You are not logged in.
My internal NIC mcp61 has been acting up on the two latest ck kernels, It takes forever to open up a NFS share and surfing is slow too. Filetransfers go down to 10 Mbit/sec from 50-60 Mbit/sec. I have had this before on some standard Arch kernels but then went by the wiki and google and found some options to the forcedeth module (modprobe.conf);
options forcedeth msi=0 msix=0
I've tried loading forcedeth from rc.conf and not loading it from there. I've tried with and without module options.
I also tried every possible setting with ethtool and mii-tool but this NIC doesn't respond at all to commands from these tools.
From logs;
Jun 6 08:46:57 localhost kernel: [ 5.719089] forcedeth: Reverse Engineered nForce ethernet driver. Version 0.64.
Jun 6 08:46:57 localhost kernel: [ 5.719243] forcedeth 0000:00:07.0: PCI INT A -> Link[LMAC] -> GSI 21 (level, low) -> IRQ 21
Jun 6 08:46:57 localhost kernel: [ 5.777490] forcedeth 0000:00:07.0: eth1: Features changed: 0x00004820 -> 0x00004020
Jun 6 08:46:57 localhost kernel: [ 5.777608] forcedeth 0000:00:07.0: ifname eth1, PHY OUI 0x732 @ 1, addr bc:ae:c5:95:53:ba
Jun 6 08:46:57 localhost kernel: [ 5.777611] forcedeth 0000:00:07.0: highdma pwrctl mgmt gbit lnktim desc-v3
But now it's all too slow on kernels kernel26-ck-k8 1:2.6.38.7-1 and 1:2.6.39-10.
For the moment I'm using a Realtech r8169 NIC wich doesn't have these particular slowdowns but poorer filetransfer speed, 20-30 Mbit/sec compared to the MCP61's 50-60 Mbit/sec, when working.
Questions:
Could it be the latest udev update? upgraded udev (168-2 -> 171-1)
Has anyone experienced the same problems with forcedeth?
Offline
More info:
modinfo forcedeth
filename: /lib/modules/2.6.39-ck/kernel/drivers/net/forcedeth.ko.gz
license: GPL
description: Reverse Engineered nForce ethernet driver
author: Manfred Spraul <manfred@colorfullife.com>
alias: pci:v000010DEd00000D7Dsv*sd*bc*sc*i*
...
alias: pci:v000010DEd000001C3sv*sd*bc*sc*i*
depends:
vermagic: 2.6.39-ck SMP preempt mod_unload K8
parm: max_interrupt_work:forcedeth maximum events handled per interrupt (int)
parm: optimization_mode:In throughput mode (0), every tx & rx packet will generate an interrupt. In CPU mode (1), interrupts are controlled by a timer. In dynamic mode (2), the mode toggles between throughput and CPU mode based on network load. (int)
parm: poll_interval:Interval determines how frequent timer interrupt is generated by [(time_in_micro_secs * 100) / (2^10)]. Min is 0 and Max is 65535. (int)
parm: msi:MSI interrupts are enabled by setting to 1 and disabled by setting to 0. (int)
parm: msix:MSIX interrupts are enabled by setting to 1 and disabled by setting to 0. (int)
parm: dma_64bit:High DMA is enabled by setting to 1 and disabled by setting to 0. (int)
parm: phy_cross:Phy crossover detection for Realtek 8201 phy is enabled by setting to 1 and disabled by setting to 0. (int)
parm: phy_power_down:Power down phy and disable link when interface is down (1), or leave phy powered up (0). (int)
Same problem today with latest kernel 2.6.39.1-1 ck. Nothing in the logs whatsoever, just the above post.
EDIT: code tags
Last edited by swanson (2011-06-09 11:29:16)
Offline
No betterment today after udev update ( upgraded udev (171-1 -> 171-2) )
Offline
@swanson - have you tried rolling back udev to test that idea? I have no experience with forcedeth, sorry. Is it okay in the 2.6.39.1 ARCH kernel?
CPU-optimized Linux-ck packages @ Repo-ck • AUR packages • Zsh and other configs
Offline
Has anyone experienced the same problems with forcedeth?
Yes.
Using the CK kernel from the AUR, the WONDERFUL repo, and last ditch effort applying the patches myself to the ABS kernel...
-Every 20-30 seconds there is a pause (mouse and keyboard)
-Video "stutters"
-File transfer is at normal speed, but "connections" are delayed (ping shows 0ms to 60ms between me and www.google.com, but even ping "hangs" for 2-3 seconds before starting)
I do not see this behavior with the stock kernel, nor with compiling my own (from ABS).
It has bothered me for quite some time, as I read time after time how the CK patch set is wonderful for desktops. With CFS-CFQ, BFS-BFQ, CFS-Deadline... same result. Without the CK patch set, it all works great. Blah, I wanted to be a cool kid too! Oh well...
There are only two ways to live your life: One is as though nothing is a miracle. The other is as though everything is a miracle. - Albert Einstein
How wonderful it is that nobody need wait a single moment before starting to improve the world. - Anne Frank
Offline
You guys really need to report this to CK via the email address on his blog. It sounds like it's definitely related to the BFS.
CPU-optimized Linux-ck packages @ Repo-ck • AUR packages • Zsh and other configs
Offline
pppoe+forcedeth works ok so far with kernel26-bfs 2.6.39.1, 6MB/s like always.
Arch64/DWM || My Dropbox referral link
Offline
I'll report it on the blog. Overall the ck kernel and graysky's compiling and packageing are great! I really do get extra snappiness on my AMD k8 system. And the forcedeth is known to act up sometimes, it seems. It did even on the standard ARCH kernel for me.
EDIT:
By the way, downgraded udev to 167.2 but no improvement. Do not downgrade udev!
Ok now with ck-k10 kernel 2.6.39.1-3 it seems. Monitoring this carefully but thinking of getting a Intel Pro 1000 GT NIC.
Last edited by swanson (2011-06-12 08:11:54)
Offline