read_file_perms vs mmap_read_file_perms

read_file_perms vs mmap_read_file_perms

[Russell Coker]

define(`read_file_perms',`{ getattr open read lock ioctl }')                                                                                                                                                                       
define(`mmap_read_file_perms',`{ getattr open map read ioctl }')

I think that the general expectation would be that mmap_read_file_perms is a 
superset of read_file_perms.  Is there any reason why mmap_read_file_perms 
doesn't include lock permission?  If not I think we should add it to avoid 
surprises.

-- 
My Main Blog         http://etbe.coker.com.au/
My Documents Blog    http://doc.coker.com.au/

Re: read_file_perms vs mmap_read_file_perms

[Chris PeBenito]

On 2/9/19 5:25 AM, Russell Coker wrote:
> define(`read_file_perms',`{ getattr open read lock ioctl }')
> define(`mmap_read_file_perms',`{ getattr open map read ioctl }')
> 
> I think that the general expectation would be that mmap_read_file_perms is a
> superset of read_file_perms.  Is there any reason why mmap_read_file_perms
> doesn't include lock permission?  If not I think we should add it to avoid
> surprises.

You are correct, it should be a proper superset.  However, my preference 
would be to go the other way and eliminate lock from read_file_perms, 
which is what I'm trying to do with mmap_read_file_perms not being a 
proper superset.

-- 
Chris PeBenito

Re: read_file_perms vs mmap_read_file_perms

[Russell Coker]

Now that we have just had a release it's a good time for changes that have the potential to break things. So removing lock now would be ok I guess.

On 10 February 2019 1:07:22 am AEDT, Chris PeBenito <pebenito@ieee.org> wrote:
>On 2/9/19 5:25 AM, Russell Coker wrote:
>> define(`read_file_perms',`{ getattr open read lock ioctl }')
>> define(`mmap_read_file_perms',`{ getattr open map read ioctl }')
>> 
>> I think that the general expectation would be that
>mmap_read_file_perms is a
>> superset of read_file_perms.  Is there any reason why
>mmap_read_file_perms
>> doesn't include lock permission?  If not I think we should add it to
>avoid
>> surprises.
>
>You are correct, it should be a proper superset.  However, my
>preference 
>would be to go the other way and eliminate lock from read_file_perms, 
>which is what I'm trying to do with mmap_read_file_perms not being a 
>proper superset.

-- 
Sent from my Huawei Mate 9 with K-9 Mail.