The current LiveCD does *not* include the sk98lin driver. This is actually maked as DEPRECATED by the gentoo tree of kernels. However, this driver (The one syskonnect provides, version 8.12 or newer) is required to initialise all Marvell Sk98xx Sk95xx PCIe Gigabit LOM and Gigabit adapters. the skge driver currently does not support any PCIe adapters, nor does the kernel-proved sk98lin. Reproducible: Always Steps to Reproduce: 1. Boot into the LiveCD 2. Attempt to initialise your network card - using the skge driver. 3. Try to load the sk98lin driver. Actual Results: The card was not even found by the skge driver. Expected Results: Using the marvel driver, the card is found, brought up, and then data about it's status is printed to /proc/net/sk98lin. The easiest thing to do, is to go to www.syskonnect.com, go to drivers, and download the Enhanced 8.1x series driver for Linux 2.6/2.4 from their page, then use it to patch the kernel sources and build sk98lin as a module.
Honestly, this is an upstream kernel bug. We would only want to patch the driver version ourselves if the driver was refused upstream by the kernel.org developers. Please file this bug at http://bugzilla.kernel.org/ instead. If they refuse to patch it before then end of June, then REOPEN this bug, and we'll patch it ourselves, so we have it in time for the next release.
Created attachment 60156 [details, diff] Sk98Lin 8.16 patch for 2.6.11 gentoo-r9
As of 2.6.11.11 (May 30, 2005), the Kernel people are REFUSING the sk98lin patch. The included file is the Integrated Marvell Adapter driver for 2.6.11 gentoo-r9. However, the file changes the patch preforms should work for all 2.6.x kernels, Marvell is using a monolithic architecture. This driver supports all previous generation sk98xx cards (PCI flavour), sk95xx cards, and the new generation Yukon Sk98xx PCIe cards.
What is "current LiveCD" in the original bug description? I'd also state this URL where the latest driver can be found: http://www.syskonnect.de/syskonnect/support/driver/ge.htm Maybe this should become the URL of this bug? I found no sk98lin driver on the 2005.0 minimal cd, would be nice for 2005.1.
Kernel guys... what version is the SK98lin and SKGE drivers? Do we need to patch ours before the next release?
The version of sk98lin included in gentoo-sources is the one included in the mainline kernel. The sk98lin patch posted here was rejected upstream so we won't be including that. I think that the mess created by that patch was the reason why skge came into existance. I'll write a mail to the netdev list asking about the extended hardware support which the new sk98lin patch seems to provide, I wasn't aware of that.
Arturo: Perhaps if you attach your "lspci -vv" output we could try modifying skge to see if it can support your card already.
Created attachment 62071 [details] info about 88E8053: lspci -vv, lspci -n and /proc/net/sk98lin I did my installation without network, so for me personally this was not a livecd issue, but I'm dependent on the extended sk98lin as well. I've got a 88E8053 on my ASUS P5GDC-V mainboard.
Created attachment 62076 [details, diff] Patch to support 88E8053 Thanks Martin. Could you please try this patch and see if the skge driver will then work for you? (Please post dmesg output after trying it) Arturo: Forgot to mention, as well as the "lspci -vv" output, please also post the "lspci -n" output like Martin has done.
Here is my LSPCI -vv output; this is an intel 925CXV Mainboard 0000:04:00.0 Ethernet controller: Marvell Technology Group Ltd.: Unknown device 4361 (rev 17) Subsystem: Intel Corp.: Unknown device 3063 Control: I/O+ Mem+ BusMaster+ SpecCycle- MemWINV- VGASnoop- ParErr- Stepping- SERR+ FastB2B- Status: Cap+ 66Mhz- UDF- FastB2B- ParErr- DEVSEL=fast >TAbort- <TAbort- <MAbort- >SERR+ <PERR- Latency: 0, cache line size 10 Interrupt: pin A routed to IRQ 17 Region 0: Memory at fe8fc000 (64-bit, non-prefetchable) [size=16K] Region 2: I/O ports at a800 [size=256] Expansion ROM at fe8c0000 [disabled] [size=128K] Capabilities: [48] Power Management version 2 Flags: PMEClk- DSI- D1+ D2+ AuxCurrent=0mA PME(D0+,D1+,D2+,D3hot+,D3cold+) Status: D0 PME-Enable- DSel=0 DScale=1 PME- Capabilities: [50] Vital Product Data Capabilities: [5c] Message Signalled Interrupts: 64bit+ Queue=0/1 Enable- Address: 0000000000000000 Data: 0000 Capabilities: [e0] #10 [0011] And I was unaware of sk98lin being such a huge mess? I thought it was a good driver...
The patch you posted here is 1.2 _megabytes_. Thats absolutely insane, and its only a patch, let alone an entire driver! Thats enough to indicate that there is something wrong with the sk98lin driver. Harald Welte was kind enough to reply to my email. He points out that the sk98lin driver patch posted here supports Yukon2 cards (in a very ugly manner) whereas skge does not. Here's his response: -------- >The user here reports that skge does not support any of the PCI Express >Marvell cards. yes, this is true. If you look at the sk98 driver, it actually supports three types of boards: Genesis, Yukon, and Yukon II. Each of them has a quite different programming model. Genesis and Yukon have more similarity to each other than Yukon II. skge currently supports Genesis and Yukon, but not Yukon II. I thought it was quite easy to add Yukon II to the driver and started coding only to find out that in fact, it is not easy. If you look at the sk98lin driver, it always has _huge_ different blocks "if (!CHIP_ID_YUKON_2()) else ...". Most importantly, a number of functions are completely different and set up via function pointers at initialization time (see sky2le.c). So I stopped and actually though that it might be better to have a totally seperate driver for yukon2, rather than bloating the neat skge driver. Unfortunately I don't have the time to do that :( -------- So, we won't be supporting the PCI-express Yukon-II cards at this point in time. Sorry for any inconvenience.
I know it's absurd, but you simply cannot choose not to support Marvell PCIe Yukons! They are commonplace in almost all of the modern PCIe boards, in fact, i did a bit of sniffing inside my kernel, and the Sk98lin is HARDLY, as massive as he patch itself: Lsmod output: Module Size Used by vmnet 33956 7 vmmon 174444 0 pcspkr 4044 0 i2c_i801 9100 0 i2c_core 22784 1 i2c_i801 snd_seq_midi 9376 0 snd_emu10k1_synth 8192 0 snd_emux_synth 39168 1 snd_emu10k1_synth snd_seq_virmidi 8576 1 snd_emux_synth snd_seq_midi_emul 7808 1 snd_emux_synth snd_emu10k1 119428 3 snd_emu10k1_synth snd_rawmidi 26528 3 snd_seq_midi,snd_seq_virmidi,snd_emu10k1 snd_ac97_codec 83576 1 snd_emu10k1 snd_util_mem 4864 2 snd_emux_synth,snd_emu10k1 snd_hwdep 9888 2 snd_emux_synth,snd_emu10k1 nvidia 3713412 20 agpgart 36944 1 nvidia sk98lin 209164 1 And YukonII will only become increasingly more common as PCIe takes off.
And, ahem... sorry, I forgot. Here's the lspci - n dump 0000:00:00.0 Class 0600: 8086:2584 (rev 04) 0000:00:01.0 Class 0604: 8086:2585 (rev 04) 0000:00:1c.0 Class 0604: 8086:2660 (rev 03) 0000:00:1c.1 Class 0604: 8086:2662 (rev 03) 0000:00:1c.2 Class 0604: 8086:2664 (rev 03) 0000:00:1c.3 Class 0604: 8086:2666 (rev 03) 0000:00:1d.0 Class 0c03: 8086:2658 (rev 03) 0000:00:1d.1 Class 0c03: 8086:2659 (rev 03) 0000:00:1d.2 Class 0c03: 8086:265a (rev 03) 0000:00:1d.3 Class 0c03: 8086:265b (rev 03) 0000:00:1d.7 Class 0c03: 8086:265c (rev 03) 0000:00:1e.0 Class 0604: 8086:244e (rev d3) 0000:00:1f.0 Class 0601: 8086:2640 (rev 03) 0000:00:1f.1 Class 0101: 8086:266f (rev 03) 0000:00:1f.2 Class 0101: 8086:2652 (rev 03) 0000:00:1f.3 Class 0c05: 8086:266a (rev 03) 0000:01:00.0 Class 0300: 10de:00f9 (rev a2) 0000:04:00.0 Class 0200: 11ab:4361 (rev 17) 0000:06:02.0 Class 0401: 1102:0004 (rev 04) 0000:06:02.1 Class 0980: 1102:7003 (rev 04) 0000:06:02.2 Class 0c00: 1102:4001 (rev 04) 0000:06:05.0 Class 0c00: 11c1:5811 (rev 61)
Sorry, I'm already adding a lot of network and storage support to the kernel for the next LiveCD, but we draw the line here. Stefan is going to look into providing unofficial precompiled module files for the new sk98lin on his website. If it works out, you'll be able to save these on floppy, boot into livecd, mount floppy, and use insmod to load the module.
ok, i created an ebuild for this driver. Can you pelase try it and report back? http://dev.gentoo.org/~genstef/files/overlay/sk98lin-8.16.2.3.ebuild Oh and there is a beta for those of you who feel lucky ;) http://dev.gentoo.org/~genstef/files/overlay/sk98lin-8.18.2.2_beta.ebuild have fun
Thank you for the Ebuild. I will try it when I get home. A word to the wise: NAPI And PCIe don't get along.
Version 8.23 is available, just renaming the ebuild works. Are you still interested in this one?
Just simply add both. Marvell's driver is Very fineky indeed. The PCIe intel ones (AKA Gallileo Technology) Work better with 8.23 than 8.16, but others may not. (Yes, This is why I know SKGE was a better bet for all you PCI users)
Created attachment 73711 [details] lspci Wonderful. Using this ebuild, my asus A8V Deluxe (http://www.asus.it/products4.aspx?modelmenu=2&model=238&l1=3&l2=15&l3=68) works fine. I write this because the standard driver on gentoo-sources gives instead a lot of problems, and it is impossible to use nfs for ltsp with it. I hope the patch be included in gentoo-sources.
Then file a bug against gentoo-sources. As it stands, this vendor driver will not be included in the kernel or in portage. And this bug is about the Yukon-2 range (PCI express adapters), whereas yours is one of the original Yukon PCI adapters.
I am talking about the YukonII Cards. Which, by the way are driven by a GPLv2 Driver. I really do not understand what the problem is with including sk98lin for YukonII, obiously skge does not work with them, and it's a bit hard to download the driver when your card is unable to be initialised!
The new sk98lin was rejected by the network driver maintainers and we tend to stick with their decisions. Just because its GPL doesn't mean everything, the code is horrible. Yes, I understand that your situation is different with the Yukon-II's being currently completely unsupported and to rectify this I plan to start shipping a new, clean, actively developed open-source driver for these cards as of 2.6.15 (it will hopefully be included in the mainline kernel shortly after). The ultimate target is to get these completely supported for 2006.0, so thanks for your patience :)
Well, if we are going to get a nice new driver for the YukonII I would certainly stop complaning. I do not know what's wrong with sk98lin - I tend not to look into the code when it's just a YouNeedtorunThis(tm) module, if you understand what I mean. :) I really hope that 2.6.15 fixes a lot of things, YukonII would be another one. :) Thanks
sky2 is included in gentoo-sources-2.6.15, please give it a try.
The sky2 driver is working fine for me under gentoo-sources-2.6.15. Additionaly, previously, this driver was working fine for me under mm-sources-2.6.15_rc1-r2.
Sky2 works with my Intel 925XCV Mainboard. Check my lspci if you need more info.
Another "work's for me" for the Marvell Yukon 88E8053 on my ASUS P5GDC-V Mainboard. Thank's a lot! I hope this will also be included in the live CD.
Now that 2006.0 is out the door, this should be resolved.
