Re: [dpdk-dev] [PATCH v2] net/virtio: fix compiling issue for vectorized NEON path

On 1/27/21 11:37 AM, Subhi Masri wrote:

Hi,

quoted

-----Original Message-----
From: Maxime Coquelin <redacted>
Sent: Wednesday, January 27, 2021 10:44 AM
To: Subhi Masri <redacted>
Cc: dev@dpdk.org; nd@arm.com; Joyce Kong <redacted>;
david.marchand@redhat.com; honnappa.nagarahalli@arm.com;
ruifeng.wang@arm.com; i.maximets@ovn.org
Subject: Re: [PATCH v2] net/virtio: fix compiling issue for vectorized NEON
path

External email: Use caution opening links or attachments


Hi Subhi,

On 1/26/21 10:57 AM, Joyce Kong wrote:

quoted

In file included from ../drivers/net/virtio/virtio_rxtx_packed.c:22:0:
../drivers/net/virtio/virtio_rxtx_packed_neon.h: In function
‘virtqueue_enqueue_batch_packed_vec’:
../drivers/net/virtio/virtio_rxtx_packed_neon.h:74:2: warning:
implicit declaration of function ‘vreinterpretq_p128_u32’
[-Wimplicit-function-declaration]
poly128_t cmp1 = vreinterpretq_p128_u32(~vceqq_u32(ref_seg,
def_ref_seg)); ^

The message shows ‘vreinterpretq_p128_u32’ intrinsic is not supported
because an old version gcc (gcc 4.8.5) is used. So fix the issue with
implementing the logic with other intrinsics.

Bugzilla ID: 621
Fixes: 530887469350 ("net/virtio: add vectorized packed ring NEON Tx")
Fixes: 5971ce5e2a59 ("net/virtio: add vectorized packed ring NEON Rx")

Reported-by: Subhi Masri <redacted>
Signed-off-by: Joyce Kong <redacted>
---
 drivers/net/virtio/virtio_rxtx_packed_neon.h | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/drivers/net/virtio/virtio_rxtx_packed_neon.h

b/drivers/net/virtio/virtio_rxtx_packed_neon.h
index 01c77b712..00dd04277 100644

--- a/drivers/net/virtio/virtio_rxtx_packed_neon.h
+++ b/drivers/net/virtio/virtio_rxtx_packed_neon.h

@@ -71,8 +71,8 @@ virtqueue_enqueue_batch_packed_vec(struct

virtnet_tx *txvq,

quoted

      uint32x4_t def_ref_seg = vdupq_n_u32(0x10001);
      /* Check refcnt and nb_segs. */
      uint32x4_t ref_seg = vreinterpretq_u32_u8(vqtbl2q_u8(mbuf,

ref_seg_msk));

quoted

-     poly128_t cmp1 = vreinterpretq_p128_u32(~vceqq_u32(ref_seg,

def_ref_seg));

quoted

-     if (unlikely(cmp1))
+     uint64x2_t cmp1 = vreinterpretq_u64_u32(~vceqq_u32(ref_seg,

def_ref_seg));

quoted

+     if (unlikely(vgetq_lane_u64(cmp1, 0) || vgetq_lane_u64(cmp1,
+ 1)))
              return -1;

      /* Check headroom is enough. */

@@ -225,10 +225,10 @@ virtqueue_dequeue_batch_packed_vec(struct

virtnet_rx *rxvq,

quoted

      if (vq->vq_packed.used_wrap_counter)
              v_used_flag = vdupq_n_u32(PACKED_FLAGS_MASK);

-     poly128_t desc_stats = vreinterpretq_p128_u32(~vceqq_u32(v_flag,

v_used_flag));

quoted

+     uint64x2_t desc_stats = vreinterpretq_u64_u32(~vceqq_u32(v_flag,
+ v_used_flag));

      /* Check all descs are used. */
-     if (desc_stats)
+     if (unlikely(vgetq_lane_u64(desc_stats, 0) ||
+ vgetq_lane_u64(desc_stats, 1)))
              return -1;

      /* Load 2 mbuf pointers per time. */

Can you confirm the patch works for you?

Yes it does , you can add my tag.
Tested-by: Subhi Masri <redacted>

Thanks!
Maxime

quoted

Thanks,
Maxime

Regards,
Subhi Masri