dialup connection freeze after a few minutes
 

Hi,
Have recently installed Fedora Core 2 from CDs borrowed from my friend. Configured almost everything including the dialup connection using toshiba internal modem (SmartLink).
Have a Satellite M30 laptop. The kernel is the one which comes with FC2. Haven't recompiled it. Downloaded the latest slamr driver and installed it successfully.
Dialup connects fine (KPPP and WVDIAL) except an occasional "Authentication Failure".

Here is the problem:

1. After some time, the networking freezes. I mean the PPPD is still running and the routing tables are present (including the default route which points to my ISP), however no communication. I can't ping IP or names. No internet access. Can this be related to modem default timeout value which is at present set to 600 somewhere in modem settings ? But in that case, shouldn't pppd also quit ?

2. If I switch to virtual consoles from Xserver or vice versa, the connection freezes almost instantly.

I have I already tried:

1. Different dialup tools, be it redhat's network config or KPPP or wvdial.
2. Adding an entry in /etc/resolv.conf for my ISP's nameserver. Without it name resolution fails even though I can ping my ISP.
3. Have included defaultroute in /etc/ppp/options.
4. Keep pinging my ISP or www(DOT)yahoo(DOT)com continuously from one terminal window.
5. Disabling my LAN and Wireless interfaces. (i.e eth0 and eth1)
5. Explored Google.

The setup:
1. Toshiba Satellite M30 Laptop.
2. Toshiba internal modem (SmartLink chipset)
3. Nvidia driver for X (GeForce). Downloaded from their website.
4. I pass acpi=on to kernel as an option in my grub.conf.


I am now running out of innovative thoughts and guess its about the right time to knock the doors of linux community.

Thanks a lot!!

Hi,
Well!! replying to my own post, I would like to add that the source of problem appears to be the latest nVIDIA drivers. After experimenting with the "nv" display driver, I found out that dialup connection keeps working properly. I am using the latest 1.0-6629 version of nvidia driver.
Has anybody else experienced this sort of problem ? Today I am going to to try the older version 1.0-6111.


--cyberjun

Another update on this one. I removed "quite" from my kernel options in GRUB. Now whenever I loose connection, I find the following in my /var/log/messages:

Mar 13 11:56:04 Mukhi modprobe: FATAL: Error running install command for sound_slot_2
Mar 13 11:57:30 Mukhi last message repeated 2 times
Mar 13 11:57:58 Mukhi last message repeated 3 times
Mar 13 11:45:19 Mukhi kernel: Debug: sleeping function called from invalid conte
xt at include/asm/semaphore.h:107
Mar 13 11:45:19 Mukhi kernel: in_atomic():1, irqs_disabled():1
Mar 13 11:45:19 Mukhi kernel: [<c0118dec>] __might_sleep+0xac/0xc0
Mar 13 11:45:19 Mukhi kernel: [<e0b4232f>] snd_ac97_update_bits+0x3f/0x80 [snd_
ac97_codec]
Mar 13 11:45:19 Mukhi kernel: [<e0a5da33>] snd_pcm_lib_read1+0x373/0x4f0 [snd_p
cm]
Mar 13 11:45:19 Mukhi kernel: [<e09ba85a>] snd_intel8x0m_pcm_trigger+0x3a/0x70 [snd_intel8x0m]
Mar 13 11:45:19 Mukhi kernel: [<e0a55165>] snd_pcm_do_stop+0x35/0x40 [snd_pcm]
Mar 13 11:45:19 Mukhi kernel: [<e0a54c22>] snd_pcm_action_group+0xa2/0x1b0 [snd _pcm]
Mar 13 11:45:19 Mukhi kernel: [<e0a54ddc>] snd_pcm_action+0x3c/0x80 [snd_pcm]
Mar 13 11:45:19 Mukhi kernel: [<e0a55f6e>] snd_pcm_drop+0x7e/0x100 [snd_pcm]
Mar 13 11:45:19 Mukhi kernel: [<e0a5840a>] snd_pcm_capture_ioctl1+0x3a/0x2c0 [s nd_pcm]
Mar 13 11:45:19 Mukhi kernel: [<e0a5870e>] snd_pcm_capture_ioctl+0x2e/0x50 [snd _pcm]
Mar 13 11:45:19 Mukhi kernel: [<c016b94a>] do_ioctl+0x6a/0x90
Mar 13 11:45:19 Mukhi kernel: [<c016bbce>] vfs_ioctl+0x5e/0x1d0
Mar 13 11:45:19 Mukhi kernel: [<c016bd9b>] sys_ioctl+0x5b/0x90
Mar 13 11:45:19 Mukhi kernel: [<c0103089>] sysenter_past_esp+0x52/0x75
Mar 13 11:45:19 Mukhi kernel: ALSA /home/mukhbens/alsa/alsa-driver-1.0.8/alsa-ke rnel/pci/intel8x0m.c:401: codec_semaphore: semaphore is not ready [0x1][0x701300 ]
Mar 13 11:45:19 Mukhi kernel: ALSA /home/mukhbens/alsa/alsa-driver-1.0.8/alsa-ke rnel/pci/intel8x0m.c:415: codec_write 1: semaphore is not ready for register 0x5 4

Please help me. This problem is driving me insane. Since my internal modem doesn't make any noise, I don't know if this is happening because of modem dropping the line. I have tried simulating this by removing the phone cord from modem. In this case also same message appears in /var/log/messages:

Thanks a lot.

--cyberjun

Now this was really some problem...
I'm afraid I don't have a good solution, just a few guesses since nobody's been able to figure out what's going on.
OK, so are they really disabled? Could it be that the connections get mixed up?
Dial-up, check output of
#ifconfig
#route
then when it freezes check those two again.
Try passing acpi=off

Those error messages says you have problem with soundcard and Alsa. That shouldn't affect networking??
You have an Intel on-board sound, is that correct?
Never had any problems with Nvidias driver and modem, but then you never know...
I'm wondering a bit about these two lines:
Can you find out what "semaphore" is?

Hi Pingu,
Thanks for replying. Well I also think there is some problem with ALSA. Networking is not related directly to sound (By the way sound works flawlessly. xmms is great!!). However, since I am using an internal modem which uses snd_intel8x0m driver from ALSA, I guess this is part alsa part kernel problem. At this moment I am not clear where I should post this error message. Does it belong to the ALSA mailing list or should it be logged as a kernel bug.
Does anybody have a suggestion ?

Thanks a lot.

--cyberjun