On Mon, 6 Nov 2000, Jeroen Ruigrok van der Werven wrote:

> -On [20001106 17:10], John W. De Boskey ([EMAIL PROTECTED]) wrote:
> >tar: cdrom/disc2/dev/acd0t39: minor number too large; not dumped
> >
> >   Is this the expected behaviour, or is it worth digging into
> >tar and making some fixes?
> 
> I am hoping to update tar soon these weeks.

Don't forget to merge all the FreeBSD changes including the ones that
print this message.  For the file:

    crw-rw----  1 root  operator    9, 0x08010002 Feb 12  1995 fd0.1200

gnu tar (v.1.13) prints the user-unfriendly message:

    tar: minor_t value 134283266 too large (max=16777215)

while FreeBSD tar prints:

    tar: fd0.1200: minor number too large; not dumped

It is not easy to see that 134283266 is 0x08010002 or that the file is
fd0.1200 (the limit 16777215 is 0xFFFFFF -- 24 bits).

gnu tar 1.13 actually dumps acd0t39, and FreeBSD tar reads back the
result correctly.  This is because gnu tar 1.13 permits 24-bit minor
numbers, while the FreeBSD version only permits 21-bit ones since it
requires a byte of null padding in the minor number field.  I seem to
have misread the gnu tar 1.13 sources in the following commit:

RCS file: /home/ncvs/src/gnu/usr.bin/tar/create.c,v
Working file: create.c
head: 1.7
...
----------------------------
revision 1.7
date: 1999/08/11 08:03:39;  author: bde;  state: Exp;  lines: +15 -2
Support 21-bit minor numbers.  Avoid wasting a byte in their octal
representation by generating the same format as tar-1.13 (use a single
space as the terminator for 7-digit octal numbers).  This is POSIX.1
conformant (2-byte terminators are just a bug or historical wart in
old versions of gnu tar).  All devices created by `MAKEDEV all' except
rsa0.ctl can now be handled by tar(1).
----------------------------

Previous versions of FreeBSD wasted 2 bytes for null padding.  The
current version still wastes (?) 1 byte relative to gnu tar 1.13.

Bruce



To Unsubscribe: send mail to [EMAIL PROTECTED]
with "unsubscribe freebsd-current" in the body of the message

Reply via email to