All of lore.kernel.org
 help / color / mirror / Atom feed
From: sashiko-bot@kernel.org
To: "James Clark" <jjc@jclark.com>
Cc: devicetree@vger.kernel.org, robh@kernel.org, conor+dt@kernel.org
Subject: Re: [PATCH net-next v2 2/2] net: ethernet: ti: am65-cpts: implement PTP_SYS_OFFSET_PRECISE
Date: Sat, 21 Nov 2026 03:47:27 +0000	[thread overview]
Message-ID: <sashiko-outbox-226601@kernel.org> (raw)
In-Reply-To: <4af45df52cdd96da920bc36d29bc341615bbe95b.1795143925.git.jjc@jclark.com>

Thank you for your contribution! Sashiko AI review found 1 potential
issue to consider:
- [Medium] am65-cpts: missing per-pop TS_PEND delay in
  am65_cpts_fifo_pop_event()
--

commit e2b0e8be4be95c8c4617e693d71c09e33009c74e
Author: James Clark <jjc@jclark.com>

net: ethernet: ti: am65-cpts: implement PTP_SYS_OFFSET_PRECISE

This commit adds hardware-supported PTP_SYS_OFFSET_PRECISE cross-timestamping
to the TI K3 CPTS driver by routing a selected bit of the Global Timebase
Counter (GTC) to a CPTS hardware timestamp input.

> diff --git a/drivers/net/ethernet/ti/am65-cpts.c b/drivers/net/ethernet/ti/am65-cpts.c
> index 3ce3b2245efba..4987909d3bc2b 100644
> --- a/drivers/net/ethernet/ti/am65-cpts.c
> +++ b/drivers/net/ethernet/ti/am65-cpts.c
[ ... ]
> @@ -797,6 +841,243 @@ static int am65_cpts_ptp_enable(struct ptp_clock_info *ptp,
[ ... ]
> +/*
> + * Read the FIFO until it is empty, then accept the next push event. One
> + * __am65_cpts_fifo_read() takes at most AM65_CPTS_FIFO_DEPTH events, and after
> + * a burst of push edges (a request that wakes up late leaves the input enabled
> + * for many periods) more can be waiting. A push event left behind would be
> + * taken for the next attempt's. With the push input disabled no new push
> + * events arrive, so after enough passes every old one has been read even if
> + * packet events keep the FIFO busy. The lock is dropped between passes; a push
> + * event read meanwhile by the IRQ thread is ignored, as nothing is armed.
> + */
> +static int am65_cpts_gtc_push_arm(struct am65_cpts *cpts)
> +{
> +	unsigned long flags;
> +	int i;
> +
> +	for (i = 0; i < AM65_CPTS_GTC_PUSH_DRAIN_PASSES; i++) {
> +		spin_lock_irqsave(&cpts->lock, flags);
> +		__am65_cpts_fifo_read(cpts);
> +		/*
> +		 * TS_PEND is valid only once four CPTS_RFT_CLK periods plus
> +		 * four CPPI_ICLK periods have passed since a pop. Read a
> +		 * register so that the last pop has reached the CPTS, then
> +		 * wait. 100 ns is enough for clocks of 80 MHz or more.
> +		 */
> +		am65_cpts_read32(cpts, intstat_raw);
> +		ndelay(100);
> +		if (!(am65_cpts_read32(cpts, intstat_raw) &
> +		      AM65_CPTS_INTSTAT_RAW_TS_PEND)) {
> +			cpts->gtc_push_armed = true;
> +			spin_unlock_irqrestore(&cpts->lock, flags);
> +			return 0;
> +		}
> +		spin_unlock_irqrestore(&cpts->lock, flags);
> +	}
> +	return -EBUSY;
> +}

[Severity: Medium]
Does __am65_cpts_fifo_read() only pop a single disarmed GTC push event per
pass in am65_cpts_gtc_push_arm()?

When __am65_cpts_fifo_read() drains a disarmed GTC push event,
am65_cpts_gtc_push_event() returns immediately and the next loop iteration
calls am65_cpts_fifo_pop_event() without waiting for the four CPTS_RFT_CLK
plus four CPPI_ICLK periods after writing event_pop:

drivers/net/ethernet/ti/am65-cpts.c:am65_cpts_fifo_pop_event() {
	u32 r = am65_cpts_read32(cpts, intstat_raw);

	if (r & AM65_CPTS_INTSTAT_RAW_TS_PEND) {
		...
		am65_cpts_write32(cpts, AM65_CPTS_EVENT_POP, event_pop);
		return false;
	}
	return true;
}

Because TS_PEND is not valid yet on that immediate re-entry,
am65_cpts_fifo_pop_event() sees TS_PEND clear and breaks out of
__am65_cpts_fifo_read() after popping just one event.  If a request wakes
up ~3 ms late so 9 to 16 push events accumulate in the 16-entry FIFO,
AM65_CPTS_GTC_PUSH_DRAIN_PASSES (8) passes only drain 8 events and
am65_cpts_gtc_push_arm() fails with -EBUSY (aborting
am65_cpts_ptp_getcrosststamp() without retrying).  A disarmed push event
sitting ahead of an AM65_CPTS_EV_PUSH event in the FIFO can similarly cause
am65_cpts_gettime() to stop reading the FIFO early and return a stale
cpts->timestamp.

Should the post-pop register read and ndelay(100) be moved into
am65_cpts_fifo_pop_event() right after writing event_pop?

-- 
Sashiko AI review · https://sashiko.dev/#/patchset/cover.1795143925.git.jjc@jclark.com?part=2

      reply	other threads:[~2026-11-21  3:47 UTC|newest]

Thread overview: 4+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-11-20  3:46 [PATCH net-next v2 0/2] net: ethernet: ti: am65-cpts: add PTP_SYS_OFFSET_PRECISE support James Clark
2026-11-20  3:47 ` [PATCH net-next v2 1/2] dt-bindings: net: ti,k3-am654-cpts: add ti,gtc-push James Clark
2026-11-20  3:47 ` [PATCH net-next v2 2/2] net: ethernet: ti: am65-cpts: implement PTP_SYS_OFFSET_PRECISE James Clark
2026-11-21  3:47   ` sashiko-bot [this message]

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=sashiko-outbox-226601@kernel.org \
    --to=sashiko-bot@kernel.org \
    --cc=conor+dt@kernel.org \
    --cc=devicetree@vger.kernel.org \
    --cc=jjc@jclark.com \
    --cc=robh@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.