Messages in this thread | | | Date | Fri, 21 May 1999 22:18:04 +0200 (CEST) | From | Andrea Arcangeli <> | Subject | Re: Bad apache perfomance wtih linux SMP |
| |
On Fri, 21 May 1999, Andi Kleen wrote:
>be fixed to have fine grained locks - but just a bandaid to fix the worst >scaling problems in 2.2 with minimal changes.
Agreed. But the changes you are talking about are not so minimal, and neither are obviously right (they are safe I agree but not obviously right), since as I just said an unlock_kernel won't release the kernel lock for sure. If you change the caller over the time, then you'll find yourself not releasing the kernel lock anymore.
If something has to be done in the meantime, according to me the _right_ solution is my one, where I use the property of all copy_from/to_user call. They all can sleep so we can safely drop the kernel lock by hand there at the higher level. Bloating the kernel with unlock_kernel all over the place in the tcp/ip ext2 etc... code would be a really a dirty choice according to me. That's at least my point of view.
Comments as always are welcome :-).
Andrea Arcangeli
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
| |