Re: [PATCH 6.12.y] mm/mmap: fix __mmap_region() error handling in rare merge failure case

["Liam R. Howlett" <Liam.Howlett@oracle.com>]

* Greg KH <gregkh@linuxfoundation.org> [241119 09:17]:
> On Mon, Nov 18, 2024 at 03:32:14PM -0500, Liam R. Howlett wrote:
> > Okay, before I get yelled at...
> > 
> > This commit is only necessary for 6.12.y until Lorenzo's other fixes to
> > older stables land (and I'll have to figure out what to do in each).
> > 
> > The commit will not work on mm-unstable, because it doesn't exist due to
> > refactoring.
> > 
> > The commit does not have a tag about "upstream commit" because there
> > isn't one - the closest thing I could point to does not have a stable
> > git id.
> > 
> > So here I am with a fix for a kernel that was released a few hours ago
> > that is not necessary in v6.13, for a bug that's out there on syzkaller.
> > 
> > Also, it's very unlikely to happen unless you inject failures like
> > syzkaller.  But hey, pretty decent turn-around on finding a fix - so
> > that's a rosy outlook.
> 
> Why isn't this needed in 6.13.y?  What's going to be different in there
> that this isn't needed?

The code has been refactored and avoids the scenario.  I'd name the
refactoring commit as the upstream commit, but it does not have a stable
git id as it's in mm-unstable.  So I'm at a bit of a loss of how to
follow the process.

> 
> Do you just want me to take this for the 6.12.y tree now?  I'll be glad
> to, just confused a bit.

Yes, please.


Thanks,
Liam