On Tue, Oct 18, 2022 at 01:07:29PM -0700, Ashok Raj wrote: > On Tue, Oct 18, 2022 at 02:33:53PM +0300, Kirill A. Shutemov wrote: > > IOMMU and SVM-capable devices know nothing about LAM and only expect > > canonical addresses. Attempt to pass down tagged pointer will lead to > > address translation failure. > > > > By default do not allow to enable both LAM and use SVM in the same > > process. > > > > The new ARCH_FORCE_TAGGED_SVM arch_prctl() overrides the limitation. > > By using the arch_prctl() userspace takes responsibility to never pass > > tagged address to the device. > > Reviewed-by: Ashok Raj <ashok.raj@xxxxxxxxx> Thanks! > > Signed-off-by: Kirill A. Shutemov <kirill.shutemov@xxxxxxxxxxxxxxx> > > --- > > arch/x86/include/asm/mmu.h | 6 ++++-- > > arch/x86/include/asm/mmu_context.h | 2 ++ > > arch/x86/include/uapi/asm/prctl.h | 1 + > > arch/x86/kernel/process_64.c | 13 +++++++++++++ > > drivers/iommu/iommu-sva-lib.c | 12 ++++++++++++ > > include/linux/mmu_context.h | 4 ++++ > > 6 files changed, 36 insertions(+), 2 deletions(-) > > > > diff --git a/arch/x86/include/asm/mmu.h b/arch/x86/include/asm/mmu.h > > index 2fdb390040b5..cce9b32b0d6d 100644 > > --- a/arch/x86/include/asm/mmu.h > > +++ b/arch/x86/include/asm/mmu.h > > @@ -9,9 +9,11 @@ > > #include <linux/bits.h> > > > > /* Uprobes on this MM assume 32-bit code */ > > -#define MM_CONTEXT_UPROBE_IA32 BIT(0) > > +#define MM_CONTEXT_UPROBE_IA32 BIT(0) > > /* vsyscall page is accessible on this MM */ > > -#define MM_CONTEXT_HAS_VSYSCALL BIT(1) > > +#define MM_CONTEXT_HAS_VSYSCALL BIT(1) > > Nit: Looks like the two above format changes got in here :-) That's side effect of keeping the new longer flag aligned to the rest. A separate patch looks like an overkill, no? -- Kiryl Shutsemau / Kirill A. Shutemov
