nvmet_pci_epf_create_cq() calls nvmet_cq_create(), which takes a
reference on the controller and installs the completion queue. If the
subsequent PCI address-space mapping fails or returns a too-small partial
mapping, the function jumps to err_internal / err_unmap_queue without
calling nvmet_cq_put(). The matching put in nvmet_pci_epf_delete_cq() is
gated on NVMET_PCI_EPF_Q_LIVE, which is only set after the mapping
succeeds, so teardown never releases these references. A remote PCI host
that drives Create IO CQ commands with a failing PRP1/pci_addr therefore
leaks the CQ and a controller reference on each attempt.

Drop the CQ reference on the mapping-failure paths. The err_internal and
err_unmap_queue labels are only reachable after nvmet_cq_create() has
succeeded, so this pairs the create/put correctly.

Fixes: 0faa0fe6f90e ("nvmet: New NVMe PCI endpoint function target driver")
Cc: stable@vger.kernel.org
Assisted-by: Claude:claude-opus-4-8
Signed-off-by: Yifei Gao <gyf161023@gmail.com>
---
 drivers/nvme/target/pci-epf.c | 1 +
 1 file changed, 1 insertion(+)

diff --git a/drivers/nvme/target/pci-epf.c b/drivers/nvme/target/pci-epf.c
index 4e9db96ebfec..794e88d1d9a5 100644
--- a/drivers/nvme/target/pci-epf.c
+++ b/drivers/nvme/target/pci-epf.c
@@ -1339,6 +1339,7 @@ static u16 nvmet_pci_epf_create_cq(struct nvmet_ctrl *tctrl,
 	nvmet_pci_epf_mem_unmap(ctrl->nvme_epf, &cq->pci_map);
 err_internal:
 	status = NVME_SC_INTERNAL | NVME_STATUS_DNR;
+	nvmet_cq_put(&cq->nvme_cq);
 err:
 	if (test_and_clear_bit(NVMET_PCI_EPF_Q_IRQ_ENABLED, &cq->flags))
 		nvmet_pci_epf_remove_irq_vector(ctrl, cq->vector);
-- 
2.43.0

On Tue, Aug 04, 2026 at 09:36:25PM +0000, Yifei Gao wrote:
> nvmet_pci_epf_create_cq() calls nvmet_cq_create(), which takes a
> reference on the controller and installs the completion queue. If the
> subsequent PCI address-space mapping fails or returns a too-small partial
> mapping, the function jumps to err_internal / err_unmap_queue without
> calling nvmet_cq_put(). The matching put in nvmet_pci_epf_delete_cq() is
> gated on NVMET_PCI_EPF_Q_LIVE, which is only set after the mapping
> succeeds, so teardown never releases these references. A remote PCI host
> that drives Create IO CQ commands with a failing PRP1/pci_addr therefore
> leaks the CQ and a controller reference on each attempt.
> 
> Drop the CQ reference on the mapping-failure paths. The err_internal and
> err_unmap_queue labels are only reachable after nvmet_cq_create() has
> succeeded, so this pairs the create/put correctly.

Thanks, applied to nvme-7.3.

On 8/5/26 06:36, Yifei Gao wrote:
> nvmet_pci_epf_create_cq() calls nvmet_cq_create(), which takes a
> reference on the controller and installs the completion queue. If the
> subsequent PCI address-space mapping fails or returns a too-small partial
> mapping, the function jumps to err_internal / err_unmap_queue without
> calling nvmet_cq_put(). The matching put in nvmet_pci_epf_delete_cq() is
> gated on NVMET_PCI_EPF_Q_LIVE, which is only set after the mapping
> succeeds, so teardown never releases these references. A remote PCI host
> that drives Create IO CQ commands with a failing PRP1/pci_addr therefore
> leaks the CQ and a controller reference on each attempt.
> 
> Drop the CQ reference on the mapping-failure paths. The err_internal and
> err_unmap_queue labels are only reachable after nvmet_cq_create() has
> succeeded, so this pairs the create/put correctly.
> 
> Fixes: 0faa0fe6f90e ("nvmet: New NVMe PCI endpoint function target driver")
> Cc: stable@vger.kernel.org
> Assisted-by: Claude:claude-opus-4-8
> Signed-off-by: Yifei Gao <gyf161023@gmail.com>

Looks OK.

Reviewed-by: Damien Le Moal <dlemoal@kernel.org>

Note: it would be nice if could explicitly mention if you actually tested this
or if this is just an AI generated fix.

-- 
Damien Le Moal
Western Digital Research