Re: seperator error in __mask_snprintf_len

[Paul Jackson <pj@sgi.com>]

> MASK_CHUNKSZ is a named constant not a variable.  Once we pick a value
> it can never change.  The specified legal values are for 1) varying to
> test for algorithmic correctness, and 2) giving the list of values from
> which we must pick the permanent value before too much more time goes by.

I can see where the testing for algorithmic correctness would be useful.

The final code should reflect the final choices.

> No.  snprintf will fill to the end of the buffer and not place the

Correct - snprintf will not nul terminate if it runs out of buffer.

But then again, I wouldn't expect a "mask_snprintf" (such as might wrap
this __mask_snprintf_len() code) to guarantee nul termination in the
full buffer case either.  Rather the caller has to watch for a returned
length that is too close to the length of the buffer, and handle it as
an error.  If you look at the cpumask_snprintf() calls, they do just
this.

snprintf-like calls should behave like snprintf calls, no fancier.

> I am pretty sure the code is within a hairsbreadth of being minimal.

We'll see.  I'm still poking at it.

-- 
                          I won't rest till it's the best ...
                          Programmer, Linux Scalability
                          Paul Jackson <pj@sgi.com> 1.650.933.1373