[dpdk-dev] [PATCH v2] net/virtio: fix Virtio-PCI ops assignment

VIRTIO_OPS() macro relies on the port ID stored in the
virtio_hw struct. Issue is that it is used before being
assigned at init time. It results in all devices setting
ops on port ID 0, causing crash later when calling ops
for port IDs other than 0.

This patch ensures port ID assignment is done at early
primary process probe time, before it is being used.

Bugzilla ID: 631
Fixes: 512e27eeb743 ("net/virtio: move PCI specific dev init to PCI ethdev init")

Reported-by: Wei Ling <redacted>
Signed-off-by: Maxime Coquelin <redacted>
---
 drivers/net/virtio/virtio_ethdev.c     | 1 -
 drivers/net/virtio/virtio_pci_ethdev.c | 1 +
 2 files changed, 1 insertion(+), 1 deletion(-)

diff --git a/drivers/net/virtio/virtio_ethdev.c b/drivers/net/virtio/virtio_ethdev.c
index e030fe31f4..333a5243a9 100644
--- a/drivers/net/virtio/virtio_ethdev.c
+++ b/drivers/net/virtio/virtio_ethdev.c

@@ -1838,7 +1838,6 @@ eth_virtio_dev_init(struct rte_eth_dev *eth_dev)
 		return -ENOMEM;
 	}
 
-	hw->port_id = eth_dev->data->port_id;
 	rte_spinlock_init(&hw->state_lock);
 
 	/* reset device and negotiate default features */

diff --git a/drivers/net/virtio/virtio_pci_ethdev.c b/drivers/net/virtio/virtio_pci_ethdev.c
index 1b818c4565..3bb5c6268b 100644
--- a/drivers/net/virtio/virtio_pci_ethdev.c
+++ b/drivers/net/virtio/virtio_pci_ethdev.c

@@ -77,6 +77,7 @@ eth_virtio_pci_init(struct rte_eth_dev *eth_dev)
 	int ret;
 
 	if (rte_eal_process_type() == RTE_PROC_PRIMARY) {
+		hw->port_id = eth_dev->data->port_id;
 		ret = vtpci_init(RTE_ETH_DEV_TO_PCI(eth_dev), dev);
 		if (ret) {
 			PMD_INIT_LOG(ERR, "Failed to init PCI device\n");

-- 
2.29.2

On Mon, Feb 1, 2021 at 9:33 PM Maxime Coquelin
[off-list ref] wrote:

VIRTIO_OPS() macro relies on the port ID stored in the
virtio_hw struct. Issue is that it is used before being
assigned at init time. It results in all devices setting
ops on port ID 0, causing crash later when calling ops
for port IDs other than 0.

This patch ensures port ID assignment is done at early
primary process probe time, before it is being used.

Bugzilla ID: 631
Fixes: 512e27eeb743 ("net/virtio: move PCI specific dev init to PCI ethdev init")

Reported-by: Wei Ling <redacted>
Signed-off-by: Maxime Coquelin <redacted>

Reviewed-by: David Marchand <redacted>


-- 
David Marchand

Tested-by: Wei Ling <redacted>

-----Original Message-----
From: Maxime Coquelin <redacted>
Sent: Tuesday, February 2, 2021 04:32 AM
To: dev@dpdk.org; Wang, Yinan <redacted>; Xia, Chenbo
[off-list ref]; amorenoz@redhat.com;
david.marchand@redhat.com; Ling, WeiX [off-list ref]; Jiang, YuX
[off-list ref]
Cc: Maxime Coquelin <redacted>
Subject: [PATCH v2] net/virtio: fix Virtio-PCI ops assignment

On 2/2/2021 10:03 AM, David Marchand wrote:

On Mon, Feb 1, 2021 at 9:33 PM Maxime Coquelin
[off-list ref] wrote:

quoted

VIRTIO_OPS() macro relies on the port ID stored in the
virtio_hw struct. Issue is that it is used before being
assigned at init time. It results in all devices setting
ops on port ID 0, causing crash later when calling ops
for port IDs other than 0.

This patch ensures port ID assignment is done at early
primary process probe time, before it is being used.

Bugzilla ID: 631
Fixes: 512e27eeb743 ("net/virtio: move PCI specific dev init to PCI ethdev init")

Reported-by: Wei Ling <redacted>
Signed-off-by: Maxime Coquelin <redacted>

Reviewed-by: David Marchand <redacted>

Tested-by: Wei Ling <redacted>

 >

Applied to dpdk-next-net/main, thanks.