Bugzilla – Bug 104460
ERROR (Media:connection failed)[couldn't connect to host]
Last modified: 2007-02-16 14:01:17 UTC
Installing SUSE-10.0-CD-OSS-i386-Beta1 on a Dell Latitude C600 from an installation server via PXE boot, and getting the following popup error after rebooting the first time to continue the installation. Unable to retrieve the remote source description ERROR (Media:connection failed)[couldn't connect to host] [Retry][Skip Refresh][Disable Source] Interestingly I am getting an IP address, the resolv.conf is populated and the default route is correct. But I cannot ping anything but my local interface. In /var/log/messages I see the following message repeated: Aug 12 12:27:41 linux kernel: eth0: Host error, FIFO diagnostic register 0000. Aug 12 12:27:41 linux kernel: eth0: PCI bus error, bus status 80000020 Aug 12 12:27:41 linux kernel: ACPI: PCI Interrupt 0000:00:10.0[A] -> Link [LNKD] -> GSI 11 (level, low) -> IRQ 11 Aug 12 12:27:41 linux kernel: eth0: Too much work in interrupt, status e003. After many reboots (about 7 reboots), eventually it is able to finish the installation.
Created attachment 45941 [details] YaST2 Logs and hopefully other relivant info tar -tvf all.tgz -rw-r--r-- root/root 110002 2005-08-12 08:30:17 y2logs.tgz -rw-r--r-- root/root 331 2005-08-12 08:29:31 route.txt -rw-r--r-- root/root 92 2005-08-12 08:46:34 resolv.conf.txt -rw-r--r-- root/root 415 2005-08-12 08:50:10 ping.txt -rw-r--r-- root/root 453 2005-08-12 08:48:23 pinglocal.txt -rw-r----- root/root 6846063 2005-08-12 08:27:42 messages -rw-r--r-- root/root 1439 2005-08-12 08:27:58 lsmod.txt -rw-r--r-- root/root 923 2005-08-12 08:28:54 ifconfig.txt -rw-r--r-- root/root 119085 2005-08-12 08:28:28 hwinfo.txt
Noticed that if I power off the laptop, then the installation is able to continue with NO errors.
Looks like a kernel problem.
Seems the 3c59x driver has problems with this card. Karsten, can you take a look please? Question to the bug submitter: apparently the card worked initially, since dhcp seems to have worked. can you pinpoint the moment it stops working?
> Question to the bug submitter: apparently the card worked initially, > since dhcp seems to have worked. can you pinpoint the moment it stops > working? Not sure when it stops working, but the first error I see is when it tries to continue the install after the first reboot. Willing to collect any data you need.
On the first glance, it looks like a chipset init/reset problem, maybe ACPI, the PCI bus error looks bad to me, so I'm not shure if it is related to the network driver at all. I'll try to dig deeper in it.
If you do a full power cycle, does the network card work, or does it only make the installation to continue ? If you remove the PCMCIA card before booting and plug it in after the system is running, does it work then ?
(In reply to comment #7) > If you do a full power cycle, does the network card work, or does it only make > the installation to continue ? > > If you remove the PCMCIA card before booting and plug it in after the system > is running, does it work then ? > During the installation I have to power off the laptop, and then power it back on in order for the installation to complete. Everything else seems to work ok after powering the laptop off and back on. I am doing the install from a http server over the network. So I would say the network card does work if I power off the laptop, and then back on during the installation. If I let the install program reboot, it fails to complete the install everytime. Once the OS is installed I have no problems. I can issue the "reboot" command, and the network card still works. This is an built-in NIC, so I cannot remove the card.
OK I was mislead by hwinfo "3Com 3c556 Hurricane CardBus [Cyclone]" So I thought it was a cardbus PCMCIA card. Since it work after installation: Maybe this is a side effect of our intensive hardware probing, I do think we can do here so much, since laptop chipsets are always a mess, regard getting informations from manufacturers. So here is always the possibility to do something wrong, some times it is enough to read from some register at the wrong time to get strange effects. I have no idea how we can improve this, Steffen do you agree to my explanation ?
There is no intensive hardware probing. :-) In any case beta-1 had still the pci config space access code used during BIOS emulation. That's gone in beta-2. Maybe it improves things here.
Any news here, did beta2 help
No reaction on this bug for more than 18 months. Closing as CANTFIX.