[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-Id: <20250310-dmem-cgroups-v1-9-2984c1bc9312@kernel.org>
Date: Mon, 10 Mar 2025 13:06:15 +0100
From: Maxime Ripard <mripard@...nel.org>
To: Andrew Morton <akpm@...ux-foundation.org>,
Marek Szyprowski <m.szyprowski@...sung.com>,
Robin Murphy <robin.murphy@....com>, Sumit Semwal <sumit.semwal@...aro.org>,
Christian König <christian.koenig@....com>,
Benjamin Gaignard <benjamin.gaignard@...labora.com>,
Brian Starkey <Brian.Starkey@....com>, John Stultz <jstultz@...gle.com>,
"T.J. Mercier" <tjmercier@...gle.com>,
Maarten Lankhorst <maarten.lankhorst@...ux.intel.com>,
Thomas Zimmermann <tzimmermann@...e.de>, David Airlie <airlied@...il.com>,
Simona Vetter <simona@...ll.ch>, Tomasz Figa <tfiga@...omium.org>,
Mauro Carvalho Chehab <mchehab@...nel.org>
Cc: Hans Verkuil <hverkuil@...all.nl>,
Laurent Pinchart <laurent.pinchart+renesas@...asonboard.com>,
linux-mm@...ck.org, linux-kernel@...r.kernel.org, iommu@...ts.linux.dev,
linux-media@...r.kernel.org, dri-devel@...ts.freedesktop.org,
linaro-mm-sig@...ts.linaro.org, Maxime Ripard <mripard@...nel.org>
Subject: [PATCH RFC 09/12] dma-buf: Clear cgroup accounting on release
In order to clean thing up when dma-heaps will allocate and register
buffers in the dev cgroup, let's uncharge a released buffer for any
(optional) cgroup controller.
Signed-off-by: Maxime Ripard <mripard@...nel.org>
---
drivers/dma-buf/dma-buf.c | 7 +++++++
include/linux/dma-buf.h | 5 +++++
2 files changed, 12 insertions(+)
diff --git a/drivers/dma-buf/dma-buf.c b/drivers/dma-buf/dma-buf.c
index 5baa83b855156516a0a766bee0789b122473efb3..a95eef17f193454b018dc8177ddfd434d7b64473 100644
--- a/drivers/dma-buf/dma-buf.c
+++ b/drivers/dma-buf/dma-buf.c
@@ -11,10 +11,11 @@
* refining of this idea.
*/
#include <linux/fs.h>
#include <linux/slab.h>
+#include <linux/cgroup_dmem.h>
#include <linux/dma-buf.h>
#include <linux/dma-fence.h>
#include <linux/dma-fence-unwrap.h>
#include <linux/anon_inodes.h>
#include <linux/export.h>
@@ -97,10 +98,16 @@ static void dma_buf_release(struct dentry *dentry)
* * dmabuf->cb_in/out.active are non-0 despite no pending fence callback
*/
BUG_ON(dmabuf->cb_in.active || dmabuf->cb_out.active);
dma_buf_stats_teardown(dmabuf);
+
+#ifdef CONFIG_CGROUP_DMEM
+ if (dmabuf->cgroup_pool)
+ dmem_cgroup_uncharge(dmabuf->cgroup_pool, dmabuf->size);
+#endif
+
dmabuf->ops->release(dmabuf);
if (dmabuf->resv == (struct dma_resv *)&dmabuf[1])
dma_resv_fini(dmabuf->resv);
diff --git a/include/linux/dma-buf.h b/include/linux/dma-buf.h
index 36216d28d8bdc01a9c9c47e27c392413f7f6c5fb..111ca5a738ae0a816ba1551313dfb0a958720b6c 100644
--- a/include/linux/dma-buf.h
+++ b/include/linux/dma-buf.h
@@ -437,10 +437,15 @@ struct dma_buf {
struct dma_fence_cb cb;
wait_queue_head_t *poll;
__poll_t active;
} cb_in, cb_out;
+
+#ifdef CONFIG_CGROUP_DMEM
+ struct dmem_cgroup_pool_state *cgroup_pool;
+#endif
+
#ifdef CONFIG_DMABUF_SYSFS_STATS
/**
* @sysfs_entry:
*
* For exposing information about this buffer in sysfs. See also
--
2.48.1
Powered by blists - more mailing lists