netdev.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
From: Chris Boot <bootc@bootc.net>
To: Nix <nix@esperi.org.uk>
Cc: e1000-devel@lists.sourceforge.net, netdev@vger.kernel.org,
	Jesse Brandeburg <jesse.brandeburg@intel.com>,
	linux-kernel@vger.kernel.org
Subject: Re: [PATCH RFC 0/2] e1000e: 82574 also needs ASPM L1 completely disabled
Date: Sun, 29 Apr 2012 19:03:52 +0100	[thread overview]
Message-ID: <4F9D8288.3000001@bootc.net> (raw)
In-Reply-To: <87y5pe1o89.fsf@spindle.srvr.nix>

On 29/04/2012 17:45, Nix wrote:
> On 24 Apr 2012, Jesse Brandeburg outgrape:
>
>> Please let us know the results of your testing, we will let you know if
>> we see any issues as well.

Right, I have finally managed to test my patch on my servers. I've had a 
really tough week with them due to my cluster falling over inexplicably 
so I didn't want to change too much too soon after everything came back up.

The patch does properly disable ASPM L1 as well as L0s as before. Unlike 
for Nix, these do remain disabled. I'll keep running with the patch now 
but I'm confident this will solve my NIC lockups just as Nix's setpci 
incantations did.

Please apply the patches. I'd also really like to have them CCed to 
stable so that Debian will pick them up in time.

> Alas, it has no effect at all here; L0s and L1 claim to be being
> disabled at boot time, but if you ask with lspci you see that they are
> not. I strongly suspect that they *are* being disabled, but then get
> re-enabled by something else, because even if I force them off with
> setpci in the boot scripts, by the time the scripts have finished
> executing and I've got to a root prompt where I can run setpci, L0s and
> L1 are always back on again.

Indeed our troubles must be different. My patch definitely disables ASPM 
fully on the NIC and the upstream device as evidenced by lspci.

Here are extracts from the boot logs and lspci before my patch:

[    3.305372] e1000e: Intel(R) PRO/1000 Network Driver - 1.5.1-k
[    3.317015] e1000e: Copyright(c) 1999 - 2011 Intel Corporation.
[    3.328436] e1000e 0000:00:19.0: PCI INT A -> GSI 20 (level, low) -> 
IRQ 20
[    3.328482] e1000e 0000:00:19.0: setting latency timer to 64
[    3.329493] e1000e 0000:00:19.0: irq 45 for MSI/MSI-X
[    3.679153] e1000e 0000:00:19.0: eth1: (PCI Express:2.5GT/s:Width x1) 
00:25:90:56:ac:d1
[    3.691391] e1000e 0000:00:19.0: eth1: Intel(R) PRO/1000 Network 
Connection
[    3.703689] e1000e 0000:00:19.0: eth1: MAC: 10, PHY: 11, PBA No: 
FFFFFF-0FF
[    3.715639] e1000e 0000:05:00.0: Disabling ASPM L0s
[    4.156806] e1000e 0000:05:00.0: PCI INT A -> GSI 16 (level, low) -> 
IRQ 16
[    4.371659] e1000e 0000:05:00.0: setting latency timer to 64
[    4.371928] e1000e 0000:05:00.0: irq 65 for MSI/MSI-X
[    4.371933] e1000e 0000:05:00.0: irq 66 for MSI/MSI-X
[    4.371937] e1000e 0000:05:00.0: irq 67 for MSI/MSI-X
[    4.485505] e1000e 0000:05:00.0: eth3: (PCI Express:2.5GT/s:Width x1) 
00:25:90:56:ac:d0
[    4.485507] e1000e 0000:05:00.0: eth3: Intel(R) PRO/1000 Network 
Connection
[    4.485647] e1000e 0000:05:00.0: eth3: MAC: 3, PHY: 8, PBA No: FFFFFF-0FF
[   14.237551] e1000e 0000:00:19.0: irq 45 for MSI/MSI-X
[   14.293193] e1000e 0000:00:19.0: irq 45 for MSI/MSI-X
[   16.160177] e1000e: eth2 NIC Link is Up 100 Mbps Full Duplex, Flow 
Control: None
[   16.174293] e1000e 0000:05:00.0: eth2: 10/100 speed: disabling TSO

tidyup ~ # lspci -vvv -s 05:00.0 | grep ASPM
                 LnkCap: Port #0, Speed 2.5GT/s, Width x1, ASPM L0s L1, 
Latency L0 <128ns, L1 <64us
                 LnkCtl: ASPM L1 Enabled; RCB 64 bytes Disabled- 
Retrain- CommClk+
tidyup ~ # lspci -vvv -s 00:1c.4 | grep ASPM
                 LnkCap: Port #5, Speed 5GT/s, Width x1, ASPM L0s L1, 
Latency L0 <512ns, L1 <4us
                 LnkCtl: ASPM L1 Enabled; RCB 64 bytes Disabled- 
Retrain- CommClk+

And now the same kernel with the patch applied:

[    3.310165] e1000e: Intel(R) PRO/1000 Network Driver - 1.5.1-k
[    3.321625] e1000e: Copyright(c) 1999 - 2011 Intel Corporation.
[    3.332996] e1000e 0000:00:19.0: PCI INT A -> GSI 20 (level, low) -> 
IRQ 20
[    3.413898] e1000e 0000:00:19.0: setting latency timer to 64
[    3.426699] e1000e 0000:00:19.0: irq 54 for MSI/MSI-X
[    3.731112] e1000e 0000:00:19.0: eth2: (PCI Express:2.5GT/s:Width x1) 
00:25:90:56:ac:d1
[    3.743437] e1000e 0000:00:19.0: eth2: Intel(R) PRO/1000 Network 
Connection
[    3.755918] e1000e 0000:00:19.0: eth2: MAC: 10, PHY: 11, PBA No: 
FFFFFF-0FF
[    3.768758] e1000e 0000:05:00.0: Disabling ASPM L0s L1
[    3.794095] e1000e 0000:05:00.0: PCI INT A -> GSI 16 (level, low) -> 
IRQ 16
[    3.794178] e1000e 0000:05:00.0: setting latency timer to 64
[    3.795074] e1000e 0000:05:00.0: irq 64 for MSI/MSI-X
[    3.795088] e1000e 0000:05:00.0: irq 65 for MSI/MSI-X
[    3.795107] e1000e 0000:05:00.0: irq 66 for MSI/MSI-X
[    3.912691] e1000e 0000:05:00.0: eth3: (PCI Express:2.5GT/s:Width x1) 
00:25:90:56:ac:d0
[    3.912693] e1000e 0000:05:00.0: eth3: Intel(R) PRO/1000 Network 
Connection
[    3.912842] e1000e 0000:05:00.0: eth3: MAC: 3, PHY: 8, PBA No: FFFFFF-0FF
[   14.454955] e1000e 0000:00:19.0: irq 54 for MSI/MSI-X
[   14.507724] e1000e 0000:00:19.0: irq 54 for MSI/MSI-X
[   15.944706] e1000e: eth2 NIC Link is Up 100 Mbps Full Duplex, Flow 
Control: None
[   15.956279] e1000e 0000:05:00.0: eth2: 10/100 speed: disabling TSO

tidyup ~ # lspci -vvv -s 05:00.0 | grep ASPM
                 LnkCap: Port #0, Speed 2.5GT/s, Width x1, ASPM L0s L1, 
Latency L0 <128ns, L1 <64us
                 LnkCtl: ASPM Disabled; RCB 64 bytes Disabled- Retrain- 
CommClk+
tidyup ~ # lspci -vvv -s 00:1c.4 | grep ASPM
                 LnkCap: Port #5, Speed 5GT/s, Width x1, ASPM L0s L1, 
Latency L0 <512ns, L1 <4us
                 LnkCtl: ASPM Disabled; RCB 64 bytes Disabled- Retrain- 
CommClk+

Cheers,
Chris

-- 
Chris Boot
bootc@bootc.net


------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and 
threat landscape has changed and how IT managers can respond. Discussions 
will include endpoint security, mobile security and the latest in malware 
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
_______________________________________________
E1000-devel mailing list
E1000-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/e1000-devel
To learn more about Intel&#174; Ethernet, visit http://communities.intel.com/community/wired

      reply	other threads:[~2012-04-29 18:03 UTC|newest]

Thread overview: 32+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2011-12-27 22:01 e1000e interface hang on 82574L Chris Boot
2011-12-27 22:33 ` Dave Taht
2011-12-31  9:31 ` Chris Boot
2012-01-03  0:02   ` Wyborny, Carolyn
2012-01-04 17:12     ` Chris Boot
2012-01-15 11:10       ` Chris Boot
2012-01-16 15:56         ` Wyborny, Carolyn
2012-01-16 16:04           ` Chris Boot
2012-03-17 15:59             ` Chris Boot
2012-03-17 17:54               ` Chris Boot
2012-03-17 23:50                 ` Nix
2012-03-19 14:59                 ` Wyborny, Carolyn
2012-03-19 16:19                   ` Nix
2012-03-19 16:29                     ` [E1000-devel] " Wyborny, Carolyn
2012-03-19 17:31                       ` Nix
2012-04-06 10:17                         ` Chris Boot
2012-04-06 12:12                           ` Bjorn Helgaas
2012-04-06 13:41                             ` Henrique de Moraes Holschuh
2012-04-06 13:48                               ` Chris Boot
2012-04-06 16:05                               ` Nix
2012-04-06 16:04                             ` Nix
2012-04-23 21:29                   ` [PATCH RFC 0/2] e1000e: 82574 also needs ASPM L1 completely disabled Chris Boot
2012-04-23 21:29                     ` [PATCH 1/2] e1000e: Disable ASPM L1 on 82574 Chris Boot
2012-04-23 23:18                       ` Jeff Kirsher
2012-04-24 11:08                       ` Nix
2012-06-01 21:17                       ` Chris Boot
2012-06-07  1:41                         ` Greg KH
2012-04-23 21:29                     ` [PATCH 2/2] e1000e: Remove special case for 82573/82574 ASPM L1 disablement Chris Boot
2012-04-23 23:18                       ` Jeff Kirsher
2012-04-23 23:11                     ` [PATCH RFC 0/2] e1000e: 82574 also needs ASPM L1 completely disabled Jesse Brandeburg
2012-04-29 16:45                       ` Nix
2012-04-29 18:03                         ` Chris Boot [this message]

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=4F9D8288.3000001@bootc.net \
    --to=bootc@bootc.net \
    --cc=e1000-devel@lists.sourceforge.net \
    --cc=jesse.brandeburg@intel.com \
    --cc=linux-kernel@vger.kernel.org \
    --cc=netdev@vger.kernel.org \
    --cc=nix@esperi.org.uk \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
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;
as well as URLs for NNTP newsgroup(s).