Re: [PATCH net-next v2] net: ti: icssg-prueth: Add ICSSG FW Stats

[Jakub Kicinski]

On Wed, 5 Mar 2025 16:46:08 +0530 MD Danish Anwar wrote:
> + - ``FW_RTU_PKT_DROP``: Diagnostic error counter which increments when RTU drops a locally injected packet due to port being disabled or rule violation.
> + - ``FW_Q0_OVERFLOW``: TX overflow counter for queue0
> + - ``FW_Q1_OVERFLOW``: TX overflow counter for queue1
> + - ``FW_Q2_OVERFLOW``: TX overflow counter for queue2
> + - ``FW_Q3_OVERFLOW``: TX overflow counter for queue3
> + - ``FW_Q4_OVERFLOW``: TX overflow counter for queue4
> + - ``FW_Q5_OVERFLOW``: TX overflow counter for queue5
> + - ``FW_Q6_OVERFLOW``: TX overflow counter for queue6
> + - ``FW_Q7_OVERFLOW``: TX overflow counter for queue7
...

Thanks for the docs, it looks good. Now, do all of these get included
in the standard stats returned by icssg_ndo_get_stats64 ?
That's the primary source of information for the user regarding packet
loss.

>  	if (prueth->pa_stats) {
>  		for (i = 0; i < ARRAY_SIZE(icssg_all_pa_stats); i++) {
> -			reg = ICSSG_FW_STATS_BASE +
> -			      icssg_all_pa_stats[i].offset *
> -			      PRUETH_NUM_MACS + slice * sizeof(u32);
> +			reg = icssg_all_pa_stats[i].offset +
> +			      slice * sizeof(u32);
>  			regmap_read(prueth->pa_stats, reg, &val);
>  			emac->pa_stats[i] += val;

This gets called by icssg_ndo_get_stats64() which is under RCU 
protection and nothing else. I don't see any locking here, and
I hope the regmap doesn't sleep. cat /proc/net/dev to test.
You probably need to send some fixes to net.
-- 
pw-bot: cr