| Message ID | 20190528161026.13193-2-steve.capper@arm.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | 52-bit kernel + user VAs | expand |
Hi Steve, On 28/05/2019 17:10, Steve Capper wrote: > We assume that the direct linear map ends at ~0 in the KVM HYP map Do we? This has stopped being the case since ed57cac83e05f ("arm64: KVM: Introduce EL2 VA randomisation"). > intersection checking code. This assumption will become invalid later on > for arm64 when the address space of the kernel is re-arranged. > > This patch introduces a new constant PAGE_OFFSET_END for both arm and > arm64 and defines it to be ~0UL > > Signed-off-by: Steve Capper <steve.capper@arm.com> > --- > arch/arm/include/asm/memory.h | 1 + > arch/arm64/include/asm/memory.h | 1 + > virt/kvm/arm/mmu.c | 4 ++-- > 3 files changed, 4 insertions(+), 2 deletions(-) > > diff --git a/arch/arm/include/asm/memory.h b/arch/arm/include/asm/memory.h > index ed8fd0d19a3e..45c211fd50da 100644 > --- a/arch/arm/include/asm/memory.h > +++ b/arch/arm/include/asm/memory.h > @@ -24,6 +24,7 @@ > > /* PAGE_OFFSET - the virtual address of the start of the kernel image */ > #define PAGE_OFFSET UL(CONFIG_PAGE_OFFSET) > +#define PAGE_OFFSET_END (~0UL) > > #ifdef CONFIG_MMU > > diff --git a/arch/arm64/include/asm/memory.h b/arch/arm64/include/asm/memory.h > index 8ffcf5a512bb..9fd387a63b9b 100644 > --- a/arch/arm64/include/asm/memory.h > +++ b/arch/arm64/include/asm/memory.h > @@ -52,6 +52,7 @@ > (UL(1) << VA_BITS) + 1) > #define PAGE_OFFSET (UL(0xffffffffffffffff) - \ > (UL(1) << (VA_BITS - 1)) + 1) > +#define PAGE_OFFSET_END (~0UL) > #define KIMAGE_VADDR (MODULES_END) > #define BPF_JIT_REGION_START (VA_START + KASAN_SHADOW_SIZE) > #define BPF_JIT_REGION_SIZE (SZ_128M) > diff --git a/virt/kvm/arm/mmu.c b/virt/kvm/arm/mmu.c > index 74b6582eaa3c..e1a777275b37 100644 > --- a/virt/kvm/arm/mmu.c > +++ b/virt/kvm/arm/mmu.c > @@ -2202,10 +2202,10 @@ int kvm_mmu_init(void) > kvm_debug("IDMAP page: %lx\n", hyp_idmap_start); > kvm_debug("HYP VA range: %lx:%lx\n", > kern_hyp_va(PAGE_OFFSET), > - kern_hyp_va((unsigned long)high_memory - 1)); > + kern_hyp_va(PAGE_OFFSET_END)); > > if (hyp_idmap_start >= kern_hyp_va(PAGE_OFFSET) && > - hyp_idmap_start < kern_hyp_va((unsigned long)high_memory - 1) && > + hyp_idmap_start < kern_hyp_va(PAGE_OFFSET_END) && > hyp_idmap_start != (unsigned long)__hyp_idmap_text_start) { > /* > * The idmap page is intersecting with the VA space, > This definitely looks like a move in the wrong direction (reverting part of the above commit). Is it that this is just an old patch that should have been dropped? Or am I completely missing the point? Thanks, M.
On Tue, May 28, 2019 at 05:27:17PM +0100, Marc Zyngier wrote: > Hi Steve, Hi Marc, > > On 28/05/2019 17:10, Steve Capper wrote: > > We assume that the direct linear map ends at ~0 in the KVM HYP map > > Do we? This has stopped being the case since ed57cac83e05f ("arm64: KVM: > Introduce EL2 VA randomisation"). > > > intersection checking code. This assumption will become invalid later on > > for arm64 when the address space of the kernel is re-arranged. > > > > This patch introduces a new constant PAGE_OFFSET_END for both arm and > > arm64 and defines it to be ~0UL > > > > Signed-off-by: Steve Capper <steve.capper@arm.com> > > --- > > arch/arm/include/asm/memory.h | 1 + > > arch/arm64/include/asm/memory.h | 1 + > > virt/kvm/arm/mmu.c | 4 ++-- > > 3 files changed, 4 insertions(+), 2 deletions(-) > > > > diff --git a/arch/arm/include/asm/memory.h b/arch/arm/include/asm/memory.h > > index ed8fd0d19a3e..45c211fd50da 100644 > > --- a/arch/arm/include/asm/memory.h > > +++ b/arch/arm/include/asm/memory.h > > @@ -24,6 +24,7 @@ > > > > /* PAGE_OFFSET - the virtual address of the start of the kernel image */ > > #define PAGE_OFFSET UL(CONFIG_PAGE_OFFSET) > > +#define PAGE_OFFSET_END (~0UL) > > > > #ifdef CONFIG_MMU > > > > diff --git a/arch/arm64/include/asm/memory.h b/arch/arm64/include/asm/memory.h > > index 8ffcf5a512bb..9fd387a63b9b 100644 > > --- a/arch/arm64/include/asm/memory.h > > +++ b/arch/arm64/include/asm/memory.h > > @@ -52,6 +52,7 @@ > > (UL(1) << VA_BITS) + 1) > > #define PAGE_OFFSET (UL(0xffffffffffffffff) - \ > > (UL(1) << (VA_BITS - 1)) + 1) > > +#define PAGE_OFFSET_END (~0UL) > > #define KIMAGE_VADDR (MODULES_END) > > #define BPF_JIT_REGION_START (VA_START + KASAN_SHADOW_SIZE) > > #define BPF_JIT_REGION_SIZE (SZ_128M) > > diff --git a/virt/kvm/arm/mmu.c b/virt/kvm/arm/mmu.c > > index 74b6582eaa3c..e1a777275b37 100644 > > --- a/virt/kvm/arm/mmu.c > > +++ b/virt/kvm/arm/mmu.c > > @@ -2202,10 +2202,10 @@ int kvm_mmu_init(void) > > kvm_debug("IDMAP page: %lx\n", hyp_idmap_start); > > kvm_debug("HYP VA range: %lx:%lx\n", > > kern_hyp_va(PAGE_OFFSET), > > - kern_hyp_va((unsigned long)high_memory - 1)); > > + kern_hyp_va(PAGE_OFFSET_END)); > > > > if (hyp_idmap_start >= kern_hyp_va(PAGE_OFFSET) && > > - hyp_idmap_start < kern_hyp_va((unsigned long)high_memory - 1) && > > + hyp_idmap_start < kern_hyp_va(PAGE_OFFSET_END) && > > hyp_idmap_start != (unsigned long)__hyp_idmap_text_start) { > > /* > > * The idmap page is intersecting with the VA space, > > > > This definitely looks like a move in the wrong direction (reverting part > of the above commit). Is it that this is just an old patch that should > have been dropped? Or am I completely missing the point? I suspect this is a case of me rebasing my series... poorly. I'll re-examine the logic here and either update the patch or the commit log to make it clearer. Cheers,
On Tue, May 28, 2019 at 05:01:29PM +0000, Steve Capper wrote: > On Tue, May 28, 2019 at 05:27:17PM +0100, Marc Zyngier wrote: > > Hi Steve, > > Hi Marc, > > > > > On 28/05/2019 17:10, Steve Capper wrote: > > > We assume that the direct linear map ends at ~0 in the KVM HYP map > > > > Do we? This has stopped being the case since ed57cac83e05f ("arm64: KVM: > > Introduce EL2 VA randomisation"). > > > > > intersection checking code. This assumption will become invalid later on > > > for arm64 when the address space of the kernel is re-arranged. > > > > > > This patch introduces a new constant PAGE_OFFSET_END for both arm and > > > arm64 and defines it to be ~0UL > > > > > > Signed-off-by: Steve Capper <steve.capper@arm.com> > > > --- > > > arch/arm/include/asm/memory.h | 1 + > > > arch/arm64/include/asm/memory.h | 1 + > > > virt/kvm/arm/mmu.c | 4 ++-- > > > 3 files changed, 4 insertions(+), 2 deletions(-) > > > > > > diff --git a/arch/arm/include/asm/memory.h b/arch/arm/include/asm/memory.h > > > index ed8fd0d19a3e..45c211fd50da 100644 > > > --- a/arch/arm/include/asm/memory.h > > > +++ b/arch/arm/include/asm/memory.h > > > @@ -24,6 +24,7 @@ > > > > > > /* PAGE_OFFSET - the virtual address of the start of the kernel image */ > > > #define PAGE_OFFSET UL(CONFIG_PAGE_OFFSET) > > > +#define PAGE_OFFSET_END (~0UL) > > > > > > #ifdef CONFIG_MMU > > > > > > diff --git a/arch/arm64/include/asm/memory.h b/arch/arm64/include/asm/memory.h > > > index 8ffcf5a512bb..9fd387a63b9b 100644 > > > --- a/arch/arm64/include/asm/memory.h > > > +++ b/arch/arm64/include/asm/memory.h > > > @@ -52,6 +52,7 @@ > > > (UL(1) << VA_BITS) + 1) > > > #define PAGE_OFFSET (UL(0xffffffffffffffff) - \ > > > (UL(1) << (VA_BITS - 1)) + 1) > > > +#define PAGE_OFFSET_END (~0UL) > > > #define KIMAGE_VADDR (MODULES_END) > > > #define BPF_JIT_REGION_START (VA_START + KASAN_SHADOW_SIZE) > > > #define BPF_JIT_REGION_SIZE (SZ_128M) > > > diff --git a/virt/kvm/arm/mmu.c b/virt/kvm/arm/mmu.c > > > index 74b6582eaa3c..e1a777275b37 100644 > > > --- a/virt/kvm/arm/mmu.c > > > +++ b/virt/kvm/arm/mmu.c > > > @@ -2202,10 +2202,10 @@ int kvm_mmu_init(void) > > > kvm_debug("IDMAP page: %lx\n", hyp_idmap_start); > > > kvm_debug("HYP VA range: %lx:%lx\n", > > > kern_hyp_va(PAGE_OFFSET), > > > - kern_hyp_va((unsigned long)high_memory - 1)); > > > + kern_hyp_va(PAGE_OFFSET_END)); > > > > > > if (hyp_idmap_start >= kern_hyp_va(PAGE_OFFSET) && > > > - hyp_idmap_start < kern_hyp_va((unsigned long)high_memory - 1) && > > > + hyp_idmap_start < kern_hyp_va(PAGE_OFFSET_END) && > > > hyp_idmap_start != (unsigned long)__hyp_idmap_text_start) { > > > /* > > > * The idmap page is intersecting with the VA space, > > > > > > > This definitely looks like a move in the wrong direction (reverting part > > of the above commit). Is it that this is just an old patch that should > > have been dropped? Or am I completely missing the point? > > I suspect this is a case of me rebasing my series... poorly. > I'll re-examine the logic here and either update the patch or the commit > log to make it clearer. > Hi Marc, Thanks, this was indeed an overzealous rebase. I've removed this patch from the series and kvmtool is happy booting guests (when rebased to v5.2-rc2). Cheers,
diff --git a/arch/arm/include/asm/memory.h b/arch/arm/include/asm/memory.h index ed8fd0d19a3e..45c211fd50da 100644 --- a/arch/arm/include/asm/memory.h +++ b/arch/arm/include/asm/memory.h @@ -24,6 +24,7 @@ /* PAGE_OFFSET - the virtual address of the start of the kernel image */ #define PAGE_OFFSET UL(CONFIG_PAGE_OFFSET) +#define PAGE_OFFSET_END (~0UL) #ifdef CONFIG_MMU diff --git a/arch/arm64/include/asm/memory.h b/arch/arm64/include/asm/memory.h index 8ffcf5a512bb..9fd387a63b9b 100644 --- a/arch/arm64/include/asm/memory.h +++ b/arch/arm64/include/asm/memory.h @@ -52,6 +52,7 @@ (UL(1) << VA_BITS) + 1) #define PAGE_OFFSET (UL(0xffffffffffffffff) - \ (UL(1) << (VA_BITS - 1)) + 1) +#define PAGE_OFFSET_END (~0UL) #define KIMAGE_VADDR (MODULES_END) #define BPF_JIT_REGION_START (VA_START + KASAN_SHADOW_SIZE) #define BPF_JIT_REGION_SIZE (SZ_128M) diff --git a/virt/kvm/arm/mmu.c b/virt/kvm/arm/mmu.c index 74b6582eaa3c..e1a777275b37 100644 --- a/virt/kvm/arm/mmu.c +++ b/virt/kvm/arm/mmu.c @@ -2202,10 +2202,10 @@ int kvm_mmu_init(void) kvm_debug("IDMAP page: %lx\n", hyp_idmap_start); kvm_debug("HYP VA range: %lx:%lx\n", kern_hyp_va(PAGE_OFFSET), - kern_hyp_va((unsigned long)high_memory - 1)); + kern_hyp_va(PAGE_OFFSET_END)); if (hyp_idmap_start >= kern_hyp_va(PAGE_OFFSET) && - hyp_idmap_start < kern_hyp_va((unsigned long)high_memory - 1) && + hyp_idmap_start < kern_hyp_va(PAGE_OFFSET_END) && hyp_idmap_start != (unsigned long)__hyp_idmap_text_start) { /* * The idmap page is intersecting with the VA space,
We assume that the direct linear map ends at ~0 in the KVM HYP map intersection checking code. This assumption will become invalid later on for arm64 when the address space of the kernel is re-arranged. This patch introduces a new constant PAGE_OFFSET_END for both arm and arm64 and defines it to be ~0UL Signed-off-by: Steve Capper <steve.capper@arm.com> --- arch/arm/include/asm/memory.h | 1 + arch/arm64/include/asm/memory.h | 1 + virt/kvm/arm/mmu.c | 4 ++-- 3 files changed, 4 insertions(+), 2 deletions(-)