[PATCH] vhost: fix compilation issue in async path

[PATCH] vhost: fix compilation issue in async path

[Maxime Coquelin]

This patch fixes a compilation issue met with GCC on
Loongarch64:

In function ‘mbuf_to_desc’,
    inlined from ‘vhost_enqueue_async_packed’ at ../../../dpdk/lib/vhost/virtio_net.c:1822:6,
    inlined from ‘virtio_dev_rx_async_packed’ at ../../../dpdk/lib/vhost/virtio_net.c:1836:6,
    inlined from ‘virtio_dev_rx_async_submit_packed’ at ../../../dpdk/lib/vhost/virtio_net.c:1895:7:
../../../dpdk/lib/vhost/virtio_net.c:1159:18: error: ‘buf_vec[0].buf_addr’ may be used uninitialized [-Werror=maybe-uninitialized]
 1159 |         buf_addr = buf_vec[vec_idx].buf_addr;
      |         ~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~
../../../dpdk/lib/vhost/virtio_net.c: In function ‘virtio_dev_rx_async_submit_packed’:
../../../dpdk/lib/vhost/virtio_net.c:1834:27: note: ‘buf_vec’ declared here
 1834 |         struct buf_vector buf_vec[BUF_VECTOR_MAX];
      |                           ^~~~~~~

It happens because the compiler assumes that 'size'
variable in vhost_enqueue_async_packed could wrap to 0 since
'size' is uint32_t and pkt->pkt_len too.

In practice, it would never happen since 'pkt->pkt_len' is
unlikely to be close to UINT32_MAX, but let's just change
'size' to uint64_t to make the compiler happy without
having to add runtime checks.

Fixes: 873e8dad6f49 ("vhost: support packed ring in async datapath")
Cc: stable@dpdk.org

Signed-off-by: Maxime Coquelin <maxime.coquelin@redhat.com>
---
 lib/vhost/virtio_net.c | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/lib/vhost/virtio_net.c b/lib/vhost/virtio_net.c
index 8f4d0f0502..b86fb26040 100644
--- a/lib/vhost/virtio_net.c
+++ b/lib/vhost/virtio_net.c
@@ -1780,7 +1780,7 @@ vhost_enqueue_async_packed(struct virtio_net *dev,
 	uint16_t buf_id = 0;
 	uint32_t len = 0;
 	uint16_t desc_count = 0;
-	uint32_t size = pkt->pkt_len + sizeof(struct virtio_net_hdr_mrg_rxbuf);
+	uint64_t size = pkt->pkt_len + sizeof(struct virtio_net_hdr_mrg_rxbuf);
 	uint32_t buffer_len[vq->size];
 	uint16_t buffer_buf_id[vq->size];
 	uint16_t buffer_desc_count[vq->size];
-- 
2.37.3

Re: [PATCH] vhost: fix compilation issue in async path

[David Marchand]

On Wed, Oct 5, 2022 at 4:49 PM Maxime Coquelin
<maxime.coquelin@redhat.com> wrote:
>
> This patch fixes a compilation issue met with GCC on

GCC 12 (it is worth detailing, since I and Thomas hit the same issue
on some other cross compiling toolchains using GCC 12).

> Loongarch64:

LoongArch64

>
> In function ‘mbuf_to_desc’,
>     inlined from ‘vhost_enqueue_async_packed’ at ../../../dpdk/lib/vhost/virtio_net.c:1822:6,
>     inlined from ‘virtio_dev_rx_async_packed’ at ../../../dpdk/lib/vhost/virtio_net.c:1836:6,
>     inlined from ‘virtio_dev_rx_async_submit_packed’ at ../../../dpdk/lib/vhost/virtio_net.c:1895:7:
> ../../../dpdk/lib/vhost/virtio_net.c:1159:18: error: ‘buf_vec[0].buf_addr’ may be used uninitialized [-Werror=maybe-uninitialized]
>  1159 |         buf_addr = buf_vec[vec_idx].buf_addr;
>       |         ~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~
> ../../../dpdk/lib/vhost/virtio_net.c: In function ‘virtio_dev_rx_async_submit_packed’:
> ../../../dpdk/lib/vhost/virtio_net.c:1834:27: note: ‘buf_vec’ declared here
>  1834 |         struct buf_vector buf_vec[BUF_VECTOR_MAX];
>       |                           ^~~~~~~
>
> It happens because the compiler assumes that 'size'
> variable in vhost_enqueue_async_packed could wrap to 0 since
> 'size' is uint32_t and pkt->pkt_len too.
>
> In practice, it would never happen since 'pkt->pkt_len' is
> unlikely to be close to UINT32_MAX, but let's just change
> 'size' to uint64_t to make the compiler happy without
> having to add runtime checks.
>
> Fixes: 873e8dad6f49 ("vhost: support packed ring in async datapath")
> Cc: stable@dpdk.org
>
> Signed-off-by: Maxime Coquelin <maxime.coquelin@redhat.com>
Reviewed-by: David Marchand <david.marchand@redhat.com>


-- 
David Marchand

Re: [PATCH] vhost: fix compilation issue in async path

[Maxime Coquelin]

On 10/5/22 16:56, David Marchand wrote:
> On Wed, Oct 5, 2022 at 4:49 PM Maxime Coquelin
> <maxime.coquelin@redhat.com> wrote:
>>
>> This patch fixes a compilation issue met with GCC on
> 
> GCC 12 (it is worth detailing, since I and Thomas hit the same issue
> on some other cross compiling toolchains using GCC 12).
> 
>> Loongarch64:
> 
> LoongArch64
> 

Oops, I did the changes you asked me off-list, but forgot to format
the patch again!

Posting v2.

>>
>> In function ‘mbuf_to_desc’,
>>      inlined from ‘vhost_enqueue_async_packed’ at ../../../dpdk/lib/vhost/virtio_net.c:1822:6,
>>      inlined from ‘virtio_dev_rx_async_packed’ at ../../../dpdk/lib/vhost/virtio_net.c:1836:6,
>>      inlined from ‘virtio_dev_rx_async_submit_packed’ at ../../../dpdk/lib/vhost/virtio_net.c:1895:7:
>> ../../../dpdk/lib/vhost/virtio_net.c:1159:18: error: ‘buf_vec[0].buf_addr’ may be used uninitialized [-Werror=maybe-uninitialized]
>>   1159 |         buf_addr = buf_vec[vec_idx].buf_addr;
>>        |         ~~~~~~~~~^~~~~~~~~~~~~~~~~~~~~~~~~~~
>> ../../../dpdk/lib/vhost/virtio_net.c: In function ‘virtio_dev_rx_async_submit_packed’:
>> ../../../dpdk/lib/vhost/virtio_net.c:1834:27: note: ‘buf_vec’ declared here
>>   1834 |         struct buf_vector buf_vec[BUF_VECTOR_MAX];
>>        |                           ^~~~~~~
>>
>> It happens because the compiler assumes that 'size'
>> variable in vhost_enqueue_async_packed could wrap to 0 since
>> 'size' is uint32_t and pkt->pkt_len too.
>>
>> In practice, it would never happen since 'pkt->pkt_len' is
>> unlikely to be close to UINT32_MAX, but let's just change
>> 'size' to uint64_t to make the compiler happy without
>> having to add runtime checks.
>>
>> Fixes: 873e8dad6f49 ("vhost: support packed ring in async datapath")
>> Cc: stable@dpdk.org
>>
>> Signed-off-by: Maxime Coquelin <maxime.coquelin@redhat.com>
> Reviewed-by: David Marchand <david.marchand@redhat.com>
> 
>