Re: [PATCH v2 0/8] Add O_DENY* support for VFS and CIFS/NFS

[Pavel Shilovsky <piastry-7qunaywFIewox3rIn2DAYQ@public.gmane.org>]

2013/2/5 J. Bruce Fields <bfields-uC3wQj2KruNg9hUCZPvPmw@public.gmane.org>:
> On Tue, Feb 05, 2013 at 03:33:21PM +0400, Pavel Shilovsky wrote:
>> 2013/1/31 J. Bruce Fields <bfields-uC3wQj2KruNg9hUCZPvPmw@public.gmane.org>:
>> > On Thu, Jan 17, 2013 at 08:52:09PM +0400, Pavel Shilovsky wrote:
>> >> This patchset adds support of O_DENY* flags for Linux fs layer. These flags can be used by any application that needs share reservations to organize a file access. VFS already has some sort of this capability - now it's done through flock/LOCK_MAND mechanis, but that approach is non-atomic. This patchset build new capabilities on top of the existing one but doesn't bring any changes into the flock call semantic.
>> >>
>> >> These flags can be used by NFS (built-in-kernel) and CIFS (Samba) servers and Wine applications through VFS (for local filesystems) or CIFS/NFS modules. This will help when e.g. Samba and NFS server share the same directory for Windows and Linux users or Wine applications use Samba/NFS share to access the same data from different clients.
>> >>
>> >> According to the previous discussions the most problematic question is how to prevent situations like DoS attacks where e.g /lib/liba.so file can be open with DENYREAD, or smth like this. That's why one extra flag O_DENYMAND is added. It indicates to underlying layer that an application want to use O_DENY* flags semantic. It allows us not affect native Linux applications (that don't use O_DENYMAND flag) - so, these flags (and the semantic of open syscall that they bring) are used only for those applications that really want it proccessed that way.
>> >
>> > Maybe that's good enough.  A mount flag might be simpler and give
>> > consistent enforcement for all users.
>> >
>> >>
>> >> So, we have four new flags:
>> >> O_DENYREAD - to prevent other opens with read access,
>> >> O_DENYWRITE - to prevent other opens with write access,
>> >> O_DENYDELETE - to prevent delete operations (this flag is not implemented in VFS and NFS part and only suitable for CIFS module),
>> >> O_DENYMAND - to switch on/off three flags above.
>> >
>> > It would be useful to have some really careful documentation of how
>> > these are meant to work.  Maybe try updating the open man page?
>>
>> Yes, that's a good idea. Do you mean smth like this?
>>
>> O_DENYMAND - used to inforce a mandatory share reservation scheme of
>> the file access. If this flag is passed,
>> the open fails with -ETXTBSY in following cases:
>>
>> 1) if O_DENYREAD flag is specified and there is another open with
>> O_DENYMAND flag and READ access to the file;
>> 2) if O_DENYWRITE flag is specified and there is another open with
>> O_DENYMAND flag and WRITE access to the file;
>> 3) if READ access is requested and there is another open with
>> O_DENYMAND and O_DENYREAD flags;
>> 4) if WRITE access is requested and there is another open with
>> O_DENYMAND and O_DENYWRITE flags;
>>
>> Also, if O_DENYDELETE flag is specified and the open succeded, any
>> further unlink operation will fail with -ETXTBSY untill this open is
>> closed. Now this flag is processed by CIFS filesystems only.
>
> Do you need to document interactions with flock() as well?

Yes, this needs to be documented, thanks. I am going to propose ver2
of this patchset (with regards to your suggestions) - will add this
part to open man page patch too.

-- 
Best regards,
Pavel Shilovsky.