From: Alex Xie AlexBin.Xie@amd.com
Add comment to describe some variables otherwise.
Signed-off-by: Alex Xie AlexBin.Xie@amd.com Reviewed-by: Christian König christian.koenig@amd.com Signed-off-by: Alex Deucher alexander.deucher@amd.com --- drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c | 32 ++++++++++++++++---------------- 1 file changed, 16 insertions(+), 16 deletions(-)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c index 48d5ad18..22fa4f8 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c @@ -748,7 +748,7 @@ static void amdgpu_vm_frag_ptes(struct amdgpu_device *adev, * @vm: requested vm * @start: start of GPU address range * @end: end of GPU address range - * @dst: destination address to map to + * @dst: destination address to map to, the next dst inside the function * @flags: mapping flags * * Update the page tables in the range @start - @end. @@ -762,43 +762,43 @@ static void amdgpu_vm_update_ptes(struct amdgpu_device *adev, { const uint64_t mask = AMDGPU_VM_PTE_COUNT - 1;
- uint64_t last_pe_start = ~0, last_pe_end = ~0, last_dst = ~0; - uint64_t addr; + uint64_t cur_pe_start = ~0, cur_pe_end = ~0, cur_dst = ~0; + uint64_t addr; /* next GPU address to be updated */
/* walk over the address space and update the page tables */ for (addr = start; addr < end; ) { uint64_t pt_idx = addr >> amdgpu_vm_block_size; struct amdgpu_bo *pt = vm->page_tables[pt_idx].entry.robj; - unsigned nptes; - uint64_t pe_start; + unsigned nptes; /* next number of ptes to be updated */ + uint64_t next_pe_start;
if ((addr & ~mask) == (end & ~mask)) nptes = end - addr; else nptes = AMDGPU_VM_PTE_COUNT - (addr & mask);
- pe_start = amdgpu_bo_gpu_offset(pt); - pe_start += (addr & mask) * 8; + next_pe_start = amdgpu_bo_gpu_offset(pt); + next_pe_start += (addr & mask) * 8;
- if (last_pe_end != pe_start) { + if (cur_pe_end != next_pe_start) {
amdgpu_vm_frag_ptes(adev, vm_update_params, - last_pe_start, last_pe_end, - last_dst, flags); + cur_pe_start, cur_pe_end, + cur_dst, flags);
- last_pe_start = pe_start; - last_pe_end = pe_start + 8 * nptes; - last_dst = dst; + cur_pe_start = next_pe_start; + cur_pe_end = next_pe_start + 8 * nptes; + cur_dst = dst; } else { - last_pe_end += 8 * nptes; + cur_pe_end += 8 * nptes; }
addr += nptes; dst += nptes * AMDGPU_GPU_PAGE_SIZE; }
- amdgpu_vm_frag_ptes(adev, vm_update_params, last_pe_start, - last_pe_end, last_dst, flags); + amdgpu_vm_frag_ptes(adev, vm_update_params, cur_pe_start, + cur_pe_end, cur_dst, flags); }
/**
From: Alex Xie AlexBin.Xie@amd.com
Use == instead of != in the if statement to make code easier understood
Signed-off-by: Alex Xie AlexBin.Xie@amd.com Reviewed-by: Christian König christian.koenig@amd.com Signed-off-by: Alex Deucher alexander.deucher@amd.com --- drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c | 11 +++++++---- 1 file changed, 7 insertions(+), 4 deletions(-)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c index 22fa4f8..0c17348 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c @@ -780,8 +780,13 @@ static void amdgpu_vm_update_ptes(struct amdgpu_device *adev, next_pe_start = amdgpu_bo_gpu_offset(pt); next_pe_start += (addr & mask) * 8;
- if (cur_pe_end != next_pe_start) { - + if (cur_pe_end == next_pe_start) { + /* The next ptb is consecutive to current ptb. + * Don't call amdgpu_vm_frag_ptes now. + * Will update two ptbs together in future. + */ + cur_pe_end += 8 * nptes; + } else { amdgpu_vm_frag_ptes(adev, vm_update_params, cur_pe_start, cur_pe_end, cur_dst, flags); @@ -789,8 +794,6 @@ static void amdgpu_vm_update_ptes(struct amdgpu_device *adev, cur_pe_start = next_pe_start; cur_pe_end = next_pe_start + 8 * nptes; cur_dst = dst; - } else { - cur_pe_end += 8 * nptes; }
addr += nptes;
From: Alex Xie AlexBin.Xie@amd.com
Initialize the variable in a straight-forward way instead of hiding the initialization inside the loop. This can also reduce one function call.
Signed-off-by: Alex Xie AlexBin.Xie@amd.com Reviewed-by: Christian König christian.koenig@amd.com Signed-off-by: Alex Deucher alexander.deucher@amd.com --- drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c | 34 ++++++++++++++++++++++++++++------ 1 file changed, 28 insertions(+), 6 deletions(-)
diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c index 0c17348..311a7c3 100644 --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_vm.c @@ -762,15 +762,36 @@ static void amdgpu_vm_update_ptes(struct amdgpu_device *adev, { const uint64_t mask = AMDGPU_VM_PTE_COUNT - 1;
- uint64_t cur_pe_start = ~0, cur_pe_end = ~0, cur_dst = ~0; + uint64_t cur_pe_start, cur_pe_end, cur_dst; uint64_t addr; /* next GPU address to be updated */ + uint64_t pt_idx; + struct amdgpu_bo *pt; + unsigned nptes; /* next number of ptes to be updated */ + uint64_t next_pe_start; + + /* initialize the variables */ + addr = start; + pt_idx = addr >> amdgpu_vm_block_size; + pt = vm->page_tables[pt_idx].entry.robj; + + if ((addr & ~mask) == (end & ~mask)) + nptes = end - addr; + else + nptes = AMDGPU_VM_PTE_COUNT - (addr & mask); + + cur_pe_start = amdgpu_bo_gpu_offset(pt); + cur_pe_start += (addr & mask) * 8; + cur_pe_end = cur_pe_start + 8 * nptes; + cur_dst = dst; + + /* for next ptb*/ + addr += nptes; + dst += nptes * AMDGPU_GPU_PAGE_SIZE;
/* walk over the address space and update the page tables */ - for (addr = start; addr < end; ) { - uint64_t pt_idx = addr >> amdgpu_vm_block_size; - struct amdgpu_bo *pt = vm->page_tables[pt_idx].entry.robj; - unsigned nptes; /* next number of ptes to be updated */ - uint64_t next_pe_start; + while (addr < end) { + pt_idx = addr >> amdgpu_vm_block_size; + pt = vm->page_tables[pt_idx].entry.robj;
if ((addr & ~mask) == (end & ~mask)) nptes = end - addr; @@ -796,6 +817,7 @@ static void amdgpu_vm_update_ptes(struct amdgpu_device *adev, cur_dst = dst; }
+ /* for next ptb*/ addr += nptes; dst += nptes * AMDGPU_GPU_PAGE_SIZE; }
dri-devel@lists.freedesktop.org
-
Alex Deucher