[Solved] libgles weird bug on Slackware

Linux-SWAT

Hardcore Member
Joined
Feb 13, 2010
Messages
8,239
Seems i have no problem with my old releases. The investigation continues...
 

Linux-SWAT

Hardcore Member
Joined
Feb 13, 2010
Messages
8,239
#using the GPU clock to detect the current CPU model
GPUCLOCK=$(cat /proc/pandora/sys_mhz_max)
 

It's before each rc.local* test.

Ok i get it, i modified rc.local_shutdown when i noticed the cpusettings was removing the link and created a new file. And i didn't reworked the first part ! Shame on me ^^.

Thx for noticing.
 

urjaman

"I Know. We're going for a ride."
Joined
Jan 6, 2009
Messages
1,111
Age
29
Location
Finland
Website
urjaman.dy.fi
If you wrote that, FYI GPUCLOCK is wrong name for that - maybe MEMCLOCK. You are using the memory bus speed to distinquish between OMAP3530 and DM3730... (TBH, I'd find a better way.)
 

Linux-SWAT

Hardcore Member
Joined
Feb 13, 2010
Messages
8,239
Ok.

Anyway, that thing is not the real culprit. I'm rebuilding a rootfs against 1.52 binaries. I'm almost certain all will work. We'll see soon.
 

Linux-SWAT

Hardcore Member
Joined
Feb 13, 2010
Messages
8,239
Oh turns out it seems that there is a random permissions problem on /dev/pvrsrvkm !!! 600 instead of 666.

That's crazy, it's supposed to be fixed by /etc/init.d/pvr-init, and until now, it always worked...
 

Linux-SWAT

Hardcore Member
Joined
Feb 13, 2010
Messages
8,239
Marked the thread as solved.

This bug appeared end of january/beginning of february.

I have no clue about the event that triggered it.

So i now chmod 666 after the pvr-init and a sync.

Tested a bit, all seems back to normal.

Thanks everyone.
 

notaz

Certified Guru
Joined
Aug 23, 2005
Messages
4,913
Location
Lithuania
Website
notaz.gp2x.de
Looks like pvr-init fails to wait until /dev/pvrsrvkm appears (differenct udev version?), can you try adding this into pvr-init just before "chmod 666 /dev/pvrsrvkm"?

Code:
        for a in `seq 20`; do
                if test -e /dev/pvrsrvkm; then 
                        break
                fi
                sleep 0.2
        done
If that works I'll push into git.
 

Linux-SWAT

Hardcore Member
Joined
Feb 13, 2010
Messages
8,239
I will.

No udev update. I just wasn't able to put my finger on that in three days :/.
 

sebt3

homebrew player (P. & C.)
Joined
Sep 9, 2008
Messages
4,747
Age
39
Location
France
Website
sebt3.openpandora.org
Thanks for pointing me on this. Turn out I have a similar issue with .next.

edit :


root@pandy:~# cat /proc/devices |grep pvr
root@pandy:~# lsmod|grep pvr
pvrsrvkm 136845 1 omaplfb

That doesnt help to get /dev/pvrsrvkm created by the pvr init script. 

Udev doesnt create it either. :(
 
Last edited by a moderator:

Linux-SWAT

Hardcore Member
Joined
Feb 13, 2010
Messages
8,239
Looks like pvr-init fails to wait until /dev/pvrsrvkm appears (differenct udev version?), can you try adding this into pvr-init just before "chmod 666 /dev/pvrsrvkm"?


for a in `seq 20`; do
if test -e /dev/pvrsrvkm; then
break
fi
sleep 0.2
done

If that works I'll push into git.
Doesn't work.
 
Top