[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <874ny2bgqa.fsf@turtle.gmx.de>
Date: Thu, 17 Nov 2011 22:07:25 +0100
From: Sven Joachim <svenjoac@....de>
To: Stephen Hemminger <shemminger@...tta.com>
Cc: davem@...emloft.net, netdev@...r.kernel.org
Subject: Re: [PATCH 4/6] sky2: reduce default Tx ring size
On 2011-11-17 00:42 +0100, Stephen Hemminger wrote:
> The default Tx ring size for the sky2 driver is quite large and could
> cause excess buffer bloat for many users. The minimum ring size
> possible and still allow handling the worst case packet on 64bit platforms
> is 38 which gets rounded up to a power of 2. But most packets only require
> a couple of ring elements.
>
> Signed-off-by: Stephen Hemminger <shemminger@...tta.com>
>
> --- a/drivers/net/ethernet/marvell/sky2.c 2011-11-16 15:19:39.518659262 -0800
> +++ b/drivers/net/ethernet/marvell/sky2.c 2011-11-16 15:19:40.990692544 -0800
> @@ -68,7 +68,7 @@
> #define MAX_SKB_TX_LE (2 + (sizeof(dma_addr_t)/sizeof(u32))*(MAX_SKB_FRAGS+1))
> #define TX_MIN_PENDING (MAX_SKB_TX_LE+1)
> #define TX_MAX_PENDING 1024
> -#define TX_DEF_PENDING 127
> +#define TX_DEF_PENDING 63
>
> #define TX_WATCHDOG (5 * HZ)
> #define NAPI_WEIGHT 64
It's hard to believe, but this innocuous-looking patch caused my system
to crash and burn as soon as I actually used the network.
Unfortunately, I have neither a digital camera nor a serial console at
hand, but from my recollections and having read sky2.c I conclude that
BUG_ON(done >= sky2->tx_ring_size);
in sky2_tx_complete() seems to have been triggered. Does this make any
sense?
Cheers,
Sven
--
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
Powered by blists - more mailing lists