jb107@prism.gatech.EDU (Jim Burns) (03/31/91)
Is the demo disk (deliberately ?) crippled, or does minix work this way too? I'm on a 12 Mhz AT clone w/512K, and I'm trying some simple things right out of the demo manual: 1) mined doesn't work unless you type 'exec mined' at the keyboard, in which case you get logged off when you finish. This also would make it useless for acting as a pager by piping something into it, if pipes worked ... 2) but pipes don't work. Simple things like ls|mined or head -15 /etc/text| tail -5, which are straight out of the demo manual, don't work. Error message is always <program-name>: not found. Fully qualifying the path just gets <program-name>: cannot execute. These are the same errors you get in 1) without prefacing mined with exec. 3) certain commands cannot be put in scripts (ls, head, tail) w/o prefacing them with exec, which makes it impossible to put anything after that in the script. Same errors as above. Problems happen whether logged in as root or ast. -- BURNS,JIM (returned student & GT Research Institute staff member) Georgia Institute of Technology, Georgia Tech Station 30178, Atlanta Georgia, 30332 | Internet: jb107@prism.gatech.edu uucp: ...!{decvax,hplabs,ncar,purdue,rutgers}!gatech!prism!jb107
ast@cs.vu.nl (Andy Tanenbaum) (04/01/91)
In article <25289@hydra.gatech.EDU> jb107@prism.gatech.EDU (Jim Burns) writes: >Is the demo disk (deliberately ?) crippled, or does minix work this way >too? I'm on a 12 Mhz AT clone w/512K, and I'm trying some simple things >right out of the demo manual: > >1) mined doesn't work unless you type 'exec mined' at the keyboard I have never heard anything like that before. I suspect you are trying this on a machine with only a tiny bit of memory. If mined runs with exec but not otherwise, you are within one shell of fitting into memory. That means something is radically wrong. Try the demo on a machine with 512K. Andy Tanenbaum (ast@cs.vu.nl)
adrie@philica.ica.philips.nl (Adrie Koolen) (04/02/91)
In article <25289@hydra.gatech.EDU> jb107@prism.gatech.EDU (Jim Burns) writes: >Is the demo disk (deliberately ?) crippled, or does minix work this way >too? I'm on a 12 Mhz AT clone w/512K, and I'm trying some simple things >right out of the demo manual: > >1) mined doesn't work unless you type 'exec mined' at the keyboard, in >which case you get logged off when you finish. This also would make it >useless for acting as a pager by piping something into it, if pipes worked >... I think that you have not enough memory to execute `big' commands like mined. Hit F1 to see how much free memory you have at a certain moment. If the shell can't fork and/or exec a program, it reports with `: not found' or `: cannot execute' messages. This also happens when the process table is (almost) full! It helps to use `exec' because the shell then removes itself from memory, thus providing somewhat more space for other programs. Adrie Koolen (adrie@ica.philips.nl) Philips Innovation Centre Aachen
jb107@prism.gatech.EDU (Jim Burns) (04/04/91)
In article <25289@hydra.gatech.EDU>, I write: > I'm on a 12 Mhz AT clone w/512K, and I'm trying some simple things > right out of the demo manual: The consensus of opinion on my problem is that I don't have enough ram to run the demo. Can any one explain why, given that the boot stats say there is 156k for minix, 192k ram disk, leaving 164k left for programs, and none of the programs on the disk are bigger than 30k? Someone suggested I move my root partition to a floppy. I was able to mkfs a floppy (tho' that's another command that had to be run as exec mkfs from the command line, logging you out when finished), and copy all my ram disk directories and their contents over to the floppy (tho' I have no way of re-creating the device files, or copying a boot file over). Naturally, this won't boot. Any way to boot off of the demo, but mount this floppy as root? The 'r' command in the boot menu doesn't cut it. Alternatively, is there any way to reduce the size of the ram disk *after* booting? Any ideas at all? Thanx to all the people who mailed me so far, including Andy Tannenbaum. At least from their surprise, it is evident that this is *not* normal behavior. -- BURNS,JIM (returned student & GT Research Institute staff member) Georgia Institute of Technology, Georgia Tech Station 30178, Atlanta Georgia, 30332 | Internet: jb107@prism.gatech.edu uucp: ...!{decvax,hplabs,ncar,purdue,rutgers}!gatech!prism!jb107
jb107@prism.gatech.EDU (Jim Burns) (04/05/91)
Kudos go to Paul Allen for the solution. He suggested 'cp /dev/ram /dev/fd0' (I picked /dev/fd1) to make a proper copy of the root partition. Thus with the copy in B: and the demo in A: and the boot menu on the screen, type 'r' to set 'f1' as the root partition, and '=' to finish booting. This freed up enough memory so that all my problems went away. (There is one minor problem with 'df' - if you leave the ram disk size at 0 when you boot, 'df' says it can't locate the superblock on /dev/ram, and if you make the ram disk size 10k, it says that /dev/ram is not a proper filesystem. Apparently, /dev/ram is hardwired into the demo kernel areas that df is looking at. The workaround is to just fully specify the file- system devices that you want stats on.) Andy Tanenbaum and Clarence Wilkerson pointed out the use of the F1 key to monitor memory usage, and I was getting a lot of fragmentation. Paul also pointed out that even tho' the procs are small, they can have up to 64k in stack. Bruce Evans made a number of interesting suggestions that unfortunately require tools that don't ship with the demo, like mknod and dd. Of particular interest was: >there any way to reduce the size of the ram disk *after* booting? Any >ideas at all? Try reducing the size at offset 0x402 in /dev/ram. These bytes should be 0xC0 0x00 now (192). Files above the new limit would be lost. It may help to remove unused files first (there aren't that many). It is best to run fsck to clean up the mess after changing the size. I supect fsck was too big to fit on the demo. Thanx to all who responded. -- BURNS,JIM (returned student & GT Research Institute staff member) Georgia Institute of Technology, Georgia Tech Station 30178, Atlanta Georgia, 30332 | Internet: jb107@prism.gatech.edu uucp: ...!{decvax,hplabs,ncar,purdue,rutgers}!gatech!prism!jb107