[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Bug-xorriso] Crash in xorriso with fix

From: Pavel Cahyna
Subject: Re: [Bug-xorriso] Crash in xorriso with fix
Date: Wed, 7 Aug 2019 10:57:37 +0200
User-agent: Mutt/1.11.3 (2019-02-01)


On Thu, Jul 25, 2019 at 11:25:11AM +0200, Thomas Schmitt wrote:
> Hi,
> Eliska Svobodova wrote:
> > Sorry about test ACL, I haven't work with it
> > before and only focused on causing the segfault.
> Insofar your script was very successful.
> The following modified loop produces valid ACL entry texts
>   for i in {1..10000}; do echo user:$i:rwx >> File_with_users; done
> These entries can be read by xorriso's -sefacl* commands, because like
> with setfacl(1) number texts are interpreted as uid numbers.
> (libisofs does this if the text is not found as user name. I am currently
>  not aware what setfacl(1) would do if user "123" has uid 1000.)
> Processing time of the -setacl_list command is about 5 seconds on 3.5 GHz
> Xeon. So it looks like the effectivity of libisofs' ACL processing with
> large lists could be improved.
> I produced an ISO image with the generously sized ACL by
>   valgrind --leak-check=full \
>     xorriso -for_backup \
>             -outdev test.iso \
>             -map . / \
>             -setfacl_list File_with_users
> and then inspected it sucessfully by
>   xorriso -for_backup -indev test.iso -getfacl_r / -- 2>&1 | less
> But i did not dare to restore Test_file to my hard disk. Not a bit curious
> to learn about potential bugs in ext4.
> Command -for_backup implies -acl "on". Without that setting ACL are neither
> stored, nor read-in, nor extracted by -osirrox "on" -extract / ./Text_dir_X
> Have a nice day :)

Eliska has been away recently, so please excuse the delay in replying.
When she will be back next week, we can continue the discussion.

We were actually interested mainly about the memory management problem
and not so much about the actual functionality of supporting such huge
ACL lists, that's why our test was minimal and not realistic, but if you
were able to make it successfuly work, even better! Concerning your
remark about ext4 support, yes, I would expect filesystems to have some
limits (hopefully not bugs, but indeed I would not dare to try on a
production system). If there is a limit in ext4, you might have better
luck by trying XFS or ZFS. I am actually surprised you did not hit some
limit in ISO9660.

Have a nice day as well :)

Regards, Pavel

reply via email to

[Prev in Thread] Current Thread [Next in Thread]