>
> On Fri, Oct 25 2002, Jens Axboe wrote:
[...]
> of course, and likewise for the cluster check. I'll cut a
> clean version
> tomorrow, I'm out for today..
>
Ok, Thanks. I'm out 'til Monday.
What originally drove me down this road was not wanting to
have:
struct scatterlist tmpsg[XXX]; /* 8 kbytes */
on the stack for every command when XXX might be 512, which
was what I was expecting when I started writing this patch.
Once I got it working, empirically I saw the most I ever
get is 64 SGs. I don't know if that's a kernel limit, or
if I just haven't beat on the system hard enough.
In hindsight, perhaps it would have been better for me to just put
it on the stack like before, only moreso, and just change the
driver to use more SGs, and only later tackle the stack problem,
if necessary. As you noted, it's not clear that the overhead of
the function call per scatter gather element isn't worse than
just reformatting the entire scatter gather list in a separate
pass, as is currently done (before my patch).
Have a good weekend.
-- steve
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
This archive was generated by hypermail 2b29 : Thu Oct 31 2002 - 22:00:29 EST