All of lore.kernel.org
 help / color / mirror / Atom feed
From: Lorenzo Bianconi <lorenzo.bianconi@oss.qualcomm.com>
To: sashiko-reviews@lists.linux.dev
Cc: bpf@vger.kernel.org
Subject: Re: [PATCH net] net: stmmac: fix num tx queues in stmmac_xdp_get_tx_queue()
Date: Wed, 18 Nov 2026 23:27:24 +0100	[thread overview]
Message-ID: <av4mTCg0yVhGXQdU@lore-qca> (raw)
In-Reply-To: <sashiko-outbox-221383@kernel.org>

[-- Attachment #1: Type: text/plain, Size: 3939 bytes --]

> Thank you for your contribution! Sashiko AI review found 1 potential
> issue to consider:
> - [Medium] net: stmmac: race on num_tx_queues in
>   stmmac_xdp_get_tx_queue() skips txtimer
> It also noticed 1 related pre-existing issue this patch does not
> introduce:
> - [High] net: stmmac: unchecked return value of
>   netif_set_real_num_tx_queues() [1]
> --
> 
> commit f55af370614a4d0bb79d7ac40f036cd36c0c9e78
> Author: Lorenzo Bianconi <lorenzo.bianconi@oss.qualcomm.com>
> 
> net: stmmac: fix num tx queues in stmmac_xdp_get_tx_queue()
> 
> This commit bounds the CPU-based TX queue selection in
> stmmac_xdp_get_tx_queue() by priv->xmit_qdisc.num_tx_queues instead of
> priv->plat->tx_queues_to_use to stay within dev->real_num_tx_queues when
> tc-mqprio offload is active.
> 
> > diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
> > index 25894771e9e9f..cf87afd273439 100644
> > --- a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
> > +++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
> > @@ -5408,13 +5408,14 @@ static int stmmac_xdp_xmit_xdpf(struct stmmac_priv *priv, int queue,
> >  static int stmmac_xdp_get_tx_queue(struct stmmac_priv *priv,
> >  				   int cpu)
> >  {
> > +	unsigned int num_tx_queues = priv->xmit_qdisc.num_tx_queues;
> >  	int index = cpu;
> >  
> >  	if (unlikely(index < 0))
> >  		index = 0;
> >  
> > -	while (index >= priv->plat->tx_queues_to_use)
> > -		index -= priv->plat->tx_queues_to_use;
> > +	while (index >= num_tx_queues)
> > +		index -= num_tx_queues;
> >  
> >  	return index;
> >  }
> 
> [Severity: Medium]
> Can stmmac_finalize_xdp_rx() arm txtimer on the wrong TX queue if
> priv->xmit_qdisc.num_tx_queues changes during an RX NAPI poll?
> 
> Unlike priv->plat->tx_queues_to_use, priv->xmit_qdisc.num_tx_queues is
> updated on a running interface by tc_setup_dwmac510_mqprio() and
> stmmac_reset_tc_mqprio() without stopping RX NAPI:
> 
> CPU0                              CPU1 (<softirq>)
>                                   stmmac_rx()
>                                     stmmac_xdp_xmit_back()
>                                       q = stmmac_xdp_get_tx_queue()
>                                       stmmac_xdp_xmit_xdpf(..., q, ...)
> tc_setup_dwmac510_mqprio()
>   priv->xmit_qdisc.num_tx_queues = n
>                                     stmmac_finalize_xdp_rx()
>                                       q' = stmmac_xdp_get_tx_queue()
>                                       stmmac_tx_timer_arm(..., q')
>                                       <-- q' != q; q's txtimer not armed
> 
> Because stmmac_xdp_xmit_xdpf() only sets interrupt-on-completion every
> priv->tx_coal_frames[queue] frames, uncoalesced XDP_TX descriptors queued
> on the old queue can be left without a completion interrupt or timer
> cleanup.

ack, right. I will look into this issue.

Regards,
Lorenzo

> 
> [Severity: High]
> This problem wasn't introduced by this patch, but stmmac_hw_setup()
> ignores the return value of netif_set_real_num_tx_queues() [1]:
> 
> drivers/net/ethernet/stmicro/stmmac/stmmac_main.c:stmmac_hw_setup() {
> 	...
> 	/* Configure real RX and TX queues */
> 	netif_set_real_num_rx_queues(dev, priv->plat->rx_queues_to_use);
> 	netif_set_real_num_tx_queues(dev, priv->xmit_qdisc.num_tx_queues);
> 
> 	/* Start the ball rolling... */
> 	stmmac_start_all_dma(priv);
> 	...
> }
> 
> Can netif_set_real_num_tx_queues() fail with -ENOMEM during stmmac_open()
> or stmmac_resume(), leaving dev->real_num_tx_queues out of sync with
> priv->xmit_qdisc.num_tx_queues while DMA is started anyway?
> 
> [1] https://sashiko.dev/#/bug/linux-1bc77f7b-c8d8-4341-a8a0-49092e607be6
> 
> -- 
> Sashiko AI review · https://sashiko.dev/#/patchset/20261117-stmmac-fix-xdp_get_tx_queue-v1-1-4cd74ee7a76c@oss.qualcomm.com?part=1

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 228 bytes --]

  reply	other threads:[~2026-11-18 22:27 UTC|newest]

Thread overview: 6+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-11-17 15:51 [PATCH net] net: stmmac: fix num tx queues in stmmac_xdp_get_tx_queue() Lorenzo Bianconi
2026-11-17 15:57 ` netdev-bot+sinfo
2026-11-17 20:30   ` Lorenzo Bianconi
2026-11-18 15:52 ` sashiko-bot
2026-11-18 22:27   ` Lorenzo Bianconi [this message]
2026-11-20  6:54 ` netdev-bot+sashiko

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=av4mTCg0yVhGXQdU@lore-qca \
    --to=lorenzo.bianconi@oss.qualcomm.com \
    --cc=bpf@vger.kernel.org \
    --cc=sashiko-reviews@lists.linux.dev \
    /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.