Anselmo-Ed@cs.yale.edu (Ed Anselmo) (02/01/90)
Foo. You don't have to apply all the patches from the R3 xperfmon, you just
have to #ifdef the call to read_names if you're running on a SPARCstation.
New, shorter, improved patch follows:
*** /tmp/,RCSt1a02137 Thu Feb 1 09:17:48 1990
--- system.c Thu Feb 1 09:17:18 1990
***************
*** 440,446 ****
--- 440,448 ----
/* I don't see a way to obtain the information that
* read_names gets so it is commented it out for now.
*/
+ #ifndef sparc
read_names();
+ #endif
#endif /* !PCS */
time(&now);
nintv = now - boottime;
--
Ed Anselmo anselmo-ed@cs.yale.edu {harvard,decvax}!yale!anselmo-edguy@auspex.auspex.com (Guy Harris) (02/18/90)
>Foo. You don't have to apply all the patches from the R3 xperfmon, you just >have to #ifdef the call to read_names if you're running on a SPARCstation. 1) do you mean "SPARCStation" or "SPARCStation-1"? I don't think the other SPARCStations have SBuses, so I'd expect the "read_names()" code to work on them. 2) even if you *did* mean "SPARCStation", "#ifdef sparc" isn't the #ifdef you want, since that'll #ifdef the code out on *any* machine with a SPARC processor, including all Sun-4s, Solbournes, etc.. Assuming it really *can* function with no "disk init info", the proper fix might simply be to, at least in the "#ifdef sun" case, have "read_names()" not bomb out if it can't find "mbdinit"....