* [PATCH] drm/amdkfd: use vma_lookup() instead of find_vma()
@ 2022-10-07 2:26 Deming Wang
2022-10-20 16:47 ` Felix Kuehling
0 siblings, 1 reply; 6+ messages in thread
From: Deming Wang @ 2022-10-07 2:26 UTC (permalink / raw)
To: airlied, daniel, Felix.Kuehling, alexander.deucher,
christian.koenig, Xinhui.Pan
Cc: amd-gfx, dri-devel, linux-kernel, Deming Wang
Using vma_lookup() verifies the start address is contained in the found
vma. This results in easier to read the code.
Signed-off-by: Deming Wang <wangdeming@inspur.com>
---
drivers/gpu/drm/amd/amdkfd/kfd_migrate.c | 8 ++++----
1 file changed, 4 insertions(+), 4 deletions(-)
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_migrate.c b/drivers/gpu/drm/amd/amdkfd/kfd_migrate.c
index 2797029bd500..3599cc931b0a 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_migrate.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_migrate.c
@@ -529,8 +529,8 @@ svm_migrate_ram_to_vram(struct svm_range *prange, uint32_t best_loc,
for (addr = start; addr < end;) {
unsigned long next;
- vma = find_vma(mm, addr);
- if (!vma || addr < vma->vm_start)
+ vma = vma_lookup(mm, addr);
+ if (!vma)
break;
next = min(vma->vm_end, end);
@@ -798,8 +798,8 @@ int svm_migrate_vram_to_ram(struct svm_range *prange, struct mm_struct *mm,
for (addr = start; addr < end;) {
unsigned long next;
- vma = find_vma(mm, addr);
- if (!vma || addr < vma->vm_start) {
+ vma = vma_lookup(mm, addr)
+ if (!vma) {
pr_debug("failed to find vma for prange %p\n", prange);
r = -EFAULT;
break;
--
2.27.0
^ permalink raw reply related [flat|nested] 6+ messages in thread
* Re: [PATCH] drm/amdkfd: use vma_lookup() instead of find_vma()
2022-10-07 2:26 [PATCH] drm/amdkfd: use vma_lookup() instead of find_vma() Deming Wang
@ 2022-10-20 16:47 ` Felix Kuehling
0 siblings, 0 replies; 6+ messages in thread
From: Felix Kuehling @ 2022-10-20 16:47 UTC (permalink / raw)
To: Deming Wang, airlied, daniel, alexander.deucher,
christian.koenig, Xinhui.Pan
Cc: amd-gfx, dri-devel, linux-kernel
Am 2022-10-06 um 22:26 schrieb Deming Wang:
> Using vma_lookup() verifies the start address is contained in the found
> vma. This results in easier to read the code.
>
> Signed-off-by: Deming Wang <wangdeming@inspur.com>
> ---
> drivers/gpu/drm/amd/amdkfd/kfd_migrate.c | 8 ++++----
> 1 file changed, 4 insertions(+), 4 deletions(-)
>
> diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_migrate.c b/drivers/gpu/drm/amd/amdkfd/kfd_migrate.c
> index 2797029bd500..3599cc931b0a 100644
> --- a/drivers/gpu/drm/amd/amdkfd/kfd_migrate.c
> +++ b/drivers/gpu/drm/amd/amdkfd/kfd_migrate.c
> @@ -529,8 +529,8 @@ svm_migrate_ram_to_vram(struct svm_range *prange, uint32_t best_loc,
> for (addr = start; addr < end;) {
> unsigned long next;
>
> - vma = find_vma(mm, addr);
> - if (!vma || addr < vma->vm_start)
> + vma = vma_lookup(mm, addr);
> + if (!vma)
> break;
>
> next = min(vma->vm_end, end);
> @@ -798,8 +798,8 @@ int svm_migrate_vram_to_ram(struct svm_range *prange, struct mm_struct *mm,
> for (addr = start; addr < end;) {
> unsigned long next;
>
> - vma = find_vma(mm, addr);
> - if (!vma || addr < vma->vm_start) {
> + vma = vma_lookup(mm, addr)
There is a semicolon missing here. I will fix this before I submit the
patch.
Please do at least a minimum amount of due diligence before posting
patches. That would include a test to make sure your code compiles.
Regards,
Felix
> + if (!vma) {
> pr_debug("failed to find vma for prange %p\n", prange);
> r = -EFAULT;
> break;
^ permalink raw reply [flat|nested] 6+ messages in thread
* Re: [PATCH] drm/amdkfd: use vma_lookup() instead of find_vma()
2022-10-20 8:20 Deming Wang
@ 2022-10-20 16:50 ` Felix Kuehling
0 siblings, 0 replies; 6+ messages in thread
From: Felix Kuehling @ 2022-10-20 16:50 UTC (permalink / raw)
To: Deming Wang, alexander.deucher, christian.koenig, Xinhui.Pan,
airlied, daniel
Cc: amd-gfx, dri-devel, linux-kernel
Am 2022-10-20 um 04:20 schrieb Deming Wang:
> Using vma_lookup() verifies the start address is contained in the found
> vma. This results in easier to read the code.
>
> Signed-off-by: Deming Wang <wangdeming@inspur.com>
Thank you. This patch is
Reviewed-by: Felix Kuehling <Felix.Kuehling@amd.com>
I squashed it with your earlier patch in kfd_migrate.c and submitted it
to our amd-staging-drm-next branch.
Regards,
Felix
> ---
> drivers/gpu/drm/amd/amdkfd/kfd_svm.c | 17 ++++++++---------
> 1 file changed, 8 insertions(+), 9 deletions(-)
>
> diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_svm.c b/drivers/gpu/drm/amd/amdkfd/kfd_svm.c
> index 64fdf63093a0..0100812478b2 100644
> --- a/drivers/gpu/drm/amd/amdkfd/kfd_svm.c
> +++ b/drivers/gpu/drm/amd/amdkfd/kfd_svm.c
> @@ -1586,8 +1586,8 @@ static int svm_range_validate_and_map(struct mm_struct *mm,
> unsigned long npages;
> bool readonly;
>
> - vma = find_vma(mm, addr);
> - if (!vma || addr < vma->vm_start) {
> + vma = vma_lookup(mm, addr);
> + if (!vma) {
> r = -EFAULT;
> goto unreserve_out;
> }
> @@ -2542,8 +2542,8 @@ svm_range_get_range_boundaries(struct kfd_process *p, int64_t addr,
> struct interval_tree_node *node;
> unsigned long start_limit, end_limit;
>
> - vma = find_vma(p->mm, addr << PAGE_SHIFT);
> - if (!vma || (addr << PAGE_SHIFT) < vma->vm_start) {
> + vma = vma_lookup(p->mm, addr << PAGE_SHIFT);
> + if (!vma) {
> pr_debug("VMA does not exist in address [0x%llx]\n", addr);
> return -EFAULT;
> }
> @@ -2871,8 +2871,8 @@ svm_range_restore_pages(struct amdgpu_device *adev, unsigned int pasid,
> /* __do_munmap removed VMA, return success as we are handling stale
> * retry fault.
> */
> - vma = find_vma(mm, addr << PAGE_SHIFT);
> - if (!vma || (addr << PAGE_SHIFT) < vma->vm_start) {
> + vma = vma_lookup(mm, addr << PAGE_SHIFT);
> + if (!vma) {
> pr_debug("address 0x%llx VMA is removed\n", addr);
> r = 0;
> goto out_unlock_range;
> @@ -3152,9 +3152,8 @@ svm_range_is_valid(struct kfd_process *p, uint64_t start, uint64_t size)
> start <<= PAGE_SHIFT;
> end = start + (size << PAGE_SHIFT);
> do {
> - vma = find_vma(p->mm, start);
> - if (!vma || start < vma->vm_start ||
> - (vma->vm_flags & device_vma))
> + vma = vma_lookup(p->mm, start);
> + if (!vma || (vma->vm_flags & device_vma))
> return -EFAULT;
> start = min(end, vma->vm_end);
> } while (start < end);
^ permalink raw reply [flat|nested] 6+ messages in thread
* [PATCH] drm/amdkfd: use vma_lookup() instead of find_vma()
@ 2022-10-20 8:20 Deming Wang
2022-10-20 16:50 ` Felix Kuehling
0 siblings, 1 reply; 6+ messages in thread
From: Deming Wang @ 2022-10-20 8:20 UTC (permalink / raw)
To: Felix.Kuehling, alexander.deucher, christian.koenig, Xinhui.Pan,
airlied, daniel
Cc: amd-gfx, dri-devel, linux-kernel, Deming Wang
Using vma_lookup() verifies the start address is contained in the found
vma. This results in easier to read the code.
Signed-off-by: Deming Wang <wangdeming@inspur.com>
---
drivers/gpu/drm/amd/amdkfd/kfd_svm.c | 17 ++++++++---------
1 file changed, 8 insertions(+), 9 deletions(-)
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_svm.c b/drivers/gpu/drm/amd/amdkfd/kfd_svm.c
index 64fdf63093a0..0100812478b2 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_svm.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_svm.c
@@ -1586,8 +1586,8 @@ static int svm_range_validate_and_map(struct mm_struct *mm,
unsigned long npages;
bool readonly;
- vma = find_vma(mm, addr);
- if (!vma || addr < vma->vm_start) {
+ vma = vma_lookup(mm, addr);
+ if (!vma) {
r = -EFAULT;
goto unreserve_out;
}
@@ -2542,8 +2542,8 @@ svm_range_get_range_boundaries(struct kfd_process *p, int64_t addr,
struct interval_tree_node *node;
unsigned long start_limit, end_limit;
- vma = find_vma(p->mm, addr << PAGE_SHIFT);
- if (!vma || (addr << PAGE_SHIFT) < vma->vm_start) {
+ vma = vma_lookup(p->mm, addr << PAGE_SHIFT);
+ if (!vma) {
pr_debug("VMA does not exist in address [0x%llx]\n", addr);
return -EFAULT;
}
@@ -2871,8 +2871,8 @@ svm_range_restore_pages(struct amdgpu_device *adev, unsigned int pasid,
/* __do_munmap removed VMA, return success as we are handling stale
* retry fault.
*/
- vma = find_vma(mm, addr << PAGE_SHIFT);
- if (!vma || (addr << PAGE_SHIFT) < vma->vm_start) {
+ vma = vma_lookup(mm, addr << PAGE_SHIFT);
+ if (!vma) {
pr_debug("address 0x%llx VMA is removed\n", addr);
r = 0;
goto out_unlock_range;
@@ -3152,9 +3152,8 @@ svm_range_is_valid(struct kfd_process *p, uint64_t start, uint64_t size)
start <<= PAGE_SHIFT;
end = start + (size << PAGE_SHIFT);
do {
- vma = find_vma(p->mm, start);
- if (!vma || start < vma->vm_start ||
- (vma->vm_flags & device_vma))
+ vma = vma_lookup(p->mm, start);
+ if (!vma || (vma->vm_flags & device_vma))
return -EFAULT;
start = min(end, vma->vm_end);
} while (start < end);
--
2.27.0
^ permalink raw reply related [flat|nested] 6+ messages in thread
* Re: [PATCH] drm/amdkfd: use vma_lookup() instead of find_vma()
2022-10-07 2:48 Deming Wang
@ 2022-10-17 19:35 ` Felix Kuehling
0 siblings, 0 replies; 6+ messages in thread
From: Felix Kuehling @ 2022-10-17 19:35 UTC (permalink / raw)
To: Deming Wang, airlied, daniel, alexander.deucher,
christian.koenig, Xinhui.Pan
Cc: amd-gfx, dri-devel, linux-kernel
On 2022-10-06 22:48, Deming Wang wrote:
> Using vma_lookup() verifies the start address is contained in the found
> vma. This results in easier to read the code.
Thank you for the patches. This and your other patch look good to me.
However, you missed one use of find_vma in svm_range_is_valid. Is that
an oversight or is there a reason why we need to use find_vma there?
If you're going to respin it, you may also squash the two patches into one.
Thanks,
Felix
>
> Signed-off-by: Deming Wang <wangdeming@inspur.com>
> ---
> drivers/gpu/drm/amd/amdkfd/kfd_svm.c | 12 ++++++------
> 1 file changed, 6 insertions(+), 6 deletions(-)
>
> diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_svm.c b/drivers/gpu/drm/amd/amdkfd/kfd_svm.c
> index 64fdf63093a0..cabcc2ca3c23 100644
> --- a/drivers/gpu/drm/amd/amdkfd/kfd_svm.c
> +++ b/drivers/gpu/drm/amd/amdkfd/kfd_svm.c
> @@ -1586,8 +1586,8 @@ static int svm_range_validate_and_map(struct mm_struct *mm,
> unsigned long npages;
> bool readonly;
>
> - vma = find_vma(mm, addr);
> - if (!vma || addr < vma->vm_start) {
> + vma = vma_lookup(mm, addr);
> + if (!vma) {
> r = -EFAULT;
> goto unreserve_out;
> }
> @@ -2542,8 +2542,8 @@ svm_range_get_range_boundaries(struct kfd_process *p, int64_t addr,
> struct interval_tree_node *node;
> unsigned long start_limit, end_limit;
>
> - vma = find_vma(p->mm, addr << PAGE_SHIFT);
> - if (!vma || (addr << PAGE_SHIFT) < vma->vm_start) {
> + vma = vma_lookup(p->mm, addr << PAGE_SHIFT);
> + if (!vma) {
> pr_debug("VMA does not exist in address [0x%llx]\n", addr);
> return -EFAULT;
> }
> @@ -2871,8 +2871,8 @@ svm_range_restore_pages(struct amdgpu_device *adev, unsigned int pasid,
> /* __do_munmap removed VMA, return success as we are handling stale
> * retry fault.
> */
> - vma = find_vma(mm, addr << PAGE_SHIFT);
> - if (!vma || (addr << PAGE_SHIFT) < vma->vm_start) {
> + vma = vma_lookup(mm, addr << PAGE_SHIFT);
> + if (!vma) {
> pr_debug("address 0x%llx VMA is removed\n", addr);
> r = 0;
> goto out_unlock_range;
^ permalink raw reply [flat|nested] 6+ messages in thread
* [PATCH] drm/amdkfd: use vma_lookup() instead of find_vma()
@ 2022-10-07 2:48 Deming Wang
2022-10-17 19:35 ` Felix Kuehling
0 siblings, 1 reply; 6+ messages in thread
From: Deming Wang @ 2022-10-07 2:48 UTC (permalink / raw)
To: airlied, daniel, Felix.Kuehling, alexander.deucher,
christian.koenig, Xinhui.Pan
Cc: amd-gfx, dri-devel, linux-kernel, Deming Wang
Using vma_lookup() verifies the start address is contained in the found
vma. This results in easier to read the code.
Signed-off-by: Deming Wang <wangdeming@inspur.com>
---
drivers/gpu/drm/amd/amdkfd/kfd_svm.c | 12 ++++++------
1 file changed, 6 insertions(+), 6 deletions(-)
diff --git a/drivers/gpu/drm/amd/amdkfd/kfd_svm.c b/drivers/gpu/drm/amd/amdkfd/kfd_svm.c
index 64fdf63093a0..cabcc2ca3c23 100644
--- a/drivers/gpu/drm/amd/amdkfd/kfd_svm.c
+++ b/drivers/gpu/drm/amd/amdkfd/kfd_svm.c
@@ -1586,8 +1586,8 @@ static int svm_range_validate_and_map(struct mm_struct *mm,
unsigned long npages;
bool readonly;
- vma = find_vma(mm, addr);
- if (!vma || addr < vma->vm_start) {
+ vma = vma_lookup(mm, addr);
+ if (!vma) {
r = -EFAULT;
goto unreserve_out;
}
@@ -2542,8 +2542,8 @@ svm_range_get_range_boundaries(struct kfd_process *p, int64_t addr,
struct interval_tree_node *node;
unsigned long start_limit, end_limit;
- vma = find_vma(p->mm, addr << PAGE_SHIFT);
- if (!vma || (addr << PAGE_SHIFT) < vma->vm_start) {
+ vma = vma_lookup(p->mm, addr << PAGE_SHIFT);
+ if (!vma) {
pr_debug("VMA does not exist in address [0x%llx]\n", addr);
return -EFAULT;
}
@@ -2871,8 +2871,8 @@ svm_range_restore_pages(struct amdgpu_device *adev, unsigned int pasid,
/* __do_munmap removed VMA, return success as we are handling stale
* retry fault.
*/
- vma = find_vma(mm, addr << PAGE_SHIFT);
- if (!vma || (addr << PAGE_SHIFT) < vma->vm_start) {
+ vma = vma_lookup(mm, addr << PAGE_SHIFT);
+ if (!vma) {
pr_debug("address 0x%llx VMA is removed\n", addr);
r = 0;
goto out_unlock_range;
--
2.27.0
^ permalink raw reply related [flat|nested] 6+ messages in thread
end of thread, other threads:[~2022-10-20 16:50 UTC | newest]
Thread overview: 6+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2022-10-07 2:26 [PATCH] drm/amdkfd: use vma_lookup() instead of find_vma() Deming Wang
2022-10-20 16:47 ` Felix Kuehling
2022-10-07 2:48 Deming Wang
2022-10-17 19:35 ` Felix Kuehling
2022-10-20 8:20 Deming Wang
2022-10-20 16:50 ` Felix Kuehling
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).