paone@topaz.rutgers.edu (Phil Paone) (11/07/88)
I found what appears to be a bug with the tos command and hard disks. I get the file starting after the firt few lines and then a get the beginning of the next file on the disk at the end. So, the byte count is right, but I am getting parts of two files. I works fine on the floppy drive though. I will look at the source code eventually and report if I find anything else. Does anyone else have this problem? -- Phil Paone paone@topaz.rutgers.edu !rutgers.edu!topaz.edu!ppaone
hcj@lzaz.ATT.COM (HC Johnson) (11/07/88)
tos.c contains several bugs. 1. only 127 clusters can be written correctly on a hard disk fix: on lines 1102 and 1103, remove the (char) 2. depending on the size of the hard disk partition, writing will both damage files and the start of a new file will be missing. This is due to the number of directory blocks (which are scaled to the file system size. fix: change lines 339-343 to +FAT_START )); 3. Very large (16mb tos file systems) don't work at all. This is due to the fat size being > 32k, and sbrk, and read using signed integers. I have fixed it, but its too complicated for this posting. 4. PLEASE note. The effect of error 1., is to permanently damage the tos file system. I was using an empty one, so fixing just required using dd to copy blocks of 0's over the fats and dir. This is DANGEROUS!
johan@nlgvax.UUCP (Johan Stevenson) (11/09/88)
In article <Nov.7.08.05.51.1988.23172@topaz.rutgers.edu> paone@topaz.rutgers.edu (Phil Paone) writes: > >I found what appears to be a bug with the tos command and hard disks. I get the >file starting after the firt few lines and then a get the beginning of >the next file on the disk at the end. So, the byte count is right, >but I am getting parts of two files. I works fine on the floppy drive >though. > >I will look at the source code eventually and report if I find >anything else. Does anyone else have this problem? >-- >Phil Paone >paone@topaz.rutgers.edu >!rutgers.edu!topaz.edu!ppaone Since already some other users have reported similar problems I must conclude that something is wrong here. I consider this a rather serious problem, since it might corrupt your TOS volumes (I do not say it does, but chances are...). So, please avoid the use of the tos(1) command on hard disks until the problems are located and fixed. If some MINIX-ST users have time to dig into this problem I would ver much appreciate that. Since I haven't looked into the tos(1) program myself before, I can offer no immediate help right now, except for putting it very high on priority list. Let us use the power of the user community (not too bg right now, but hopefully growing fast). Please report any progress to me (or to this newsgroup), so that I can sanction a solution as the "official" one, much like Andy Tanenbaum does for MINIX-PC. A note on my mail address: in the announcement of MINIX-ST Andy Tanenbaum mentioned johan@nlgvax.nl as my e-mail address. This is incorrect. It is either ...!mcvax!nlgvax!johan or johan@nlgvax.uucp or, preferably, johan@pcg.philips.nl Sorry for the confusion. Johan W. Stevenson, Philips Research Laboratories, Project Centre Geldrop, Building XR, Room 15, Willem Alexanderlaan 7B, 5664 AN Geldrop, The Netherlands. phone: +31 40 892205 email: johan@pcg.philips.nl