wee@iris (Chris Wee) (05/03/88)
I have tried to use mkfs in minix to construct a file system us- ing a proto file -- it does not work for me. Is anyone aware of any problems with mkfs? Even the examples in the text and the example files on the master distribution do not work. I get an "unexpected end of file" error on line #4. It is always line #4, no matter what file is being used. The command line I use is, # mkfs /dev/at0 proto just like it says in the MINIX book. Please, if anyone can help, send mail directly me. Thanks in advance :-). Chris Wee wee@iris.ucdavis.edu
driscoll@eecae.UUCP (Mike Driscoll) (05/03/88)
in article <1867@ucdavis.ucdavis.edu>, wee@iris (Chris Wee) says: > > I have tried to use mkfs in minix to construct a file system us- > ing a proto file -- it does not work for me. Is anyone aware of > any problems with mkfs? Even the examples in the text and the > example files on the master distribution do not work. > > I get an "unexpected end of file" error on line #4. It is always > line #4, no matter what file is being used. > > The command line I use is, > # mkfs /dev/at0 proto > just like it says in the MINIX book. > > Please, if anyone can help, send mail directly me. > Thanks in advance :-). > > Chris Wee wee@iris.ucdavis.edu Sorry to post, but I couldn't find a valid path to Andy. I have just had the same problem. I have a copy of the 640K PC distribution, that I want to get up and running on my 512K AT. I will need to remake the kernel at some point to use at_wini.c. As a first step I proceeded to create the AT disk set from the proto files in the proto.a archive on the tools diskette and mkfs (running under MINIX). I get precisely the same error as Chris no matter what proto file is used. Even a minimal proto file (e.g. only a root directory and an empty subdirectory) creates only the root directory before giving the error message. I then made an empty root file system using the mkfs option on the startup menu and copied the neccesary files to create a root file system of the correct size. But, as pointed out in the READ_ME file in the distribution, since mknod cannot add sizes to the special files it creates, all the files in /dev have size 0, instead of 512 (for mem, kmem, etc.), 1200 for at0, at1, and so on. Is there a fix? Thanks, Mike -- Michael A. Driscoll UUCP: ...ihnp4!msudoc!eecae!driscoll Dept. of Electrical Engineering ARPA: driscoll@eecae.ee.msu.edu (35.8.8.151) Michigan State University Office: (517) 353-5337 E. Lansing, MI, 48824
lmjm@doc.ic.ac.uk (Lee McLoughlin) (05/05/88)
I too was unable to use mkfs with a proto type file. The ``fix'' is to rebuild it and reinstall it. Worked for me. Lee.
hall@nosc.NOSC.MIL (Robert R. Hall) (05/05/88)
The problem isn't with your proto file but with mkfs. I replaced mine with version 1.0 which worked fine. Then I put the 1.2 upgrades in , recompiled all my library element, then compiled and relinked mkfs. After all this it now works fine. Sorry I can't tell you which of all my changes corrected the problem. Robert R. Hall hall@nosc.mil
br@laura.UUCP (Bodo Rueskamp) (05/06/88)
The mkfs binary of the Minix 1.2 640k PC distribution is bad. After compiling the source, mkfs works now without errors. -- Bodo Rueskamp br@unido.uucp
baumann@hope.UUCP (Michael Baumann) (05/06/88)
In article <7930@eecae.UUCP> driscoll@eecae.UUCP (Mike Driscoll) writes: >in article <1867@ucdavis.ucdavis.edu>, wee@iris (Chris Wee) says: >> >> I have tried to use mkfs in minix to construct a file system us- >> ing a proto file -- it does not work for me. Is anyone aware of >> any problems with mkfs? Even the examples in the text and the >> example files on the master distribution do not work. >> >> I get an "unexpected end of file" error on line #4. It is always >> line #4, no matter what file is being used. >> stuff deleted >the tools diskette and mkfs (running under MINIX). I get precisely >the same error as Chris no matter what proto file is used. Even a > > Is there a fix? You bet there is, and you're gonna kick yourself when you hear it (I know, 'cause we did :-)). On page 387 of The Book it says, " If the compiler (or, in fact, almost any program) begins acting strange, it is almost always due to its running out of space,..." The solution is to use chmem on mkfs. I can't remember what the correct value is supposed to be, look in the file changemem, I think it's in the tools directory. I don't know why but just about everything was correct in the allocation of stack but this one. In truth, since it is unlikely that you will be running anything else when you do mkfs, I'd just give it the maximum stack it will take: chmem =64000 mkfs. this should solve your problem. ----------------------------------------------------------------------------- "Life is full of little suprises." -- Pandora (as quoted by Robert Asprin) UUCP: {ucbvax!ucdavis,ucsd,ucivax}!ucrmath!hope!baumann or !ucrmath!jinx!baumann
ast@cs.vu.nl (Andy Tanenbaum) (05/08/88)
In article <7930@eecae.UUCP> driscoll@eecae.UUCP (Mike Driscoll) writes: >in article <1867@ucdavis.ucdavis.edu>, wee@iris (Chris Wee) says: >> >> I have tried to use mkfs in minix to construct a file system us- >> ing a proto file -- it does not work for me. Is anyone aware of >> any problems with mkfs? Even the examples in the text and the >> example files on the master distribution do not work. The only thing I can think of, is give mkfs more stack space with chmem. For example, chmem =40000 /usr/bin/mkfs. Many weird problems are caused by not having enough stack space. Andy Tanenbaum (ast@cs.vu.nl)
driscoll@eecae.UUCP (Mike Driscoll) (05/08/88)
in article <701@ast.cs.vu.nl>, ast@cs.vu.nl (Andy Tanenbaum) says: > > In article <7930@eecae.UUCP> driscoll@eecae.UUCP (Mike Driscoll) writes: >>in article <1867@ucdavis.ucdavis.edu>, wee@iris (Chris Wee) says: [ Problem using mkfs with proto files is described. ] > > The only thing I can think of, is give mkfs more stack space with chmem. > For example, chmem =40000 /usr/bin/mkfs. Many weird problems are caused by > not having enough stack space. > > Andy Tanenbaum (ast@cs.vu.nl) As several people pointed out, this does fix the problem. I used chmem =20000 mkfs and it worked fine. The commands makefile uses chmem =2048 for whatever command you are making. This is obviously not correct for mkfs. The book says that whenever something acts funny, try chmem. So, of course, I didn't try that until someone told me to :-) Thanks for the help, everyone! Mike -- Michael A. Driscoll UUCP: ...ihnp4!msudoc!eecae!driscoll Dept. of Electrical Engineering ARPA: driscoll@eecae.ee.msu.edu (35.8.8.151) Michigan State University Office: (517) 353-5337 E. Lansing, MI, 48824
frank@morgan.com (Frank Wortner) (05/09/88)
Mike Driscoll writes: > As several people pointed out, this does fix the problem. I used >chmem =20000 mkfs and it worked fine. The commands makefile uses >chmem =2048 for whatever command you are making. This is obviously >not correct for mkfs. The book says that whenever something >acts funny, try chmem. One thing the book doesn't tell you about is the "changemem" shell script in the distribution. It contains suggested chmem commands for many commands which need more than a miserly 2048 bytes. I've found that it pays to update, improve, and use this script. Frank