Re: [stable] [PATCH] IA64,sparc: local DoS with corrupted ELFs

[Willy Tarreau <w@1wt.eu>]

On Wed, Sep 06, 2006 at 12:25:11PM -0700, Greg KH wrote:
> On Wed, Sep 06, 2006 at 01:12:16PM -0600, Matthew Wilcox wrote:
> > On Wed, Sep 06, 2006 at 11:45:09AM -0700, Greg KH wrote:
> > > On Wed, Sep 06, 2006 at 12:27:33PM -0600, Matthew Wilcox wrote:
> > > > On Wed, Sep 06, 2006 at 11:24:05AM -0700, Linus Torvalds wrote:
> > > > > If MIPS and parisc don't matter for the stable tree (very possible - there 
> > > > > are no big commercial distributions for them), then dammit, neither should 
> > > > > ia64 and sparc (there are no big commercial distros for them either). 
> > > > 
> > > > Erm, RHEL and SLES both support ia64.
> > > 
> > > Yes, but the -stable developers don't build for those arches, that's why
> > > it was missed here.
> > 
> > What's the easiest way to get coverage here?  Sending a parisc
> > workstation or server to someone?  Giving accounts to some/all of the
> > stable team?  Finding someone who cares about parisc to join the stable
> > team?
> 
> How about: Someone from that arch trying out the -stable release
> canidates to make sure it doesn't break anything on their arches /
> favorite machine?

IMHO it's even simpler than that. You already announce release candidates
with what you intend to push into next -stable. Those who complain that
-stable breaks on them just get what they deserve. They're free to announce
the problem and even provide a patch in order to fix the problem in next
-stable ASAP, but I find it a bit easy to complain about the -stable team
that some fixes break a few rarely tested pieces of software !

I'd prefer that we get slightly more -stable releases with a few ones
potentially wrong on rare occasions, than fewer ones which get released
only once everyone agrees (ie mostly never).

> And no, I really don't want a parisc machine here :)

I have one right here serving my web pages, but I have to check that
my toolchain is still OK (I don't build on it - 32 MB, NFS root). You
don't know what you're missing :-)

> thanks,
> 
> greg k-h

Regards,
Willy