]> git.ipfire.org Git - thirdparty/linux.git/commit
usb: gadget: f_fs: Fix DMA fence leak
authorPaul Cercueil <paul@crapouillou.net>
Tue, 9 Jun 2026 15:29:05 +0000 (17:29 +0200)
committerGreg Kroah-Hartman <gregkh@linuxfoundation.org>
Thu, 25 Jun 2026 14:11:56 +0000 (15:11 +0100)
commitbaa6b6068a3f2bf2ed525a1cb37975905dadc658
tree29175d65492c6962e5b5fbf4090fbec8930b96b7
parentd092d7edf8faefa3e27b9fc7f0e7904b06c833a2
usb: gadget: f_fs: Fix DMA fence leak

In ffs_dmabuf_transfer(), a ffs_dma_fence object is kmalloc'd, with the
underlying dma_fence later initialized by dma_fence_init(), which sets
its kref counter to 1. Then, dma_resv_add_fence() gets a second
reference, and a pointer to the ffs_dma_fence is passed as the
usb_request's "context" field.

The dma-resv mechanism will manage the second reference, but the first
reference is never properly released; the ffs_dmabuf_cleanup() function
decreases the reference count, but only to balance with the reference
grab in ffs_dmabuf_signal_done().

The code will then slowly leak memory as more ffs_dma_fence objects are
created without being ever freed.

Address this issue by transferring ownership of the fence to the DMA
reservation object, by calling dma_fence_put() right after
dma_resv_add_fence(). The ffs_dma_fence then gets properly discarded
after being signalled.

Fixes: 7b07a2a7ca02 ("usb: gadget: functionfs: Add DMABUF import interface")
Cc: stable <stable@kernel.org>
Signed-off-by: Paul Cercueil <paul@crapouillou.net>
Tested-by: Nuno Sá <nuno.sa@analog.com>
Reviewed-by: Nuno Sá <nuno.sa@analog.com>
Link: https://patch.msgid.link/20260609152905.729328-1-paul@crapouillou.net
Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
drivers/usb/gadget/function/f_fs.c