...making Linux just a little more fun!
MNZ [mnzaki at gmail.com]
Hi TAG, I have recently bought a NIC that uses the rtl8139 chip. I can't seem to get it working. I load the 8139too module but I still don't get /dev/eth0. I'm not sure how to go about fixing this so any help would be appreciated.
-- //MNZ\\
René Pfeiffer [lynx at luchs.at]
On Feb 07, 2008 at 1347 +0200, MNZ appeared and said:
> Hi TAG, > I have recently bought a NIC that uses the rtl8139 chip. I can't seem > to get it working. I load the 8139too module but I still don't get > /dev/eth0. I'm not sure how to go about fixing this so any help > would be appreciated.
Can you send the output of lspci and the kernel version your are using? The 8139too module handles a couple of chips.
Why do you check "/dev/eth0"? The device should show up by checking with "ifconfig" or "ip link show".
Best, René.
MNZ [mnzaki at gmail.com]
On Feb 7, 2008 2:29 PM, Ren? Pfeiffer <lynx@luchs.at> wrote:
> Can you send the output of lspci and the kernel version your are using? > The 8139too module handles a couple of chips.
Kernel:
Linux mnzbox 2.6.19.1-rt15 #24 SMP PREEMPT Wed Feb 6 22:26:15 EET 2008 i686 pentium4 i386 GNU/Linux lspci identifies the NIC as: 02:03.0 Ethernet controller: Hangzhou Silan Microelectronics Co., Ltd.: Unknown device 2031 (rev 01)
Hangzhou whatever, I'm sure it's an RTL8139 chip. Also wind0ze picks it up as a Realtech RTL8139 Card.
> Why do you check "/dev/eth0"? The device should show up by checking with > "ifconfig" or "ip link show".
Well, ip link show only picks up the loopback.
-- //MNZ\\
René Pfeiffer [lynx at luchs.at]
On Feb 07, 2008 at 1450 +0200, MNZ appeared and said:
> On Feb 7, 2008 2:29 PM, René Pfeiffer <lynx@luchs.at> wrote: > > Can you send the output of lspci and the kernel version your are using? > > The 8139too module handles a couple of chips. > > Kernel: > Linux mnzbox 2.6.19.1-rt15 #24 SMP PREEMPT Wed Feb 6 22:26:15 > EET 2008 i686 pentium4 i386 GNU/Linux > > lspci identifies the NIC as: > 02:03.0 Ethernet controller: Hangzhou Silan Microelectronics Co., Ltd.: > Unknown device 2031 (rev 01)
It seems the kernel doesn't know its PCI ID.
> Hangzhou whatever, I'm sure it's an RTL8139 chip. Also wind0ze picks > it up as a Realtech RTL8139 Card.
Yes, Windows does, but Linux apparently not. I just found this: http://vishalmanohar.wordpress.com/2007/07/03/trouble-with-fake-lan-card/
It seems to be a 8139D chipset: http://www.mail-archive.com/netdev@vger.kernel.org/msg27232.html
I've never heard of this problem before, but I do know that=20
> > Why do you check "/dev/eth0"? The device should show up by checking with > > "ifconfig" or "ip link show". >=20 > Well, ip link show only picks up the loopback.
Ok, then the driver doesn't initialise it. I was confused by
"/dev/eth0". Accessing everything by file works only with Plan 9. 
Best, René.
MNZ [mnzaki at gmail.com]
On Feb 7, 2008 3:06 PM, Ren? Pfeiffer <lynx@luchs.at> wrote:
> On Feb 07, 2008 at 1450 +0200, MNZ appeared and said: > > On Feb 7, 2008 2:29 PM, Ren? Pfeiffer <lynx@luchs.at> wrote: > > > Can you send the output of lspci and the kernel version your are using? > > > The 8139too module handles a couple of chips. > > > > Kernel: > > Linux mnzbox 2.6.19.1-rt15 #24 SMP PREEMPT Wed Feb 6 22:26:15 > > EET 2008 i686 pentium4 i386 GNU/Linux > > > > lspci identifies the NIC as: > > 02:03.0 Ethernet controller: Hangzhou Silan Microelectronics Co., Ltd.: > > Unknown device 2031 (rev 01) > > It seems the kernel doesn't know its PCI ID. > > > Hangzhou whatever, I'm sure it's an RTL8139 chip. Also wind0ze picks > > it up as a Realtech RTL8139 Card. > > Yes, Windows does, but Linux apparently not. I just found this: > http://vishalmanohar.wordpress.com/2007/07/03/trouble-with-fake-lan-card/
Counterfeit?? great.....
> It seems to be a 8139D chipset: > http://www.mail-archive.com/netdev@vger.kernel.org/msg27232.html > > I've never heard of this problem before, but I do know that
Thanks, I'll try the patches right away.
-- //MNZ\\
MNZ [mnzaki at gmail.com]
On Feb 7, 2008 3:20 PM, MNZ <mnzaki@gmail.com> wrote:
I tried the driver you specified, it does work though I get a whole bunch of kernel debug messages all over the console and logs every time I run dhclient or access the network. They had something about bogus return values and bad irqs and conflict with snd_hda_intel...... I reinserted the module and produced the same messages several times. too bad I cleared the log to get a clean capture for the last time but it stopped happening ;)
mplayer was running while I was inserting the module for the last time and everything just worked without a hitch..... what did I do? is it because it was using the snd_hda_intel module somehow? This is kinda strange.
-- //MNZ\\
Ben Okopnik [ben at linuxgazette.net]
On Thu, Feb 07, 2008 at 04:22:16PM +0200, MNZ wrote:
> On Feb 7, 2008 3:20 PM, MNZ <mnzaki@gmail.com> wrote: > > I tried the driver you specified, it does work though I get a whole bunch > of kernel debug messages all over the console and logs every time I > run dhclient or access the network. They had something about bogus > return values and bad irqs and conflict with snd_hda_intel...... I > reinserted the module and produced the same messages several times. > too bad I cleared the log to get a clean capture for the last time but it > stopped happening ;) > > mplayer was running while I was inserting the module for the last time > and everything just worked without a hitch..... what did I do? is it > because it was using the snd_hda_intel module somehow? This is > kinda strange.
Possibly. If your /faux/ NIC usually tries to use IRQ5 (which would be a
little odd but possible), and the sound driver used IRQ5 first (which
would be pretty common), the NIC would have to use another IRQ - which
would eliminate the conflicts. That's a possible theory, at least.
You might want to try troubleshooting it the old-fashioned way - i.e., load the NIC module first and check '/proc/interrupts' to see what's being used; then, reload it the "good" way (sound driver first) and check '/proc/interrupts' again to figure out which IRQ is being used by the NIC. Last of all, check to see if the NIC module can take an 'irq=' or similar argument when loading. Presto, end of conflict.
-- * Ben Okopnik * Editor-in-Chief, Linux Gazette * http://LinuxGazette.NET *
MNZ [mnzaki at gmail.com]
On Feb 8, 2008 12:03 AM, Ben Okopnik <ben@linuxgazette.net> wrote:
> > On Thu, Feb 07, 2008 at 04:22:16PM +0200, MNZ wrote: > > On Feb 7, 2008 3:20 PM, MNZ <mnzaki@gmail.com> wrote: > > > > I tried the driver you specified, it does work though I get a whole bunch > > of kernel debug messages all over the console and logs every time I > > run dhclient or access the network. They had something about bogus > > return values and bad irqs and conflict with snd_hda_intel...... I > > reinserted the module and produced the same messages several times. > > too bad I cleared the log to get a clean capture for the last time but it > > stopped happening ;) > > > > mplayer was running while I was inserting the module for the last time > > and everything just worked without a hitch..... what did I do? is it > > because it was using the snd_hda_intel module somehow? This is > > kinda strange. > > Possibly. If your /faux/ NIC usually tries to use IRQ5 (which would be a > little odd but possible), and the sound driver used IRQ5 first (which > would be pretty common), the NIC would have to use another IRQ - which > would eliminate the conflicts. That's a possible theory, at least.
/proc/interrupts says:
16: 3196 0 IO-APIC-fasteoi HDA Intel, eth0
It does not take an 'irq=' parameter (snd_hda_intel too). I tried inserting the NIC into another PCI slot, now it conflicts with my ATI driver :-/
-- //MNZ\\
MNZ [mnzaki at gmail.com]
On Feb 8, 2008 11:00 AM, MNZ <mnzaki@gmail.com> wrote:
> /proc/interrupts says: > 16: 3196 0 IO-APIC-fasteoi HDA Intel, eth0 > > It does not take an 'irq=' parameter (snd_hda_intel too). > I tried inserting the NIC into another PCI slot, now it conflicts with my > ATI driver :-/
I just got ndiswrapper and it works perfectly. Still conflicts with the ATI driver but neither one of them is complaining about it, not sure why.
I think I'll just stick with this. Thanks for the help.
-- //MNZ\\
Ben Okopnik [ben at linuxgazette.net]
On Fri, Feb 08, 2008 at 01:30:58PM +0200, MNZ wrote:
> On Feb 8, 2008 11:00 AM, MNZ <mnzaki@gmail.com> wrote: > > /proc/interrupts says: > > 16: 3196 0 IO-APIC-fasteoi HDA Intel, eth0 > > > > It does not take an 'irq=' parameter (snd_hda_intel too). > > I tried inserting the NIC into another PCI slot, now it conflicts with my > > ATI driver :-/ > > I just got ndiswrapper and it works perfectly. Still conflicts with the ATI > driver but neither one of them is complaining about it, not sure why. > > I think I'll just stick with this. Thanks for the help.
"ndiswrapper" is a pretty good approach for wonky hardware that only wants to work under Wind0ws - I've been using it for my laptop's WiFi (Broadcom's BCM4306, *bleh*) for several years now. It leaves a lot of mouse droppings in my log (hundreds of messages showing it dropping packets, etc.) - but it allows me to have WiFi even though there's no Linux module/driver for this chip.
The "right" answer is, of course, to be prescient enough to know that
all the hardware on a prospective laptop is Linux-ready... but I can't
say that I've ever managed that degree of clairvoyance or perfection.
-- * Ben Okopnik * Editor-in-Chief, Linux Gazette * http://LinuxGazette.NET *
Martin J Hooper [martinjh at blueyonder.co.uk]
Ben Okopnik wrote:
> The "right" answer is, of course, to be prescient enough to know that > all the hardware on a prospective laptop is Linux-ready... but I can't > say that I've ever managed that degree of clairvoyance or perfection.
Ben you looked at those Ubuntu loaded Dells??
Ben Okopnik [ben at linuxgazette.net]
On Fri, Feb 08, 2008 at 04:39:40PM +0000, Martin J Hooper wrote:
> Ben Okopnik wrote: > > The "right" answer is, of course, to be prescient enough to know that > > all the hardware on a prospective laptop is Linux-ready... but I can't > > say that I've ever managed that degree of clairvoyance or perfection.
I've owned several Dells; most of those had wonky hardware too (nVidia video and some APM-hostile chipset that prevented 'suspend' from working.) In addition, my truly crapful experiences with Dell's service and tech support convinced me to avoid them unless they were absolutely the best deal on the planet.
After vgrepping my 'Sent_mail', I now recall what it was that turned me off to Dell. I quote:
I've had hardware problems for the last day and a half or so - a fried backlight driver module in the LCD panel (temporarily patched, and a new laptop coming tomorrow.) Given Dell's insistence on charging me $500+ for a new LCD panel instead of selling me a $10 module, I hereby wholeheartedly curse them for the greedy grasping grabby spawn of leprous louse-infested licentious wombats crossed with a spastic scheming spectral tarsier's spew [1] that they are, and declare them to be apogenous, bovaristic, coprolalial, dasypygal, excerebrose, facinorous, gnathonic, hircine, ithyphallic, jumentous, kyphotic, labrose, mephitic, napiform, oligophrenial, papuliferous, quisquilian, rebarbative, saponaceous, thersitical, unguinous, ventripotent, wlatsome, xylocephalous, yirning zoophytes [2]. They also have no rhythm and their mama dresses them funny. And I'm never doing business with them again, and recommending the same to my customers. [1] [Yawn] Oh, I'm getting alliterative again. Must be well past my bedtime. [2] Credit for the abecedary insult goes to Peter Bowler.
-- * Ben Okopnik * Editor-in-Chief, Linux Gazette * http://LinuxGazette.NET *
René Pfeiffer [lynx at luchs.at]
On Feb 08, 2008 at 1129 -0500, Ben Okopnik appeared and said:
> On Fri, Feb 08, 2008 at 01:30:58PM +0200, MNZ wrote: > > On Feb 8, 2008 11:00 AM, MNZ <mnzaki@gmail.com> wrote: > > > /proc/interrupts says: > > > 16: 3196 0 IO-APIC-fasteoi HDA Intel, eth0 > > > > > > It does not take an 'irq=3D' parameter (snd_hda_intel too). > > > I tried inserting the NIC into another PCI slot, now it conflicts with my > > > ATI driver :-/ > > > > I just got ndiswrapper and it works perfectly. Still conflicts with the ATI > > driver but neither one of them is complaining about it, not sure why. > > > > I think I'll just stick with this. Thanks for the help. > > "ndiswrapper" is a pretty good approach for wonky hardware that only > wants to work under Wind0ws - I've been using it for my laptop's WiFi > (Broadcom's BCM4306, *bleh*) for several years now. It leaves a lot of > mouse droppings in my log (hundreds of messages showing it dropping > packets, etc.) - but it allows me to have WiFi even though there's no > Linux module/driver for this chip.
You might want to give 2.6.24 a try:
lynx@nephtys:~$ grep BCM /boot/config-2.6.24 CONFIG_BT_HCIBCM203X=m CONFIG_BCM43XX=m CONFIG_BCM43XX_DEBUG=y CONFIG_BCM43XX_DMA=y CONFIG_BCM43XX_PIO=y CONFIG_BCM43XX_DMA_AND_PIO_MODE=y # CONFIG_BCM43XX_DMA_MODE is not set # CONFIG_BCM43XX_PIO_MODE is not set lynx@nephtys:~$Cheers, René, who doesn't like ndiswrapper and running certain code on his machines.
Ben Okopnik [ben at linuxgazette.net]
On Fri, Feb 08, 2008 at 05:39:50PM +0100, Ren? Pfeiffer wrote:
> On Feb 08, 2008 at 1129 -0500, Ben Okopnik appeared and said: > > > > "ndiswrapper" is a pretty good approach for wonky hardware that only > > wants to work under Wind0ws - I've been using it for my laptop's WiFi > > (Broadcom's BCM4306, *bleh*) for several years now. It leaves a lot of > > mouse droppings in my log (hundreds of messages showing it dropping > > packets, etc.) - but it allows me to have WiFi even though there's no > > Linux module/driver for this chip. > > You might want to give 2.6.24 a try: > > lynx@nephtys:~$ grep BCM /boot/config-2.6.24 > CONFIG_BT_HCIBCM203X=m > CONFIG_BCM43XX=m > CONFIG_BCM43XX_DEBUG=y > CONFIG_BCM43XX_DMA=y > CONFIG_BCM43XX_PIO=y > CONFIG_BCM43XX_DMA_AND_PIO_MODE=y > # CONFIG_BCM43XX_DMA_MODE is not set > # CONFIG_BCM43XX_PIO_MODE is not set
I've been hoping that they'd get around to it one day. Although I
doubt that I can use it yet:
ben@Tyr:~$ grep 'model name' /proc/cpuinfo model name : AMD Athlon(tm) 64 Processor 3200+
So, I'm usually at the tail end of the line when they hand out the goodies.
> who doesn't like ndiswrapper and running certain code on his machines. >Me either - but paying high bucks for (theoretically) a high-end laptop and then not being able to use its core features can be a pretty good motivator. :\
-- * Ben Okopnik * Editor-in-Chief, Linux Gazette * http://LinuxGazette.NET *
René Pfeiffer [lynx at luchs.at]
On Feb 08, 2008 at 1152 -0500, Ben Okopnik appeared and said:
> On Fri, Feb 08, 2008 at 05:39:50PM +0100, René Pfeiffer wrote: > > [...] > > You might want to give 2.6.24 a try: > > > > lynx@nephtys:~$ grep BCM /boot/config-2.6.24 > > CONFIG_BT_HCIBCM203X=m > > CONFIG_BCM43XX=m > > CONFIG_BCM43XX_DEBUG=y > > CONFIG_BCM43XX_DMA=y > > CONFIG_BCM43XX_PIO=y > > CONFIG_BCM43XX_DMA_AND_PIO_MODE=y > > # CONFIG_BCM43XX_DMA_MODE is not set > > # CONFIG_BCM43XX_PIO_MODE is not set > > I've been hoping that they'd get around to it one day.
Why? Linux runs on many platforms, you know. ;)
> > who doesn't like ndiswrapper and running certain code on his machines. > >
I usually hand the "high bucks" or "high euros" only to vendors that give me working and documented hardware in return. I know that this is not always possible, but it works for me. Maybe this is due to my low demands (getting some xterms going is almost possible on any hardware ;).
Best, René.
Ben Okopnik [ben at linuxgazette.net]
On Fri, Feb 08, 2008 at 06:05:04PM +0100, Ren? Pfeiffer wrote:
> On Feb 08, 2008 at 1152 -0500, Ben Okopnik appeared and said: > > > > I've been hoping that they'd get around to it one day.
That's true - although there are some small differences in what works where. a 64-bit AMD is a daily education in those differences. [GRRRR...]
> > > who doesn't like ndiswrapper and running certain code on his machines. > > >
That's pretty tough with laptops, where the vendors change the hardware arbitrarily, randomly, and (I suspect) with a great amount of glee at the projected discomfort this will cause to the buyers.
What would be really awesome is if every laptop came with an 'lspci' printout. [sigh]
> I know that this is > not always possible, but it works for me. Maybe this is due to my low > demands (getting some xterms going is almost possible on any hardware > ;).
Well, yes. Although that would make using FlightGear really, really
difficult.
-- * Ben Okopnik * Editor-in-Chief, Linux Gazette * http://LinuxGazette.NET *
Jim Jackson [jj at franjam.org.uk]
On Thu, 7 Feb 2008, MNZ wrote:
> Hi TAG, > I have recently bought a NIC that uses the rtl8139 chip. I can't seem > to get it working. I load the 8139too module but I still don't get > /dev/eth0.
??? What makes you think you'll get that device?
What is the output of...
/sbin/ifconfig eth0
MNZ [mnzaki at gmail.com]
On Feb 7, 2008 2:47 PM, Jim Jackson <jj@franjam.org.uk> wrote:
> On Thu, 7 Feb 2008, MNZ wrote: > > > Hi TAG, > > I have recently bought a NIC that uses the rtl8139 chip. I can't seem > > to get it working. I load the 8139too module but I still don't get > > /dev/eth0. > > ??? What makes you think you'll get that device?
I figured I ought to have it :-/
> What is the output of... > > /sbin/ifconfig eth0
SIOCGIFFLAGS: No such device Error: cannot enable interface eth0.
-- //MNZ\\
Ben Okopnik [ben at linuxgazette.net]
On Thu, Feb 07, 2008 at 12:47:35PM +0000, Jim Jackson wrote:
> > > > On Thu, 7 Feb 2008, MNZ wrote: > > > Hi TAG, > > I have recently bought a NIC that uses the rtl8139 chip. I can't seem > > to get it working. I load the 8139too module but I still don't get > > /dev/eth0. > > ??? What makes you think you'll get that device? > > What is the output of... > > /sbin/ifconfig eth0
Since the name of the device is in doubt, I would think that
ifconfig -a
(i.e., "show all devices") would be a bit more useful.
-- * Ben Okopnik * Editor-in-Chief, Linux Gazette * http://LinuxGazette.NET *
Jim Jackson [jj at franjam.org.uk]
>>> I have recently bought a NIC that uses the rtl8139 chip. I can't seem >>> to get it working. I load the 8139too module but I still don't get >>> /dev/eth0. >> >> ??? What makes you think you'll get that device? >> >> What is the output of... >> >> /sbin/ifconfig eth0 > > Since the name of the device is in doubt, I would think that > > `` > ifconfig -a > '' > > (i.e., "show all devices") would be a bit more useful.
Indeed.
But I didn't think the name of the device was in question??? Just what to look for.
Ben Okopnik [ben at linuxgazette.net]
On Thu, Feb 07, 2008 at 10:35:57PM +0000, Jim Jackson wrote:
> > >>> I have recently bought a NIC that uses the rtl8139 chip. I can't seem > >>> to get it working. I load the 8139too module but I still don't get > >>> /dev/eth0. > >> > >> ??? What makes you think you'll get that device? > >> > >> What is the output of... > >> > >> /sbin/ifconfig eth0 > > > > Since the name of the device is in doubt, I would think that > > > > `` > > ifconfig -a > > '' > > > > (i.e., "show all devices") would be a bit more useful. > > Indeed. > > But I didn't think the name of the device was in question??? > Just what to look for.
I quote, from the above:
??? What makes you think you'll get that device?
It wouldn't be unreasonable to expect it to show up as another device - so it's worth checking out the entire list without assuming a specific device name.
-- * Ben Okopnik * Editor-in-Chief, Linux Gazette * http://LinuxGazette.NET *
Jim Jackson [jj at franjam.org.uk]
On Thu, 7 Feb 2008, Ben Okopnik wrote:
>>>> >>>> ??? What makes you think you'll get that device? >>>> >>>> What is the output of... >>>> >>>> /sbin/ifconfig eth0 >>> >>> Since the name of the device is in doubt, I would think that >>> >>> `` >>> ifconfig -a >>> '' >>> >>> (i.e., "show all devices") would be a bit more useful. >> >> Indeed. >> >> But I didn't think the name of the device was in question??? >> Just what to look for. > > I quote, from the above: > > `` > ??? What makes you think you'll get that device? > ''
Ben, I was clumsily trying to point out ethernet drivers do not create a linux device (i.e. something you can create a filesystem entry for, with a device number etc) - they create a network interface. Hence you wouldn't get any device. In my urge to reply quickly I badly worded my question to him. I should have asked "What makes you think you'll get a device?"
He seems to have had some familiarity with plan 9, when indeed everything is (or can be represented as) a file, and that had lead him to look for a /dev/eth0, instead of simply doing ifconfig eth0.
cheers Jim
Ben Okopnik [ben at linuxgazette.net]
On Fri, Feb 08, 2008 at 03:21:44PM +0000, Jim Jackson wrote:
> On Thu, 7 Feb 2008, Ben Okopnik wrote: > > > > I quote, from the above: > > > > `` > > ??? What makes you think you'll get that device? > > '' > > Ben, I was clumsily trying to point out ethernet drivers do not create a > linux device (i.e. something you can create a filesystem entry for, with a > device number etc) - they create a network interface. Hence you wouldn't > get any device. In my urge to reply quickly I badly worded my question to > him. I should have asked "What makes you think you'll get a device?"
Ah - you're right, I misunderstood. Well, a little extra info never
hurts.
-- * Ben Okopnik * Editor-in-Chief, Linux Gazette * http://LinuxGazette.NET *
MNZ [mnzaki at gmail.com]
On Feb 8, 2008 5:21 PM, Jim Jackson <jj@franjam.org.uk> wrote:
> >> But I didn't think the name of the device was in question??? > >> Just what to look for. > > > > I quote, from the above: > > > > `` > > ??? What makes you think you'll get that device? > > '' > > Ben, I was clumsily trying to point out ethernet drivers do not create a > linux device (i.e. something you can create a filesystem entry for, with a > device number etc) - they create a network interface. Hence you wouldn't > get any device. In my urge to reply quickly I badly worded my question to > him. I should have asked "What makes you think you'll get a device?"
I know they create network interfaces but I didn't know that they didn't normally create /dev/eth0 (because I've always had that node)
> He seems to have had some familiarity with plan 9, when indeed everything > is (or can be represented as) a file, and that had lead him to look for a > /dev/eth0, instead of simply doing ifconfig eth0.
First time to hear the word plan 9 for me. I did my googling though.
-- //MNZ\\
MNZ [mnzaki at gmail.com]
On Feb 7, 2008 7:00 PM, Kapil Hari Paranjape <kapil@imsc.res.in> wrote:
> Hello, > > > On Thu, 07 Feb 2008, MNZ wrote: > > I have recently bought a NIC that uses the rtl8139 chip. I can't seem > > to get it working. I load the 8139too module but I still don't get > > /dev/eth0. I'm not sure how to go about fixing this so any help > > would be appreciated. > > I suppose you already tried the 8139cp module.
Actually no, but it appears it's not an rtl8139 so that wouldn't help, I think? I don't see to have the module anyway
-- //MNZ\\//MNZ\\