There is a guard hole at the beginning of kernel address space, also used by hypervisors. It occupies 16 PGD entries. We do not state the reserved range directly, but calculate it relative to other entities: direct mapping and user space ranges. The calculation got broken by recent change in kernel memory layout: LDT remap range is now mapped before direct mapping and makes the calculation invalid. The breakage leads to crash on Xen dom0 boot[1]. State the reserved range directly. It's part of kernel ABI (hypervisors expect it to be stable) and must not depend on changes in the rest of kernel memory layout. [1] https://lists.xenproject.org/archives/html/xen-devel/2018-11/msg03313.html Signed-off-by: Kirill A. Shutemov <kirill.shutemov@xxxxxxxxxxxxxxx> Reported-by: Hans van Kranenburg <Hans.van.Kranenburg@xxxxxxxxxx> Fixes: d52888aa2753 ("x86/mm: Move LDT remap out of KASLR region on 5-level paging") --- arch/x86/include/asm/pgtable_64_types.h | 5 +++++ arch/x86/mm/dump_pagetables.c | 8 ++++---- arch/x86/xen/mmu_pv.c | 11 ++++++----- 3 files changed, 15 insertions(+), 9 deletions(-) diff --git a/arch/x86/include/asm/pgtable_64_types.h b/arch/x86/include/asm/pgtable_64_types.h index 84bd9bdc1987..13aef22cee18 100644 --- a/arch/x86/include/asm/pgtable_64_types.h +++ b/arch/x86/include/asm/pgtable_64_types.h @@ -111,6 +111,11 @@ extern unsigned int ptrs_per_p4d; */ #define MAXMEM (1UL << MAX_PHYSMEM_BITS) +#define GUARD_HOLE_PGD_ENTRY -256UL +#define GUARD_HOLE_SIZE (16UL << PGDIR_SHIFT) +#define GUARD_HOLE_BASE_ADDR (LDT_PGD_ENTRY << PGDIR_SHIFT) +#define GUARD_HOLE_END_ADDR (GUARD_HOLE_BASE_ADDR + GUARD_HOLE_SIZE) + #define LDT_PGD_ENTRY -240UL #define LDT_BASE_ADDR (LDT_PGD_ENTRY << PGDIR_SHIFT) #define LDT_END_ADDR (LDT_BASE_ADDR + PGDIR_SIZE) diff --git a/arch/x86/mm/dump_pagetables.c b/arch/x86/mm/dump_pagetables.c index fc37bbd23eb8..dad153e5a427 100644 --- a/arch/x86/mm/dump_pagetables.c +++ b/arch/x86/mm/dump_pagetables.c @@ -512,11 +512,11 @@ static inline bool is_hypervisor_range(int idx) { #ifdef CONFIG_X86_64 /* - * ffff800000000000 - ffff87ffffffffff is reserved for - * the hypervisor. + * A hole in the beginning of kernel address space reserved + * for a hypervisor. */ - return (idx >= pgd_index(__PAGE_OFFSET) - 16) && - (idx < pgd_index(__PAGE_OFFSET)); + return (idx >= pgd_index(GUARD_HOLE_BASE_ADDR)) && + (idx < pgd_index(GUARD_HOLE_END_ADDR)); #else return false; #endif diff --git a/arch/x86/xen/mmu_pv.c b/arch/x86/xen/mmu_pv.c index a5d7ed125337..0f4fe206dcc2 100644 --- a/arch/x86/xen/mmu_pv.c +++ b/arch/x86/xen/mmu_pv.c @@ -648,19 +648,20 @@ static int __xen_pgd_walk(struct mm_struct *mm, pgd_t *pgd, unsigned long limit) { int i, nr, flush = 0; - unsigned hole_low, hole_high; + unsigned hole_low = 0, hole_high = 0; /* The limit is the last byte to be touched */ limit--; BUG_ON(limit >= FIXADDR_TOP); +#ifdef CONFIG_X86_64 /* * 64-bit has a great big hole in the middle of the address - * space, which contains the Xen mappings. On 32-bit these - * will end up making a zero-sized hole and so is a no-op. + * space, which contains the Xen mappings. */ - hole_low = pgd_index(USER_LIMIT); - hole_high = pgd_index(PAGE_OFFSET); + hole_low = pgd_index(GUARD_HOLE_BASE_ADDR); + hole_high = pgd_index(GUARD_HOLE_END_ADDR); +#endif nr = pgd_index(limit) + 1; for (i = 0; i < nr; i++) { -- 2.19.2
