[PATCH v2 4/8] vdpa/ifc: write queue count to MQ register

[Pei, Andy]

Hi Chenbo, 

For BLK devices, ifc driver does not have VHOST_USER_SET_VRING_ENABLE msg.
So it does not need to dynamic configuration.

But now I think it is fine to use the same logic.

> -----Original Message-----
> From: Xia, Chenbo <chenbo.xia at intel.com>
> Sent: Friday, September 9, 2022 3:35 PM
> To: Pei, Andy <andy.pei at intel.com>; dev at dpdk.org
> Cc: Xu, Rosen <rosen.xu at intel.com>; Huang, Wei <wei.huang at intel.com>; Cao,
> Gang <gang.cao at intel.com>; maxime.coquelin at redhat.com; Huang Wei
> <wei_huang at intel.com>
> Subject: RE: [PATCH v2 4/8] vdpa/ifc: write queue count to MQ register
> 
> > -----Original Message-----
> > From: Pei, Andy <andy.pei at intel.com>
> > Sent: Thursday, September 8, 2022 1:54 PM
> > To: dev at dpdk.org
> > Cc: Xia, Chenbo <chenbo.xia at intel.com>; Xu, Rosen
> > <rosen.xu at intel.com>; Huang, Wei <wei.huang at intel.com>; Cao, Gang
> > <gang.cao at intel.com>; maxime.coquelin at redhat.com; Huang Wei
> > <wei_huang at intel.com>
> > Subject: [PATCH v2 4/8] vdpa/ifc: write queue count to MQ register
> >
> > Write queue count to IFCVF_MQ_OFFSET register to enable multi queue
> > feature.
> >
> > Signed-off-by: Andy Pei <andy.pei at intel.com>
> > Signed-off-by: Huang Wei <wei_huang at intel.com>
> > ---
> >  drivers/vdpa/ifc/base/ifcvf.c | 33 +++++++++++++++++++++++++++++++++
> >  1 file changed, 33 insertions(+)
> >
> > diff --git a/drivers/vdpa/ifc/base/ifcvf.c
> > b/drivers/vdpa/ifc/base/ifcvf.c index 34c8226..0444d74 100644
> > --- a/drivers/vdpa/ifc/base/ifcvf.c
> > +++ b/drivers/vdpa/ifc/base/ifcvf.c
> > @@ -198,6 +198,38 @@
> >  	IFCVF_WRITE_REG32(val >> 32, hi);
> >  }
> >
> > +STATIC void
> > +ifcvf_enable_multiqueue(struct ifcvf_hw *hw) {
> > +	u8 *mq_cfg;
> > +	int qid;
> > +	int nr_queue_pair = 0;
> > +
> > +	if (hw->device_type == IFCVF_NET) {
> > +		for (qid = 0; qid < hw->nr_vring; qid++) {
> > +			if (!hw->vring[qid].enable)
> > +				continue;
> > +			nr_queue_pair++;
> > +		}
> > +
> > +		if (nr_queue_pair == 0) {
> > +			WARNINGOUT("no enabled vring\n");
> > +			return;
> > +		}
> > +
> > +		nr_queue_pair = (nr_queue_pair + 1) / 2;
> > +	} else if (hw->device_type == IFCVF_BLK) {
> > +		nr_queue_pair = hw->nr_vring;
> > +	}
> 
> Why above logic will be different for NET/BLK, could you help me understand it?
> 
> Thanks,
> Chenbo
> 
> > +
> > +	mq_cfg = hw->mq_cfg;
> > +	if (mq_cfg) {
> > +		*(u32 *)mq_cfg = nr_queue_pair;
> > +		RTE_LOG(INFO, PMD, "%d queue pairs are enabled\n",
> > +		nr_queue_pair);
> > +	}
> > +}
> > +
> >  STATIC int
> >  ifcvf_hw_enable(struct ifcvf_hw *hw)
> >  {
> > @@ -215,6 +247,7 @@
> >  		return -1;
> >  	}
> >
> > +	ifcvf_enable_multiqueue(hw);
> >  	for (i = 0; i < hw->nr_vring; i++) {
> >  		IFCVF_WRITE_REG16(i, &cfg->queue_select);
> >  		io_write64_twopart(hw->vring[i].desc, &cfg->queue_desc_lo,
> > --
> > 1.8.3.1