🚀 go-pugleaf

>In article <1992Sep25.084830.10568@cs.sfu.ca> oneill@cs.sfu.ca (Richard O'Neill) writes:
>>Matthew Dillon <dillon@overload.Berkeley.CA.US> writes:
>
>This sound horribly rude, but do you know what you are doing when you
>attempt to use mmap (i.e. have you just mmap before on another Unix, such
>as SunOS)?

    Yes, I'm quite familar with mmap() on SysV and, in fact, the test worked
    on a SysV machine.

    Note that there could have been a problem... I'm not sure whether BSD
    mmap() semantics are the same as SysV mmap() semantics and, since I
    could find no prototype for mmap() on the NeXT, it could be they are
    different and my arguments were horribly wrong for the call.

    Still, I think that if NeXT had a working mmap() (i.e. shared writable
    pages), they would hype it up, and they haven't peeped.

>I don't know whether mmap is broken or not in NS3.0, but I can say
>that the mmap system call causes a real routine in the kernel (not
>some dummy stub), a routine called '_smmap' for those who like
>rumaging about in /sdmach, and the code there certainly looks like it
>is at the very least *trying* to implement the desired functionality.
>
>If someone has a small piece of test code for mmap (that they know
>works other machines), I'm happy to try it out and post the results.

    I would be interested too.	I really do want to get writable shared
    memory working between unassociated processes!

					-Matt


>Best Regards,
>
>    Richard.
>--
>Composing a suitably apt and witty | Use one of the addresses below, no others
>signature and disclaimer is left   | oneill@fornax.UUCP | NeXTmail just about
>as an exercise for the reader.     | oneill@cs.sfu.ca	 | OK, but not >100K !

--

    Matthew Dillon		dillon@Overload.Berkeley.CA.US
    1005 Apollo Way		uunet.uu.net!overload!dillon
    Incline Village, NV. 89451	ham: KC6LVW (no mail drop)
    USA 			Sandel-Avery Engineering (702)831-8000