+ else
+ cmd->cmd_id = GFX_CMD_ID_SETUP_TMR;
cmd->cmd.cmd_setup_tmr.buf_phy_addr_lo = lower_32_bits(tmr_mc);
cmd->cmd.cmd_setup_tmr.buf_phy_addr_hi = upper_32_bits(tmr_mc);
cmd->cmd.cmd_setup_tmr.buf_size = size;
@@ -192,7 +204,7 @@ static int psp_tmr_load(struct psp_context *psp)
if (!cmd)
return -ENOMEM;
- psp_prep_tmr_cmd_buf(cmd, psp->tmr_mc_addr, PSP_TMR_SIZE);
+ psp_prep_tmr_cmd_buf(psp, cmd, psp->tmr_mc_addr, PSP_TMR_SIZE);
DRM_INFO("reserve 0x%x from 0x%llx for PSP TMR SIZE\n",
PSP_TMR_SIZE, psp->tmr_mc_addr);
@@ -536,8 +548,10 @@ static int psp_load_fw(struct amdgpu_device *adev)
int ret;
struct psp_context *psp = &adev->psp;
- if (amdgpu_sriov_vf(adev) && adev->in_gpu_reset != 0)
+ if (amdgpu_sriov_vf(adev) && adev->in_gpu_reset) {
+ psp_ring_destroy(psp, PSP_RING_TYPE__KM);
goto skip_memalloc;
+ }
psp->cmd = kzalloc(sizeof(struct psp_gfx_cmd_resp), GFP_KERNEL);
if (!psp->cmd)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h
index 9ec5d1a666a6..10decf70c9aa 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_psp.h
@@ -217,6 +217,7 @@ extern const struct amdgpu_ip_block_version psp_v10_0_ip_block;
int psp_gpu_reset(struct amdgpu_device *adev);
int psp_xgmi_invoke(struct psp_context *psp, uint32_t ta_cmd_id);
+bool psp_support_vmr_ring(struct psp_context *psp);
extern const struct amdgpu_ip_block_version psp_v11_0_ip_block;
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c
index 5b75bdc8dc28..335a0edf114b 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ring.c
@@ -397,7 +397,7 @@ bool amdgpu_ring_soft_recovery(struct amdgpu_ring *ring, unsigned int vmid,
{
ktime_t deadline = ktime_add_us(ktime_get(), 10000);
- if (!ring->funcs->soft_recovery)
+ if (!ring->funcs->soft_recovery || !fence)
return false;
atomic_inc(&ring->adev->gpu_reset_counter);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
index 58a2363040dd..fc91f3e54a87 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c
@@ -617,7 +617,8 @@ void amdgpu_vm_get_pd_bo(struct amdgpu_vm *vm,
{
entry->priority = 0;
entry->tv.bo = &vm->root.base.bo->tbo;
- entry->tv.shared = true;
+ /* One for the VM updates, one for TTM and one for the CS job */
+ entry->tv.num_shared = 3;
entry->user_pages = NULL;
list_add(&entry->tv.head, validated);
}
@@ -773,10 +774,6 @@ static int amdgpu_vm_clear_bo(struct amdgpu_device *adev,
ring = container_of(vm->entity.rq->sched, struct amdgpu_ring, sched);
- r = reservation_object_reserve_shared(bo->tbo.resv, 1);
- if (r)
- return r;
-
r = ttm_bo_validate(&bo->tbo, &bo->placement, &ctx);
if (r)
goto error;
@@ -1842,10 +1839,6 @@ static int amdgpu_vm_bo_update_mapping(struct amdgpu_device *adev,
if (r)
goto error_free;
- r = reservation_object_reserve_shared(vm->root.base.bo->tbo.resv, 1);
- if (r)
- goto error_free;
-
r = amdgpu_vm_update_ptes(&params, start, last + 1, addr, flags);
if (r)
goto error_free;
@@ -3026,6 +3019,10 @@ int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm,
if (r)
goto error_free_root;
+ r = reservation_object_reserve_shared(root->tbo.resv, 1);
+ if (r)
+ goto error_unreserve;
+
r = amdgpu_vm_clear_bo(adev, vm, root,
adev->vm_manager.root_level,
vm->pte_support_ats);
@@ -3055,7 +3052,6 @@ int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm,
}
INIT_KFIFO(vm->faults);
- vm->fault_credit = 16;
return 0;
@@ -3268,42 +3264,6 @@ void amdgpu_vm_fini(struct amdgpu_device *adev, struct amdgpu_vm *vm)
}
/**
- * amdgpu_vm_pasid_fault_credit - Check fault credit for given PASID
- *
- * @adev: amdgpu_device pointer
- * @pasid: PASID do identify the VM
- *
- * This function is expected to be called in interrupt context.
- *
- * Returns:
- * True if there was fault credit, false otherwise
- */
-bool amdgpu_vm_pasid_fault_credit(struct amdgpu_device *adev,
- unsigned int pasid)
-{
- struct amdgpu_vm *vm;
-
- spin_lock(&adev->vm_manager.pasid_lock);
- vm = idr_find(&adev->vm_manager.pasid_idr, pasid);
- if (!vm) {
- /* VM not found, can't track fault credit */
- spin_unlock(&adev->vm_manager.pasid_lock);
- return true;
- }
-
- /* No lock needed. only accessed by IRQ handler */
- if (!vm->fault_credit) {
- /* Too many faults in this VM */
- spin_unlock(&adev->vm_manager.pasid_lock);
- return false;
- }
-
- vm->fault_credit--;
- spin_unlock(&adev->vm_manager.pasid_lock);
- return true;
-}
-
-/**
* amdgpu_vm_manager_init - init the VM manager
*
* @adev: amdgpu_device pointer
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h
index 2a8898d19c8b..e8dcfd59fc93 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.h
@@ -229,9 +229,6 @@ struct amdgpu_vm {
/* Up to 128 pending retry page faults */
DECLARE_KFIFO(faults, u64, 128);
- /* Limit non-retry fault storms */
- unsigned int fault_credit;
-
/* Points to the KFD process VM info */
struct amdkfd_process_info *process_info;
@@ -299,8 +296,6 @@ int amdgpu_vm_init(struct amdgpu_device *adev, struct amdgpu_vm *vm,
int amdgpu_vm_make_compute(struct amdgpu_device *adev, struct amdgpu_vm *vm, unsigned int pasid);
void amdgpu_vm_release_compute(struct amdgpu_device *adev, struct amdgpu_vm *vm);
void amdgpu_vm_fini(struct amdgpu_device *adev, struct amdgpu_vm *vm);
-bool amdgpu_vm_pasid_fault_credit(struct amdgpu_device *adev,
- unsigned int pasid);
void amdgpu_vm_get_pd_bo(struct amdgpu_vm *vm,
struct list_head *validated,
struct amdgpu_bo_list_entry *entry);
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c
index fb37e69f1bba..0b263a9857c6 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.c
@@ -78,7 +78,7 @@ int amdgpu_xgmi_update_topology(struct amdgpu_hive_info *hive, struct amdgpu_dev
adev->gmc.xgmi.node_id,
adev->gmc.xgmi.hive_id, ret);
else
- dev_info(adev->dev, "XGMI: Add node %d to hive 0x%llx.\n",
+ dev_info(adev->dev, "XGMI: Set topology for node %d, hive 0x%llx.\n",
adev->gmc.xgmi.physical_node_id,
adev->gmc.xgmi.hive_id);
@@ -94,9 +94,9 @@ int amdgpu_xgmi_add_device(struct amdgpu_device *adev)
int count = 0, ret = -EINVAL;
- if ((adev->asic_type < CHIP_VEGA20) ||
- (adev->flags & AMD_IS_APU) )
+ if (!adev->gmc.xgmi.supported)
return 0;
+
adev->gmc.xgmi.node_id = psp_xgmi_get_node_id(&adev->psp);
adev->gmc.xgmi.hive_id = psp_xgmi_get_hive_id(&adev->psp);
@@ -135,3 +135,23 @@ exit:
mutex_unlock(&xgmi_mutex);
return ret;
}
+
+void amdgpu_xgmi_remove_device(struct amdgpu_device *adev)
+{
+ struct amdgpu_hive_info *hive;
+
+ if (!adev->gmc.xgmi.supported)
+ return;
+
+ mutex_lock(&xgmi_mutex);
+
+ hive = amdgpu_get_xgmi_hive(adev);
+ if (!hive)
+ goto exit;
+
+ if (!(hive->number_devices--))
+ mutex_destroy(&hive->hive_lock);
+
+exit:
+ mutex_unlock(&xgmi_mutex);
+}
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.h b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.h
index 6335bfdcc51d..6151eb9c8ad3 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.h
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_xgmi.h
@@ -35,5 +35,6 @@ struct amdgpu_hive_info {
struct amdgpu_hive_info *amdgpu_get_xgmi_hive(struct amdgpu_device *adev);
int amdgpu_xgmi_update_topology(struct amdgpu_hive_info *hive, struct amdgpu_device *adev);
int amdgpu_xgmi_add_device(struct amdgpu_device *adev);
+void amdgpu_xgmi_remove_device(struct amdgpu_device *adev);
#endif
diff --git a/drivers/gpu/drm/amd/amdgpu/cik_ih.c b/drivers/gpu/drm/amd/amdgpu/cik_ih.c
index b5775c6a857b..8a8b4967a101 100644
--- a/drivers/gpu/drm/amd/amdgpu/cik_ih.c
+++ b/drivers/gpu/drm/amd/amdgpu/cik_ih.c
@@ -228,34 +228,6 @@ static u32 cik_ih_get_wptr(struct amdgpu_device *adev)
* [127:96] - reserved
*/
-/**
- * cik_ih_prescreen_iv - prescreen an interrupt vector
- *
- * @adev: amdgpu_device pointer
- *
- * Returns true if the interrupt vector should be further processed.
- */
-static bool cik_ih_prescreen_iv(struct amdgpu_device *adev)
-{
- u32 ring_index = adev->irq.ih.rptr >> 2;
- u16 pasid;
-
- switch (le32_to_cpu(adev->irq.ih.ring[ring_index]) & 0xff) {
- case 146:
- case 147:
- pasid = le32_to_cpu(adev->irq.ih.ring[ring_index + 2]) >> 16;
- if (!pasid || amdgpu_vm_pasid_fault_credit(adev, pasid))
- return true;
- break;
- default:
- /* Not a VM fault */
- return true;
- }
-
- adev->irq.ih.rptr += 16;
- return false;
-}
-
/**
* cik_ih_decode_iv - decode an interrupt vector
*
@@ -461,7 +433,6 @@ static const struct amd_ip_funcs cik_ih_ip_funcs = {
static const struct amdgpu_ih_funcs cik_ih_funcs = {
.get_wptr = cik_ih_get_wptr,
- .prescreen_iv = cik_ih_prescreen_iv,
.decode_iv = cik_ih_decode_iv,
.set_rptr = cik_ih_set_rptr
};
diff --git a/drivers/gpu/drm/amd/amdgpu/cz_ih.c b/drivers/gpu/drm/amd/amdgpu/cz_ih.c
index df5ac4d85a00..9d3ea298e116 100644
--- a/drivers/gpu/drm/amd/amdgpu/cz_ih.c
+++ b/drivers/gpu/drm/amd/amdgpu/cz_ih.c
@@ -208,34 +208,6 @@ static u32 cz_ih_get_wptr(struct amdgpu_device *adev)
}
/**
- * cz_ih_prescreen_iv - prescreen an interrupt vector
- *
- * @adev: amdgpu_device pointer
- *
- * Returns true if the interrupt vector should be further processed.
- */
-static bool cz_ih_prescreen_iv(struct amdgpu_device *adev)
-{
- u32 ring_index = adev->irq.ih.rptr >> 2;
- u16 pasid;
-
- switch (le32_to_cpu(adev->irq.ih.ring[ring_index]) & 0xff) {
- case 146:
- case 147:
- pasid = le32_to_cpu(adev->irq.ih.ring[ring_index + 2]) >> 16;
- if (!pasid || amdgpu_vm_pasid_fault_credit(adev, pasid))
- return true;
- break;
- default:
- /* Not a VM fault */
- return true;
- }
-
- adev->irq.ih.rptr += 16;
- return false;
-}
-
-/**
* cz_ih_decode_iv - decode an interrupt vector
*
* @adev: amdgpu_device pointer
@@ -442,7 +414,6 @@ static const struct amd_ip_funcs cz_ih_ip_funcs = {
static const struct amdgpu_ih_funcs cz_ih_funcs = {
.get_wptr = cz_ih_get_wptr,
- .prescreen_iv = cz_ih_prescreen_iv,
.decode_iv = cz_ih_decode_iv,
.set_rptr = cz_ih_set_rptr
};
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c
index 1454fc306783..381f593b0cda 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v8_0.c
@@ -4068,6 +4068,11 @@ static void gfx_v8_0_rlc_start(struct amdgpu_device *adev)
static int gfx_v8_0_rlc_resume(struct amdgpu_device *adev)
{
+ if (amdgpu_sriov_vf(adev)) {
+ gfx_v8_0_init_csb(adev);
+ return 0;
+ }
+
adev->gfx.rlc.funcs->stop(adev);
adev->gfx.rlc.funcs->reset(adev);
gfx_v8_0_init_pg(adev);
@@ -4947,14 +4952,13 @@ static bool gfx_v8_0_check_soft_reset(void *handle)
static int gfx_v8_0_pre_soft_reset(void *handle)
{
struct amdgpu_device *adev = (struct amdgpu_device *)handle;
- u32 grbm_soft_reset = 0, srbm_soft_reset = 0;
+ u32 grbm_soft_reset = 0;
if ((!adev->gfx.grbm_soft_reset) &&
(!adev->gfx.srbm_soft_reset))
return 0;
grbm_soft_reset = adev->gfx.grbm_soft_reset;
- srbm_soft_reset = adev->gfx.srbm_soft_reset;
/* stop the rlc */
adev->gfx.rlc.funcs->stop(adev);
@@ -5051,14 +5055,13 @@ static int gfx_v8_0_soft_reset(void *handle)
static int gfx_v8_0_post_soft_reset(void *handle)
{
struct amdgpu_device *adev = (struct amdgpu_device *)handle;
- u32 grbm_soft_reset = 0, srbm_soft_reset = 0;
+ u32 grbm_soft_reset = 0;
if ((!adev->gfx.grbm_soft_reset) &&
(!adev->gfx.srbm_soft_reset))
return 0;
grbm_soft_reset = adev->gfx.grbm_soft_reset;
- srbm_soft_reset = adev->gfx.srbm_soft_reset;
if (REG_GET_FIELD(grbm_soft_reset, GRBM_SOFT_RESET, SOFT_RESET_CP) ||
REG_GET_FIELD(grbm_soft_reset, GRBM_SOFT_RESET, SOFT_RESET_CPF) ||
diff --git a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c
index af8ccb014be3..f62d570a81a8 100644
--- a/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gfx_v9_0.c
@@ -86,6 +86,7 @@ MODULE_FIRMWARE("amdgpu/picasso_me.bin");
MODULE_FIRMWARE("amdgpu/picasso_mec.bin");
MODULE_FIRMWARE("amdgpu/picasso_mec2.bin");
MODULE_FIRMWARE("amdgpu/picasso_rlc.bin");
+MODULE_FIRMWARE("amdgpu/picasso_rlc_am4.bin");
MODULE_FIRMWARE("amdgpu/raven2_ce.bin");
MODULE_FIRMWARE("amdgpu/raven2_pfp.bin");
@@ -645,7 +646,20 @@ static int gfx_v9_0_init_microcode(struct amdgpu_device *adev)
adev->gfx.ce_fw_version = le32_to_cpu(cp_hdr->header.ucode_version);
adev->gfx.ce_feature_version = le32_to_cpu(cp_hdr->ucode_feature_version);
- snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_rlc.bin", chip_name);
+ /*
+ * For Picasso && AM4 SOCKET board, we use picasso_rlc_am4.bin
+ * instead of picasso_rlc.bin.
+ * Judgment method:
+ * PCO AM4: revision >= 0xC8 && revision <= 0xCF
+ * or revision >= 0xD8 && revision <= 0xDF
+ * otherwise is PCO FP5
+ */
+ if (!strcmp(chip_name, "picasso") &&
+ (((adev->pdev->revision >= 0xC8) && (adev->pdev->revision <= 0xCF)) ||
+ ((adev->pdev->revision >= 0xD8) && (adev->pdev->revision <= 0xDF))))
+ snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_rlc_am4.bin", chip_name);
+ else
+ snprintf(fw_name, sizeof(fw_name), "amdgpu/%s_rlc.bin", chip_name);
err = request_firmware(&adev->gfx.rlc_fw, fw_name, adev->dev);
if (err)
goto out;
diff --git a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
index 3a4e5d8d5162..ce150de723c9 100644
--- a/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/gmc_v9_0.c
@@ -244,6 +244,62 @@ static int gmc_v9_0_vm_fault_interrupt_state(struct amdgpu_device *adev,
return 0;
}
+/**
+ * vega10_ih_prescreen_iv - prescreen an interrupt vector
+ *
+ * @adev: amdgpu_device pointer
+ *
+ * Returns true if the interrupt vector should be further processed.
+ */
+static bool gmc_v9_0_prescreen_iv(struct amdgpu_device *adev,
+ struct amdgpu_iv_entry *entry,
+ uint64_t addr)
+{
+ struct amdgpu_vm *vm;
+ u64 key;
+ int r;
+
+ /* No PASID, can't identify faulting process */
+ if (!entry->pasid)
+ return true;
+
+ /* Not a retry fault */
+ if (!(entry->src_data[1] & 0x80))
+ return true;
+
+ /* Track retry faults in per-VM fault FIFO. */
+ spin_lock(&adev->vm_manager.pasid_lock);
+ vm = idr_find(&adev->vm_manager.pasid_idr, entry->pasid);
+ if (!vm) {
+ /* VM not found, process it normally */
+ spin_unlock(&adev->vm_manager.pasid_lock);
+ return true;
+ }
+
+ key = AMDGPU_VM_FAULT(entry->pasid, addr);
+ r = amdgpu_vm_add_fault(vm->fault_hash, key);
+
+ /* Hash table is full or the fault is already being processed,
+ * ignore further page faults
+ */
+ if (r != 0) {
+ spin_unlock(&adev->vm_manager.pasid_lock);
+ return false;
+ }
+ /* No locking required with single writer and single reader */
+ r = kfifo_put(&vm->faults, key);
+ if (!r) {
+ /* FIFO is full. Ignore it until there is space */
+ amdgpu_vm_clear_fault(vm->fault_hash, key);
+ spin_unlock(&adev->vm_manager.pasid_lock);
+ return false;
+ }
+
+ spin_unlock(&adev->vm_manager.pasid_lock);
+ /* It's the first fault for this address, process it normally */
+ return true;
+}
+
static int gmc_v9_0_process_interrupt(struct amdgpu_device *adev,
struct amdgpu_irq_src *source,
struct amdgpu_iv_entry *entry)
@@ -255,6 +311,9 @@ static int gmc_v9_0_process_interrupt(struct amdgpu_device *adev,
addr = (u64)entry->src_data[0] << 12;
addr |= ((u64)entry->src_data[1] & 0xf) << 44;
+ if (!gmc_v9_0_prescreen_iv(adev, entry, addr))
+ return 1; /* This also prevents sending it to KFD */
+
if (!amdgpu_sriov_vf(adev)) {
status = RREG32(hub->vm_l2_pro_fault_status);
WREG32_P(hub->vm_l2_pro_fault_cntl, 1, ~1);
@@ -902,6 +961,9 @@ static int gmc_v9_0_sw_init(void *handle)
/* This interrupt is VMC page fault.*/
r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_VMC, VMC_1_0__SRCID__VM_FAULT,
&adev->gmc.vm_fault);
+ if (r)
+ return r;
+
r = amdgpu_irq_add_id(adev, SOC15_IH_CLIENTID_UTCL2, UTCL2_1_0__SRCID__FAULT,
&adev->gmc.vm_fault);
@@ -934,7 +996,7 @@ static int gmc_v9_0_sw_init(void *handle)
}
adev->need_swiotlb = drm_get_max_iomem() > ((u64)1 << dma_bits);
- if (adev->asic_type == CHIP_VEGA20) {
+ if (adev->gmc.xgmi.supported) {
r = gfxhub_v1_1_get_xgmi_info(adev);
if (r)
return r;
diff --git a/drivers/gpu/drm/amd/amdgpu/iceland_ih.c b/drivers/gpu/drm/amd/amdgpu/iceland_ih.c
index cf0fc61aebe6..a3984d10b604 100644
--- a/drivers/gpu/drm/amd/amdgpu/iceland_ih.c
+++ b/drivers/gpu/drm/amd/amdgpu/iceland_ih.c
@@ -208,34 +208,6 @@ static u32 iceland_ih_get_wptr(struct amdgpu_device *adev)
}
/**
- * iceland_ih_prescreen_iv - prescreen an interrupt vector
- *
- * @adev: amdgpu_device pointer
- *
- * Returns true if the interrupt vector should be further processed.
- */
-static bool iceland_ih_prescreen_iv(struct amdgpu_device *adev)
-{
- u32 ring_index = adev->irq.ih.rptr >> 2;
- u16 pasid;
-
- switch (le32_to_cpu(adev->irq.ih.ring[ring_index]) & 0xff) {
- case 146:
- case 147:
- pasid = le32_to_cpu(adev->irq.ih.ring[ring_index + 2]) >> 16;
- if (!pasid || amdgpu_vm_pasid_fault_credit(adev, pasid))
- return true;
- break;
- default:
- /* Not a VM fault */
- return true;
- }
-
- adev->irq.ih.rptr += 16;
- return false;
-}
-
-/**
* iceland_ih_decode_iv - decode an interrupt vector
*
* @adev: amdgpu_device pointer
@@ -440,7 +412,6 @@ static const struct amd_ip_funcs iceland_ih_ip_funcs = {
static const struct amdgpu_ih_funcs iceland_ih_funcs = {
.get_wptr = iceland_ih_get_wptr,
- .prescreen_iv = iceland_ih_prescreen_iv,
.decode_iv = iceland_ih_decode_iv,
.set_rptr = iceland_ih_set_rptr
};
diff --git a/drivers/gpu/drm/amd/amdgpu/psp_gfx_if.h b/drivers/gpu/drm/amd/amdgpu/psp_gfx_if.h
index 882bd83a28c4..0de00fbe9233 100644
--- a/drivers/gpu/drm/amd/amdgpu/psp_gfx_if.h
+++ b/drivers/gpu/drm/amd/amdgpu/psp_gfx_if.h
@@ -43,6 +43,8 @@ enum psp_gfx_crtl_cmd_id
GFX_CTRL_CMD_ID_ENABLE_INT = 0x00050000, /* enable PSP-to-Gfx interrupt */
GFX_CTRL_CMD_ID_DISABLE_INT = 0x00060000, /* disable PSP-to-Gfx interrupt */
GFX_CTRL_CMD_ID_MODE1_RST = 0x00070000, /* trigger the Mode 1 reset */
+ GFX_CTRL_CMD_ID_CONSUME_CMD = 0x000A0000, /* send interrupt to psp for updating write pointer of vf */
+ GFX_CTRL_CMD_ID_DESTROY_GPCOM_RING = 0x000C0000, /* destroy GPCOM ring */
GFX_CTRL_CMD_ID_MAX = 0x000F0000, /* max command ID */
};
@@ -89,7 +91,8 @@ enum psp_gfx_cmd_id
GFX_CMD_ID_LOAD_IP_FW = 0x00000006, /* load HW IP FW */
GFX_CMD_ID_DESTROY_TMR = 0x00000007, /* destroy TMR region */
GFX_CMD_ID_SAVE_RESTORE = 0x00000008, /* save/restore HW IP FW */
-
+ GFX_CMD_ID_SETUP_VMR = 0x00000009, /* setup VMR region */
+ GFX_CMD_ID_DESTROY_VMR = 0x0000000A, /* destroy VMR region */
};
diff --git a/drivers/gpu/drm/amd/amdgpu/psp_v11_0.c b/drivers/gpu/drm/amd/amdgpu/psp_v11_0.c
index e5dd052d9e06..6c9a1b748ca7 100644
--- a/drivers/gpu/drm/amd/amdgpu/psp_v11_0.c
+++ b/drivers/gpu/drm/amd/amdgpu/psp_v11_0.c
@@ -171,8 +171,11 @@ static int psp_v11_0_bootloader_load_sysdrv(struct psp_context *psp)
* are already been loaded.
*/
sol_reg = RREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_81);
- if (sol_reg)
+ if (sol_reg) {
+ psp->sos_fw_version = RREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_58);
+ printk("sos fw version = 0x%x.\n", psp->sos_fw_version);
return 0;
+ }
/* Wait for bootloader to signify that is ready having bit 31 of C2PMSG_35 set to 1 */
ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_35),
@@ -296,26 +299,47 @@ static int psp_v11_0_ring_create(struct psp_context *psp,
struct psp_ring *ring = &psp->km_ring;
struct amdgpu_device *adev = psp->adev;
- /* Write low address of the ring to C2PMSG_69 */
- psp_ring_reg = lower_32_bits(ring->ring_mem_mc_addr);
- WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_69, psp_ring_reg);
- /* Write high address of the ring to C2PMSG_70 */
- psp_ring_reg = upper_32_bits(ring->ring_mem_mc_addr);
- WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_70, psp_ring_reg);
- /* Write size of ring to C2PMSG_71 */
- psp_ring_reg = ring->ring_size;
- WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_71, psp_ring_reg);
- /* Write the ring initialization command to C2PMSG_64 */
- psp_ring_reg = ring_type;
- psp_ring_reg = psp_ring_reg << 16;
- WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_64, psp_ring_reg);
-
- /* there might be handshake issue with hardware which needs delay */
- mdelay(20);
-
- /* Wait for response flag (bit 31) in C2PMSG_64 */
- ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_64),
- 0x80000000, 0x8000FFFF, false);
+ if (psp_support_vmr_ring(psp)) {
+ /* Write low address of the ring to C2PMSG_102 */
+ psp_ring_reg = lower_32_bits(ring->ring_mem_mc_addr);
+ WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_102, psp_ring_reg);
+ /* Write high address of the ring to C2PMSG_103 */
+ psp_ring_reg = upper_32_bits(ring->ring_mem_mc_addr);
+ WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_103, psp_ring_reg);
+
+ /* Write the ring initialization command to C2PMSG_101 */
+ WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_101,
+ GFX_CTRL_CMD_ID_INIT_GPCOM_RING);
+
+ /* there might be handshake issue with hardware which needs delay */
+ mdelay(20);
+
+ /* Wait for response flag (bit 31) in C2PMSG_101 */
+ ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_101),
+ 0x80000000, 0x8000FFFF, false);
+
+ } else {
+ /* Write low address of the ring to C2PMSG_69 */
+ psp_ring_reg = lower_32_bits(ring->ring_mem_mc_addr);
+ WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_69, psp_ring_reg);
+ /* Write high address of the ring to C2PMSG_70 */
+ psp_ring_reg = upper_32_bits(ring->ring_mem_mc_addr);
+ WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_70, psp_ring_reg);
+ /* Write size of ring to C2PMSG_71 */
+ psp_ring_reg = ring->ring_size;
+ WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_71, psp_ring_reg);
+ /* Write the ring initialization command to C2PMSG_64 */
+ psp_ring_reg = ring_type;
+ psp_ring_reg = psp_ring_reg << 16;
+ WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_64, psp_ring_reg);
+
+ /* there might be handshake issue with hardware which needs delay */
+ mdelay(20);
+
+ /* Wait for response flag (bit 31) in C2PMSG_64 */
+ ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_64),
+ 0x80000000, 0x8000FFFF, false);
+ }
return ret;
}
@@ -326,15 +350,24 @@ static int psp_v11_0_ring_stop(struct psp_context *psp,
int ret = 0;
struct amdgpu_device *adev = psp->adev;
- /* Write the ring destroy command to C2PMSG_64 */
- WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_64, GFX_CTRL_CMD_ID_DESTROY_RINGS);
+ /* Write the ring destroy command*/
+ if (psp_support_vmr_ring(psp))
+ WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_101,
+ GFX_CTRL_CMD_ID_DESTROY_GPCOM_RING);
+ else
+ WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_64,
+ GFX_CTRL_CMD_ID_DESTROY_RINGS);
/* there might be handshake issue with hardware which needs delay */
mdelay(20);
- /* Wait for response flag (bit 31) in C2PMSG_64 */
- ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_64),
- 0x80000000, 0x80000000, false);
+ /* Wait for response flag (bit 31) */
+ if (psp_support_vmr_ring(psp))
+ ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_101),
+ 0x80000000, 0x80000000, false);
+ else
+ ret = psp_wait_for(psp, SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_64),
+ 0x80000000, 0x80000000, false);
return ret;
}
@@ -373,7 +406,10 @@ static int psp_v11_0_cmd_submit(struct psp_context *psp,
uint32_t rb_frame_size_dw = sizeof(struct psp_gfx_rb_frame) / 4;
/* KM (GPCOM) prepare write pointer */
- psp_write_ptr_reg = RREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_67);
+ if (psp_support_vmr_ring(psp))
+ psp_write_ptr_reg = RREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_102);
+ else
+ psp_write_ptr_reg = RREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_67);
/* Update KM RB frame pointer to new frame */
/* write_frame ptr increments by size of rb_frame in bytes */
@@ -402,7 +438,11 @@ static int psp_v11_0_cmd_submit(struct psp_context *psp,
/* Update the write Pointer in DWORDs */
psp_write_ptr_reg = (psp_write_ptr_reg + rb_frame_size_dw) % ring_size_dw;
- WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_67, psp_write_ptr_reg);
+ if (psp_support_vmr_ring(psp)) {
+ WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_102, psp_write_ptr_reg);
+ WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_101, GFX_CTRL_CMD_ID_CONSUME_CMD);
+ } else
+ WREG32_SOC15(MP0, 0, mmMP0_SMN_C2PMSG_67, psp_write_ptr_reg);
return 0;
}
@@ -547,7 +587,7 @@ static int psp_v11_0_mode1_reset(struct psp_context *psp)
/*send the mode 1 reset command*/
WREG32(offset, GFX_CTRL_CMD_ID_MODE1_RST);
- mdelay(1000);
+ msleep(500);
offset = SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_33);
diff --git a/drivers/gpu/drm/amd/amdgpu/psp_v3_1.c b/drivers/gpu/drm/amd/amdgpu/psp_v3_1.c
index 7efb823dd3b1..7357fd56e614 100644
--- a/drivers/gpu/drm/amd/amdgpu/psp_v3_1.c
+++ b/drivers/gpu/drm/amd/amdgpu/psp_v3_1.c
@@ -592,7 +592,7 @@ static int psp_v3_1_mode1_reset(struct psp_context *psp)
/*send the mode 1 reset command*/
WREG32(offset, GFX_CTRL_CMD_ID_MODE1_RST);
- mdelay(1000);
+ msleep(500);
offset = SOC15_REG_OFFSET(MP0, 0, mmMP0_SMN_C2PMSG_33);
diff --git a/drivers/gpu/drm/amd/amdgpu/si_ih.c b/drivers/gpu/drm/amd/amdgpu/si_ih.c
index b3d7d9f83202..2938fb9f17cc 100644
--- a/drivers/gpu/drm/amd/amdgpu/si_ih.c
+++ b/drivers/gpu/drm/amd/amdgpu/si_ih.c
@@ -118,19 +118,6 @@ static u32 si_ih_get_wptr(struct amdgpu_device *adev)
return (wptr & adev->irq.ih.ptr_mask);
}
-/**
- * si_ih_prescreen_iv - prescreen an interrupt vector
- *
- * @adev: amdgpu_device pointer
- *
- * Returns true if the interrupt vector should be further processed.
- */
-static bool si_ih_prescreen_iv(struct amdgpu_device *adev)
-{
- /* Process all interrupts */
- return true;
-}
-
static void si_ih_decode_iv(struct amdgpu_device *adev,
struct amdgpu_iv_entry *entry)
{
@@ -301,7 +288,6 @@ static const struct amd_ip_funcs si_ih_ip_funcs = {
static const struct amdgpu_ih_funcs si_ih_funcs = {
.get_wptr = si_ih_get_wptr,
- .prescreen_iv = si_ih_prescreen_iv,
.decode_iv = si_ih_decode_iv,
.set_rptr = si_ih_set_rptr
};
diff --git a/drivers/gpu/drm/amd/amdgpu/soc15.c b/drivers/gpu/drm/amd/amdgpu/soc15.c
index 83624e150ca7..8849b74078d6 100644
--- a/drivers/gpu/drm/amd/amdgpu/soc15.c
+++ b/drivers/gpu/drm/amd/amdgpu/soc15.c
@@ -507,6 +507,9 @@ int soc15_set_ip_blocks(struct amdgpu_device *adev)
return -EINVAL;
}
+ if (adev->asic_type == CHIP_VEGA20)
+ adev->gmc.xgmi.supported = true;
+
if (adev->flags & AMD_IS_APU)
adev->nbio_funcs = &nbio_v7_0_funcs;
else if (adev->asic_type == CHIP_VEGA20)
diff --git a/drivers/gpu/drm/amd/amdgpu/tonga_ih.c b/drivers/gpu/drm/amd/amdgpu/tonga_ih.c
index dcdbb4d72472..15da06ddeb75 100644
--- a/drivers/gpu/drm/amd/amdgpu/tonga_ih.c
+++ b/drivers/gpu/drm/amd/amdgpu/tonga_ih.c
@@ -219,34 +219,6 @@ static u32 tonga_ih_get_wptr(struct amdgpu_device *adev)
}
/**
- * tonga_ih_prescreen_iv - prescreen an interrupt vector
- *
- * @adev: amdgpu_device pointer
- *
- * Returns true if the interrupt vector should be further processed.
- */
-static bool tonga_ih_prescreen_iv(struct amdgpu_device *adev)
-{
- u32 ring_index = adev->irq.ih.rptr >> 2;
- u16 pasid;
-
- switch (le32_to_cpu(adev->irq.ih.ring[ring_index]) & 0xff) {
- case 146:
- case 147:
- pasid = le32_to_cpu(adev->irq.ih.ring[ring_index + 2]) >> 16;
- if (!pasid || amdgpu_vm_pasid_fault_credit(adev, pasid))
- return true;
- break;
- default:
- /* Not a VM fault */
- return true;
- }
-
- adev->irq.ih.rptr += 16;
- return false;
-}
-
-/**
* tonga_ih_decode_iv - decode an interrupt vector
*
* @adev: amdgpu_device pointer
@@ -506,7 +478,6 @@ static const struct amd_ip_funcs tonga_ih_ip_funcs = {
static const struct amdgpu_ih_funcs tonga_ih_funcs = {
.get_wptr = tonga_ih_get_wptr,
- .prescreen_iv = tonga_ih_prescreen_iv,
.decode_iv = tonga_ih_decode_iv,
.set_rptr = tonga_ih_set_rptr
};
diff --git a/drivers/gpu/drm/amd/amdgpu/vega10_ih.c b/drivers/gpu/drm/amd/amdgpu/vega10_ih.c
index d84b687240d1..2c250b01a903 100644
--- a/drivers/gpu/drm/amd/amdgpu/vega10_ih.c
+++ b/drivers/gpu/drm/amd/amdgpu/vega10_ih.c
@@ -220,90 +220,6 @@ static u32 vega10_ih_get_wptr(struct amdgpu_device *adev)
}
/**
- * vega10_ih_prescreen_iv - prescreen an interrupt vector
- *
- * @adev: amdgpu_device pointer
- *
- * Returns true if the interrupt vector should be further processed.
- */
-static bool vega10_ih_prescreen_iv(struct amdgpu_device *adev)
-{
- u32 ring_index = adev->irq.ih.rptr >> 2;
- u32 dw0, dw3, dw4, dw5;
- u16 pasid;
- u64 addr, key;
- struct amdgpu_vm *vm;
- int r;
-
- dw0 = le32_to_cpu(adev->irq.ih.ring[ring_index + 0]);
- dw3 = le32_to_cpu(adev->irq.ih.ring[ring_index + 3]);
- dw4 = le32_to_cpu(adev->irq.ih.ring[ring_index + 4]);
- dw5 = le32_to_cpu(adev->irq.ih.ring[ring_index + 5]);
-
- /* Filter retry page faults, let only the first one pass. If
- * there are too many outstanding faults, ignore them until
- * some faults get cleared.
- */
- switch (dw0 & 0xff) {
- case SOC15_IH_CLIENTID_VMC:
- case SOC15_IH_CLIENTID_UTCL2:
- break;
- default:
- /* Not a VM fault */
- return true;
- }
-
- pasid = dw3 & 0xffff;
- /* No PASID, can't identify faulting process */
- if (!pasid)
- return true;
-
- /* Not a retry fault, check fault credit */
- if (!(dw5 & 0x80)) {
- if (!amdgpu_vm_pasid_fault_credit(adev, pasid))
- goto ignore_iv;
- return true;
- }
-
- /* Track retry faults in per-VM fault FIFO. */
- spin_lock(&adev->vm_manager.pasid_lock);
- vm = idr_find(&adev->vm_manager.pasid_idr, pasid);
- addr = ((u64)(dw5 & 0xf) << 44) | ((u64)dw4 << 12);
- key = AMDGPU_VM_FAULT(pasid, addr);
- if (!vm) {
- /* VM not found, process it normally */
- spin_unlock(&adev->vm_manager.pasid_lock);
- return true;
- } else {
- r = amdgpu_vm_add_fault(vm->fault_hash, key);
-
- /* Hash table is full or the fault is already being processed,
- * ignore further page faults
- */
- if (r != 0) {
- spin_unlock(&adev->vm_manager.pasid_lock);
- goto ignore_iv;
- }
- }
- /* No locking required with single writer and single reader */
- r = kfifo_put(&vm->faults, key);
- if (!r) {
- /* FIFO is full. Ignore it until there is space */
- amdgpu_vm_clear_fault(vm->fault_hash, key);
- spin_unlock(&adev->vm_manager.pasid_lock);
- goto ignore_iv;
- }
-
- spin_unlock(&adev->vm_manager.pasid_lock);
- /* It's the first fault for this address, process it normally */
- return true;
-
-ignore_iv:
- adev->irq.ih.rptr += 32;
- return false;
-}
-
-/**
* vega10_ih_decode_iv - decode an interrupt vector
*
* @adev: amdgpu_device pointer
@@ -487,7 +403,6 @@ const struct amd_ip_funcs vega10_ih_ip_funcs = {
static const struct amdgpu_ih_funcs vega10_ih_funcs = {
.get_wptr = vega10_ih_get_wptr,
- .prescreen_iv = vega10_ih_prescreen_iv,
.decode_iv = vega10_ih_decode_iv,
.set_rptr = vega10_ih_set_rptr
};