Martín Ferrari wrote: > On Mon, Dec 8, 2008 at 09:30, Thiemo Seufer <[EMAIL PROTECTED]> wrote: > > >> Could you tell me if the assembler code generated by gcc looks as a > >> reasonable replacement for the offending code you've patched? > > > > I think __sync_lock_test_and_set should be good enough, AFAICS there's > > no need to do a full c-a-s. > > I was afraid by the warnings in the gcc docs, but seeing that in sparc > has to be implemented in an external subroutine, I'll go with > test-and-set. > > > Another consideration is that upstream might not like to require gcc-4.3, > > and the IRIX platform can't use the same code in that case. > > >From what I could find, the atomic builtins were introduced in 4.1,
For MIPS they are only implemented since 4.3. Thiemo -- To UNSUBSCRIBE, email to [EMAIL PROTECTED] with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]