[MODERATED] Re: [PATCH 0/5] Slow Randomizing Boosts DoS v4.4.219 backport 0

On Mon, Apr 27, 2020 at 09:47:11PM +0200, speck for Thomas Gleixner wrote:
> Thomas Gleixner <[email protected]> writes:
> > speck for mark gross <[email protected]> writes:
> >> From: mark gross <[email protected]>
> >> Subject: [PATCH 0/5] Slow Ranomizing Boosts DoS linux-4.4.y backport
> >>
> >> This is a backport of the backports done by Thomas for linux-4.14.y to
> >> linux-4.4.y.
> >>
> >> The backport needed a backport of one additional dependency for
> >> cpu_data-x86_mask getting renamed to x86_stepping.
> >>
> >> This was tested by the test engineer who tested the other SRBDS implementations
> >> and it seems to work.
> >>
> >>
> >> Jia Zhang (1):
> >>   x86/cpu: Rename cpu_data.x86_mask to cpu_data.x86_stepping
> >
> > This one is missing and rightfully so. Maybe you get your mail setup for
> > sending patches right some day in the future.
> >
> > This one has in the mail headers:
> >
> >   In-Reply-To: <[email protected]>
> >   References: <[email protected]>
> >   From: Jia Zhang <[email protected]>
> >   Date: Mon, 1 Jan 2018 09:52:10 +0800
> >
> > While it's correct that the patch is from Jia and was written on Jan. 1
> > 2018, neither the From: nor the Date: header in the mail are correct.
> >
> > That's commit b399151cb48db30ad1e0e93dd40d68c6d007b637 upstream. Do we
> > really want to backport that thing to 4.4 or just live with the uglies
> > of x86_cpuinfo::x86_mask ?
> 
> Fished it out from the moderator mail, applied it to 4.4.y and then ran:
> 
> # git grep 'x86_mask'
> arch/x86/kernel/cpu/amd.c:      if (c->x86_model <= 1 && c->x86_mask <= 1)
yeah, I caught that in my backport that failed to send for some reason.  There
were a few places needing fixups.

> I clearly can see how that was tested.
I didn't test your backport of course.
But, I did do a allyesconfig build that hit that and then I followed up with a
similar git grep.


FWIW :
1915  git checkout srbs-linux-4.4.y
1916  git cherry-pick 06be007aa436a

Then I did a few fixups including the same git grep x86_mask and allyes and
allmod builds.

--mark