Re: [PATCH 0/4] arch, mm: improve robustness of direct map manipulation

From: "Edgecombe, Rick P" <rick.p.edgecombe@intel.com>
To: "david@redhat.com" <david@redhat.com>,
	"rppt@kernel.org" <rppt@kernel.org>
Cc: "rientjes@google.com" <rientjes@google.com>,
	"cl@linux.com" <cl@linux.com>,
	"gor@linux.ibm.com" <gor@linux.ibm.com>,
	"hpa@zytor.com" <hpa@zytor.com>,
	"peterz@infradead.org" <peterz@infradead.org>,
	"catalin.marinas@arm.com" <catalin.marinas@arm.com>,
	"dave.hansen@linux.intel.com" <dave.hansen@linux.intel.com>,
	"borntraeger@de.ibm.com" <borntraeger@de.ibm.com>,
	"penberg@kernel.org" <penberg@kernel.org>,
	"linux-mm@kvack.org" <linux-mm@kvack.org>,
	"iamjoonsoo.kim@lge.com" <iamjoonsoo.kim@lge.com>,
	"will@kernel.org" <will@kernel.org>,
	"aou@eecs.berkeley.edu" <aou@eecs.berkeley.edu>,
	"kirill@shutemov.name" <kirill@shutemov.name>,
	"rppt@linux.ibm.com" <rppt@linux.ibm.com>,
	"paulus@samba.org" <paulus@samba.org>,
	"hca@linux.ibm.com" <hca@linux.ibm.com>,
	"bp@alien8.de" <bp@alien8.de>, "pavel@ucw.cz" <pavel@ucw.cz>,
	"sparclinux@vger.kernel.org" <sparclinux@vger.kernel.org>,
	"akpm@linux-foundation.org" <akpm@linux-foundation.org>,
	"luto@kernel.org" <luto@kernel.org>,
	"davem@davemloft.net" <davem@davemloft.net>,
	"mpe@ellerman.id.au" <mpe@ellerman.id.au>,
	"benh@kernel.crashing.org" <benh@kernel.crashing.org>,
	"linuxppc-dev@lists.ozlabs.org" <linuxppc-dev@lists.ozlabs.org>,
	"linux-riscv@lists.infradead.org"
	<linux-riscv@lists.infradead.org>,
	"tglx@linutronix.de" <tglx@linutronix.de>,
	"rjw@rjwysocki.net" <rjw@rjwysocki.net>,
	"x86@kernel.org" <x86@kernel.org>,
	"linux-pm@vger.kernel.org" <linux-pm@vger.kernel.org>,
	"linux-arm-kernel@lists.infradead.org" 
	<linux-arm-kernel@lists.infradead.org>,
	"palmer@dabbelt.com" <palmer@dabbelt.com>,
	"Brown, Len" <len.brown@intel.com>,
	"mingo@redhat.com" <mingo@redhat.com>,
	"linux-s390@vger.kernel.org" <linux-s390@vger.kernel.org>,
	"linux-kernel@vger.kernel.org" <linux-kernel@vger.kernel.org>,
	"paul.walmsley@sifive.com" <paul.walmsley@sifive.com>
Subject: Re: [PATCH 0/4] arch, mm: improve robustness of direct map manipulation
Date: Wed, 28 Oct 2020 18:31:48 +0000	[thread overview]
Message-ID: <0471581f783632648ab5d38753f16ed34ef0d941.camel@intel.com> (raw)
In-Reply-To: <20201028110945.GE1428094@kernel.org>

On Wed, 2020-10-28 at 13:09 +0200, Mike Rapoport wrote:
> On Tue, Oct 27, 2020 at 09:46:35AM +0100, David Hildenbrand wrote:
> > On 27.10.20 09:38, Mike Rapoport wrote:
> > > On Mon, Oct 26, 2020 at 06:05:30PM +0000, Edgecombe, Rick P
> > > wrote:
> > > 
> > > > Beyond whatever you are seeing, for the latter case of new
> > > > things
> > > > getting introduced to an interface with hidden dependencies...
> > > > Another
> > > > edge case could be a new caller to set_memory_np() could result
> > > > in
> > > > large NP pages. None of the callers today should cause this
> > > > AFAICT, but
> > > > it's not great to rely on the callers to know these details.
> > > A caller of set_memory_*() or set_direct_map_*() should expect a
> > > failure
> > > and be ready for that. So adding a WARN to safe_copy_page() is
> > > the first
> > > step in that direction :)
> > > 
> > 
> > I am probably missing something important, but why are we
> > saving/restoring
> > the content of pages that were explicitly removed from the identity
> > mapping
> > such that nobody will access them?
> 
> Actually, we should not be saving/restoring free pages during
> hibernation as there are several calls to mark_free_pages() that
> should
> exclude the free pages from the snapshot. I've tried to find why the
> fix
> that maps/unmaps a page to save it was required at the first place,
> but
> I could not find bug reports.
> 
> The closest I've got is an email from Rafael that asked to update
> "hibernate: handle DEBUG_PAGEALLOC" patch:
> 
> https://lore.kernel.org/linux-pm/200802200133.44098.rjw@sisk.pl/
> 
> Could it be that safe_copy_page() tries to workaround a non-existent
> problem?

It looks like inside page_alloc.c it unmaps the page before it actually
frees it, so to hibernate it could look like the page is still
allocated even though it's unmapped? Maybe that small window is what it
cared about initially.

There is also now the vmalloc case, which I am actually working on
expanding. So I think the re-mapping logic is needed.