Re: [PATCH net-next RFC v4 2/5] net: Expose available time stamping layers to user space.

From: Jacob Keller <jacob.e.keller@intel.com>
To: Jakub Kicinski <kuba@kernel.org>,
	Vladimir Oltean <vladimir.oltean@nxp.com>
Cc: "Russell King (Oracle)" <linux@armlinux.org.uk>,
	"Andrew Lunn" <andrew@lunn.ch>,
	"Köry Maincent" <kory.maincent@bootlin.com>,
	netdev@vger.kernel.org, glipus@gmail.com,
	maxime.chevallier@bootlin.com, vadim.fedorenko@linux.dev,
	richardcochran@gmail.com, gerhard@engleder-embedded.com,
	thomas.petazzoni@bootlin.com, krzysztof.kozlowski+dt@linaro.org,
	robh+dt@kernel.org
Subject: Re: [PATCH net-next RFC v4 2/5] net: Expose available time stamping layers to user space.
Date: Wed, 17 May 2023 15:01:34 -0700	[thread overview]
Message-ID: <1571c8f5-b36a-5af1-d56c-d38f69406840@intel.com> (raw)
In-Reply-To: <20230511161625.2e3f0161@kernel.org>

On 5/11/2023 4:16 PM, Jakub Kicinski wrote:
> On Fri, 12 May 2023 02:07:17 +0300 Vladimir Oltean wrote:
>> AFAIU from igc_ptp_tx_hwtstamp(), it's just that the igc DMA
>> controller did not bother to transport the timestamps from the MAC
>> back into the descriptor, leaving it up to software to do it out of
>> band, which of course may cause correlation bugs and limits
>> throughput. Surely they can do better.
> 

In myy understanding for for MAC timestamping, the igc hardware stores
the timestamp in a register and assumes you only have one outstanding
timestamp request at a time.

The MAC timestamp is captured well after the Tx descriptor is written
back as complete. There's no completion queue on this device to send
such a message out-of-band of the Tx writeback. Delaying the writeback
until after the Tx timestamp is captured would have its own challenges.

Obviously it *can* be done better than this, but thats not how the igc
hardware was designed.