Simon Burge writes:

> How far should we take this?  Looking at my 1.5 alpha kernel with
> RaidFrame I see some obvious ones:
[snip]
> thoreau:~ 15648> strings /netbsd | grep 'raidframe error at line %d file %s$'
>  | wc
>       18     126     630

This is related to the RF_PANIC() macro, which I'm hoping to eliminate a 
bunch of them... (there are 51 instances of RF_PANIC in the code).

> thoreau:~ 15649> strings /netbsd | grep 'raidframe error at line %d file %s '
>  | wc
>       44     440    2508

This is related to the RF_ASSERT() macro...  of which there are some 769
instances!  It would be interesting to see how much smaller a kernel gets
by adding the line:

 #define NOASSERT

to the beginning of sys/dev/raidframe/rf_general.h  (and, of course, 
comparing benchmarks of that new kernel against the old one would be 
interesting too :) )

In any event, moving the 'sprintf()' calls from the RF_PANIC() and RF_ASSERT()
macros into their own function(s?) would help a lot here... I'll take care of 
that :)

Later...

Greg Oster