Received-SPF: pass (google.com: best guess record for domain of linux-kernel-owner@vger.kernel.org designates 209.132.180.67 as permitted sender) client-ip=209.132.180.67;
MIME-Version: 1.0
References: <154180834787.2060925.7738215365584115230.stgit@dwillia2-desk3.amr.corp.intel.com>
 <7590EF40-B0CF-40BD-9D29-FB731A2A2E3A@amacapital.net> <CAPcyv4hXzBT1LkEdBYw_mkrKV+SK1LRE2PwJ5=T+e7GxWbXaNQ@mail.gmail.com>
 <9DFD717D-857D-493D-A606-B635D72BAC21@amacapital.net>
In-Reply-To: <9DFD717D-857D-493D-A606-B635D72BAC21@amacapital.net>
From:   Dan Williams <dan.j.williams@intel.com>
Date:   Sat, 10 Nov 2018 16:31:45 -0800
Message-ID: <CAPcyv4i7m+FuwFKLyQ6StNL2BCiNv7=O_Y9J4Ojqek7nC9mNjw@mail.gmail.com>
Subject: Re: [PATCH] x86/mm/pat: Fix missing preemption disable for __native_flush_tlb()
To:     Andy Lutomirski <luto@amacapital.net>
Cc:     Thomas Gleixner <tglx@linutronix.de>,
        Sebastian Andrzej Siewior <bigeasy@linutronix.de>,
        Andy Lutomirski <luto@kernel.org>,
        Dave Hansen <dave.hansen@linux.intel.com>,
        Peter Zijlstra <peterz@infradead.org>,
        Borislav Petkov <bp@alien8.de>,
        stable <stable@vger.kernel.org>, X86 ML <x86@kernel.org>,
        Linux Kernel Mailing List <linux-kernel@vger.kernel.org>,
        "Kirill A. Shutemov" <kirill.shutemov@linux.intel.com>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

[ added Kirill ]

On Sat, Nov 10, 2018 at 4:19 PM Andy Lutomirski <luto@amacapital.net> wrote=
:
> > On Nov 10, 2018, at 3:57 PM, Dan Williams <dan.j.williams@intel.com> wr=
ote:
> >
> >> On Fri, Nov 9, 2018 at 4:22 PM Andy Lutomirski <luto@amacapital.net> w=
rote:
> >>
> >>
> >>
> >>> On Nov 9, 2018, at 4:05 PM, Dan Williams <dan.j.williams@intel.com> w=
rote:
> >>>
> >>> Commit f77084d96355 "x86/mm/pat: Disable preemption around
> >>> __flush_tlb_all()" addressed a case where __flush_tlb_all() is called
> >>> without preemption being disabled. It also left a warning to catch ot=
her
> >>> cases where preemption is not disabled. That warning triggers for the
> >>> memory hotplug path which is also used for persistent memory enabling=
:
> >>
> >> I don=E2=80=99t think I agree with the patch. If you call __flush_tlb_=
all() in a context where you might be *migrated*, then there=E2=80=99s a bu=
g. We could change the code to allow this particular use by checking that w=
e haven=E2=80=99t done SMP init yet, perhaps.
> >
> > Hmm, are saying the entire kernel_physical_mapping_init() sequence
> > needs to run with pre-emption disabled?
>
> If it indeed can run late in boot or after boot, then it sure looks buggy=
. Either the __flush_tlb_all() should be removed or it should be replaced w=
ith flush_tlb_kernel_range(). It=E2=80=99s unclear to me why a flush is nee=
ded at all, but if it=E2=80=99s needed, surely all CPUs need flushing.

Yeah, I don't think __flush_tlb_all() is needed at
kernel_physical_mapping_init() time, and at
kernel_physical_mapping_remove() time we do a full flush_tlb_all().

Kirill?