OESF Portables Forum

Everything Else => General Support and Discussion => Zaurus General Forums => Archived Forums => Software => Topic started by: PeterL on September 17, 2004, 02:05:12 am

Title: good console ?
Post by: PeterL on September 17, 2004, 02:05:12 am

Hi,

I'm running on an SL-C860.  If I use the standard terminal application from the CDROM, or if I use QKonsole, tasks that are running in the console/terminal window are stopped when I power off.  Even if I simply su to root, the console comes back up with

[1]+ Stopped      su

After that, lots of weird stuff happens before functioning goes back to normal.  Keyboard doesn't seem to work right etc.

Now, this isn't so much of a problem when running just su or other small stuff.  It's a bit of a hassle.  But when using vi, emacs, HNB or other editors, things get quite complicated.  I always have to think about saving the file, just in case the system suspends itself.  Then, after suspend, it works best if I just kill whatever I was doing & restart it all :S  Quite annoying.

Is there any terminal program that doesn't exhibit this behaviour ?  Or any way around this ?

Thanks,


Peter.

Title: good console ?
Post by: Mickeyl on September 17, 2004, 05:40:41 am

This is not a terminal issue, but a kernel one. 2.4.18-embedix sends SIGSTOP on suspend.

Title: good console ?
Post by: PeterL on September 19, 2004, 01:43:59 pm

Hi Mickey,

Thanks for your clarification.  But it still is a hassle.  Any way around it, so that I can keep programs open in the terminal window even when power off ?  On batteries, I like a quick auto power off, so it tends to get irritating

Peter.

Title: good console ?
Post by: Mickeyl on September 19, 2004, 04:00:47 pm

Patch the kernel. The offending part is in /arch/arm/mach-sa1100/[collie|sharpsl]_apm.c. Grep for SIGSTOP

Title: good console ?
Post by: guylhem on September 20, 2004, 06:42:42 am

Won't that cause problem with apps depending on SIGSTOP at suspend for say saving data?

Title: good console ?
Post by: Mickeyl on September 20, 2004, 12:30:08 pm

Not sure about that. Maybe we need a kernel patch offering a /proc/ entry where one can supply binary names which should not receive a SIGSTOP on suspend/resume.