Re: Invalid filenames found by opreport

[John L. <le...@mo...>]

On Tue, Jun 09, 2009 at 09:46:16PM +0200, Robert Richter wrote:

> > > In sync_buffer() the data is copied to the event buffer. Since the
> > > buffer state is reset at buffer boundaries (when no buffer data is
> > > available), there could be corrupt data too.
> > 
> > I'm not clear on the problem, could you explain it further? As the CPU
> > buffer is a ring buffer, when reading, it's not possible for anything
> > else to be writing to those buffer slots.
> 
> Assuming a backtrace of a sequence is not yet written completely and
> the variable 'available' points in the middle of it, next time

available is just a count of the slots ready to read from the CPU
buffer. It doesn't point anywhere. Do you mean tail_pos?

Yes, this looks like a problem. buffer_sync() is starting off in a
kernel context. It shouldn't be calling add_sample() until we've had a
clear indication of kernel mode or we get an mm to use. Good spot.

regards
john