Bug#836075: libfile-stripnondeterminism-perl: hard-link is destroyed after normalization

[AYANOKOUZI, Ryuunosuke]

Package: libfile-stripnondeterminism-perl
Version: 0.023-2
Severity: normal
Tags: patch
User: reproducible-builds at lists.alioth.debian.org
Usertags: toolchain
X-Debbugs-Cc: reproducible-builds at lists.alioth.debian.org

Dear maintainer,

Current file handlers is considering file mode, but other file attributes
(Ex. user ID, group ID, inode number, link counts, ...) are not considered.
It means that a file with same file name before and after normalization
could have different file attributes except for file mode.
This would cause some problem.

Suppose that we have a debian package which is packaged by "debhelper"
and that "debian/rule" execute "rdfind -makehardlinks true"
in "override_dh_auto_install" target to reduce package size
using hard-link of duplicated files in the package [1].

[1] https://wiki.debian.org/dedup.debian.net

"dh_strip_nondeterminism" target is executed
after "override_dh_auto_install" (hard-linking).
In the target, File::StripNondeterminism module
create new normalized file which has different inode number from target file
and rename() the normalized file to the target file it data has changed.
Because inode number has changed after rename().
hard-link status of files by "rdfind" could be changed [2].
This cause duplicated files in final package
in spite of "rdfind" invocation in "dh_strip_nondeterminism" target.

[2] hard-link status could be changed.
$ (umask 0006; convert xc:none /tmp/preserve_file_attributes.v1YRi3/ln_target.png)
$ ln /tmp/preserve_file_attributes.v1YRi3/ln_target.png /tmp/preserve_file_attributes.v1YRi3/ln_hardlink.png
$ stat /tmp/preserve_file_attributes.v1YRi3/ln_hardlink.png /tmp/preserve_file_attributes.v1YRi3/ln_target.png
  File: ‘/tmp/preserve_file_attributes.v1YRi3/ln_hardlink.png’
  Size: 272             Blocks: 8          IO Block: 4096   regular file
Device: fe01h/65025d    Inode: 498297      Links: 2
Access: (0660/-rw-rw----)  Uid: ( 1000/********)   Gid: ( 1000/********)
Access: 2016-08-30 21:53:27.562341916 +0900
Modify: 2016-08-30 21:53:27.558341948 +0900
Change: 2016-08-30 21:53:27.558341948 +0900
 Birth: -
  File: ‘/tmp/preserve_file_attributes.v1YRi3/ln_target.png’
  Size: 272             Blocks: 8          IO Block: 4096   regular file
Device: fe01h/65025d    Inode: 498297      Links: 2
Access: (0660/-rw-rw----)  Uid: ( 1000/********)   Gid: ( 1000/********)
Access: 2016-08-30 21:53:27.562341916 +0900
Modify: 2016-08-30 21:53:27.558341948 +0900
Change: 2016-08-30 21:53:27.558341948 +0900
 Birth: -
$ perl -I lib/ bin/strip-nondeterminism /tmp/preserve_file_attributes.v1YRi3/ln_hardlink.png
$ stat /tmp/preserve_file_attributes.v1YRi3/ln_hardlink.png /tmp/preserve_file_attributes.v1YRi3/ln_target.png
  File: ‘/tmp/preserve_file_attributes.v1YRi3/ln_hardlink.png’
  Size: 155             Blocks: 8          IO Block: 4096   regular file
Device: fe01h/65025d    Inode: 498310      Links: 1
Access: (0660/-rw-rw----)  Uid: ( 1000/********)   Gid: ( 1000/********)
Access: 2016-08-30 21:53:27.850341897 +0900
Modify: 2016-08-30 21:53:27.846341898 +0900
Change: 2016-08-30 21:53:27.846341898 +0900
 Birth: -
  File: ‘/tmp/preserve_file_attributes.v1YRi3/ln_target.png’
  Size: 272             Blocks: 8          IO Block: 4096   regular file
Device: fe01h/65025d    Inode: 498297      Links: 1
Access: (0660/-rw-rw----)  Uid: ( 1000/********)   Gid: ( 1000/********)
Access: 2016-08-30 21:53:27.850341897 +0900
Modify: 2016-08-30 21:53:27.558341948 +0900
Change: 2016-08-30 21:53:27.846341898 +0900
 Birth: -

To fix this issue and
to preserve as much file attribute information of target file as possible,
I prepared a patch to "master" branch of official repository [3].
I would like someone to request review the patch and merge it to the branch.

[3] https://anonscm.debian.org/git/reproducible/strip-nondeterminism.git/log/?h=master

-- System Information:
Debian Release: stretch/sid
  APT prefers unstable
  APT policy: (500, 'unstable')
Architecture: amd64 (x86_64)

Kernel: Linux 3.16.0-4-amd64 (SMP w/2 CPU cores)
Locale: LANG=C, LC_CTYPE=C (charmap=ANSI_X3.4-1968)
Shell: /bin/sh linked to /bin/dash
Init: unable to detect

Versions of packages libfile-stripnondeterminism-perl depends on:
ii  libarchive-zip-perl  1.59-1
ii  perl                 5.22.2-3

libfile-stripnondeterminism-perl recommends no packages.

libfile-stripnondeterminism-perl suggests no packages.

-- no debconf information

Sincerely yours,
Ryuunosuke Ayanokouzi
-- 
AYANOKOUZI, Ryuunosuke <i38w7i3 at yahoo.co.jp>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: 0001-Preserve-file-attribute-information-of-target-file.patch
Type: application/octet-stream
Size: 9403 bytes
Desc: not available
URL: <http://lists.alioth.debian.org/pipermail/reproducible-builds/attachments/20160830/b98615c3/attachment.obj>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 473 bytes
Desc: OpenPGP Digital Signature
URL: <http://lists.alioth.debian.org/pipermail/reproducible-builds/attachments/20160830/b98615c3/attachment.sig>