From: Leon Romanovsky <leon@kernel.org>
To: Marek Szyprowski <m.szyprowski@samsung.com>,
Robin Murphy <robin.murphy@arm.com>, Jon Mason <jdmason@kudzu.us>,
Dave Jiang <dave.jiang@intel.com>,
Allen Hubbe <allenbh@gmail.com>
Cc: iommu@lists.linux.dev, linux-kernel@vger.kernel.org, ntb@lists.linux.dev
Subject: [PATCH v2 6/6] dma-debug: Ensure mappings are created and released with matching attributes
Date: Fri, 1 May 2026 09:35:10 +0300 [thread overview]
Message-ID: <20260501-dma-attrs-debug-v2-6-8dbac75cd501@nvidia.com> (raw)
In-Reply-To: <20260501-dma-attrs-debug-v2-0-8dbac75cd501@nvidia.com>
From: Leon Romanovsky <leonro@nvidia.com>
The DMA API expects that callers use the same attributes when mapping
and unmapping. Add tracking to verify this and catch mismatches.
Signed-off-by: Leon Romanovsky <leonro@nvidia.com>
---
kernel/dma/debug.c | 23 +++++++++++++++++++++++
1 file changed, 23 insertions(+)
diff --git a/kernel/dma/debug.c b/kernel/dma/debug.c
index 3dfed51c3d9aa..c38efc1ac8d6c 100644
--- a/kernel/dma/debug.c
+++ b/kernel/dma/debug.c
@@ -1074,6 +1074,29 @@ static void check_unmap(struct dma_debug_entry *ref)
type2name[entry->type]);
}
+ /*
+ * This may be no bug in reality - but DMA API still expects
+ * that entry is unmapped with same attributes as it was mapped.
+ *
+ * DMA_ATTR_UNMAP_VALID lists the attributes that must be identical
+ * between map and unmap. Any attribute outside this set (e.g.
+ * DMA_ATTR_NO_WARN, DMA_ATTR_SKIP_CPU_SYNC) is allowed to differ.
+ */
+#define DMA_ATTR_UNMAP_VALID \
+ (DMA_ATTR_NO_KERNEL_MAPPING | DMA_ATTR_FORCE_CONTIGUOUS | \
+ DMA_ATTR_MMIO | DMA_ATTR_REQUIRE_COHERENT | DMA_ATTR_PRIVILEGED | \
+ DMA_ATTR_CC_SHARED)
+ if ((ref->attrs & DMA_ATTR_UNMAP_VALID) !=
+ (entry->attrs & DMA_ATTR_UNMAP_VALID)) {
+ err_printk(ref->dev, entry,
+ "device driver frees "
+ "DMA memory with different attributes "
+ "[device address=0x%016llx] [size=%llu bytes] "
+ "[mapped with 0x%lx] [unmapped with 0x%lx]\n",
+ ref->dev_addr, ref->size, entry->attrs, ref->attrs);
+ }
+#undef DMA_ATTR_UNMAP_VALID
+
hash_bucket_del(entry);
put_hash_bucket(bucket, flags);
--
2.53.0
prev parent reply other threads:[~2026-05-01 6:35 UTC|newest]
Thread overview: 9+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-05-01 6:35 [PATCH v2 0/6] Add DMA attributes tracking Leon Romanovsky
2026-05-01 6:35 ` [PATCH v2 1/6] ntb: Store original DMA address for future release Leon Romanovsky
2026-05-01 17:00 ` Dave Jiang
2026-05-01 6:35 ` [PATCH v2 2/6] ntb: Use consistent DMA attributes when freeing DMA mappings Leon Romanovsky
2026-05-01 17:00 ` Dave Jiang
2026-05-01 6:35 ` [PATCH v2 3/6] dma-debug: Remove unused DMA attribute parameter Leon Romanovsky
2026-05-01 6:35 ` [PATCH v2 4/6] dma-debug: Record DMA attributes in debug entry Leon Romanovsky
2026-05-01 6:35 ` [PATCH v2 5/6] dma-debug: Feed DMA attribute for unmapping flows too Leon Romanovsky
2026-05-01 6:35 ` Leon Romanovsky [this message]
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=20260501-dma-attrs-debug-v2-6-8dbac75cd501@nvidia.com \
--to=leon@kernel.org \
--cc=allenbh@gmail.com \
--cc=dave.jiang@intel.com \
--cc=iommu@lists.linux.dev \
--cc=jdmason@kudzu.us \
--cc=linux-kernel@vger.kernel.org \
--cc=m.szyprowski@samsung.com \
--cc=ntb@lists.linux.dev \
--cc=robin.murphy@arm.com \
/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