wietse@eurifb.UUCP (Wietse Venema) (09/26/87)
In article <105@suprt.UUCP>, jmsully@suprt.UUCP (John M. Sully UUCP: ...!{sun | ucbvax | ihnp4}!amdcad!uport!techs) writes: > In article <207@eurifb.UUCP>, wietse@eurifb.UUCP (Wietse Venema) writes: > > [problem with large-buffer floppy writes/reads in a program that > > spreads the input/output of tar across several floppies] > > As of September 22, 1987 this bug had not been reported to us. We have a > program which we use in-house which writes to the floppy in ~60K chunks > and we do not have a problem with this. > > [remarks about raw devices and tar blocking factors] (the original article appeared in comp.unix.wizards but this seems like a more appropriate newsgroup) In the following examples the kernel starts to print error messages when buffers of 10-20 kb are read/written from/to the raw 360kb floppy device: tar cvf - $* | while dd ibs=512 count=720 obs=20k of=/dev/rdsk/fd148 do (echo "\7next flop: \c"; read answer) >/dev/tty </dev/tty done and: while dd ibs=20k count=18 obs=512 if=/dev/rdsk/fd148 do (echo "\7next flop: \c"; read answer) >/dev/tty </dev/tty done | tar tvf - $* Results are not always the same; sometimes data remains unharmed, sometimes tdd aborts due to an i/o error, and sometimes the kernel dies with a double error. Wietse Venema (uucp: mcvax!eutwc1!wietse) (bitnet: wswietse@heithe5)