| Message ID | 20230103182556.29080-1-fmdefrancesco@gmail.com |
|---|---|
| State | New |
| Headers | show |
| Series | scsi: ipr: Replace kmap() with kmap_local_page() | expand |
Fabio, > Therefore, replace kmap() with kmap_local_page() in > ipr_copy_ucode_buffer() and, instead of open-coding local mappings + > memcpy() + local un-mappings, use the better suited memcpy_to_page() > helper. Applied to 6.3/scsi-staging, thanks!
On Tue, 03 Jan 2023 19:25:56 +0100, Fabio M. De Francesco wrote: > The use of kmap() is deprecated in favor of kmap_local_page(). > > There are two main problems with kmap(): (1) It comes with an overhead as > the 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. > > [...] Applied to 6.3/scsi-queue, thanks! [1/1] scsi: ipr: Replace kmap() with kmap_local_page() https://git.kernel.org/mkp/scsi/c/a3e2e248fd77
diff --git a/drivers/scsi/ipr.c b/drivers/scsi/ipr.c index 2022ffb45041..7dff517a0858 100644 --- a/drivers/scsi/ipr.c +++ b/drivers/scsi/ipr.c @@ -3912,7 +3912,6 @@ static int ipr_copy_ucode_buffer(struct ipr_sglist *sglist, { int bsize_elem, i, result = 0; struct scatterlist *sg; - void *kaddr; /* Determine the actual number of bytes per element */ bsize_elem = PAGE_SIZE * (1 << sglist->order); @@ -3923,9 +3922,7 @@ static int ipr_copy_ucode_buffer(struct ipr_sglist *sglist, buffer += bsize_elem) { struct page *page = sg_page(sg); - kaddr = kmap(page); - memcpy(kaddr, buffer, bsize_elem); - kunmap(page); + memcpy_to_page(page, 0, buffer, bsize_elem); sg->length = bsize_elem; @@ -3938,9 +3935,7 @@ static int ipr_copy_ucode_buffer(struct ipr_sglist *sglist, if (len % bsize_elem) { struct page *page = sg_page(sg); - kaddr = kmap(page); - memcpy(kaddr, buffer, len % bsize_elem); - kunmap(page); + memcpy_to_page(page, 0, buffer, len % bsize_elem); sg->length = len % bsize_elem; }
The use of kmap() is deprecated in favor of kmap_local_page(). There are two main problems with kmap(): (1) It comes with an overhead as the 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 still valid. Therefore, replace kmap() with kmap_local_page() in ipr_copy_ucode_buffer() and, instead of open-coding local mappings + memcpy() + local un-mappings, use the better suited memcpy_to_page() helper. Suggested-by: Ira Weiny <ira.weiny@intel.com> Signed-off-by: Fabio M. De Francesco <fmdefrancesco@gmail.com> --- drivers/scsi/ipr.c | 9 ++------- 1 file changed, 2 insertions(+), 7 deletions(-)