Strange behavior of kernel module (output terminated)

[John Baldwin]

On Tuesday 23 March 2010 7:22:47 am Dag-Erling Smørgrav wrote:
> Dmitry Krivenok <krivenok.dmitry at gmail.com> writes:
> > /* The function called at load/unload. */
> > static int event_handler(struct module *module, int event, void *arg)
> > {
> >   int e = 0; /* Error, 0 for normal return status */
> >   switch (event)
> >     {
> >       case MOD_LOAD:
> >         uprintf("Hello FreeBSD kernel!\n");
> 
> I'm not sure it's such a good idea to use uprintf() here.  The event
> handler can be called in non-process context.

If you are doing a kldload post-boot it is actually done from some sort of 
process context.  We run module handlers synchronously from the kldload(2) 
syscall.

> >         int i = 0;
> >         for(i = 0; i < 1000; i++)
> >           {
> >             uprintf("%3d ", i);
> >             if(! (i % 10) ) uprintf("\n");
> >           }
> 
> (i % 10) is not a predicate.  The test should be if (i % 10 == 0).
> 
> If you want to work on FreeBSD, I recommend you get used to the
> FreeBSD coding style; see 'man 9 style'.

True, but that doesn't explain the behavior he sees.

> > As you can see the loop was terminated after i==466.  I tried to
> > load/unload the module many times but the last printed number was
> > always 466.
> 
> You filled up a buffer somewhere...

Does uprintf() require the caller to flush the output to the tty somehow?  If 
so, that seems to be a bug.  Nothing in the uprintf(9) manpage suggests that 
the output should be manually flushed.

-- 
John Baldwin