Netdev List
 help / color / mirror / Atom feed
From: Pavel Begunkov <asml.silence@gmail.com>
To: io-uring@vger.kernel.org
Cc: asml.silence@gmail.com, netdev@vger.kernel.org
Subject: [PATCH review-only 09/17] io_uring/zcrx: split dmabuf unmap and release
Date: Sat, 11 Jul 2026 10:11:32 +0100	[thread overview]
Message-ID: <81b362f58b0f8489c17f03f2d984f499e8ab74c4.1783616211.git.asml.silence@gmail.com> (raw)
In-Reply-To: <cover.1783616211.git.asml.silence@gmail.com>

Until now unmapping and destroying dmabuf were the same thing. To keep
it consistent with non-dmabuf, split it into two separate helpers. Unmap
destroys mappings and attachements as it should, and release only
putting down the dmabuf fd reference.

Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
---
 io_uring/zcrx.c | 17 +++++++++++------
 1 file changed, 11 insertions(+), 6 deletions(-)

diff --git a/io_uring/zcrx.c b/io_uring/zcrx.c
index 79099a78f8cd..86e8046e98c4 100644
--- a/io_uring/zcrx.c
+++ b/io_uring/zcrx.c
@@ -121,21 +121,25 @@ static int io_populate_area_dma(struct io_zcrx_ifq *ifq,
 	return 0;
 }
 
-static void io_release_dmabuf(struct io_zcrx_mem *mem)
+static void io_unmap_dmabuf(struct io_zcrx_mem *mem)
 {
 	if (!IS_ENABLED(CONFIG_DMA_SHARED_BUFFER))
 		return;
-
 	if (mem->sgt)
 		dma_buf_unmap_attachment_unlocked(mem->attach, mem->sgt,
 						  DMA_FROM_DEVICE);
 	if (mem->attach)
 		dma_buf_detach(mem->dmabuf, mem->attach);
-	if (mem->dmabuf)
-		dma_buf_put(mem->dmabuf);
-
 	mem->sgt = NULL;
 	mem->attach = NULL;
+}
+
+static void io_release_dmabuf(struct io_zcrx_mem *mem)
+{
+	if (!IS_ENABLED(CONFIG_DMA_SHARED_BUFFER))
+		return;
+	if (mem->dmabuf)
+		dma_buf_put(mem->dmabuf);
 	mem->dmabuf = NULL;
 }
 
@@ -190,6 +194,7 @@ static int io_import_dmabuf(struct io_zcrx_ifq *ifq,
 	mem->size = len;
 	return 0;
 err:
+	io_unmap_dmabuf(mem);
 	io_release_dmabuf(mem);
 	return ret;
 }
@@ -317,7 +322,7 @@ static void io_zcrx_unmap_area(struct io_zcrx_ifq *ifq,
 	}
 
 	if (area->mem.is_dmabuf) {
-		io_release_dmabuf(&area->mem);
+		io_unmap_dmabuf(&area->mem);
 	} else {
 		dma_unmap_sgtable(ifq->dev, &area->mem.page_sg_table,
 				  DMA_FROM_DEVICE, IO_DMA_ATTR);
-- 
2.54.0


WARNING: multiple messages have this Message-ID (diff)
From: Pavel Begunkov <asml.silence@gmail.com>
To: io-uring@vger.kernel.org
Cc: asml.silence@gmail.com, netdev@vger.kernel.org
Subject: [PATCH review-only 09/17] io_uring/zcrx: split dmabuf unmap and release
Date: Sat, 11 Jul 2026 11:40:02 +0100	[thread overview]
Message-ID: <81b362f58b0f8489c17f03f2d984f499e8ab74c4.1783616211.git.asml.silence@gmail.com> (raw)
Message-ID: <20260711104002.UCYhKy95e-2rm_gM-6uZPuI7Ird7Q10FbRt3CNo7K58@z> (raw)
In-Reply-To: <cover.1783616211.git.asml.silence@gmail.com>

Until now unmapping and destroying dmabuf were the same thing. To keep
it consistent with non-dmabuf, split it into two separate helpers. Unmap
destroys mappings and attachements as it should, and release only
putting down the dmabuf fd reference.

Signed-off-by: Pavel Begunkov <asml.silence@gmail.com>
---
 io_uring/zcrx.c | 17 +++++++++++------
 1 file changed, 11 insertions(+), 6 deletions(-)

diff --git a/io_uring/zcrx.c b/io_uring/zcrx.c
index 79099a78f8cd..86e8046e98c4 100644
--- a/io_uring/zcrx.c
+++ b/io_uring/zcrx.c
@@ -121,21 +121,25 @@ static int io_populate_area_dma(struct io_zcrx_ifq *ifq,
 	return 0;
 }
 
-static void io_release_dmabuf(struct io_zcrx_mem *mem)
+static void io_unmap_dmabuf(struct io_zcrx_mem *mem)
 {
 	if (!IS_ENABLED(CONFIG_DMA_SHARED_BUFFER))
 		return;
-
 	if (mem->sgt)
 		dma_buf_unmap_attachment_unlocked(mem->attach, mem->sgt,
 						  DMA_FROM_DEVICE);
 	if (mem->attach)
 		dma_buf_detach(mem->dmabuf, mem->attach);
-	if (mem->dmabuf)
-		dma_buf_put(mem->dmabuf);
-
 	mem->sgt = NULL;
 	mem->attach = NULL;
+}
+
+static void io_release_dmabuf(struct io_zcrx_mem *mem)
+{
+	if (!IS_ENABLED(CONFIG_DMA_SHARED_BUFFER))
+		return;
+	if (mem->dmabuf)
+		dma_buf_put(mem->dmabuf);
 	mem->dmabuf = NULL;
 }
 
@@ -190,6 +194,7 @@ static int io_import_dmabuf(struct io_zcrx_ifq *ifq,
 	mem->size = len;
 	return 0;
 err:
+	io_unmap_dmabuf(mem);
 	io_release_dmabuf(mem);
 	return ret;
 }
@@ -317,7 +322,7 @@ static void io_zcrx_unmap_area(struct io_zcrx_ifq *ifq,
 	}
 
 	if (area->mem.is_dmabuf) {
-		io_release_dmabuf(&area->mem);
+		io_unmap_dmabuf(&area->mem);
 	} else {
 		dma_unmap_sgtable(ifq->dev, &area->mem.page_sg_table,
 				  DMA_FROM_DEVICE, IO_DMA_ATTR);
-- 
2.54.0


  parent reply	other threads:[~2026-07-11  9:12 UTC|newest]

Thread overview: 36+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-07-11  9:11 [PATCH review-only 00/17] zcrx RQ improvements and dynamic memory provisioning Pavel Begunkov
2026-07-11  9:11 ` [PATCH review-only 01/17] io_uring/zcrx: scale refilling with large pages Pavel Begunkov
2026-07-11 10:39   ` Pavel Begunkov
2026-07-11  9:11 ` [PATCH review-only 02/17] io_uring/zcrx: move RQ head/tail to separate cache lines Pavel Begunkov
2026-07-11 10:39   ` Pavel Begunkov
2026-07-11  9:11 ` [PATCH review-only 03/17] io_uring/zcrx: add RQ iterator Pavel Begunkov
2026-07-11 10:39   ` Pavel Begunkov
2026-07-11  9:11 ` [PATCH review-only 04/17] io_uring/zcrx: cache RQ tail Pavel Begunkov
2026-07-11 10:39   ` Pavel Begunkov
2026-07-11  9:11 ` [PATCH review-only 05/17] io_uring/zcrx: coalesce same-niov RQEs on refill Pavel Begunkov
2026-07-11 10:39   ` Pavel Begunkov
2026-07-11  9:11 ` [PATCH review-only 06/17] io_uring/zcrx: constify area_reg on import Pavel Begunkov
2026-07-11 10:39   ` Pavel Begunkov
2026-07-11  9:11 ` [PATCH review-only 07/17] io_uring/zcrx: add helper for deriving area token Pavel Begunkov
2026-07-11 10:40   ` Pavel Begunkov
2026-07-11  9:11 ` [PATCH review-only 08/17] io_uring/zcrx: don't pass ifq_reg to area creation Pavel Begunkov
2026-07-11 10:40   ` Pavel Begunkov
2026-07-11  9:11 ` Pavel Begunkov [this message]
2026-07-11 10:40   ` [PATCH review-only 09/17] io_uring/zcrx: split dmabuf unmap and release Pavel Begunkov
2026-07-11  9:11 ` [PATCH review-only 10/17] io_uring/zcrx: unmap under netdev lock Pavel Begunkov
2026-07-11 10:40   ` Pavel Begunkov
2026-07-11  9:11 ` [PATCH review-only 11/17] io_uring/zcrx: split append out of area creation Pavel Begunkov
2026-07-11 10:40   ` Pavel Begunkov
2026-07-11  9:11 ` [PATCH review-only 12/17] io_uring/zcrx: move freelist lock to struct zcrx Pavel Begunkov
2026-07-11 10:40   ` Pavel Begunkov
2026-07-11  9:11 ` [PATCH review-only 13/17] io_uring/zcrx: array of areas Pavel Begunkov
2026-07-11 10:40   ` Pavel Begunkov
2026-07-11  9:11 ` [PATCH review-only 14/17] io_uring/zcrx: pass area_id to __zcrx_create_area() Pavel Begunkov
2026-07-11 10:40   ` Pavel Begunkov
2026-07-11  9:11 ` [PATCH review-only 15/17] io_uring/zcrx: add dynamic area creation Pavel Begunkov
2026-07-11 10:40   ` Pavel Begunkov
2026-07-11  9:11 ` [PATCH review-only 16/17] io_urint/zcrx: narrow var scope in io_zcrx_recv_skb() Pavel Begunkov
2026-07-11 10:40   ` Pavel Begunkov
2026-07-11 10:39 ` [PATCH review-only 00/17] zcrx RQ improvements and dynamic memory provisioning Pavel Begunkov
2026-07-11 10:39 ` [PATCH RESEND " Pavel Begunkov
2026-07-11 10:40 ` [PATCH review-only 17/17] io_uring/zcrx: don't reload skb_shinfo Pavel Begunkov

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=81b362f58b0f8489c17f03f2d984f499e8ab74c4.1783616211.git.asml.silence@gmail.com \
    --to=asml.silence@gmail.com \
    --cc=io-uring@vger.kernel.org \
    --cc=netdev@vger.kernel.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox