10 Dec 2007 11:24
Re: vax needs atomic ops
Johnny Billquist <bqt <at> softjar.se>
2007-12-10 10:24:35 GMT
2007-12-10 10:24:35 GMT
Well, if someone can fix the latest gcc, which broke the VAX again, I could maybe look at this. (But why do you want compare-and-swap? Why wouldn't compare-and-set suffice?) Johnny Andrew Doran skrev: > Hi, > > We're nearly ready with atomic ops in the kernel, however the code for vax > needs to be written. The minimum that needs to be provided is an atomic > compare and swap function and the rest can be provided by C code. > > There is already and atomic compare and _set_, for the reader/writer locks > in lock_stubs.S. The description of how the CAS function should behave is > below. It needs to be known by a few different names, this is from the arm > port: > > STRONG_ALIAS(_atomic_cas_ulong,_lock_cas) > STRONG_ALIAS(atomic_cas_ulong,_lock_cas) > STRONG_ALIAS(_atomic_cas_32,_lock_cas) > STRONG_ALIAS(atomic_cas_32,_lock_cas) > STRONG_ALIAS(_atomic_cas_uint,_lock_cas) > STRONG_ALIAS(atomic_cas_uint,_lock_cas) > STRONG_ALIAS(_atomic_cas_ptr,_lock_cas) > STRONG_ALIAS(atomic_cas_ptr,_lock_cas) > > Any takers? >(Continue reading)
RSS Feed