[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: xorriso, create an iso archive from a directory which has NTFS ACLs/
From: |
Thomas Schmitt |
Subject: |
Re: xorriso, create an iso archive from a directory which has NTFS ACLs/xattr |
Date: |
Wed, 14 Sep 2022 10:31:24 +0200 |
Hi,
Kristaps Esterliņš wrote:
> service:/srv/backup/samba/service/service-13-09-2022# /home/src/acl_test
> service/Lietotājs/
> ACL_TYPE_DEFAULT :
> user::rwx
> ...
> ACL_TYPE_ACCESS :
> user::rwx
> ...
No error from acl_get_file(3) or acl_to_text(3). That's strange.
I had expected that a failure of these functions causes the xorriso error
message (actually the error is reported by libisofs).
Did you try with one of the files which were reported by xorriso ?
E.g.
/home/src/acl_test
/srv/backup/samba/service/service-13-09-2022/service/folder1/garantijuveidi.pdf
I will ponder about better diagnosis on the way to the error message
and about other possible reasons for the problem.
----------------------------------------------------------------------
Test proposal to find out whether the problem is with ACL at all:
After the original xorriso run with -for_backup, which reports the errors,
do you see ACL recorded with the problematic files ?
xorriso -for_backup \
-indev /srv/samba/service/folder/service-13.09.2022.iso \
-getfacl /folder1/garantijuveidi.pdf --
(Or
-getfacl 'folder1/garantiju veidi.pdf' --
if there is indeed a blank in the name. Your HTML text says:
> '/srv/backup/samba/service/service-13-09-2022/service/folder1/garantij=
u<br>
> veidi.pdf'<br>
)
If ACL were determined successfully, then the result should roughly look
like the result of getfacl(1) of ./acl_test with the original file .
If xorriso sees the non-trivial ACLs in the ISO then the problem was not
with ACL but possibly with xattr.
In this case try a xorriso ISO production run with
-for_backup -xattr off
which is equivalent to
-acl on -md5 on -hardlinks on
If this run emits no "Error with reading ACL or xattr", then the attempt to
read xattr by listxattr(2) would be the new suspect.
I would then make a test program similar to acl_test.c which exercises
the listxattr gesture from libisofs/aaip-os-linux.c .
Have a nice day :)
Thomas