/tmp on tmpfs (was: Re: Summary/Minutes for today's FESCo meeting (2012-04-02))

[Lennart Poettering]

On Mon, 02.04.12 20:58, Richard W.M. Jones (rjones at redhat.com) wrote:
Heya,

> The feature page is wrong about "The user experience should barely
> change.  This is mostly a low-level change that has little visibility
> to the user."

Well, i'd claim this is not really user visible if implemented
correctly. This is however visible to the developer.

> tmpfs is different in a number of important ways:
> 
>  - it's very limited in space compared to a real disk

Yes, large files need to be placed on /var/tmp, which is explained in
the feature page.

>  - it doesn't support O_DIRECT

Well, all code using O_DIRECT should probably have a fallback to work
without, anyway, and very likely has. O_DIRECT doesn't really make sense
for tmpfs, it isn't really a feature that currently isn't supported and
could be implemented, it is something that genuinly makes little
sense for tmpfs...

>  - it doesn't support user extended attrs; and not very old kernels
>    didn't support any xattrs at all, meaning things like SELinux
>    labels don't work

We do this change for F18, not for old Fedora with old kernels. SELinux
labels work fine on tmpfs. User xattr patches have recently been posted
on lkml, but indeed are not supported right now.

> All this means it's going to need a bit more testing, since
> potentially any package that stores a file on /tmp should be tested
> and may need to be fixed.

But yes, this is absolutely true. We do expect breakages. That's why our
plan is: Turn on early in the F18 cycle. Fix bugs which appear as they
show up. If they are too many, revert to turned off.

I will post a longer annoucnement of this with suggested fixes to
fedora-devel when we make the upload to turn this on for F18.

Thanks,

Lennart

-- 
Lennart Poettering - Red Hat, Inc.