Received-SPF: pass (google.com: domain of linux-kernel+bounces-210263-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:4601:e00::3 as permitted sender) client-ip=2604:1380:4601:e00::3;
Message-ID: <7eccba8f-2b4d-42cc-8a12-8badc41e1c01@linux.dev>
Date: Tue, 11 Jun 2024 12:43:45 -0400
Precedence: bulk
MIME-Version: 1.0
Subject: Re: [PATCH net-next 3/3] net: xilinx: axienet: Add statistics support
To: Andrew Lunn <andrew@lunn.ch>
Cc: Radhey Shyam Pandey <radhey.shyam.pandey@amd.com>,
 netdev@vger.kernel.org, linux-arm-kernel@lists.infradead.org,
 Michal Simek <michal.simek@amd.com>, Jakub Kicinski <kuba@kernel.org>,
 Russell King <linux@armlinux.org.uk>, Paolo Abeni <pabeni@redhat.com>,
 Eric Dumazet <edumazet@google.com>, linux-kernel@vger.kernel.org,
 "David S . Miller" <davem@davemloft.net>
References: <20240610231022.2460953-1-sean.anderson@linux.dev>
 <20240610231022.2460953-4-sean.anderson@linux.dev>
 <7c06c9d7-ad11-4acd-8c80-fbeb902da40d@lunn.ch>
 <d2cc10a6-0c6a-471a-bd5b-3e939905fc41@lunn.ch>
Content-Language: en-US
From: Sean Anderson <sean.anderson@linux.dev>
In-Reply-To: <d2cc10a6-0c6a-471a-bd5b-3e939905fc41@lunn.ch>
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 7bit

On 6/10/24 20:29, Andrew Lunn wrote:
> On Tue, Jun 11, 2024 at 02:13:40AM +0200, Andrew Lunn wrote:
>> On Mon, Jun 10, 2024 at 07:10:22PM -0400, Sean Anderson wrote:
>> > Add support for reading the statistics counters, if they are enabled.
>> > The counters may be 64-bit, but we can't detect this as there's no
>> > ability bit for it and the counters are read-only. Therefore, we assume
>> > the counters are 32-bits.
>> 
>> > +static void axienet_stats_update(struct axienet_local *lp)
>> > +{
>> > +	enum temac_stat stat;
>> > +
>> > +	lockdep_assert_held(&lp->stats_lock);
>> > +
>> > +	u64_stats_update_begin(&lp->hw_stat_sync);
>> > +	for (stat = 0; stat < STAT_COUNT; stat++) {
>> > +		u32 counter = axienet_ior(lp, XAE_STATS_OFFSET + stat * 8);
>> 
>> The * 8 here suggests the counters are spaced so that they could be 64
>> bit wide, even when only 32 bits are used. Does the documentation say
>> anything about the upper 32 bits when the counters are only 32 bits?
>> Are they guaranteed to read as zero? I'm just wondering if the code
>> should be forward looking and read all 64 bits? 
> 
> Actually, if you read the upper 32 bits and they are not 0, you know
> you have 64 bit counters. You can then kill off your period task, it
> is not needed because your software counters will wrap around the same
> time as the hardware counters.

Yes, but then our stats remain stale forever, because we don't refresh
stats before reading them as detailed in my other response.

--Sean