Some comments on the i386 implementation of smp_lock.h/locks.S
which recently appeared in the kernel:
I agree with most of your commentary. And I bet some of the
improvements will show up. The best test is lmbench's lat_syscall on
an SMP system to see what lock_kernel()'s true overhead really is.
I'll try to merge your stuff in for 2.1.26 if Alan doesn't get to it
first ;-)
---------------------------------------------////
Yow! 11.26 MB/s remote host TCP bandwidth & ////
199 usec remote TCP latency over 100Mb/s ////
ethernet. Beat that! ////
-----------------------------------------////__________ o
David S. Miller, davem@caip.rutgers.edu /_____________/ / // /_/ ><