[sane-devel] Mac OS X: make errors

[Louis Lagendijk]

On Thu, 2009-03-26 at 12:16 -0700, Peter Schoenrank wrote:
> My apologies for my tardy reply. I recently upgraded from Mac OS X  
> 10.3 to 10.5, and I have been busy learning about the (sometimes  
> idiotic) differences.
> 
> 
> On 09-Mar-25, at 14:00, Nicolas Martin wrote:
> 
> > Are you sure you got all the log trace in the scanimage.log file after
> > the scanimage -T command ?
> 
> No.
> 
> I am sure that Activity Monitor showed scanimage using 0.0 CPU for a  
> very long time before I restarted my Mac.
> 

> I had to restart: nothing else I tried would stop the scanimage  
> process. kill -9 and sudo killall didn’t work. Quit and Force Quit in  
> Activity Monitor did not work. Closing the Terminal window didn’t  
> work. Logging out and back in didn’t work.
> 
> 
This makes it sound like a case where libusb hangs somewhere in a
OS-call where the OS crashes/hangs. From what I have seen on Linux an
application may become unresponsive when an OS-call does not return. I
seem to recall that all Macs have a multi-core processor. Could you
please check the system log AFTER the hang BEFORE you restart the Mac? 
Mattias, do you have seen other libusb/Mac kernel error reports?

> > Looks like it suddenly hangs (in libusb), but no error is output.
> >
See above, Sane would not get back in control after a kernel crash/oops
type of error. 

> > Do you get a core dump in /var/log/syslog or in /var/log/messages ?
> 
> No. I do not have a /var/log/syslog or a /var/log/messages.
> 
> /var/log/system.log does not go back that far and I don’t seem to have  
> any old system.logs. Mac OS X 10.3 kept old logs around; why 10.5  
> doesn’t seem to is yet another 10.5 peculiarity for me to investigate,  
> later.
Could you please check the system log WHILE starting the scan? I would
guess that we are hanging in the kernel space (hence the
non-interruptable state of the application) I would hope that there is
something in the system log BEFORE the reboot

best regards,
Louis