paul@ukpoit.co.uk (Paul Wood) (12/12/90)
I recently posted the following ... > My mac has 4mb of memory, no inits, etc. > While making the kernel, actually making idle.c, I got the following message: > "/usr/include/mac/Resources.h", line 54: fatal error - out of memory > make: Error Code 256 > > I would have thought that 4mb was enough. BTW I tried it with disk cache off > as well as disk cache on with 1mb. > > Help Please! I have investigated a bit further, so here are my findings ... It appeared that cc was running out of memory, so quickly turning to the manual page I discovered the -F option, perhaps this would help, unfortuately it is marked [IBM], so I did the next best thing, I amended the Makefile-s in /usr/src/kernel, fs and mm, as follows: CPP=/usr/lib/cpp CPPFLAGS=-DACK CFLAGS=-DACK -T. .c.o: $(CPP) $(CPPFLAGS) $*.c > $*.i $(CC) $(CFLAGS) $*.i -c -o $*.o rm -f $*.i .s.o: $(CC) $(CFLAGS) $*.s -c -o $*.o This had the effect of splitting up the cc into a cpp followed by a cc without the cpp phase (I hope! The manual is somewhat lacking in a full description of the parameters which each phase of cc uses. Can someone post this info?). Also I changed the sizes of the following: /usr/lib/cpp 100000 /usr/lib/cem 150000 /usr/lib/cv 50000 /usr/bin/rmaker 300000 and eventually 600000 (see below) Does anybody have any better recommendations for the size of the above. I guess that these would be machine specific. I have a MacPlus with 4MB memory and a 20MB harddisk (root 200KB, /usr 13MB, /tmp 1MB). With the above changes the make process went much smoother. This is not to say that nothing went wrong. After a make failure I entered another make, and if that failed I entered the following: make -n > /tmp/m sh /tmp/m This seems to suggest that the problem lies with where make calls the various programs, and subsequently receives their exit codes (I think anyway). I had problems with rmaker. Applying the fix which was posted recently solved some of the problems, but it was difficult to decide what size to make it, so I amended the fatal call as follows: fatal("Exhausted mem. Requested Mem=%ld\n",size); Then at least I knew how much memory was being requested (sometimes as much as 375000 bytes). chmem-ing rmaker solved these problems. Then I had finally made the NEWminixboot program. Wheeew! Next I changed the minix/config.h, but that is another story ... ... ... Paul Wood | UUCP Mail: paul@ukpoit.co.uk | 31 Buttermere Drive | Bang-Style: ...!ukc!ukpoit!paul | Dronfield Woodhouse | Voice: +44 246 214256 | Sheffield | FAX: +44 246 214353 | England S18 5PX |