wheels@mks.UUCP (Gerry Wheeler) (11/17/88)
I have just recently recompiled MINIX-ST, and thought I would mention some of the steps I had to use. The first thing I came across was a lack of stack space in cem when compiling files in the kernel. I have now increased cem's stack+heap space to 70000. Of course, that made cem too large for my memory, because I had set up a 512K ram disk. So, I trimmed it down to 200K. Now I have found that fsck won't compile because of lack of space in /tmp. You can't win, eh? I suspect there is a balance between stack space in cem and ram disk size that will work. (Or maybe I'll use the -T option to put the temp files elsewhere when compiling fsck.) ------------------- The other day I mentioned that I had tricked the BMS formatter into giving me a 19 megabyte partition. I have now used the Supra utilities to examine the partition table created. It is correct, but there was one thing I did notice -- the BMS utilities tend to round the numbers to the nearest megabyte, whereas the Supra utilities report the partition size in blocks. I now realise that the partition is smaller than I thought. Why did mkfs allow me to make a file system larger than the partition? Well, I guess I'll peek at the source to answer that myself. In general, things are going well. Happy MINIXing. -- Gerry Wheeler Phone: (519)884-2251 Mortice Kern Systems Inc. UUCP: uunet!watmath!mks!wheels 35 King St. North BIX: join mks Waterloo, Ontario N2J 2W9 CompuServe: 73260,1043
johan@nlgvax.UUCP (Johan Stevenson) (11/18/88)
In article <561@mks.UUCP> wheels@mks.UUCP (Gerry Wheeler) writes: > >I have just recently recompiled MINIX-ST, and thought I would mention >some of the steps I had to use. > >The first thing I came across was a lack of stack space in cem when >compiling files in the kernel. I have now increased cem's stack+heap >space to 70000. You are advised to do so in the user manual (Section 2.5, page 9). > >Of course, that made cem too large for my memory, because I had set up a >512K ram disk. So, I trimmed it down to 200K. Even with 512K ram disk and 70000 stack space for cem you can compile. You can NOT just say 'make'. You have to use the 'make -n' trick: $ make -n >x $ . x Unfortunately this is NOT documented in the user manual. However, the user manual (Section 3.4, page 25) describes the 'cc -vn' trick that works in even more restricted situations, but is a little more cumbersome: $ >job $ make CC='cc -vn 2>>job' $ . job Hope this clarification helps. > >Now I have found that fsck won't compile because of lack of space in >/tmp. You can't win, eh? I suspect there is a balance between stack >space in cem and ram disk size that will work. (Or maybe I'll use the >-T option to put the temp files elsewhere when compiling fsck.) This is indeed the right solution once you have trimmed down the ROOT file system (see Section 3.4, page 23). -- Johan W. Stevenson johan@pcg.philips.nl Philips Research