From mboxrd@z Thu Jan 1 00:00:00 1970 From: "Kirill A. Shutemov" Date: Fri, 13 Sep 2019 08:51:35 +0000 Subject: Re: [PATCH] mm/pgtable/debug: Fix test validating architecture page table helpers Message-Id: <20190913085135.rfr3zrabghi2qo2t@box> List-Id: References: <1892b37d1fd9a4ed39e76c4b999b6556077201c0.1568355752.git.christophe.leroy@c-s.fr> <527dd29d-45fa-4d83-1899-6cbf268dd749@arm.com> In-Reply-To: MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: quoted-printable To: Anshuman Khandual Cc: Mark Rutland , linux-ia64@vger.kernel.org, linux-sh@vger.kernel.org, Peter Zijlstra , James Hogan , Heiko Carstens , Michal Hocko , linux-mm@kvack.org, Paul Mackerras , sparclinux@vger.kernel.org, Thomas Gleixner , linux-s390@vger.kernel.org, Jason Gunthorpe , x86@kernel.org, Russell King - ARM Linux , Matthew Wilcox , Steven Price , Tetsuo Handa , Gerald Schaefer , linux-snps-arc@lists.infradead.org, linux-arm-kernel@lists.infradead.org, Kees Cook , Masahiro Yamada , Mark Brown , Dan Williams , Vlastimil Babka , Christophe Leroy , Sri Krishna chowdary , Dave Hansen , Greg Kroah-Hartman , Ard Biesheuvel , linux-mips@vger.kernel.org, Ralf Baechle , linux-kernel@vger.kernel.org, Paul Burton , Mike Rapoport , Vineet Gupta , Martin Schwidefsky , Andrew Morton , linuxppc-dev@lists.ozlabs.org, "David S. Miller" On Fri, Sep 13, 2019 at 02:12:45PM +0530, Anshuman Khandual wrote: >=20 >=20 > On 09/13/2019 12:41 PM, Christophe Leroy wrote: > >=20 > >=20 > > Le 13/09/2019 =E0 09:03, Christophe Leroy a =E9crit=A0: > >> > >> > >> Le 13/09/2019 =E0 08:58, Anshuman Khandual a =E9crit=A0: > >>> On 09/13/2019 11:53 AM, Christophe Leroy wrote: > >>>> Fix build failure on powerpc. > >>>> > >>>> Fix preemption imbalance. > >>>> > >>>> Signed-off-by: Christophe Leroy > >>>> --- > >>>> =A0 mm/arch_pgtable_test.c | 3 +++ > >>>> =A0 1 file changed, 3 insertions(+) > >>>> > >>>> diff --git a/mm/arch_pgtable_test.c b/mm/arch_pgtable_test.c > >>>> index 8b4a92756ad8..f2b3c9ec35fa 100644 > >>>> --- a/mm/arch_pgtable_test.c > >>>> +++ b/mm/arch_pgtable_test.c > >>>> @@ -24,6 +24,7 @@ > >>>> =A0 #include > >>>> =A0 #include > >>>> =A0 #include > >>>> +#include > >>> > >>> This is okay. > >>> > >>>> =A0 #include > >>>> =A0 #include > >>>> @@ -400,6 +401,8 @@ static int __init arch_pgtable_tests_init(void) > >>>> =A0=A0=A0=A0=A0 p4d_clear_tests(p4dp); > >>>> =A0=A0=A0=A0=A0 pgd_clear_tests(mm, pgdp); > >>>> +=A0=A0=A0 pte_unmap(ptep); > >>>> + > >>> > >>> Now the preemption imbalance via pte_alloc_map() path i.e > >>> > >>> pte_alloc_map() -> pte_offset_map() -> kmap_atomic() > >>> > >>> Is not this very much powerpc 32 specific or this will be applicable > >>> for all platform which uses kmap_XXX() to map high memory ? > >>> > >> > >> See https://elixir.bootlin.com/linux/v5.3-rc8/source/include/linux/hig= hmem.h#L91 > >> > >> I think it applies at least to all arches using the generic implementa= tion. > >> > >> Applies also to arm: > >> https://elixir.bootlin.com/linux/v5.3-rc8/source/arch/arm/mm/highmem.c= #L52 > >> > >> Applies also to mips: > >> https://elixir.bootlin.com/linux/v5.3-rc8/source/arch/mips/mm/highmem.= c#L47 > >> > >> Same on sparc: > >> https://elixir.bootlin.com/linux/v5.3-rc8/source/arch/sparc/mm/highmem= .c#L52 > >> > >> Same on x86: > >> https://elixir.bootlin.com/linux/v5.3-rc8/source/arch/x86/mm/highmem_3= 2.c#L34 > >> > >> I have not checked others, but I guess it is like that for all. > >> > >=20 > >=20 > > Seems like I answered too quickly. All kmap_atomic() do preempt_disable= (), but not all pte_alloc_map() call kmap_atomic(). > >=20 > > However, for instance ARM does: > >=20 > > https://elixir.bootlin.com/linux/v5.3-rc8/source/arch/arm/include/asm/p= gtable.h#L200 > >=20 > > And X86 as well: > >=20 > > https://elixir.bootlin.com/linux/v5.3-rc8/source/arch/x86/include/asm/p= gtable_32.h#L51 > >=20 > > Microblaze also: > >=20 > > https://elixir.bootlin.com/linux/v5.3-rc8/source/arch/microblaze/includ= e/asm/pgtable.h#L495 >=20 > All the above platforms checks out to be using k[un]map_atomic(). I am wo= ndering whether > any of the intermediate levels will have similar problems on any these 32= bit platforms > or any other platforms which might be using generic k[un]map_atomic(). No. Kernel only allocates pte page table from highmem. All other page tables are always visible in kernel address space. --=20 Kirill A. Shutemov