From: Michael Breuer <mbreuer@majjas•com>
To: Mike McCormack <mikem@ring3k•org>
Cc: Jarek Poplawski <jarkao2@gmail•com>,
David Miller <davem@davemloft•net>,
shemminger@vyatta•com, flyboy@gmail•com, rjw@sisk•pl,
netdev@vger•kernel.org
Subject: Re: [PATCH] sky2: safer transmit ring cleaning (v4)
Date: Thu, 14 Jan 2010 11:46:15 -0500 [thread overview]
Message-ID: <4B4F4A57.5050708@majjas.com> (raw)
In-Reply-To: <4B4F19EB.5080207@ring3k.org>
On 1/14/2010 8:19 AM, Mike McCormack wrote:
> Jarek Poplawski wrote:
>
>> On Thu, Jan 14, 2010 at 03:20:09AM -0800, David Miller wrote:
>>
>>> From: Jarek Poplawski<jarkao2@gmail•com>
>>> Date: Thu, 14 Jan 2010 11:16:36 +0000
>>>
>>>
>>>> So, now I really ;-) agree with David: this needs a proper fix.
>>>>
>>> Now Jarek, do you now see my dirty little secret?
>>>
>>> When people disagree with me, I just silently sit around waiting for
>>> them to eventually change their mind.
>>>
>>> Isn't it brilliant? -)
>>>
>> If it were that easy... (it never works for me :-()
>>
>> Probably, there is something more... ;-)
>>
> Here's what was sitting in my tree...
>
>
> Subject: [PATCH] sky2: Don't detach device when restarting
>
> Block the tx queue from transmitting when restarting
> rather than trying to take the device offline.
>
> Signed-off-by: Mike McCormack<mikem@ring3k•org>
> ---
> drivers/net/sky2.c | 13 +++++++++----
> 1 files changed, 9 insertions(+), 4 deletions(-)
>
> diff --git a/drivers/net/sky2.c b/drivers/net/sky2.c
> index d76d907..061f6f2 100644
> --- a/drivers/net/sky2.c
> +++ b/drivers/net/sky2.c
> @@ -1580,6 +1580,8 @@ static int sky2_up(struct net_device *dev)
> if (netif_msg_ifup(sky2))
> printk(KERN_INFO PFX "%s: enabling interface\n", dev->name);
>
> + netif_start_queue(dev);
> +
> return 0;
>
> err_out:
> @@ -1596,6 +1598,8 @@ static inline int tx_inuse(const struct sky2_port *sky2)
> /* Number of list elements available for next tx */
> static inline int tx_avail(const struct sky2_port *sky2)
> {
> + if (unlikely(!sky2->tx_ring))
> + return 0;
> return sky2->tx_pending - tx_inuse(sky2);
> }
>
> @@ -1925,7 +1929,9 @@ static int sky2_down(struct net_device *dev)
> sky2_read32(hw, B0_IMSK);
>
> synchronize_irq(hw->pdev->irq);
> - napi_synchronize(&hw->napi);
> + netif_tx_lock_bh(dev);
> + napi_disable(&hw->napi);
> + netif_stop_queue(dev);
>
> spin_lock_bh(&sky2->phy_lock);
> sky2_phy_power_down(hw, port);
> @@ -1939,6 +1945,8 @@ static int sky2_down(struct net_device *dev)
> sky2_rx_clean(sky2);
>
> sky2_free_buffers(sky2);
> + napi_enable(&hw->napi);
> + netif_tx_unlock_bh(dev);
>
> return 0;
> }
> @@ -3177,9 +3185,6 @@ static void sky2_reset(struct sky2_hw *hw)
> static void sky2_detach(struct net_device *dev)
> {
> if (netif_running(dev)) {
> - netif_tx_lock_bh(dev);
> - netif_device_detach(dev); /* stop txq */
> - netif_tx_unlock_bh(dev);
> sky2_down(dev);
> }
> }
> -- 1.5.6.5
>
Ok - no obvious difference with this patch + Stephen's.
I still see:
No reported errors; decent throughput; earlier issues with IRQ resolved.
But... still seeing DHCP DISCOVER/OFFER (no REQUEST/ACK) while under
load. I can try to sniff this... but given that it's under load, I'd
probably have to filter out non DHCP stuff and might miss whatever is
really going on. Again - main point here is that absent these patches I
don't see this issue. It's also possible that these two patches allow
the load to be heavier thus actually causing a different problem.
next prev parent reply other threads:[~2010-01-14 16:46 UTC|newest]
Thread overview: 59+ messages / expand[flat|nested] mbox.gz Atom feed top
2010-01-10 22:27 2.6.33-rc3-git3: Reported regressions from 2.6.32 Rafael J. Wysocki
2010-01-10 22:32 ` [Bug #14925] sky2 panic under load Rafael J. Wysocki
2010-01-11 0:36 ` Berck E. Nash
2010-01-11 13:26 ` Jarek Poplawski
2010-01-11 19:32 ` Rafael J. Wysocki
2010-01-11 20:31 ` Jarek Poplawski
2010-01-11 20:50 ` Rafael J. Wysocki
2010-01-11 21:02 ` Berck E. Nash
2010-01-11 21:47 ` Jarek Poplawski
2010-01-11 14:03 ` Mike McCormack
2010-01-11 16:45 ` Stephen Hemminger
2010-01-11 22:07 ` Jarek Poplawski
2010-01-12 0:14 ` David Miller
2010-01-12 7:50 ` Jarek Poplawski
2010-01-12 8:08 ` David Miller
2010-01-12 8:56 ` Jarek Poplawski
2010-01-12 9:42 ` David Miller
2010-01-12 10:31 ` Jarek Poplawski
2010-01-12 10:56 ` David Miller
2010-01-12 11:04 ` Jarek Poplawski
2010-01-12 15:39 ` Stephen Hemminger
2010-01-12 16:15 ` [PATCH] sky2: safer transmit ring cleaning Stephen Hemminger
2010-01-12 16:32 ` Michael Breuer
2010-01-12 17:02 ` Stephen Hemminger
2010-01-12 18:04 ` Jarek Poplawski
2010-01-12 18:13 ` Stephen Hemminger
2010-01-12 18:24 ` Jarek Poplawski
2010-01-12 18:49 ` [PATCH] sky2: safer transmit ring cleaning (v2) Stephen Hemminger
2010-01-12 19:16 ` Jarek Poplawski
2010-01-12 19:23 ` Stephen Hemminger
2010-01-12 19:50 ` Jarek Poplawski
2010-01-13 1:23 ` Stephen Hemminger
2010-01-12 19:34 ` Michael Breuer
2010-01-12 18:35 ` [PATCH] sky2: safer transmit ring cleaning Michael Breuer
2010-01-12 18:42 ` Michael Breuer
2010-01-12 20:31 ` Michael Breuer
2010-01-13 4:10 ` [PATCH] sky2: safer transmit ring cleaning (v3) Stephen Hemminger
2010-01-13 4:31 ` Michael Breuer
2010-01-13 7:35 ` Jarek Poplawski
2010-01-13 16:04 ` Michael Breuer
2010-01-14 3:41 ` [PATCH] sky2: safer transmit ring cleaning (v4) Stephen Hemminger
2010-01-14 10:14 ` Jarek Poplawski
2010-01-14 11:16 ` Jarek Poplawski
2010-01-14 11:20 ` David Miller
2010-01-14 11:26 ` Jarek Poplawski
2010-01-14 13:19 ` Mike McCormack
2010-01-14 15:43 ` Michael Breuer
2010-01-14 16:46 ` Michael Breuer [this message]
2010-01-14 17:51 ` Stephen Hemminger
2010-01-14 17:52 ` Stephen Hemminger
2010-01-14 23:51 ` Michael Breuer
2010-01-16 18:35 ` sky2 DHCPOFFER packet loss under load (Was Re: [PATCH] sky2: safer transmit ring cleaning (v4)) Michael Breuer
2010-01-14 15:46 ` [PATCH] sky2: safer transmit ring cleaning (v4) Michael Breuer
2010-01-11 22:31 ` [Bug #14925] sky2 panic under load Jarek Poplawski
2010-01-11 16:00 ` 2.6.33-rc3-git3: Reported regressions from 2.6.32 Luis R. Rodriguez
2010-01-11 21:47 ` Nick Bowler
2010-01-11 22:10 ` Rafael J. Wysocki
[not found] ` <omZ8aBZReyD.A.rz.COlSLB@chimera>
2010-01-28 23:18 ` [PATCH net-2.6] cdc_ether: Partially revert "usbnet: Set link down initially ..." Ben Hutchings
2010-01-29 5:37 ` David Miller
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=4B4F4A57.5050708@majjas.com \
--to=mbreuer@majjas$(echo .)com \
--cc=davem@davemloft$(echo .)net \
--cc=flyboy@gmail$(echo .)com \
--cc=jarkao2@gmail$(echo .)com \
--cc=mikem@ring3k$(echo .)org \
--cc=netdev@vger$(echo .)kernel.org \
--cc=rjw@sisk$(echo .)pl \
--cc=shemminger@vyatta$(echo .)com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox