Thread (56 messages) 56 messages, 5 authors, 2025-09-08

Re: [PATCH v4 14/16] block-dma: migrate to dma_map_phys instead of map_page

From: "Leon Romanovsky" <leon@kernel.org>
Date: 2025-08-19 18:49:29
Also in: linux-block, linux-doc, linux-iommu, linux-mm, linux-nvme, linux-trace-kernel, lkml, rust-for-linux, virtualization, xen-devel


On Tue, Aug 19, 2025, at 20:20, Keith Busch wrote:
On Tue, Aug 19, 2025 at 08:36:58PM +0300, Leon Romanovsky wrote:
quoted
 static bool blk_dma_map_direct(struct request *req, struct device *dma_dev,
 		struct blk_dma_iter *iter, struct phys_vec *vec)
 {
-	iter->addr = dma_map_page(dma_dev, phys_to_page(vec->paddr),
-			offset_in_page(vec->paddr), vec->len, rq_dma_dir(req));
+	iter->addr = dma_map_phys(dma_dev, vec->paddr, vec->len,
+			rq_dma_dir(req), 0);
Looks good.

Reviewed-by: Keith Busch <kbusch@kernel.org>

Just a random thought when I had to double back to check what the "0"
means: many dma_ api's have a default macro without an "attrs" argument,
then an _attrs() version for when you need it. Not sure if you want to
strictly follow that pattern, but merely a suggestion.
At some point,  I had both functions with and without attrs, but Christoph said that it is an artefact and I should introduce one function which accepts attrs but without _attrs in the name.

Thanks 
Keyboard shortcuts
hback out one level
jnext message in thread
kprevious message in thread
ldrill in
Escclose help / fold thread tree
?toggle this help