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;
Date:   Mon, 15 Jul 2019 17:44:18 +0200
From:   Joerg Roedel <jroedel@suse.de>
To:     Thomas Gleixner <tglx@linutronix.de>
Cc:     Joerg Roedel <joro@8bytes.org>,
        Dave Hansen <dave.hansen@linux.intel.com>,
        Andy Lutomirski <luto@kernel.org>,
        Peter Zijlstra <peterz@infradead.org>,
        Ingo Molnar <mingo@redhat.com>, Borislav Petkov <bp@alien8.de>,
        Andrew Morton <akpm@linux-foundation.org>,
        linux-kernel@vger.kernel.org, linux-mm@kvack.org
Subject: Re: [PATCH 1/3] x86/mm: Check for pfn instead of page in
 vmalloc_sync_one()
Message-ID: <20190715154418.GA13091@suse.de>
References: <20190715110212.18617-1-joro@8bytes.org>
 <20190715110212.18617-2-joro@8bytes.org>
 <alpine.DEB.2.21.1907151508210.1722@nanos.tec.linutronix.de>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Disposition: inline
In-Reply-To: <alpine.DEB.2.21.1907151508210.1722@nanos.tec.linutronix.de>
User-Agent: Mutt/1.10.1 (2018-07-13)
Sender: linux-kernel-owner@vger.kernel.org
Precedence: bulk

On Mon, Jul 15, 2019 at 03:08:42PM +0200, Thomas Gleixner wrote:
> On Mon, 15 Jul 2019, Joerg Roedel wrote:
> 
> > From: Joerg Roedel <jroedel@suse.de>
> > 
> > Do not require a struct page for the mapped memory location
> > because it might not exist. This can happen when an
> > ioremapped region is mapped with 2MB pages.
> > 
> > Signed-off-by: Joerg Roedel <jroedel@suse.de>
> 
> Lacks a Fixes tag, hmm?

Yeah, right, the question is, which commit to put in there. The problem
results from two changes:

	1) Introduction of !SHARED_KERNEL_PMD path in x86-32. In itself
	   this is not a problem, and the path was only enabled for
	   Xen-PV.

	2) Huge IORemapings which use the PMD level. Also not a problem
	   by itself, but together with !SHARED_KERNEL_PMD problematic
	   because it requires to sync the PMD entries between all
	   page-tables, and that was not implemented.

Before PTI-x32 was merged this problem did not show up, maybe because
the 32-bit Xen-PV users did not trigger it. But with PTI-x32 all PAE
users run with !SHARED_KERNEL_PMD and the problem popped up.

For the last patch I put the PTI-x32 enablement commit in the fixes tag,
because that was the one that showed up during bisection. But more
correct would probably be

	5d72b4fba40e ('x86, mm: support huge I/O mapping capability I/F')

Or do I miss something?

Regards,

	Joerg