[hsflinux] khsfd/modem no stopped

[Linuxant support (Jonathan)]

Hi,

we are aware of this problem on a few machines and we are trying to 
reproduce this problem in order to fix it in a future release of the HSF 
modem driver. In the meantime, the only known workaround is to remove 
the HSF driver modules from the kernel before you suspend the machine 
and all these commands in a root shell should be able to do it:

---
hsfstop
hsfconfig --rcstop
---

Regards,


Jonathan
Technical specialist / Linuxant
www.linuxant.com
support at linuxant.com



Henrik Nilsson wrote:
> Hi,
> 
> Michel Kerszberg wrote:
> 
>  >Hi,
>  >
>  > Since installing a new kernel on my IBM T42 and installing
>  > hsfmodem-7.18.00.03full_k2.6.11.4_20a-default-1suse.i586
>  > I cannot use suspend to disk or to RAM, which worked before.
>  > Obviously, a task cannot be stopped, namely I get the message
>  > "strange, khsfd/modem not stopped".
>  >
>  >What to do?
> 
> Jonathan replied:
> 
>  > suspend to disk or RAM was working fine with which version of the HSF
>  > driver and kernel version?
> 
> And suggested the following workaround:
> 
>  > For your current installation, you can try this workaround. Before you
>  > suspend the system, shut down the modem connection and then run the
>  > 'hsfstop' command in a root shell. The 'hsfstop' command will remove
>  > the HSF modules from the kernel and it will probably fix the suspend
>  > problem.
> 
> I'm running Fedora Core 4 on a Dell Latitude D600. My kernel is
> 2.6.12-1.1398_FC4 configured as shipped
> (configs/kernel-2.6.12-i686.config), except that the touchpad/stick
> driver has been modified a tiny little bit to make the touchpad/stick
> work when the machine is restarted after a suspend (to RAM).
> 
> I'm seeing exactly the same problem as Michael.
> 
> I discovered that I could work around the problem by either
> manually removing all HSF kernel modules (using "rmmod"), or
> by issuing the command "hsfconfig --rcstop". This must be
> pretty much the same as the workaround above suggested by Jonathan,
> and I can thus confirm that the workaround should work.
> 
> While not a huge problem (e.g. one can automate by configuring the ACPI
> daemon "acpid" to do the work on a sleep event), I wondered if the
> problem has been further investigated and if a proper fix might be
> forthcoming.
> 
> Incidentally, I'm having a somewhat similar problem with DriverLoader,
> used to load a driver for my TrueMobile 1300 wireless ethernet card.
> In this case, five processes refuse to stop:
> 
>     kdldrd/normwork
>     kdldrd/critwork
>     kdldrd/hyperwork
>     kdldrd/dpc
>     kdldrd/wrk
> 
> However, that problem goes away completely as soon as I've issued
> the command "dldrconf --rcstop" once. I.e. after this, I can enable
> the network and resume and suspend as often as I like without any
> further action. But once I reboot, I again am unable to suspend
> until I've issued the commend "dldrconf --rcstop" once.
> 
> I'm also seeing what appears to be some other side effects of
> the driver loader installation. I'm going to send a separate mail
> to the driver loader list with all the details.
> 
> Best,
> 
> /Henrik
>