linux-kernel.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
* [PATCH] drm/amd/amdgpu: Replace kmap() with kmap_local_page()
@ 2022-08-12 17:57 Fabio M. De Francesco
  2022-08-14 17:14 ` Christian König
  0 siblings, 1 reply; 2+ messages in thread
From: Fabio M. De Francesco @ 2022-08-12 17:57 UTC (permalink / raw)
  To: Alex Deucher, Christian König, Pan, Xinhui, David Airlie,
	Daniel Vetter, Sumit Semwal, Felix Kuehling, Nirmoy Das,
	Jonathan Kim, Philip Yang, Matthew Auld, Zack Rusin, amd-gfx,
	dri-devel, linux-kernel, linux-media, linaro-mm-sig
  Cc: Fabio M. De Francesco, Ira Weiny

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.

Suggested-by: Ira Weiny <ira.weiny@intel.com>
Signed-off-by: Fabio M. De Francesco <fmdefrancesco@gmail.com>
---
 drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
index 3b4c19412625..c11657b5915f 100644
--- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
+++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
@@ -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


^ permalink raw reply related	[flat|nested] 2+ messages in thread

* Re: [PATCH] drm/amd/amdgpu: Replace kmap() with kmap_local_page()
  2022-08-12 17:57 [PATCH] drm/amd/amdgpu: Replace kmap() with kmap_local_page() Fabio M. De Francesco
@ 2022-08-14 17:14 ` Christian König
  0 siblings, 0 replies; 2+ messages in thread
From: Christian König @ 2022-08-14 17:14 UTC (permalink / raw)
  To: Fabio M. De Francesco, Alex Deucher, Christian König, Pan,
	Xinhui, David Airlie, Daniel Vetter, Sumit Semwal,
	Felix Kuehling, Nirmoy Das, Jonathan Kim, Philip Yang,
	Matthew Auld, Zack Rusin, amd-gfx, dri-devel, linux-kernel,
	linux-media, linaro-mm-sig
  Cc: Ira Weiny

Am 12.08.22 um 19:57 schrieb 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.
>
> Suggested-by: Ira Weiny <ira.weiny@intel.com>
> Signed-off-by: Fabio M. De Francesco <fmdefrancesco@gmail.com>

For those two use cases that looks good to me. Feel free to add an 
Acked-by: Christian König <christian.koenig@amd.com>

Regards,
Christian.

> ---
>   drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c | 8 ++++----
>   1 file changed, 4 insertions(+), 4 deletions(-)
>
> diff --git a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
> index 3b4c19412625..c11657b5915f 100644
> --- a/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
> +++ b/drivers/gpu/drm/amd/amdgpu/amdgpu_ttm.c
> @@ -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;
>   


^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2022-08-14 17:14 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2022-08-12 17:57 [PATCH] drm/amd/amdgpu: Replace kmap() with kmap_local_page() Fabio M. De Francesco
2022-08-14 17:14 ` Christian König

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).