drm/amdgpu/gfx10: set SH_MEM_CONFIG.INITIAL_INST_PREFETCH

Prefetch mode 0 is not supported and can lead to hangs with certain very
specific code patterns. Set a sound prefetch mode for all VMIDs rather
than forcing all shaders to set the prefetch mode at the beginning.

Reduce code duplication a bit while we're at it. Note that the 64-bit
address mode enum and the retry all enum are both 0, so the only
functional change is in the INITIAL_INST_PREFETCH field.

Signed-off-by: Nicolai Hähnle <nicolai.haehnle@amd.com>
Reviewed-by: Marek Olšák <marek.olsak@amd.com>
Acked-by: Felix Kuehling <Felix.Kuehling@amd.com>
Signed-off-by: Alex Deucher <alexander.deucher@amd.com>
This commit is contained in:
Nicolai Hähnle 2019-07-16 13:11:22 -05:00 committed by Alex Deucher
parent 597292eb73
commit 08473888e0

View File

@ -109,6 +109,13 @@ static const struct soc15_reg_golden golden_settings_gc_10_0_nv10[] =
/* Pending on emulation bring up */ /* Pending on emulation bring up */
}; };
#define DEFAULT_SH_MEM_CONFIG \
((SH_MEM_ADDRESS_MODE_64 << SH_MEM_CONFIG__ADDRESS_MODE__SHIFT) | \
(SH_MEM_ALIGNMENT_MODE_UNALIGNED << SH_MEM_CONFIG__ALIGNMENT_MODE__SHIFT) | \
(SH_MEM_RETRY_MODE_ALL << SH_MEM_CONFIG__RETRY_MODE__SHIFT) | \
(3 << SH_MEM_CONFIG__INITIAL_INST_PREFETCH__SHIFT))
static void gfx_v10_0_set_ring_funcs(struct amdgpu_device *adev); static void gfx_v10_0_set_ring_funcs(struct amdgpu_device *adev);
static void gfx_v10_0_set_irq_funcs(struct amdgpu_device *adev); static void gfx_v10_0_set_irq_funcs(struct amdgpu_device *adev);
static void gfx_v10_0_set_gds_init(struct amdgpu_device *adev); static void gfx_v10_0_set_gds_init(struct amdgpu_device *adev);
@ -1415,7 +1422,6 @@ static u32 gfx_v10_0_init_pa_sc_tile_steering_override(struct amdgpu_device *ade
static void gfx_v10_0_init_compute_vmid(struct amdgpu_device *adev) static void gfx_v10_0_init_compute_vmid(struct amdgpu_device *adev)
{ {
int i; int i;
uint32_t sh_mem_config;
uint32_t sh_mem_bases; uint32_t sh_mem_bases;
/* /*
@ -1426,15 +1432,11 @@ static void gfx_v10_0_init_compute_vmid(struct amdgpu_device *adev)
*/ */
sh_mem_bases = DEFAULT_SH_MEM_BASES | (DEFAULT_SH_MEM_BASES << 16); sh_mem_bases = DEFAULT_SH_MEM_BASES | (DEFAULT_SH_MEM_BASES << 16);
sh_mem_config = SH_MEM_ADDRESS_MODE_64 |
SH_MEM_ALIGNMENT_MODE_UNALIGNED <<
SH_MEM_CONFIG__ALIGNMENT_MODE__SHIFT;
mutex_lock(&adev->srbm_mutex); mutex_lock(&adev->srbm_mutex);
for (i = FIRST_COMPUTE_VMID; i < LAST_COMPUTE_VMID; i++) { for (i = FIRST_COMPUTE_VMID; i < LAST_COMPUTE_VMID; i++) {
nv_grbm_select(adev, 0, 0, 0, i); nv_grbm_select(adev, 0, 0, 0, i);
/* CP and shaders */ /* CP and shaders */
WREG32_SOC15(GC, 0, mmSH_MEM_CONFIG, sh_mem_config); WREG32_SOC15(GC, 0, mmSH_MEM_CONFIG, DEFAULT_SH_MEM_CONFIG);
WREG32_SOC15(GC, 0, mmSH_MEM_BASES, sh_mem_bases); WREG32_SOC15(GC, 0, mmSH_MEM_BASES, sh_mem_bases);
} }
nv_grbm_select(adev, 0, 0, 0, 0); nv_grbm_select(adev, 0, 0, 0, 0);
@ -1527,17 +1529,8 @@ static void gfx_v10_0_constants_init(struct amdgpu_device *adev)
for (i = 0; i < adev->vm_manager.id_mgr[AMDGPU_GFXHUB].num_ids; i++) { for (i = 0; i < adev->vm_manager.id_mgr[AMDGPU_GFXHUB].num_ids; i++) {
nv_grbm_select(adev, 0, 0, 0, i); nv_grbm_select(adev, 0, 0, 0, i);
/* CP and shaders */ /* CP and shaders */
if (i == 0) { WREG32_SOC15(GC, 0, mmSH_MEM_CONFIG, DEFAULT_SH_MEM_CONFIG);
tmp = REG_SET_FIELD(0, SH_MEM_CONFIG, ALIGNMENT_MODE, if (i != 0) {
SH_MEM_ALIGNMENT_MODE_UNALIGNED);
tmp = REG_SET_FIELD(tmp, SH_MEM_CONFIG, RETRY_MODE, 0);
WREG32_SOC15(GC, 0, mmSH_MEM_CONFIG, tmp);
WREG32_SOC15(GC, 0, mmSH_MEM_BASES, 0);
} else {
tmp = REG_SET_FIELD(0, SH_MEM_CONFIG, ALIGNMENT_MODE,
SH_MEM_ALIGNMENT_MODE_UNALIGNED);
tmp = REG_SET_FIELD(tmp, SH_MEM_CONFIG, RETRY_MODE, 0);
WREG32_SOC15(GC, 0, mmSH_MEM_CONFIG, tmp);
tmp = REG_SET_FIELD(0, SH_MEM_BASES, PRIVATE_BASE, tmp = REG_SET_FIELD(0, SH_MEM_BASES, PRIVATE_BASE,
(adev->gmc.private_aperture_start >> 48)); (adev->gmc.private_aperture_start >> 48));
tmp = REG_SET_FIELD(tmp, SH_MEM_BASES, SHARED_BASE, tmp = REG_SET_FIELD(tmp, SH_MEM_BASES, SHARED_BASE,