diff options
author | Alex Waterman <alexw@nvidia.com> | 2018-07-02 20:14:27 -0400 |
---|---|---|
committer | mobile promotions <svcmobile_promotions@nvidia.com> | 2018-09-09 20:20:43 -0400 |
commit | 3cf92ec89ba8deac77d726f02d79cba7c0e73e4d (patch) | |
tree | 9b20d30e81ab45d85eddbdf1cdd36901fd5a4ad2 /drivers/gpu/nvgpu/common/mm/vm_area.c | |
parent | 2dd9bb03dd56ca86b0e61b89fab38d38a58ecddf (diff) |
gpu: nvgpu: Fix several issues with the buddy allocator
The issues are:
1. Non-fixed allocs must take into account explicit PTE size
requests. Previously the PTE size was determines from the
allocation size which was incorect. To do this, the PTE size
is now plumbed through all GPU VA allocations. This is what
the new alloc_pte() op does.
2. Fix buddy PTE size assignment. This changes a '<=' into a
'<' in the buddy allocation logic. Effectively this is now
leaving the PTE size for buddy blocks equal to the PDE block
size as 'ANY'.
This prevents a buddy block of PDE size which has yet to be
allocated from having a specific PDE size. Without this its
possible to do a fixed alloc that fails unexpectedly due to
mismatching PDE sizes.
Consider two PDE block sized fixed allocs that are contained
in one buddy twice the size of a PDE block. Let's call these
fixed allocs S and B (small and big). Let's assume that two
fixed allocs are done, each targeting S and B, in that order.
With the current logic the first alloc, when we create the
two buddies S and B, causes both S and B to have a PTE size of
SMALL. Now when the second alloc happens we attempt to find
a buddy B with a PTE size of either BIG or ANY. But we cannot
becasue B already has size SMALL. This casues us to appear
like we have a conflicting fixed alloc despite this not being
the case.
3. Misc cleanups & bug fixes:
- Clean up some MISRA issues
- Delete an extraneous unlock that could have caused a
deadlock.
Bug 200105199
Change-Id: Ib5447ec6705a5a289ac0cf3d5e90c79b5d67582d
Signed-off-by: Alex Waterman <alexw@nvidia.com>
Reviewed-on: https://git-master.nvidia.com/r/1768582
Reviewed-by: mobile promotions <svcmobile_promotions@nvidia.com>
Tested-by: mobile promotions <svcmobile_promotions@nvidia.com>
Diffstat (limited to 'drivers/gpu/nvgpu/common/mm/vm_area.c')
-rw-r--r-- | drivers/gpu/nvgpu/common/mm/vm_area.c | 24 |
1 files changed, 18 insertions, 6 deletions
diff --git a/drivers/gpu/nvgpu/common/mm/vm_area.c b/drivers/gpu/nvgpu/common/mm/vm_area.c index c2c0d569..d096de5d 100644 --- a/drivers/gpu/nvgpu/common/mm/vm_area.c +++ b/drivers/gpu/nvgpu/common/mm/vm_area.c | |||
@@ -99,11 +99,22 @@ int nvgpu_vm_area_alloc(struct vm_gk20a *vm, u32 pages, u32 page_size, | |||
99 | struct nvgpu_allocator *vma; | 99 | struct nvgpu_allocator *vma; |
100 | struct nvgpu_vm_area *vm_area; | 100 | struct nvgpu_vm_area *vm_area; |
101 | u64 vaddr_start = 0; | 101 | u64 vaddr_start = 0; |
102 | u64 our_addr = *addr; | ||
102 | u32 pgsz_idx = GMMU_PAGE_SIZE_SMALL; | 103 | u32 pgsz_idx = GMMU_PAGE_SIZE_SMALL; |
103 | 104 | ||
105 | /* | ||
106 | * If we have a fixed address then use the passed address in *addr. This | ||
107 | * corresponds to the o_a field in the IOCTL. But since we do not | ||
108 | * support specific alignments in the buddy allocator we ignore the | ||
109 | * field if it isn't a fixed offset. | ||
110 | */ | ||
111 | if ((flags & NVGPU_VM_AREA_ALLOC_FIXED_OFFSET) != 0U) { | ||
112 | our_addr = *addr; | ||
113 | } | ||
114 | |||
104 | nvgpu_log(g, gpu_dbg_map, | 115 | nvgpu_log(g, gpu_dbg_map, |
105 | "ADD vm_area: pgsz=%#-8x pages=%-9u addr=%#-14llx flags=0x%x", | 116 | "ADD vm_area: pgsz=%#-8x pages=%-9u a/o=%#-14llx flags=0x%x", |
106 | page_size, pages, *addr, flags); | 117 | page_size, pages, our_addr, flags); |
107 | 118 | ||
108 | for (; pgsz_idx < GMMU_NR_PAGE_SIZES; pgsz_idx++) { | 119 | for (; pgsz_idx < GMMU_NR_PAGE_SIZES; pgsz_idx++) { |
109 | if (vm->gmmu_page_sizes[pgsz_idx] == page_size) { | 120 | if (vm->gmmu_page_sizes[pgsz_idx] == page_size) { |
@@ -133,14 +144,15 @@ int nvgpu_vm_area_alloc(struct vm_gk20a *vm, u32 pages, u32 page_size, | |||
133 | 144 | ||
134 | vma = vm->vma[pgsz_idx]; | 145 | vma = vm->vma[pgsz_idx]; |
135 | if (flags & NVGPU_VM_AREA_ALLOC_FIXED_OFFSET) { | 146 | if (flags & NVGPU_VM_AREA_ALLOC_FIXED_OFFSET) { |
136 | vaddr_start = nvgpu_alloc_fixed(vma, *addr, | 147 | vaddr_start = nvgpu_alloc_fixed(vma, our_addr, |
137 | (u64)pages * | 148 | (u64)pages * |
138 | (u64)page_size, | 149 | (u64)page_size, |
139 | page_size); | 150 | page_size); |
140 | } else { | 151 | } else { |
141 | vaddr_start = nvgpu_alloc(vma, | 152 | vaddr_start = nvgpu_alloc_pte(vma, |
142 | (u64)pages * | 153 | (u64)pages * |
143 | (u64)page_size); | 154 | (u64)page_size, |
155 | page_size); | ||
144 | } | 156 | } |
145 | 157 | ||
146 | if (!vaddr_start) { | 158 | if (!vaddr_start) { |