Re: [PATCH 0/3] Reduce impact of overlayfs fake path files

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

 



On Fri, 9 Jun 2023 at 09:32, Amir Goldstein <amir73il@xxxxxxxxx> wrote:
>
> Miklos,
>
> This is the solution that we discussed for removing FMODE_NONOTIFY
> from overlayfs real files.
>
> My branch [1] has an extra patch for remove FMODE_NONOTIFY, but
> I am still testing the ovl-fsnotify interaction, so we can defer
> that step to later.
>
> I wanted to post this series earlier to give more time for fsdevel
> feedback and if these patches get your blessing and the blessing of
> vfs maintainers, it is probably better that they will go through the
> vfs tree.
>
> I've tested that overlay "fake" path are still shown in /proc/self/maps
> and in the /proc/self/exe and /proc/self/map_files/ symlinks.
>
> The audit and tomoyo use of file_fake_path() is not tested
> (CC maintainers), but they both look like user displayed paths,
> so I assumed they's want to preserve the existing behavior
> (i.e. displaying the fake overlayfs path).

I did an audit of all ->vm_file  and found a couple of missing ones:

dump_common_audit_data
ima_file_mprotect
common_file_perm (I don't understand the code enough to know whether
it needs fake dentry or not)
aa_file_perm
__file_path_perm
print_bad_pte
file_path
seq_print_user_ip
__mnt_want_write_file
__mnt_drop_write_file
file_dentry_name

Didn't go into drivers/ and didn't follow indirect calls (e.g.
f_op->fsysnc).  I also may have missed something along the way, but my
guess is that I did catch most cases.

Thanks,
Miklos



[Index of Archives]     [Linux Filesystems Devel]     [Linux NFS]     [Linux NILFS]     [Linux USB Devel]     [Linux Audio Users]     [Yosemite News]     [Linux Kernel]     [Linux SCSI]

  Powered by Linux