diff mbox series

[v2] module: Replace kmap() with kmap_local_page()

Message ID 20220720161932.9567-1-fmdefrancesco@gmail.com (mailing list archive)
State New, archived
Headers show
Series [v2] module: Replace kmap() with kmap_local_page() | expand

Commit Message

Fabio M. De Francesco July 20, 2022, 4:19 p.m. UTC
kmap() is being deprecated in favor of kmap_local_page().

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).
Tasks can be preempted and, when scheduled to run again, the kernel
virtual addresses are restored and still valid.

kmap_local_page() is faster than kmap() in kernels with HIGHMEM enabled.

Since the use of kmap_local_page() in module_gzip_decompress() and in
module_xz_decompress() is safe (i.e., it does not break the strict rules
of use), it should be preferred over kmap().

Therefore, replace kmap() with kmap_local_page().

Tested on a QEMU/KVM x86_32 VM with 4GB RAM, booting kernels with
HIGHMEM64GB enabled. Modules compressed with XZ or GZIP decompress
properly.

Cc: Matthew Wilcox <willy@infradead.com>
Suggested-by: Ira Weiny <ira.weiny@intel.com>
Signed-off-by: Fabio M. De Francesco <fmdefrancesco@gmail.com>
---

v1->v2: Add information which says that the pointers returned by
kmap_local_page() are still valid if tasks are preempted and then
rescheduled to run again (thanks to Luis Chamberlain).

 kernel/module/decompress.c | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

Comments

Luis Chamberlain July 20, 2022, 9:29 p.m. UTC | #1
On Wed, Jul 20, 2022 at 06:19:32PM +0200, Fabio M. De Francesco wrote:
> kmap() is being deprecated in favor of kmap_local_page().
> 
> 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).
> Tasks can be preempted and, when scheduled to run again, the kernel
> virtual addresses are restored and still valid.
> 
> kmap_local_page() is faster than kmap() in kernels with HIGHMEM enabled.
> 
> Since the use of kmap_local_page() in module_gzip_decompress() and in
> module_xz_decompress() is safe (i.e., it does not break the strict rules
> of use), it should be preferred over kmap().
> 
> Therefore, replace kmap() with kmap_local_page().
> 
> Tested on a QEMU/KVM x86_32 VM with 4GB RAM, booting kernels with
> HIGHMEM64GB enabled. Modules compressed with XZ or GZIP decompress
> properly.
> 
> Cc: Matthew Wilcox <willy@infradead.com>
> Suggested-by: Ira Weiny <ira.weiny@intel.com>
> Signed-off-by: Fabio M. De Francesco <fmdefrancesco@gmail.com>

Thanks for the enhanced commit log! Queued onto modules-testing, if that
doesn't blow something up I'll move to modules-next afterwards.

  Luis
diff mbox series

Patch

diff --git a/kernel/module/decompress.c b/kernel/module/decompress.c
index 2fc7081dd7c1..4d0bcb3d9e44 100644
--- a/kernel/module/decompress.c
+++ b/kernel/module/decompress.c
@@ -119,10 +119,10 @@  static ssize_t module_gzip_decompress(struct load_info *info,
 			goto out_inflate_end;
 		}
 
-		s.next_out = kmap(page);
+		s.next_out = kmap_local_page(page);
 		s.avail_out = PAGE_SIZE;
 		rc = zlib_inflate(&s, 0);
-		kunmap(page);
+		kunmap_local(s.next_out);
 
 		new_size += PAGE_SIZE - s.avail_out;
 	} while (rc == Z_OK);
@@ -178,11 +178,11 @@  static ssize_t module_xz_decompress(struct load_info *info,
 			goto out;
 		}
 
-		xz_buf.out = kmap(page);
+		xz_buf.out = kmap_local_page(page);
 		xz_buf.out_pos = 0;
 		xz_buf.out_size = PAGE_SIZE;
 		xz_ret = xz_dec_run(xz_dec, &xz_buf);
-		kunmap(page);
+		kunmap_local(xz_buf.out);
 
 		new_size += xz_buf.out_pos;
 	} while (xz_buf.out_pos == PAGE_SIZE && xz_ret == XZ_OK);