GNU bug report logs - #53209
stat shows wrong, non-existing device number

Previous Next

To add a comment to this bug, you must first unarchive it, by sending
a message to control AT debbugs.gnu.org, with unarchive 53209 in the body.
You can then email your comments to 53209 AT debbugs.gnu.org in the normal way.

Toggle the display of automated, internal messages from the tracker.

Message #5 received at submit <at> debbugs.gnu.org (full text, mbox):

From: Joachim Wagner <jo2l <at> mqzmail.eu>
To: bug-coreutils <at> gnu.org
Subject: stat shows wrong, non-existing device number
Date: Wed, 12 Jan 2022 11:13:21 +0000

Issue: The device number shown in `stat` output does not match the device 
numbers of any of the devices used by the filesystem.

On my system:

$ stat --version | head -n 1 
stat (GNU coreutils) 8.32

$ cat /etc/os-release | fgrep -i pretty
PRETTY_NAME="openSUSE Leap 15.3"

$ ls -l /dev/nvme0n1p6 /dev/nvme1n1p3
brw-rw---- 1 root disk 259,  6 Jan 12 09:29 /dev/nvme0n1p6
brw-rw---- 1 root disk 259, 12 Jan 12 09:29 /dev/nvme1n1p3

$ fgrep home /etc/crypttab 
cr_home0   /dev/disk/by-id/nvme-RETRACTED1-part6
cr_home1   /dev/disk/by-id/nvme-RETRACTED2-part3

$ ls -l /dev/mapper/cr_home?
lrwxrwxrwx 1 root root 7 Jan 12 09:29 /dev/mapper/cr_home0 -> ../dm-6
lrwxrwxrwx 1 root root 7 Jan 12 09:30 /dev/mapper/cr_home1 -> ../dm-5

$ ls -l /dev/dm-[56]
brw-rw---- 1 root disk 254, 5 Jan 12 09:30 /dev/dm-5
brw-rw---- 1 root disk 254, 6 Jan 12 09:29 /dev/dm-6

$ fgrep home /etc/fstab 
/dev/mapper/cr_home0                       /home        btrfs  defaults              
0  0

$ history | fgrep cr_home | tail -n 3 | head -n 2
  108  2022-01-05 17:01:22 mkfs.btrfs /dev/mapper/cr_home0
  141  2022-01-05 17:16:49 btrfs device add /dev/mapper/cr_home1 /home

$ touch /home/test

$ stat /home/test | fgrep -i device
Device: 3ch/60d Inode: 2143351     Links: 1

$ ls -l /dev | grep "^b" | fgrep 60 | wc -l
0

Expected behaviour: Device should be one (or more) of 

fe06h/65030d  <-- most expected as this is the primary btrfs device
fe05h/65029d  <-- the 2nd device (/dev/cr_home1)
10306h/66310d  <-- backing device for cr_home0 (/dev/nvme0n1p6)
1030ch/66316d  <-- backing device for cr_home1 (/dev/nvme1n1p3)
10300h/66304d  <-- SSD 1 (/dev/nvme0n1)
10309h/66313d  <-- SSD 2 (/dev/nvme1n1)

Speculation: Searching the mailing list for stat + major on https://
www.gnu.org/software/coreutils/ shows results indicating that stat uses only 8 
bit for major and minor. However, since kernel 2.6.0 according to https://
stackoverflow.com/questions/14833467/maximum-values-of-major-and-minor-
numbers-in-linux , the major has 12 bit and the minor 20 bit. Output above 
shows a major with at least 9 bits that may confuse `stat` if it doesn't stay 
with the immediate device number of the top level dm-crypt device.

Message #8 received at submit <at> debbugs.gnu.org (full text, mbox):

From: Joachim Wagner <jo2l <at> mqzmail.eu>
To: bug-coreutils <at> gnu.org
Subject: Re: stat shows wrong, non-existing device number
Date: Wed, 12 Jan 2022 16:14:04 +0000

> and only filesystems

and only some filesystems

Message #11 received at 53209 <at> debbugs.gnu.org (full text, mbox):

From: Joachim Wagner <jo2l <at> mqzmail.eu>
To: 53209 <at> debbugs.gnu.org
Subject: Update on 53209 "stat shows wrong, non-existing device number"
Date: Wed, 12 Jan 2022 16:25:16 +0000

Further reading suggests this a not a bug but a feature and only some 
filesystems choose (major<<8)+minor of the primary device specified when 
mounting the filesystem as the filesystem's device number.

A sentence in the man page explaining this would be helpful.

Message #14 received at submit <at> debbugs.gnu.org (full text, mbox):

From: Joachim Wagner <jo2l <at> mqzmail.eu>
To: bug-coreutils <at> gnu.org
Subject: Re: stat shows wrong, non-existing device number
Date: Wed, 12 Jan 2022 15:35:26 +0000

> Issue: The device number shown in `stat` output does not match the device
> numbers of any of the devices used by the filesystem.

Further reading suggests this a not a bug but a feature and only filesystems 
choose (major<<8)+minor of the primary device specified when mounting the 
filesystem.

I cloned git://git.sv.gnu.org/coreutils to add a sentence on this to stat's 
man page but coreutils/man/stat.x is only a template, missing what I see in 
`man stat` on my system. Trying to find out how to contribute to 
documentation, I encountered the following cycle with no answer:

https://www.gnu.org/help/
https://www.gnu.org/philosophy/free-doc.html
https://www.gnu.org/doc/doc.html
https://www.gnu.org/help/help.html#helpgnu

Joachim

Message #19 received at 53209 <at> debbugs.gnu.org (full text, mbox):

From: Bernhard Voelker <mail <at> bernhard-voelker.de>
To: Joachim Wagner <jo2l <at> mqzmail.eu>, 53209 <at> debbugs.gnu.org
Subject: Re: bug#53209: stat shows wrong, non-existing device number
Date: Wed, 12 Jan 2022 18:37:49 +0100

On 1/12/22 16:35, Joachim Wagner wrote:
> I cloned git://git.sv.gnu.org/coreutils to add a sentence on this to stat's 
> man page but coreutils/man/stat.x is only a template, missing what I see in 
> `man stat` on my system. Trying to find out how to contribute to 
> documentation, I encountered the following cycle with no answer:
> 
> https://www.gnu.org/help/
> https://www.gnu.org/philosophy/free-doc.html
> https://www.gnu.org/doc/doc.html
> https://www.gnu.org/help/help.html#helpgnu

Typically, the first thing to look at are the README* files, especially
'README-hacking'.

Re. man pages of the coreutils:

a) they are generated during the build from the output of "$UTIL --help"
plus the information in "man/$UTIL.x".

b) The information in the man pages should be kept as terse as possible,
and more detailed information should go into the Texinfo manual (which also
gets uploaded to https://www.gnu.org/software/coreutils/manual/ in HTML,
PDF and other formats after each release).

Have a nice day,
Berny

Previous Next

GNU bug tracking system
Copyright (C) 1999 Darren O. Benham, 1997,2003 nCipher Corporation Ltd, 1994-97 Ian Jackson.