All of lore.kernel.org
 help / color / mirror / Atom feed
From: Sriram Yagnaraman <sriram.yagnaraman@est.tech>
To: Akihiko Odaki <akihiko.odaki@daynix.com>
Cc: "qemu-devel@nongnu.org" <qemu-devel@nongnu.org>,
	Jason Wang <jasowang@redhat.com>,
	Dmitry Fleytman <dmitry.fleytman@gmail.com>,
	"Michael S . Tsirkin" <mst@redhat.com>,
	Marcel Apfelbaum <marcel.apfelbaum@gmail.com>
Subject: RE: [PATCH 5/9] igb: respect E1000_VMOLR_RSSE
Date: Mon, 30 Jan 2023 12:07:20 +0000	[thread overview]
Message-ID: <DBBP189MB143380F3BC336D9A80B251A995D39@DBBP189MB1433.EURP189.PROD.OUTLOOK.COM> (raw)
In-Reply-To: <e6a03b04-2565-7ae3-ace5-1486369d000b@daynix.com>

> -----Original Message-----
> From: Akihiko Odaki <akihiko.odaki@daynix.com>
> Sent: Sunday, 29 January 2023 08:25
> To: Sriram Yagnaraman <sriram.yagnaraman@est.tech>
> Cc: qemu-devel@nongnu.org; Jason Wang <jasowang@redhat.com>; Dmitry
> Fleytman <dmitry.fleytman@gmail.com>; Michael S . Tsirkin
> <mst@redhat.com>; Marcel Apfelbaum <marcel.apfelbaum@gmail.com>
> Subject: Re: [PATCH 5/9] igb: respect E1000_VMOLR_RSSE
> 
> On 2023/01/28 22:46, Sriram Yagnaraman wrote:
> > RSS for VFs is only enabled if VMOLR[n].RSSE is set.
> >
> > Signed-off-by: Sriram Yagnaraman <sriram.yagnaraman@est.tech>
> > ---
> >   hw/net/igb_core.c | 18 +++++++++++++-----
> >   1 file changed, 13 insertions(+), 5 deletions(-)
> >
> > diff --git a/hw/net/igb_core.c b/hw/net/igb_core.c index
> > 1eb7ba168f..e4fd4a1a5f 100644
> > --- a/hw/net/igb_core.c
> > +++ b/hw/net/igb_core.c
> > @@ -69,7 +69,7 @@ typedef struct IGBTxPktVmdqCallbackContext {
> >
> >   static ssize_t
> >   igb_receive_internal(IGBCore *core, const struct iovec *iov, int iovcnt,
> > -                     bool has_vnet, bool *assigned);
> > +                     bool has_vnet, bool *external_tx);
> 
> I admit external_tx is somewhat confusing, but it is more than just telling if it
> is assigned to Rx queue or not. If external_tx is not NULL, it indicates it is part
> of Tx packet switching. In that case, a bool value which describes whether the
> packet should be routed to external LAN must be assigned. The value can be
> false even if the packet is assigned to Rx queues; it will be always false if it is
> multicast, for example.

Yes, I undestand the purpose of external_tx. I merely changed the parameter name in the function declaration to match it's defintion. Anyhow, I will remove this change since it was purely comsetic.

> 
> >
> >   static inline void
> >   igb_set_interrupt_cause(IGBCore *core, uint32_t val); @@ -942,7
> > +942,7 @@ static uint16_t igb_receive_assign(IGBCore *core, const
> > struct eth_header *ehdr,
> >
> >       if (core->mac[MRQC] & 1) {
> >           if (is_broadcast_ether_addr(ehdr->h_dest)) {
> > -            for (i = 0; i < 8; i++) {
> > +            for (i = 0; i < IGB_MAX_VF_FUNCTIONS; i++) {
> 
> I just left it as 8 because VMDq is not specific to VF. Perhaps it is better to
> have another macro to denote the number of VMDq pools, but it is not done
> yet.

Ok, I agree. I will introduce a new IGB_MAX_VM_POOLS macro instead.

> 
> >                   if (core->mac[VMOLR0 + i] & E1000_VMOLR_BAM) {
> >                       queues |= BIT(i);
> >                   }
> > @@ -976,7 +976,7 @@ static uint16_t igb_receive_assign(IGBCore *core,
> const struct eth_header *ehdr,
> >                   f = ta_shift[(rctl >> E1000_RCTL_MO_SHIFT) & 3];
> >                   f = (((ehdr->h_dest[5] << 8) | ehdr->h_dest[4]) >> f) & 0xfff;
> >                   if (macp[f >> 5] & (1 << (f & 0x1f))) {
> > -                    for (i = 0; i < 8; i++) {
> > +                    for (i = 0; i < IGB_MAX_VF_FUNCTIONS; i++) {
> >                           if (core->mac[VMOLR0 + i] & E1000_VMOLR_ROMPE) {
> >                               queues |= BIT(i);
> >                           }
> > @@ -999,7 +999,7 @@ static uint16_t igb_receive_assign(IGBCore *core,
> const struct eth_header *ehdr,
> >                       }
> >                   }
> >               } else {
> > -                for (i = 0; i < 8; i++) {
> > +                for (i = 0; i < IGB_MAX_VF_FUNCTIONS; i++) {
> >                       if (core->mac[VMOLR0 + i] & E1000_VMOLR_AUPE) {
> >                           mask |= BIT(i);
> >                       }
> > @@ -1018,7 +1018,15 @@ static uint16_t igb_receive_assign(IGBCore
> *core, const struct eth_header *ehdr,
> >           queues &= core->mac[VFRE];
> >           igb_rss_parse_packet(core, core->rx_pkt, external_tx != NULL,
> rss_info);
> >           if (rss_info->queue & 1) {
> > -            queues <<= 8;
> > +            for (i = 0; i < IGB_MAX_VF_FUNCTIONS; i++) {
> > +                if (!(queues & BIT(i))) {
> > +                    continue;
> > +                }
> > +                if (core->mac[VMOLR0 + i] & E1000_VMOLR_RSSE) {
> > +                    queues |= BIT(i + IGB_MAX_VF_FUNCTIONS);
> > +                    queues &= ~BIT(i);
> > +                }
> > +            }
> >           }
> >       } else {
> >           switch (net_rx_pkt_get_packet_type(core->rx_pkt)) {

  reply	other threads:[~2023-01-30 12:08 UTC|newest]

Thread overview: 20+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2023-01-28 13:46 [PATCH 0/9] igb: add missing feature set from Sriram Yagnaraman
2023-01-28 13:46 ` [PATCH 1/9] MAINTAINERS: Add Sriram Yagnaraman as a igb reviewer Sriram Yagnaraman
2023-01-28 13:46 ` [PATCH 2/9] igb: handle PF/VF reset properly Sriram Yagnaraman
2023-01-29  5:58   ` Akihiko Odaki
2023-01-29  6:15     ` Akihiko Odaki
2023-01-28 13:46 ` [PATCH 3/9] igb: implement VFRE and VFTE registers Sriram Yagnaraman
2023-01-29  9:15   ` Akihiko Odaki
2023-01-30 10:16     ` Sriram Yagnaraman
2023-01-30 13:20       ` Akihiko Odaki
2023-01-28 13:46 ` [PATCH 4/9] igb: check oversized packets for VMDq Sriram Yagnaraman
2023-01-29  7:06   ` Akihiko Odaki
2023-01-28 13:46 ` [PATCH 5/9] igb: respect E1000_VMOLR_RSSE Sriram Yagnaraman
2023-01-29  7:24   ` Akihiko Odaki
2023-01-30 12:07     ` Sriram Yagnaraman [this message]
2023-01-30 13:20       ` Akihiko Odaki
2023-01-28 13:46 ` [PATCH 6/9] igb: add ICR_RXDW Sriram Yagnaraman
2023-01-28 13:46 ` [PATCH 7/9] igb: implement VF Tx and Rx stats Sriram Yagnaraman
2023-01-28 13:46 ` [PATCH 8/9] igb: respect VT_CTL ignore MAC field Sriram Yagnaraman
2023-01-28 13:46 ` [PATCH 9/9] igb: respect VMVIR and VMOLR for VLAN Sriram Yagnaraman
2023-01-29  8:13   ` Akihiko Odaki

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=DBBP189MB143380F3BC336D9A80B251A995D39@DBBP189MB1433.EURP189.PROD.OUTLOOK.COM \
    --to=sriram.yagnaraman@est.tech \
    --cc=akihiko.odaki@daynix.com \
    --cc=dmitry.fleytman@gmail.com \
    --cc=jasowang@redhat.com \
    --cc=marcel.apfelbaum@gmail.com \
    --cc=mst@redhat.com \
    --cc=qemu-devel@nongnu.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.