netdev - Re: [PATCH] b44: add 64 bit stats

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:	Sun, 15 Jul 2012 09:26:05 +0200
From:	Eric Dumazet <eric.dumazet@...il.com>
To:	Kevin Groeneveld <kgroeneveld@...il.com>
Cc:	netdev@...r.kernel.org
Subject: Re: [PATCH] b44: add 64 bit stats

On Sat, 2012-07-14 at 21:51 -0400, Kevin Groeneveld wrote:
> From: Kevin Groeneveld <kgroeneveld@...il.com>
> 
> Add support for 64 bit stats to Broadcom b44 ethernet driver.
> 
> Signed-off-by: Kevin Groeneveld <kgroeneveld@...il.com>
> ---
>  drivers/net/ethernet/broadcom/b44.c |   96 ++++++++++++++++++++---------------
>  drivers/net/ethernet/broadcom/b44.h |    3 +-
>  2 files changed, 58 insertions(+), 41 deletions(-)
> 
> diff --git a/drivers/net/ethernet/broadcom/b44.c
> b/drivers/net/ethernet/broadcom/b44.c
> index 46b8b7d..c799b5c 100644
> --- a/drivers/net/ethernet/broadcom/b44.c
> +++ b/drivers/net/ethernet/broadcom/b44.c
> @@ -483,9 +483,11 @@ out:

b44_stats_update() is run from timer, (softirq), check b44_timer()

>  static void b44_stats_update(struct b44 *bp)
>  {
>  	unsigned long reg;
> -	u32 *val;
> +	u64 *val;
> 
>  	val = &bp->hw_stats.tx_good_octets;
> +	u64_stats_update_begin(&bp->hw_stats.syncp);
> +
>  	for (reg = B44_TX_GOOD_O; reg <= B44_TX_PAUSE; reg += 4UL) {
>  		*val++ += br32(bp, reg);
>  	}
> @@ -496,6 +498,8 @@ static void b44_stats_update(struct b44 *bp)
>  	for (reg = B44_RX_GOOD_O; reg <= B44_RX_NPAUSE; reg += 4UL) {
>  		*val++ += br32(bp, reg);
>  	}
> +
> +	u64_stats_update_end(&bp->hw_stats.syncp);
>  }
> 
>  static void b44_link_report(struct b44 *bp)
> @@ -1635,44 +1639,49 @@ static int b44_close(struct net_device *dev)
>  	return 0;
>  }
> 
> -static struct net_device_stats *b44_get_stats(struct net_device *dev)
> +static struct rtnl_link_stats64 *b44_get_stats64(struct net_device *dev,
> +					struct rtnl_link_stats64 *nstat)
>  {
>  
> -	nstat->tx_aborted_errors = hwstat->tx_underruns;
> +	unsigned int start;
> +
> +	do {
> +		start = u64_stats_fetch_begin(&hwstat->syncp);

So you must use u64_stats_fetch_begin_bh()


Because on 32bit, uniprocessor, u64_stats_fetch_begin() only disables
preemption. (there is no seqlock in syncp)

So softirq are allowed to interrupt you and corrupt your stats while you
read them, and you dont notice you have to retry.

> +
> +		/* Convert HW stats into rtnl_link_stats64 stats. */
> +		nstat->rx_packets = hwstat->rx_pkts;
> +		nstat->tx_packets = hwstat->tx_pkts;
> +		nstat->rx_bytes   = hwstat->rx_octets;
> +		nstat->tx_bytes   = hwstat->tx_octets;
> +		nstat->tx_errors  = (hwstat->tx_jabber_pkts +
> +				     hwstat->tx_oversize_pkts +
> +				     hwstat->tx_underruns +
> +				     hwstat->tx_excessive_cols +
> +				     hwstat->tx_late_cols);
> +		nstat->multicast  = hwstat->tx_multicast_pkts;
> +		nstat->collisions = hwstat->tx_total_cols;
> +
> +		nstat->rx_length_errors = (hwstat->rx_oversize_pkts +
> +					   hwstat->rx_undersize);
> +		nstat->rx_over_errors   = hwstat->rx_missed_pkts;
> +		nstat->rx_frame_errors  = hwstat->rx_align_errs;
> +		nstat->rx_crc_errors    = hwstat->rx_crc_errs;
> +		nstat->rx_errors        = (hwstat->rx_jabber_pkts +
> +					   hwstat->rx_oversize_pkts +
> +					   hwstat->rx_missed_pkts +
> +					   hwstat->rx_crc_align_errs +
> +					   hwstat->rx_undersize +
> +					   hwstat->rx_crc_errs +
> +					   hwstat->rx_align_errs +
> +					   hwstat->rx_symbol_errs);
> +
> +		nstat->tx_aborted_errors = hwstat->tx_underruns;
>  #if 0
> -	/* Carrier lost counter seems to be broken for some devices */
> -	nstat->tx_carrier_errors = hwstat->tx_carrier_lost;
> +		/* Carrier lost counter seems to be broken for some devices */
> +		nstat->tx_carrier_errors = hwstat->tx_carrier_lost;
>  #endif
> +	} while (u64_stats_fetch_retry(&hwstat->syncp, start));


u64_stats_fetch_retry_bh()


--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html