[PATCH] afs: convert kmap() to kmap_local_page() in mntpt.c

[PATCH] afs: convert kmap() to kmap_local_page() in mntpt.c

[zhaoxiao]

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 mntpt.c is safe everywhere, it should be preferred.

Therefore, replace kmap() with kmap_local_page() in mntpt.c.

Tested in a QEMU/KVM x86_32 VM, 6GB RAM, booting a kernel with
HIGHMEM64GB enabled.

Signed-off-by: zhaoxiao <zhaoxiao@uniontech.com>
---
 fs/afs/mntpt.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/fs/afs/mntpt.c b/fs/afs/mntpt.c
index 97f50e9fd9eb..7e3cacb11780 100644
--- a/fs/afs/mntpt.c
+++ b/fs/afs/mntpt.c
@@ -132,11 +132,11 @@ static int afs_mntpt_set_params(struct fs_context *fc, struct dentry *mntpt)
 		if (IS_ERR(page))
 			return PTR_ERR(page);
 
-		buf = kmap(page);
+		buf = kmap_local_page(page);
 		ret = -EINVAL;
 		if (buf[size - 1] == '.')
 			ret = vfs_parse_fs_string(fc, "source", buf, size - 1);
-		kunmap(page);
+		kunmap_local(buf);
 		put_page(page);
 		if (ret < 0)
 			return ret;
-- 
2.20.1

Re: [PATCH] afs: convert kmap() to kmap_local_page() in mntpt.c

[Fabio M. De Francesco]

On venerdì 12 agosto 2022 10:04:32 CEST zhaoxiao wrote:
> 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 mntpt.c is safe everywhere, it should be preferred.
> 
> Therefore, replace kmap() with kmap_local_page() in mntpt.c.
> 
> Tested in a QEMU/KVM x86_32 VM, 6GB RAM, booting a kernel with
> HIGHMEM64GB enabled.
> 
> Signed-off-by: zhaoxiao <zhaoxiao@uniontech.com>
> ---
>  fs/afs/mntpt.c | 4 ++--
>  1 file changed, 2 insertions(+), 2 deletions(-)
> 
> diff --git a/fs/afs/mntpt.c b/fs/afs/mntpt.c
> index 97f50e9fd9eb..7e3cacb11780 100644
> --- a/fs/afs/mntpt.c
> +++ b/fs/afs/mntpt.c
> @@ -132,11 +132,11 @@ static int afs_mntpt_set_params(struct fs_context *fc,
> struct dentry *mntpt) if (IS_ERR(page))
>  			return PTR_ERR(page);
> 
> -		buf = kmap(page);
> +		buf = kmap_local_page(page);
>  		ret = -EINVAL;
>  		if (buf[size - 1] == '.')
>  			ret = vfs_parse_fs_string(fc, "source", buf, 
size - 1);
> -		kunmap(page);
> +		kunmap_local(buf);
>  		put_page(page);
>  		if (ret < 0)
>  			return ret;
> --
> 2.20.1

I doubt that zhaoxiao is your legal name.
If I'm not wrong, can you please submit a v2 with the name that you use to 
sign documents with legal value? Otherwise, please discard this warning.

Aside from the above, this conversion looks good to me, so feel free to 
forward my tag to next version...

Reviewed-by: Fabio M. De Francesco <fmdefrancesco@gmail.com>

Thanks,

Fabio

P.S.: I'm adding Ira to the list of recipients. Ira and I have been
working through converting kmap() calls to kmap_local_page(). Please let Ira 
or me know if you plan on tacking more conversions so we can mark it
off the list and avoid duplicate works.