Kernel module cx88 hangs at powering down

[cyberpatrol]

I've just switched from Arch Linux to Salix and encountered a bug regarding the cx88 kernel module.

I'm using a cx88 based DVB-T card. When I shut down the computer, the system is firstly shut down as usual, but when the computer actually should power down I get those error messages repeatedly instead:

Code: Select all

Powering down.
<4>[     86.93B221] cx88[0]:   risc3: 0xffffffff [ INVALID sol eol irq2 irq1 23 22 21 20 19 18 cnt1 cnt0 resync 14 13 12<4>[   88.325664] cx88[0]:    iq9: 0xffffffff [ INVALID sol eol irq2 irq1 23 22 21 20 19 18 cnt1 cnt0 resync 14 13 12<4>[   88.346940] cx88[0]:   iq5: 0xffffffff [ INVALID sol eol irq2 irq1 23 22 21 20 19 18 cnt1 cnt0 resync 14 13 12<7>[   88.357366] cx88[0]: irq mpeg  [0xffffffff] ts_risci1* 2* 3* ts_risci2* 5* 6* 7* ts_oflow* 9* 18* 11* ts_sync* 13* 14* 15* opc_err* par_err* rip_err* pci_abort* ts_err?* 21* 22* 23* 24* 25* 26* 27* 28* 29* 30* 31*

Adding

Code: Select all

rmmod cx8800

to /etc/rc.d/rc.local_shutdown solved the issue, and the computer powers down properly without those error messages.

Since I never had this issue with Arch Linux nor with Gentoo, and it can be solved by this rmmod workaround, I think it's likely to be a distribution specific bug, probably either in the kernel, in /sbin/poweroff resp. /sbin/halt or in /etc/rc.d/rc.6.

[thenktor]

Looks like a kernel bug. As Salix uses vanilla kernels this need to be reported upstream. But maybe it's already fixed in more recent kernel versions.

[cyberpatrol]

I used this kernel version as well as earlier and later ones with Gentoo and Arch Linux before, and never had this issue. And Arch Linux doesn't patch the kernel too heavily. So if it's a kernel bug then I think it only occurs with a specific kernel configuration resp. a combination of kernel configurations which is not done in the other distributions.

But I hadn't had time to look into the kernel configuration, the rc scripts and the shutdown process in detail since I'm still moving to Salix. And asking upstream indeed couldn't hurt, too. I will do this as soon as I'm finished setting up my system..

But due to the workaround it's a minor issue anyway.