> Alexander Klimov <ask@wisdom.weizmann.ac.il> writes:
>> Suddenly I obtain access to
>> ULTRIX black 4.3 1 RISC
> Uh ... what kind of processor is that? Offhand I don't see any
> indication that any of the entries in s_lock.h are supposed to work
> for Ultrix.
On closer look I notice that the putative support for machines without
a TEST_AND_SET implementation got broken by careless rearrangement of
the declarations in s_lock.h :-(. I have repaired this, and if you
update from CVS you should find that things compile.
However, you don't really want to run without TEST_AND_SET support;
it'll be dog-slow. Furthermore, the support for machines without
TEST_AND_SET is fairly new. I doubt it existed when the Ultrix port
was last reported to work. So the question above still stands.
I suspect that some one of the implementations in s_lock.h was intended
to be usable on Ultrix, and we've somehow dropped the declarations
needed to make it go. You might want to pull down an old tarball (6.3
or before) and look at how it compiles the s_lock support on Ultrix.
Please send in a patch if you find that one is necessary for s_lock
support on Ultrix.
regards, tom lane