netdev.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Jakub Kicinski <kuba@kernel.org>
To: Sunil Kovvuri <sunil.kovvuri@gmail.com>
Cc: Linux Netdev List <netdev@vger.kernel.org>,
	"David S. Miller" <davem@davemloft.net>,
	Michal Kubecek <mkubecek@suse.cz>,
	Sunil Goutham <sgoutham@marvell.com>,
	Geetha sowjanya <gakula@marvell.com>
Subject: Re: [PATCH v4 07/17] octeontx2-pf: Add packet transmission support
Date: Thu, 23 Jan 2020 06:24:25 -0800	[thread overview]
Message-ID: <20200123062425.41e6d8ae@cakuba> (raw)
In-Reply-To: <CA+sq2CdX31cqsSc=qRhbcZ5fOk2zGOrhTMGqhsPddbhW=YQPCQ@mail.gmail.com>

On Thu, 23 Jan 2020 01:20:51 +0530, Sunil Kovvuri wrote:
> > Why do you care about big endian bitfields tho, if you don't care about
> > endianness of the data itself?  
> 
> At this point of time we are not addressing big endian functionality,
> so few things
> might be broken in that aspect. If it's preferred to remove, i will remove it.

Yes, I think removing it would be cleaner than having it partially
working. The driver depends on ARM64, anyway.

> > > +};
> > > +
> > >  #endif /* OTX2_STRUCT_H */
> > > diff --git a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_txrx.c b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_txrx.c
> > > index e6be18d..f416603 100644
> > > --- a/drivers/net/ethernet/marvell/octeontx2/nic/otx2_txrx.c
> > > +++ b/drivers/net/ethernet/marvell/octeontx2/nic/otx2_txrx.c
> > > @@ -32,6 +32,78 @@ static struct nix_cqe_hdr_s *otx2_get_next_cqe(struct otx2_cq_queue *cq)
> > >       return cqe_hdr;
> > >  }
> > >  
> >> > +static void otx2_sqe_flush(struct otx2_snd_queue *sq, int size)  
> > > +{
> > > +     u64 status;
> > > +
> > > +     /* Packet data stores should finish before SQE is flushed to HW */  
> >
> > Packet data is synced by the dma operations the barrier shouldn't be
> > needed AFAIK (and if it would be, dma_wmb() would not be the one, as it
> > only works for iomem AFAIU).
> >  
> > > +     dma_wmb();  
> 
> Due to out of order execution by CPU, HW folks have suggested add a barrier
> to avoid scenarios where packet is transmitted before all stores from
> CPU are committed.
> On arm64 a dmb() is less costlier than a dsb() barrier and as per HW
> folks a dmb(st)
> is sufficient to ensure all stores from CPU are committed. And
> dma_wmb() uses dmb(st)
> hence it is used. It's more of choice of architecture specific
> instruction rather than the API.

Hmm.. I'm not a DMA API expert but AFAIU it'd be a serious bug if
dma_map..() APIs didn't guarantee that packet data is written out.
Not only out of the CPU store buffer but also the caches to DRAM.

  reply	other threads:[~2020-01-23 14:24 UTC|newest]

Thread overview: 36+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2020-01-21 13:21 [PATCH v4 00/17] octeontx2-pf: Add network driver for physical function sunil.kovvuri
2020-01-21 13:21 ` [PATCH v4 01/17] octeontx2-pf: Add Marvell OcteonTX2 NIC driver sunil.kovvuri
2020-01-21 16:00   ` Jakub Kicinski
2020-01-21 13:21 ` [PATCH v4 02/17] octeontx2-pf: Mailbox communication with AF sunil.kovvuri
2020-01-21 16:00   ` Jakub Kicinski
2020-01-22 19:27     ` Sunil Kovvuri
2020-01-23 14:14       ` Jakub Kicinski
2020-01-24  8:33   ` Maciej Fijalkowski
2020-01-24 17:15     ` Sunil Kovvuri
2020-01-21 13:21 ` [PATCH v4 03/17] octeontx2-pf: Attach NIX and NPA block LFs sunil.kovvuri
2020-01-21 16:00   ` Jakub Kicinski
2020-01-22 19:27     ` Sunil Kovvuri
2020-01-21 13:21 ` [PATCH v4 04/17] octeontx2-pf: Initialize and config queues sunil.kovvuri
2020-01-21 16:00   ` Jakub Kicinski
2020-01-22 19:29     ` Sunil Kovvuri
2020-01-23 14:20       ` Jakub Kicinski
2020-01-24 11:21         ` Sunil Kovvuri
2020-01-21 13:21 ` [PATCH v4 05/17] octeontx2-pf: Setup interrupts and NAPI handler sunil.kovvuri
2020-01-21 13:21 ` [PATCH v4 06/17] octeontx2-pf: Receive packet handling support sunil.kovvuri
2020-01-21 16:33   ` Jakub Kicinski
2020-01-22 19:34     ` Sunil Kovvuri
2020-01-24 10:14   ` Maciej Fijalkowski
2020-01-21 13:21 ` [PATCH v4 07/17] octeontx2-pf: Add packet transmission support sunil.kovvuri
2020-01-21 16:54   ` Jakub Kicinski
2020-01-22 19:50     ` Sunil Kovvuri
2020-01-23 14:24       ` Jakub Kicinski [this message]
2020-01-21 13:21 ` [PATCH v4 08/17] octeontx2-pf: Register and handle link notifications sunil.kovvuri
2020-01-21 13:21 ` [PATCH v4 09/17] octeontx2-pf: MTU, MAC and RX mode config support sunil.kovvuri
2020-01-21 13:21 ` [PATCH v4 10/17] octeontx2-pf: Error handling support sunil.kovvuri
2020-01-21 13:21 ` [PATCH v4 11/17] octeontx2-pf: Receive side scaling support sunil.kovvuri
2020-01-21 13:21 ` [PATCH v4 12/17] octeontx2-pf: TCP segmentation offload support sunil.kovvuri
2020-01-21 13:21 ` [PATCH v4 13/17] octeontx2-pf: Add ndo_get_stats64 sunil.kovvuri
2020-01-21 13:21 ` [PATCH v4 14/17] octeontx2-pf: Add basic ethtool support sunil.kovvuri
2020-01-21 13:21 ` [PATCH v4 15/17] octeontx2-pf: ethtool RSS config support sunil.kovvuri
2020-01-21 13:21 ` [PATCH v4 16/17] Documentation: net: octeontx2: Add RVU HW and drivers overview sunil.kovvuri
2020-01-21 13:21 ` [PATCH v4 17/17] MAINTAINERS: Add entry for Marvell OcteonTX2 Physical Function driver sunil.kovvuri

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=20200123062425.41e6d8ae@cakuba \
    --to=kuba@kernel.org \
    --cc=davem@davemloft.net \
    --cc=gakula@marvell.com \
    --cc=mkubecek@suse.cz \
    --cc=netdev@vger.kernel.org \
    --cc=sgoutham@marvell.com \
    --cc=sunil.kovvuri@gmail.com \
    /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 a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).