Hi,
Daniel Lewart wrote:
> I did find 215 MiB of exact duplicate files.
> ...
> 24724624 Oct 30 10:07 firmware/firmware-atheros_20240909-2_all.deb
> 24724624 Oct 30 10:07
> pool/non-free-firmware/f/firmware-nonfree/firmware-atheros_20240909-2_all.deb
These two most probably share their content in the ISO. So their
duplication would cost less than a kilobyte for the additional directory
record. Reason would be that both files had the same dev_t,ino_t
numbers when the ISO was created. In this case libisofs lets both
directory records point to the same content in the ISO.
You may verify this by a xorriso run:
xorriso -indev debian-testing-amd64-netinst.iso \
-find / -name firmware-atheros_20240909-2_all.deb \
-exec report_lba --
When doing the equivalent with firmware-atheros*.deb in
debian-12.7.0-amd64-netinst.iso i get on stdout:
Report layout: xt , Startlba , Blocks , Filesize , ISO image path
File data lba: 0 , 18483 , 8217 , 16827872 ,
'/firmware/firmware-atheros_20230210-5_all.deb'
File data lba: 0 , 18483 , 8217 , 16827872 ,
'/pool/non-free-firmware/f/firmware-nonfree/firmware-atheros_20230210-5_all.deb'
Decisive is that both files have the same Startlba.
(One could get the content by: dd bs=2048 skip=18483 count=8217 )
> 10024896 Dec 8 21:20 install.amd/gtk/vmlinuz
> 10024896 Dec 8 21:20 install.amd/vmlinuz
> 10024896 Dec 8 21:20 install.amd/xen/vmlinuz
I get from debian-12.7.0-amd64-netinst.iso :
Report layout: xt , Startlba , Blocks , Filesize , ISO image path
File data lba: 0 , 94511 , 3993 , 8177600 ,
'/install.amd/gtk/vmlinuz'
File data lba: 0 , 94511 , 3993 , 8177600 , '/install.amd/vmlinuz'
File data lba: 0 , 94511 , 3993 , 8177600 ,
'/install.amd/xen/vmlinuz'
So these three share their content, too.
debian-12.7.0-amd64-netinst.iso has a size of 631 MiB.
The Linux kernel is not aware of shared content. When mounted as /mnt/iso
i see:
$ du -s /mnt/iso
766644 /mnt/iso
So there is a deduplication of about 135 MiB.
Have a nice day :)
Thomas
