Eric Lee / smarc-ti-linux-kernel

Download as
Browse Code ยป

Commit 0a4ae727d6aa459247b027387edb6ff99f657792

Authored by Dave Airlie
2 parents ba6f582606 91b0275c0e
Exists in ti-lsk-linux-4.1.y and in 12 other branches dlt-processor-sdk-linux-03.00.00.04, processor-sdk-linux-02.00.01, smarc-ti-linux-3.15.y, smarc-ti-lsk-linux-4.1.y, smarct3x-processor-sdk-04.01.00.06, smarct3x-processor-sdk-linux-02.00.01, smarct3x-processor-sdk-linux-03.00.00.04, smarct4x-800-processor-sdk-linux-02.00.01, smarct4x-processor-sdk-04.01.00.06, smarct4x-processor-sdk-linux-02.00.01, smarct4x-processor-sdk-linux-03.00.00.04, ti-linux-3.15.y

Merge branch 'drm-fixes-3.15' of git://people.freedesktop.org/~deathsimple/linux into drm-fixes 

The first one is a one liner fixing a stupid typo in the VM handling code and is only relevant if play with one of the VM defines.

The other two switches CIK to use the CPDMA instead of the SDMA for buffer moves, as it turned out the SDMA is still sometimes not 100% reliable.

* 'drm-fixes-3.15' of git://people.freedesktop.org/~deathsimple/linux:
  drm/radeon: use the CP DMA on CIK
  drm/radeon: sync page table updates
  drm/radeon: fix vm buffer size estimation

Showing 2 changed files Side-by-side Diff

drivers/gpu/drm/radeon/radeon_asic.c
Diff comments   View file @ 0a4ae72
... ... @@ -2049,8 +2049,8 @@
2049 2049 .blit_ring_index = RADEON_RING_TYPE_GFX_INDEX,
2050 2050 .dma = &cik_copy_dma,
2051 2051 .dma_ring_index = R600_RING_TYPE_DMA_INDEX,
2052   - .copy = &cik_copy_dma,
2053   - .copy_ring_index = R600_RING_TYPE_DMA_INDEX,
  2052 + .copy = &cik_copy_cpdma,
  2053 + .copy_ring_index = RADEON_RING_TYPE_GFX_INDEX,
2054 2054 },
2055 2055 .surface = {
2056 2056 .set_reg = r600_set_surface_reg,
drivers/gpu/drm/radeon/radeon_vm.c
Diff comments   View file @ 0a4ae72
... ... @@ -132,7 +132,7 @@
132 132 struct radeon_cs_reloc *list;
133 133 unsigned i, idx;
134 134  
135   - list = kmalloc_array(vm->max_pde_used + 1,
  135 + list = kmalloc_array(vm->max_pde_used + 2,
136 136 sizeof(struct radeon_cs_reloc), GFP_KERNEL);
137 137 if (!list)
138 138 return NULL;
... ... @@ -585,7 +585,8 @@
585 585 {
586 586 static const uint32_t incr = RADEON_VM_PTE_COUNT * 8;
587 587  
588   - uint64_t pd_addr = radeon_bo_gpu_offset(vm->page_directory);
  588 + struct radeon_bo *pd = vm->page_directory;
  589 + uint64_t pd_addr = radeon_bo_gpu_offset(pd);
589 590 uint64_t last_pde = ~0, last_pt = ~0;
590 591 unsigned count = 0, pt_idx, ndw;
591 592 struct radeon_ib ib;
... ... @@ -642,6 +643,7 @@
642 643 incr, R600_PTE_VALID);
643 644  
644 645 if (ib.length_dw != 0) {
  646 + radeon_semaphore_sync_to(ib.semaphore, pd->tbo.sync_obj);
645 647 radeon_semaphore_sync_to(ib.semaphore, vm->last_id_use);
646 648 r = radeon_ib_schedule(rdev, &ib, NULL);
647 649 if (r) {
648 650  
649 651  
... ... @@ -689,15 +691,18 @@
689 691 /* walk over the address space and update the page tables */
690 692 for (addr = start; addr < end; ) {
691 693 uint64_t pt_idx = addr >> RADEON_VM_BLOCK_SIZE;
  694 + struct radeon_bo *pt = vm->page_tables[pt_idx].bo;
692 695 unsigned nptes;
693 696 uint64_t pte;
694 697  
  698 + radeon_semaphore_sync_to(ib->semaphore, pt->tbo.sync_obj);
  699 +
695 700 if ((addr & ~mask) == (end & ~mask))
696 701 nptes = end - addr;
697 702 else
698 703 nptes = RADEON_VM_PTE_COUNT - (addr & mask);
699 704  
700   - pte = radeon_bo_gpu_offset(vm->page_tables[pt_idx].bo);
  705 + pte = radeon_bo_gpu_offset(pt);
701 706 pte += (addr & mask) * 8;
702 707  
703 708 if ((last_pte + 8 * count) != pte) {