sky2 still broken

Message

xanas3712 · Post by **xanas3712** » Sun Apr 22, 2007 5:21 pm

gregf are you using gigabit switch/router or are you on a 10/100? I have the problem of the card *8056 rev12* mysteriously dying if I transfer files over gigabit for a certain amount of time. To fix I have to /etc/init.d/net.eth0 restart, or switch to 100 mode instead of gigabit.

I'm using vanilla 21-r7 now but was previously using gentoo 20-r5

I haven't tried git4 yet.

HeXiLeD · Post by **HeXiLeD** » Sun Apr 22, 2007 5:38 pm

kernel 2.6.21-rc7 changelog

sky2: phy workarounds for Yukon EC-U A1

The workaround Yukon EC-U wasn't comparing with correct
version and wasn't doing correct setup. Without it, 88e8056
throws all sorts of errors.

sky2: turn on clocks when doing resume

Some of these chips are disabled until clock is enabled.
This fixes:
http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=404107

sky2: turn carrier off when down

Driver needs to turn off carrier when down.

working for any of you ?

l_bratch · Post by **l_bratch** » Sun Apr 22, 2007 5:39 pm

2.6.21-rc7 still doesn't work for me, but I'm going to try out -git5 tonight.

dangertools · Post by **dangertools** » Wed Apr 25, 2007 8:25 pm

updates from ndiswrapper:

giri made support for sky2 now fully working, meaning if there are still problems with the native drivers please don't worry to checkout ndiswrapper from svn (only svn code has needed changes) and give it a try.

best regards

Georg Schild

HeXiLeD · Post by **HeXiLeD** » Wed Apr 25, 2007 8:43 pm

are you talking about the wirless nic and drivers or the regular nic ?

dangertools · Post by **dangertools** » Wed Apr 25, 2007 8:48 pm

I'm talking about the wired NIC, driven by the sky2 drivers, ndiswrapper also supports some wired drivers where sky2 is proven to work.

Georg Schild

VoVaN · Post by **VoVaN** » Thu Apr 26, 2007 8:08 am

kernel 2.6.21 is out, but... there's bad news for 88E8056 NIC
from LKLM:

Stephen Hemminger (7):
sky2: disable support for 88E8056
sky2: handle descriptor errors
sky2: disable ASF on all chip types
sky2: EC-U performance and jumbo support
sky2: no jumbo on Yukon FE
sky2: version 1.14
[TCP]: Congestion control initialization.

from sky2.c:

Code: Select all

#ifdef broken
        /* This device causes data corruption problems that are not resolved */
        { PCI_DEVICE(PCI_VENDOR_ID_MARVELL, 0x4364) }, /* 88E8056 */
#endif

So, no kernel support for 88E8056....?

l_bratch · Post by **l_bratch** » Thu Apr 26, 2007 8:12 am

Hmm, that sucks.

On the plus side, 2.6.21-rc7 is going better than I thought it was. It failed after 20 hours the first day of running it, but since then it has been running for another 4 days, and hasn't gone wrong since. Perhaps the first problem was a random other problem?

I basically have had 4 days perfect sky2 behaviour with this card:

Code: Select all

02:00.0 Ethernet controller: Marvell Technology Group Ltd. 88E8053 PCI-E Gigabit Ethernet Controller (rev 20)

VoVaN · Post by **VoVaN** » Thu Apr 26, 2007 8:31 am

l_bratch wrote:Hmm, that sucks.

On the plus side, 2.6.21-rc7 is going better than I thought it was. It failed after 20 hours the first day of running it, but since then it has been running for another 4 days, and hasn't gone wrong since. Perhaps the first problem was a random other problem?

I basically have had 4 days perfect sky2 behaviour with this card:
Code: Select all
02:00.0 Ethernet controller: Marvell Technology Group Ltd. 88E8053 PCI-E Gigabit Ethernet Controller (rev 20)

But, you have the 88E8053 NIC, so it is supported. All the changes in 2.6.21 only concerning 88E8056. Personally I'm using 88E8056 with 2.6.21-rc7 for more then a week and don't have any problems, maybe only so far...

Post by **dsd** » Fri Apr 27, 2007 9:56 pm

if you have an 88E8056 and it worked fine with sky2 on 2.6.20 or newer, then please file a bug at http://bugs.gentoo.org with "lspci -vvxx" output and we will enable it again

Paapaa · Post by **Paapaa** » Fri Apr 27, 2007 10:36 pm

dsd wrote:if you have an 88E8056 and it worked fine with sky2 on 2.6.20 or newer, then please file a bug at http://bugs.gentoo.org with "lspci -vvxx" output and we will enable it again

And here is mine, for the record:

http://bugs.gentoo.org/show_bug.cgi?id=176219

Cyker · Post by **Cyker** » Tue May 01, 2007 9:35 pm

Phrack!

After a few weeks of light use, I hadn't had much trouble.

Today I did a bulk transfer at gigabit speeds and abotu half-way through the damned network stopped working again

I was still getting link lights and both the Linux box and the Windows box that was transferring reported their cards were up and okay.

But no IP traffic passed at all; Just getting "Destination Unreachable".

Annotated Log wrote: --- Logged in and connected ---
May 1 15:48:35 CyServ sky2 eth0: Link is up at 1000 Mbps, full duplex, flow control both
May 1 15:48:35 CyServ br0: port 1(eth0) entering learning state
May 1 15:48:40 CyServ br0: topology change detected, propagating
May 1 15:48:40 CyServ br0: port 1(eth0) entering forwarding state
--- Started transfer about 2 hours before this point ---
--- Link goes a bit funny, feels very slow ---
May 1 20:45:43 CyServ sky2 eth0: Link is down.
May 1 20:45:44 CyServ br0: port 1(eth0) entering disabled state
May 1 20:45:57 CyServ sky2 eth0: Link is up at 1000 Mbps, full duplex, flow control both
May 1 20:45:57 CyServ br0: port 1(eth0) entering learning state
May 1 20:46:02 CyServ br0: topology change detected, propagating
May 1 20:46:02 CyServ br0: port 1(eth0) entering forwarding state
May 1 20:47:21 CyServ sky2 eth0: Link is down.
May 1 20:47:21 CyServ br0: port 1(eth0) entering disabled state
May 1 20:47:38 CyServ sky2 eth0: Link is up at 1000 Mbps, full duplex, flow control both
May 1 20:47:39 CyServ br0: port 1(eth0) entering learning state
May 1 20:47:44 CyServ br0: topology change detected, propagating
May 1 20:47:44 CyServ br0: port 1(eth0) entering forwarding state
May 1 20:50:00 CyServ sky2 eth0: Link is down.
May 1 20:50:00 CyServ br0: port 1(eth0) entering disabled state
May 1 20:50:09 CyServ sky2 eth0: Link is up at 1000 Mbps, full duplex, flow control both
May 1 20:50:09 CyServ br0: port 1(eth0) entering learning state
May 1 20:50:14 CyServ br0: topology change detected, propagating
May 1 20:50:14 CyServ br0: port 1(eth0) entering forwarding state
May 1 20:51:07 CyServ sky2 eth0: Link is down.
May 1 20:51:07 CyServ br0: port 1(eth0) entering disabled state
May 1 20:51:14 CyServ sky2 eth0: Link is up at 1000 Mbps, full duplex, flow control both
May 1 20:51:14 CyServ br0: port 1(eth0) entering learning state
May 1 20:51:19 CyServ br0: topology change detected, propagating
May 1 20:51:19 CyServ br0: port 1(eth0) entering forwarding state
May 1 20:51:36 CyServ sky2 eth0: Link is down.
May 1 20:51:36 CyServ br0: port 1(eth0) entering disabled state
May 1 20:51:38 CyServ sky2 eth0: Link is up at 1000 Mbps, full duplex, flow control both
May 1 20:51:38 CyServ br0: port 1(eth0) entering learning state
May 1 20:51:43 CyServ br0: topology change detected, propagating
May 1 20:51:43 CyServ br0: port 1(eth0) entering forwarding state
May 1 20:51:52 CyServ sky2 eth0: Link is down.
May 1 20:51:36 CyServ br0: port 1(eth0) entering disabled state
May 1 20:51:38 CyServ sky2 eth0: Link is up at 1000 Mbps, full duplex, flow control both
May 1 20:51:38 CyServ br0: port 1(eth0) entering learning state
May 1 20:51:43 CyServ br0: topology change detected, propagating
May 1 20:51:43 CyServ br0: port 1(eth0) entering forwarding state
May 1 20:51:52 CyServ sky2 eth0: Link is down.
May 1 20:51:52 CyServ br0: port 1(eth0) entering disabled state
--- By this point it's not working at all ---
May 1 20:53:45 CyServ sky2 eth0: Link is up at 10 Mbps, half duplex, flow control none
May 1 20:53:45 CyServ br0: port 1(eth0) entering learning state
May 1 20:53:50 CyServ br0: topology change detected, propagating
May 1 20:53:50 CyServ br0: port 1(eth0) entering forwarding state
May 1 20:53:51 CyServ sky2 eth0: Link is down.
May 1 20:53:51 CyServ br0: port 1(eth0) entering disabled state
May 1 20:54:12 CyServ sky2 eth0: Link is up at 1000 Mbps, full duplex, flow control both
May 1 20:54:12 CyServ br0: port 1(eth0) entering learning state
May 1 20:54:17 CyServ br0: topology change detected, propagating
May 1 20:54:17 CyServ br0: port 1(eth0) entering forwarding state
May 1 20:55:02 CyServ sky2 eth0: Link is down.
May 1 20:55:02 CyServ br0: port 1(eth0) entering disabled state
May 1 20:55:04 CyServ sky2 eth0: Link is up at 1000 Mbps, full duplex, flow control both
May 1 20:55:04 CyServ br0: port 1(eth0) entering learning state
May 1 20:55:09 CyServ br0: topology change detected, propagating
May 1 20:55:09 CyServ br0: port 1(eth0) entering forwarding state
May 1 20:57:04 CyServ sky2 eth0: Link is down.
May 1 20:57:04 CyServ br0: port 1(eth0) entering disabled state
May 1 20:58:16 CyServ sky2 eth0: Link is up at 1000 Mbps, full duplex, flow control both
May 1 20:58:16 CyServ br0: port 1(eth0) entering learning state
May 1 20:58:21 CyServ br0: topology change detected, propagating
May 1 20:58:21 CyServ br0: port 1(eth0) entering forwarding state
May 1 20:59:43 CyServ sky2 eth0: Link is down.
May 1 20:59:43 CyServ br0: port 1(eth0) entering disabled state
May 1 21:08:30 CyServ sky2 eth0: Link is up at 1000 Mbps, full duplex, flow control both
May 1 21:08:31 CyServ br0: port 1(eth0) entering learning state
May 1 21:08:36 CyServ br0: topology change detected, propagating
May 1 21:08:36 CyServ br0: port 1(eth0) entering forwarding state
May 1 21:09:56 CyServ br0: port 1(eth0) entering disabled state
--- Console: /etc/init.d/net.eth0 restart ---
May 1 21:09:56 CyServ sky2 eth0: disabling interface
May 1 21:09:56 CyServ sky2 eth0: enabling interface
May 1 21:09:56 CyServ sky2 eth0: ram buffer 48K
May 1 21:09:59 CyServ sky2 eth0: Link is up at 1000 Mbps, full duplex, flow control both
May 1 21:09:59 CyServ br0: port 1(eth0) entering learning state
May 1 21:10:04 CyServ br0: topology change detected, propagating
May 1 21:10:04 CyServ br0: port 1(eth0) entering forwarding state
--- Working again... ---

For reference:
sky2 v1.10 addr 0xdbefc000 irq 16 Yukon-EC (0xb6) rev 2
sky2 eth0: addr 00:18:f3:4a:74:b8

02:00.0 Ethernet controller: Marvell Technology Group Ltd. 88E8053 PCI-E Gigabit Ethernet Controller (rev 15)

Anybody else getting this behaviour?

As I said, it'd been fine until today when I dumped about 78GB of data down the wire via Samba, then suddenly poof! Nothing; Not even a ping packet.
Link light was all there, Yukon diagnostics on the Windows machine checked out okay, reported link as active and functioning, but no packets...

Post by **Ken69267** » Wed May 02, 2007 11:23 pm

Cyker wrote:Phrack!

After a few weeks of light use, I hadn't had much trouble.

Today I did a bulk transfer at gigabit speeds and abotu half-way through the damned network stopped working again

I was still getting link lights and both the Linux box and the Windows box that was transferring reported their cards were up and okay.

But no IP traffic passed at all; Just getting "Destination Unreachable".

For reference:
sky2 v1.10 addr 0xdbefc000 irq 16 Yukon-EC (0xb6) rev 2
sky2 eth0: addr 00:18:f3:4a:74:b8

02:00.0 Ethernet controller: Marvell Technology Group Ltd. 88E8053 PCI-E Gigabit Ethernet Controller (rev 15)

Anybody else getting this behaviour?

As I said, it'd been fine until today when I dumped about 78GB of data down the wire via Samba, then suddenly poof! Nothing; Not even a ping packet.
Link light was all there, Yukon diagnostics on the Windows machine checked out okay, reported link as active and functioning, but no packets...

I get this all the time under heavy load

syslog reveals something sending a SIGTERM to dhcpcd...

(I have the same NIC btw, its on my gigabyte ds3 mobo)

zorth · Post by **zorth** » Tue May 08, 2007 3:26 pm

hi, i have a gigabyte dq6 mother board with the marvel yucom eth card with the same problem. my solution? buy a pci eth card like my new d-link 10/100/1000. she runs fine, any problem like with the marvel yucon integrated in board. i disabled this in bios

a fast eth0 → D-Link System Inc DGE-528T Gigabit Ethernet Adapter

sorry for my english

regards

ningo · Post by **ningo** » Tue May 08, 2007 3:36 pm

OpenBSD's msk works perfectly.

jniklast · Post by **jniklast** » Tue May 08, 2007 3:41 pm

So does Marvell's sk98lin driver...
But that's not what this thread is about!

ningo · Post by **ningo** » Tue May 08, 2007 3:44 pm

Lies! This thread is about Marvell Yukon 2 chips!

jniklast · Post by **jniklast** » Tue May 08, 2007 3:47 pm

Really? I thought the title read "sky2 still broken"...
Anyway, you don't even own that chip. So what do you want here, apart from flaming?

ningo · Post by **ningo** » Tue May 08, 2007 3:50 pm

I'm not flaming, I just give helpful remarks, since sky is obviously in an unusable state.
And who told you that I don't own that chip?!

jniklast · Post by **jniklast** » Tue May 08, 2007 4:04 pm

Well you told it yourself, right here by posting your dmesg: http://www.bsdforen.de/showthread.php?p ... post118546
Your Marvell chip would use the skge module and is clearly no Yukon 2...
As you obviously made your point by telling us you like OpenBSD more than Linux, you can now leave this thread, unless you have useful information. I for my part won't answer any more of your silly posts...

ningo · Post by **ningo** » Tue May 08, 2007 4:10 pm

jniklast wrote:Well you told it yourself, right here by posting your dmesg: http://www.bsdforen.de/showthread.php?p ... post118546
Your Marvell chip would use the skge module and is clearly no Yukon 2...

That's my PC, but I actually own more than one Computer. (As you are very well aware of!)

As you obviously made your point by telling us you like OpenBSD more than Linux, you can now leave this thread, unless you have useful information. I for my part won't answer any more of your silly posts...

Blatant lies! But what to expect from a hardcore gentoo fanboi?

Earthwings · Post by **Earthwings** » Tue May 08, 2007 8:56 pm

Please stay calm and friendly, folks. Thanks.

l_bratch · Post by **l_bratch** » Tue May 08, 2007 10:11 pm

l_bratch wrote:Hmm, that sucks.

On the plus side, 2.6.21-rc7 is going better than I thought it was. It failed after 20 hours the first day of running it, but since then it has been running for another 4 days, and hasn't gone wrong since. Perhaps the first problem was a random other problem?

I basically have had 4 days perfect sky2 behaviour with this card:
Code: Select all
02:00.0 Ethernet controller: Marvell Technology Group Ltd. 88E8053 PCI-E Gigabit Ethernet Controller (rev 20)

Sadly this wasn't the case - it failed after 14 days uptime.

Cyker · Post by **Cyker** » Tue May 08, 2007 10:30 pm

What are you guys using, 10, 100 or 1000?

I have to say, on mine it's been pretty good so far - The only time it bugs out is when I hit it with a sustained bi-directional saturation of data, but then getting the damned thing back up is a PITA.

Most of the time, I can connect via the other NIC (The nVidia one - Stable as a rock!) and net.eth0 restart the NIC to get it back up.

On rare occasions (Well, once so far), I've had to power cycle the box to get the NIC back up....

l_bratch · Post by **l_bratch** » Tue May 08, 2007 10:32 pm

I'm using 100 (see above post for results).

Code: Select all

# ethtool eth0
Settings for eth0:
        Supported ports: [ TP ]
        Supported link modes:   10baseT/Half 10baseT/Full 
                                100baseT/Half 100baseT/Full 
                                1000baseT/Half 1000baseT/Full 
        Supports auto-negotiation: Yes
        Advertised link modes:  10baseT/Half 10baseT/Full 
                                100baseT/Half 100baseT/Full 
                                1000baseT/Half 1000baseT/Full 
        Advertised auto-negotiation: Yes
        Speed: 100Mb/s
        Duplex: Full
        Port: Twisted Pair
        PHYAD: 0
        Transceiver: internal
        Auto-negotiation: on
        Supports Wake-on: pg
        Wake-on: d
        Current message level: 0x000000ff (255)
        Link detected: yes