From: Ahmad Fatoum <a.fatoum@barebox.org>
To: barebox@lists.infradead.org
Cc: Ahmad Fatoum <a.fatoum@barebox.org>
Subject: [PATCH 1/2] nvme: fix PRP pool resize before first use
Date: Thu, 9 Jul 2026 09:22:51 +0200 [thread overview]
Message-ID: <20260709072255.510962-1-a.fatoum@barebox.org> (raw)
It's not safe to call dma_free_coherent on a NULL pointer. Implement
dma_realloc_coherent, which correctly handles reallocating a buffer
without leaking memory and use it to ensure only actually allocated
buffers are freed.
Fixes: aedcb568afe4 ("drivers: Import a very basic NVME implementation from Linux")
Reported-by: Ben Pye # Matrix
Signed-off-by: Ahmad Fatoum <a.fatoum@barebox.org>
---
drivers/dma/map.c | 21 +++++++++++++++++++++
drivers/nvme/host/pci.c | 17 +++++++++++------
include/dma.h | 5 +++++
3 files changed, 37 insertions(+), 6 deletions(-)
diff --git a/drivers/dma/map.c b/drivers/dma/map.c
index 2980d4c2e790..11671f9f663d 100644
--- a/drivers/dma/map.c
+++ b/drivers/dma/map.c
@@ -68,3 +68,24 @@ void dma_unmap_single(struct device *dev, dma_addr_t dma_addr,
debug_dma_unmap(dev, dma_addr, size, dir);
}
EXPORT_SYMBOL(dma_unmap_single);
+
+void *dma_realloc_coherent(struct device *dev, void *oldbuf,
+ size_t oldsize, size_t newsize,
+ dma_addr_t *dma_handle)
+{
+ dma_addr_t newdma = DMA_ERROR_CODE;
+ void *newbuf = ZERO_SIZE_PTR;
+
+ if (newsize) {
+ newbuf = dma_alloc_coherent(dev, newsize, &newdma);
+ if (!newbuf)
+ return NULL;
+ }
+
+ if (oldbuf)
+ dma_free_coherent(dev, oldbuf, *dma_handle, oldsize);
+
+ *dma_handle = newdma;
+ return newbuf;
+}
+EXPORT_SYMBOL(dma_realloc_coherent);
diff --git a/drivers/nvme/host/pci.c b/drivers/nvme/host/pci.c
index 6343af25026b..29a6db76e0f6 100644
--- a/drivers/nvme/host/pci.c
+++ b/drivers/nvme/host/pci.c
@@ -93,13 +93,18 @@ static int nvme_pci_setup_prps(struct nvme_dev *dev,
nprps = DIV_ROUND_UP(length, page_size);
if (nprps > dev->prp_pool_size) {
- dma_free_coherent(DMA_DEVICE_BROKEN,
- dev->prp_pool, dev->prp_dma,
- dev->prp_pool_size * sizeof(u64));
+ __le64 *prp_pool;
+
+ prp_pool = dma_realloc_coherent(DMA_DEVICE_BROKEN,
+ dev->prp_pool,
+ dev->prp_pool_size * sizeof(*prp_pool),
+ nprps * sizeof(*prp_pool),
+ &dev->prp_dma);
+ if (!prp_pool)
+ return -ENOMEM;
+
+ dev->prp_pool = prp_pool;
dev->prp_pool_size = nprps;
- dev->prp_pool = dma_alloc_coherent(DMA_DEVICE_BROKEN,
- nprps * sizeof(u64),
- &dev->prp_dma);
}
prp_list = dev->prp_pool;
diff --git a/include/dma.h b/include/dma.h
index 60937a69aa55..c285b2095a58 100644
--- a/include/dma.h
+++ b/include/dma.h
@@ -161,4 +161,9 @@ static inline bool dma_map_buf_is_aligned(struct device *dev, const void *buf, s
return PTR_IS_ALIGNED(buf, ARCH_DMA_MINALIGN) &&
IS_ALIGNED(size, ARCH_DMA_MINALIGN);
}
+
+void *dma_realloc_coherent(struct device *dev, void *oldbuf,
+ size_t oldsize, size_t newsize,
+ dma_addr_t *dma_handle);
+
#endif /* __DMA_H */
--
2.47.3
next reply other threads:[~2026-07-09 7:24 UTC|newest]
Thread overview: 3+ messages / expand[flat|nested] mbox.gz Atom feed top
2026-07-09 7:22 Ahmad Fatoum [this message]
2026-07-09 7:22 ` [PATCH 2/2] nvme: pass device to dma dma_alloc/free_coherent Ahmad Fatoum
2026-07-10 21:38 ` [PATCH 1/2] nvme: fix PRP pool resize before first use Sascha Hauer
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=20260709072255.510962-1-a.fatoum@barebox.org \
--to=a.fatoum@barebox.org \
--cc=barebox@lists.infradead.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