opkg upgrade = endless IRQ messages

Brian C brianwc at OCF.Berkeley.EDU
Wed Jul 16 11:09:16 CEST 2008

Brian C wrote:
> 1. opkg upgrade is busily upgrading packages, then
> 2. the x server appears to shut down
> 3. the following never-ending text message appears:
> pcf50633_irq: entering)irq=53, pcf=c7d34c00): scheduling work
> pcf50633_work: INT1:0x80 INT2=0x00 INT3=0x00 INT4=0x00 INT5=0x00
> (although sometimes "SECOND" is replaced by "ADCADY"

Some off-list advice may have resolved this.  I think the solution is:

1. Don't upgrade dropbear via ssh.
2. Don't upgrade xserver-* packages via FreeRunner's terminal.

So I logged in via ssh and did this:

root at om-gta02:~# opkg list_installed|grep xserver
xserver-kdrive - 1: -
xserver-kdrive-common - 0.1-r27.01 -
xserver-kdrive-fbdev - 1: -
xserver-kdrive-glamo - 1: -
xserver-nodm-init - 1.0-r8 -

root at om-gta02:~# opkg install xserver-kdrive
Package xserver-kdrive (1: installed in root is up to date.

root at om-gta02:~# opkg install xserver-kdrive-common
Upgrading xserver-kdrive-common on root from 0.1-r27.01 to 0.1-r28...
Configuring xserver-kdrive-common

root at om-gta02:~# opkg install xserver-kdrive-fbdev
Package xserver-kdrive-fbdev (1: installed in root is up to

root at om-gta02:~# opkg install xserver-kdrive-glamo
Upgrading xserver-kdrive-glamo on root from 1: to
Configuring xserver-kdrive-glamo

root at om-gta02:~# opkg install xserver-nodm-init
Upgrading xserver-nodm-init on root from 1.0-r8 to 1.0-r9...
Configuring xserver-nodm-init
 System startup links for /etc/init.d/xserver-nodm already exist.
Stopping XServer
Starting Xserver

After this last one, the Xserver restarted fine.  Then I went back to
the FreeRunner's terminal and did another round of opkg upgrade and it
seemed to be able to do all the rest and come to a normal completion.

(There was one weird error re ncurses wanting to install a file already
provided by another package, but if I'm now down to just a single
package that isn't upgrading cleanly, that's a huge improvement over the
50+ that seemed to be affected before.)


More information about the support mailing list