Skip to contentSkip to conversation

drm/amd/amdgpu: Replace kmap() with kmap_local_page()

PATCH RESEND Applied Fabio M. De Francesco 2 participants 4 messages Oct 16, 2022 – Nov 2, 2022

Subsystems (2)

v1 of 2 Applied Latest is v2 →

lore
Share & tools
lore
#0 lore raw

PATCH RESEND 1/1 drm/amd/amdgpu: Replace kmap() with kmap_local_page()

kmap() is being deprecated in favor of kmap_local_page().

There are two main problems with kmap(): (1) It comes with an overhead as mapping space is restricted and protected by a global lock for synchronization and (2) it also requires global TLB invalidation when the kmap’s pool wraps and it might block when the mapping space is fully utilized until a slot becomes available.

With kmap_local_page() the mappings are per thread, CPU local, can take page faults, and can be called from any context (including interrupts). It is faster than kmap() in kernels with HIGHMEM enabled. Furthermore, the tasks can be preempted and, when they are scheduled to run again, the kernel virtual addresses are restored and are still valid.

Since its use in amdgpu/amdgpu_ttm.c is safe, it should be preferred.

Therefore, replace kmap() with kmap_local_page() in amdgpu/amdgpu_ttm.c.

---

I'm resending because I suspect that this patch might have been lost. In the meantime I added an "Acked-by" tag from Christian K.. Obviviously, there are no further changes in the code.

 drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)
drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c +4 -4
index 3b4c19412625..c11657b5915f 100644
@@ -2301,9 +2301,9 @@ static ssize_t amdgpu_iomem_read(struct file *f, char __user *buf,
 		if (p->mapping != adev->mman.bdev.dev_mapping)
 			return -EPERM;
 
-		ptr = kmap(p);
+		ptr = kmap_local_page(p);
 		r = copy_to_user(buf, ptr + off, bytes);
-		kunmap(p);
+		kunmap_local(ptr);
 		if (r)
 			return -EFAULT;
 
@@ -2352,9 +2352,9 @@ static ssize_t amdgpu_iomem_write(struct file *f, const char __user *buf,
 		if (p->mapping != adev->mman.bdev.dev_mapping)
 			return -EPERM;
 
-		ptr = kmap(p);
+		ptr = kmap_local_page(p);
 		r = copy_from_user(ptr + off, buf, bytes);
-		kunmap(p);
+		kunmap_local(ptr);
 		if (r)
 			return -EFAULT;
 
Signature · 1 line
2.37.1
Applied
#1 lore raw

Applied. Thanks!

Quoted 62 lines · Fabio M. De Francesco

kmap() is being deprecated in favor of kmap_local_page().

There are two main problems with kmap(): (1) It comes with an overhead as mapping space is restricted and protected by a global lock for synchronization and (2) it also requires global TLB invalidation when the kmap’s pool wraps and it might block when the mapping space is fully utilized until a slot becomes available.

With kmap_local_page() the mappings are per thread, CPU local, can take page faults, and can be called from any context (including interrupts). It is faster than kmap() in kernels with HIGHMEM enabled. Furthermore, the tasks can be preempted and, when they are scheduled to run again, the kernel virtual addresses are restored and are still valid.

Since its use in amdgpu/amdgpu_ttm.c is safe, it should be preferred.

Therefore, replace kmap() with kmap_local_page() in amdgpu/amdgpu_ttm.c.

---

I'm resending because I suspect that this patch might have been lost. In the meantime I added an "Acked-by" tag from Christian K.. Obviviously, there are no further changes in the code.

 drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)
drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c +4 -5
index 3b4c19412625..c11657b5915f 100644
@@ -2301,9 +2301,9 @@ static ssize_t amdgpu_iomem_read(struct file *f, char __user *buf,
                if (p->mapping != adev->mman.bdev.dev_mapping)
                        return -EPERM;
 
-               ptr = kmap(p);
+               ptr = kmap_local_page(p);
                r = copy_to_user(buf, ptr + off, bytes);
-               kunmap(p);
+               kunmap_local(ptr);
                if (r)
                        return -EFAULT;
 
@@ -2352,9 +2352,9 @@ static ssize_t amdgpu_iomem_write(struct file *f, const char __user *buf,
                if (p->mapping != adev->mman.bdev.dev_mapping)
                        return -EPERM;
 
-               ptr = kmap(p);
+               ptr = kmap_local_page(p);
                r = copy_from_user(ptr + off, buf, bytes);
-               kunmap(p);
+               kunmap_local(ptr);
                if (r)
                        return -EFAULT;
 
--

2.37.1

#3 lore raw
Quoted 5 lines · Fabio M. De Francesco
Quoted 2 lines · On lunedì 17 ottobre 2022 18:53:24 CET Alex Deucher wrote:

Applied. Thanks!

Quoted 7 lines

The same report about which I just wrote in my previous email to you is also referring to this patch which later changed status to "Not Applicable".

It points to https://patchwork.linuxtv.org/project/linux-media/patch/
[email protected]/

Can you please let me understand why?

I'm not sure I understand what you are asking. The patch is applied:
https://gitlab.freedesktop.org/agd5f/linux/-/commit/a2c554262d39f81be7422fd8bee2f2fe3779f7f5

Alex

Quoted 7 lines

Thanks,

Fabio