Messages in this thread | | | Date | Fri, 10 Feb 2012 15:50:02 +0200 | Subject | Re: [PATCH] Reduce the number of expensive division instructions done by _parse_integer() | From | Alexey Dobriyan <> |
| |
On Thu, Feb 9, 2012 at 10:14 PM, Linus Torvalds <torvalds@linux-foundation.org> wrote:
Ugh, I'm late to the party.
> Because quite frankly, I think we only support bases 8/10/16 in the > kernel, and if you really have some case where this all is expensive, > it might be better to simply have three different functions for the > three bases. That would turn the multiplies into constants too, and > also simplify the character tests.
base 2 is used to sort of autolimit input to 0/1 characters.
> That said, I can't really see how this could ever be all that hot a > function. Did you ever see it in a profile, or was this all just from > looking at the code?
That's why there was no such check from the beginning -- not performance critical. We could maintain small table from which digit overflow can happen, but since this is already committed...
There is CONFIG_TEST_KSTRTOX, does it still passes?
| |