| Message ID | 1444821634-1689-3-git-send-email-suzuki.poulose@arm.com (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
On Wed, Oct 14, 2015 at 12:20:25PM +0100, Suzuki K. Poulose wrote: > We use section maps with 4K page size to create the swapper/idmaps. > So far we have used !64K or 4K checks to handle the case where we > use the section maps. > This patch adds a new symbol, ARM64_SWAPPER_USES_SECTION_MAPS, to > handle cases where we use section maps, instead of using the page size > symbols. > > Cc: Ard Biesheuvel <ard.biesheuvel@linaro.org> > Cc: Mark Rutland <mark.rutland@arm.com> > Cc: Catalin Marinas <catalin.marinas@arm.com> > Cc: Will Deacon <will.deacon@arm.com> > Signed-off-by: Suzuki K. Poulose <suzuki.poulose@arm.com> > --- > arch/arm64/include/asm/kernel-pgtable.h | 31 ++++++++----- > arch/arm64/mm/mmu.c | 72 ++++++++++++++----------------- > 2 files changed, 52 insertions(+), 51 deletions(-) > > diff --git a/arch/arm64/include/asm/kernel-pgtable.h b/arch/arm64/include/asm/kernel-pgtable.h > index 622929d..5876a36 100644 > --- a/arch/arm64/include/asm/kernel-pgtable.h > +++ b/arch/arm64/include/asm/kernel-pgtable.h > @@ -19,6 +19,13 @@ > #ifndef __ASM_KERNEL_PGTABLE_H > #define __ASM_KERNEL_PGTABLE_H > > +/* With 4K pages, we use section maps. */ > +#ifdef CONFIG_ARM64_4K_PAGES > +#define ARM64_SWAPPER_USES_SECTION_MAPS 1 > +#else > +#define ARM64_SWAPPER_USES_SECTION_MAPS 0 > +#endif The comment is somewhat redunant. It would be better to state why we do this for 4K and not 64K (or 16K). > @@ -406,14 +407,11 @@ static void __init map_mem(void) > * memory addressable from the initial direct kernel mapping. > * > * The initial direct kernel mapping, located at swapper_pg_dir, gives > - * us PUD_SIZE (4K pages) or PMD_SIZE (64K pages) memory starting from > - * PHYS_OFFSET (which must be aligned to 2MB as per > - * Documentation/arm64/booting.txt). > + * us PUD_SIZE (with SECTION maps, i.e, 4K) or PMD_SIZE (without > + * SECTION maps, i.e, 64K pages) memory starting from PHYS_OFFSET > + * (which must be aligned to 2MB as per Documentation/arm64/booting.txt). This didn't seem to get updated for 16K later in the series, unless I missed something. Perhaps drop the mention of 4K / 64K entirely here? > @@ -551,7 +552,7 @@ int kern_addr_valid(unsigned long addr) > return pfn_valid(pte_pfn(*pte)); > } > #ifdef CONFIG_SPARSEMEM_VMEMMAP > -#ifdef CONFIG_ARM64_64K_PAGES > +#if !ARM64_SWAPPER_USES_SECTION_MAPS This leaves the comments on the #else and #endif stale. Please update those too. Otherwise this looks good! Thanks, Mark.
On 14/10/15 13:06, Mark Rutland wrote: > On Wed, Oct 14, 2015 at 12:20:25PM +0100, Suzuki K. Poulose wrote: >> We use section maps with 4K page size to create the swapper/idmaps. >> So far we have used !64K or 4K checks to handle the case where we >> use the section maps. >> This patch adds a new symbol, ARM64_SWAPPER_USES_SECTION_MAPS, to >> handle cases where we use section maps, instead of using the page size >> symbols. >> >> Cc: Ard Biesheuvel <ard.biesheuvel@linaro.org> >> Cc: Mark Rutland <mark.rutland@arm.com> >> Cc: Catalin Marinas <catalin.marinas@arm.com> >> Cc: Will Deacon <will.deacon@arm.com> >> Signed-off-by: Suzuki K. Poulose <suzuki.poulose@arm.com> >> --- >> arch/arm64/include/asm/kernel-pgtable.h | 31 ++++++++----- >> arch/arm64/mm/mmu.c | 72 ++++++++++++++----------------- >> 2 files changed, 52 insertions(+), 51 deletions(-) >> >> diff --git a/arch/arm64/include/asm/kernel-pgtable.h b/arch/arm64/include/asm/kernel-pgtable.h >> index 622929d..5876a36 100644 >> --- a/arch/arm64/include/asm/kernel-pgtable.h >> +++ b/arch/arm64/include/asm/kernel-pgtable.h >> @@ -19,6 +19,13 @@ >> #ifndef __ASM_KERNEL_PGTABLE_H >> #define __ASM_KERNEL_PGTABLE_H >> >> +/* With 4K pages, we use section maps. */ >> +#ifdef CONFIG_ARM64_4K_PAGES >> +#define ARM64_SWAPPER_USES_SECTION_MAPS 1 >> +#else >> +#define ARM64_SWAPPER_USES_SECTION_MAPS 0 >> +#endif > > The comment is somewhat redunant. It would be better to state why we do > this for 4K and not 64K (or 16K). Something like : /* * ARM64 kernel is guaranteed to be loaded at 2M aligned * address (as per booting requirements). Hence we can use * section mapping with 4K (section size = 2M) and not with * 16K(section size = 32M) or 64K (section size = 512M). */ >> + * us PUD_SIZE (with SECTION maps, i.e, 4K) or PMD_SIZE (without >> + * SECTION maps, i.e, 64K pages) memory starting from PHYS_OFFSET >> + * (which must be aligned to 2MB as per Documentation/arm64/booting.txt). > > This didn't seem to get updated for 16K later in the series, unless I > missed something. > > Perhaps drop the mention of 4K / 64K entirely here? You are right, I missed it. We can drop the pagesize info entirely. > >> @@ -551,7 +552,7 @@ int kern_addr_valid(unsigned long addr) >> return pfn_valid(pte_pfn(*pte)); >> } >> #ifdef CONFIG_SPARSEMEM_VMEMMAP >> -#ifdef CONFIG_ARM64_64K_PAGES >> +#if !ARM64_SWAPPER_USES_SECTION_MAPS > > This leaves the comments on the #else and #endif stale. Please update > those too. Done. Thanks Suzuki
> >>+/* With 4K pages, we use section maps. */ > >>+#ifdef CONFIG_ARM64_4K_PAGES > >>+#define ARM64_SWAPPER_USES_SECTION_MAPS 1 > >>+#else > >>+#define ARM64_SWAPPER_USES_SECTION_MAPS 0 > >>+#endif > > > >The comment is somewhat redunant. It would be better to state why we do > >this for 4K and not 64K (or 16K). > > Something like : > > /* > * ARM64 kernel is guaranteed to be loaded at 2M aligned > * address (as per booting requirements). Hence we can use > * section mapping with 4K (section size = 2M) and not with > * 16K(section size = 32M) or 64K (section size = 512M). > */ That sounds much better. I hadn't figured out why myself, so thanks for the explanation :) However, there's one minor nit: the start of memory below the kernel is 2M aligned, but the offset means that the kernel itself is not loaded at a 2M aligned address. So how about: /* * The linear mapping and the start of memory are both 2M aligned (per * the arm64 booting.txt requirements). Hence we can use section mapping * with 4K (section size = 2M) but not with 16K (section size = 32M) or * 64K (section size = 512M). */ > >>+ * us PUD_SIZE (with SECTION maps, i.e, 4K) or PMD_SIZE (without > >>+ * SECTION maps, i.e, 64K pages) memory starting from PHYS_OFFSET > >>+ * (which must be aligned to 2MB as per Documentation/arm64/booting.txt). > > > >This didn't seem to get updated for 16K later in the series, unless I > >missed something. > > > >Perhaps drop the mention of 4K / 64K entirely here? > > You are right, I missed it. We can drop the pagesize info entirely. Ok. Sounds good. > >>@@ -551,7 +552,7 @@ int kern_addr_valid(unsigned long addr) > >> return pfn_valid(pte_pfn(*pte)); > >> } > >> #ifdef CONFIG_SPARSEMEM_VMEMMAP > >>-#ifdef CONFIG_ARM64_64K_PAGES > >>+#if !ARM64_SWAPPER_USES_SECTION_MAPS > > > >This leaves the comments on the #else and #endif stale. Please update > >those too. > > Done. Great. Thanks, Mark.
On 14/10/15 15:51, Mark Rutland wrote: >>>> +/* With 4K pages, we use section maps. */ >> /* >> * ARM64 kernel is guaranteed to be loaded at 2M aligned >> * address (as per booting requirements). Hence we can use >> * section mapping with 4K (section size = 2M) and not with >> * 16K(section size = 32M) or 64K (section size = 512M). >> */ > > That sounds much better. I hadn't figured out why myself, so thanks for > the explanation :) > > However, there's one minor nit: the start of memory below the kernel is > 2M aligned, but the offset means that the kernel itself is not loaded at > a 2M aligned address. Oh yes > > So how about: > > /* > * The linear mapping and the start of memory are both 2M aligned (per > * the arm64 booting.txt requirements). Hence we can use section mapping > * with 4K (section size = 2M) but not with 16K (section size = 32M) or > * 64K (section size = 512M). > */ Will add this Thanks Suzuki
On Wed, Oct 14, 2015 at 04:08:43PM +0100, Suzuki K. Poulose wrote: > On 14/10/15 15:51, Mark Rutland wrote: > >>>>+/* With 4K pages, we use section maps. */ > > >>/* > >> * ARM64 kernel is guaranteed to be loaded at 2M aligned > >> * address (as per booting requirements). Hence we can use > >> * section mapping with 4K (section size = 2M) and not with > >> * 16K(section size = 32M) or 64K (section size = 512M). > >> */ > > > >That sounds much better. I hadn't figured out why myself, so thanks for > >the explanation :) > > > >However, there's one minor nit: the start of memory below the kernel is > >2M aligned, but the offset means that the kernel itself is not loaded at > >a 2M aligned address. > > Oh yes > > > > >So how about: > > > >/* > > * The linear mapping and the start of memory are both 2M aligned (per > > * the arm64 booting.txt requirements). Hence we can use section mapping > > * with 4K (section size = 2M) but not with 16K (section size = 32M) or > > * 64K (section size = 512M). > > */ > > Will add this Great. FWIW, with the changes from this thread applied: Acked-by: Mark Rutland <mark.rutland@arm.com> Thanks, Mark.
diff --git a/arch/arm64/include/asm/kernel-pgtable.h b/arch/arm64/include/asm/kernel-pgtable.h index 622929d..5876a36 100644 --- a/arch/arm64/include/asm/kernel-pgtable.h +++ b/arch/arm64/include/asm/kernel-pgtable.h @@ -19,6 +19,13 @@ #ifndef __ASM_KERNEL_PGTABLE_H #define __ASM_KERNEL_PGTABLE_H +/* With 4K pages, we use section maps. */ +#ifdef CONFIG_ARM64_4K_PAGES +#define ARM64_SWAPPER_USES_SECTION_MAPS 1 +#else +#define ARM64_SWAPPER_USES_SECTION_MAPS 0 +#endif + /* * The idmap and swapper page tables need some space reserved in the kernel * image. Both require pgd, pud (4 levels only) and pmd tables to (section) @@ -28,26 +35,28 @@ * could be increased on the fly if system RAM is out of reach for the default * VA range, so 3 pages are reserved in all cases. */ -#ifdef CONFIG_ARM64_64K_PAGES -#define SWAPPER_PGTABLE_LEVELS (CONFIG_PGTABLE_LEVELS) -#else +#if ARM64_SWAPPER_USES_SECTION_MAPS #define SWAPPER_PGTABLE_LEVELS (CONFIG_PGTABLE_LEVELS - 1) +#else +#define SWAPPER_PGTABLE_LEVELS (CONFIG_PGTABLE_LEVELS) #endif #define SWAPPER_DIR_SIZE (SWAPPER_PGTABLE_LEVELS * PAGE_SIZE) #define IDMAP_DIR_SIZE (3 * PAGE_SIZE) /* Initial memory map size */ -#ifdef CONFIG_ARM64_64K_PAGES -#define SWAPPER_BLOCK_SHIFT PAGE_SHIFT -#define SWAPPER_BLOCK_SIZE PAGE_SIZE -#define SWAPPER_TABLE_SHIFT PMD_SHIFT -#else +#if ARM64_SWAPPER_USES_SECTION_MAPS #define SWAPPER_BLOCK_SHIFT SECTION_SHIFT #define SWAPPER_BLOCK_SIZE SECTION_SIZE #define SWAPPER_TABLE_SHIFT PUD_SHIFT +#else +#define SWAPPER_BLOCK_SHIFT PAGE_SHIFT +#define SWAPPER_BLOCK_SIZE PAGE_SIZE +#define SWAPPER_TABLE_SHIFT PMD_SHIFT #endif +/* The size of the initial kernel direct mapping */ +#define SWAPPER_INIT_MAP_SIZE (_AC(1, UL) << SWAPPER_TABLE_SHIFT) /* * Initial memory map attributes. @@ -55,10 +64,10 @@ #define SWAPPER_PTE_FLAGS PTE_TYPE_PAGE | PTE_AF | PTE_SHARED #define SWAPPER_PMD_FLAGS PMD_TYPE_SECT | PMD_SECT_AF | PMD_SECT_S -#ifdef CONFIG_ARM64_64K_PAGES -#define SWAPPER_MM_MMUFLAGS PTE_ATTRINDX(MT_NORMAL) | SWAPPER_PTE_FLAGS -#else +#if ARM64_SWAPPER_USES_SECTION_MAPS #define SWAPPER_MM_MMUFLAGS PMD_ATTRINDX(MT_NORMAL) | SWAPPER_PMD_FLAGS +#else +#define SWAPPER_MM_MMUFLAGS PTE_ATTRINDX(MT_NORMAL) | SWAPPER_PTE_FLAGS #endif diff --git a/arch/arm64/mm/mmu.c b/arch/arm64/mm/mmu.c index eed6d52..bc4373a 100644 --- a/arch/arm64/mm/mmu.c +++ b/arch/arm64/mm/mmu.c @@ -32,6 +32,7 @@ #include <asm/cputype.h> #include <asm/fixmap.h> +#include <asm/kernel-pgtable.h> #include <asm/sections.h> #include <asm/setup.h> #include <asm/sizes.h> @@ -406,14 +407,11 @@ static void __init map_mem(void) * memory addressable from the initial direct kernel mapping. * * The initial direct kernel mapping, located at swapper_pg_dir, gives - * us PUD_SIZE (4K pages) or PMD_SIZE (64K pages) memory starting from - * PHYS_OFFSET (which must be aligned to 2MB as per - * Documentation/arm64/booting.txt). + * us PUD_SIZE (with SECTION maps, i.e, 4K) or PMD_SIZE (without + * SECTION maps, i.e, 64K pages) memory starting from PHYS_OFFSET + * (which must be aligned to 2MB as per Documentation/arm64/booting.txt). */ - if (IS_ENABLED(CONFIG_ARM64_64K_PAGES)) - limit = PHYS_OFFSET + PMD_SIZE; - else - limit = PHYS_OFFSET + PUD_SIZE; + limit = PHYS_OFFSET + SWAPPER_INIT_MAP_SIZE; memblock_set_current_limit(limit); /* map all the memory banks */ @@ -424,21 +422,24 @@ static void __init map_mem(void) if (start >= end) break; -#ifndef CONFIG_ARM64_64K_PAGES - /* - * For the first memory bank align the start address and - * current memblock limit to prevent create_mapping() from - * allocating pte page tables from unmapped memory. - * When 64K pages are enabled, the pte page table for the - * first PGDIR_SIZE is already present in swapper_pg_dir. - */ - if (start < limit) - start = ALIGN(start, PMD_SIZE); - if (end < limit) { - limit = end & PMD_MASK; - memblock_set_current_limit(limit); + if (ARM64_SWAPPER_USES_SECTION_MAPS) { + /* + * For the first memory bank align the start address and + * current memblock limit to prevent create_mapping() from + * allocating pte page tables from unmapped memory. With + * the section maps, if the first block doesn't end on section + * size boundary, create_mapping() will try to allocate a pte + * page, which may be returned from an unmapped area. + * When section maps are not used, the pte page table for the + * current limit is already present in swapper_pg_dir. + */ + if (start < limit) + start = ALIGN(start, SECTION_SIZE); + if (end < limit) { + limit = end & SECTION_MASK; + memblock_set_current_limit(limit); + } } -#endif __map_memblock(start, end); } @@ -551,7 +552,7 @@ int kern_addr_valid(unsigned long addr) return pfn_valid(pte_pfn(*pte)); } #ifdef CONFIG_SPARSEMEM_VMEMMAP -#ifdef CONFIG_ARM64_64K_PAGES +#if !ARM64_SWAPPER_USES_SECTION_MAPS int __meminit vmemmap_populate(unsigned long start, unsigned long end, int node) { return vmemmap_populate_basepages(start, end, node); @@ -691,7 +692,7 @@ void *__init fixmap_remap_fdt(phys_addr_t dt_phys) { const u64 dt_virt_base = __fix_to_virt(FIX_FDT); pgprot_t prot = PAGE_KERNEL | PTE_RDONLY; - int granularity, size, offset; + int size, offset; void *dt_virt; /* @@ -717,24 +718,15 @@ void *__init fixmap_remap_fdt(phys_addr_t dt_phys) */ BUILD_BUG_ON(dt_virt_base % SZ_2M); - if (IS_ENABLED(CONFIG_ARM64_64K_PAGES)) { - BUILD_BUG_ON(__fix_to_virt(FIX_FDT_END) >> PMD_SHIFT != - __fix_to_virt(FIX_BTMAP_BEGIN) >> PMD_SHIFT); - - granularity = PAGE_SIZE; - } else { - BUILD_BUG_ON(__fix_to_virt(FIX_FDT_END) >> PUD_SHIFT != - __fix_to_virt(FIX_BTMAP_BEGIN) >> PUD_SHIFT); - - granularity = PMD_SIZE; - } + BUILD_BUG_ON(__fix_to_virt(FIX_FDT_END) >> SWAPPER_TABLE_SHIFT != + __fix_to_virt(FIX_BTMAP_BEGIN) >> SWAPPER_TABLE_SHIFT); - offset = dt_phys % granularity; + offset = dt_phys % SWAPPER_BLOCK_SIZE; dt_virt = (void *)dt_virt_base + offset; /* map the first chunk so we can read the size from the header */ - create_mapping(round_down(dt_phys, granularity), dt_virt_base, - granularity, prot); + create_mapping(round_down(dt_phys, SWAPPER_BLOCK_SIZE), dt_virt_base, + SWAPPER_BLOCK_SIZE, prot); if (fdt_check_header(dt_virt) != 0) return NULL; @@ -743,9 +735,9 @@ void *__init fixmap_remap_fdt(phys_addr_t dt_phys) if (size > MAX_FDT_SIZE) return NULL; - if (offset + size > granularity) - create_mapping(round_down(dt_phys, granularity), dt_virt_base, - round_up(offset + size, granularity), prot); + if (offset + size > SWAPPER_BLOCK_SIZE) + create_mapping(round_down(dt_phys, SWAPPER_BLOCK_SIZE), dt_virt_base, + round_up(offset + size, SWAPPER_BLOCK_SIZE), prot); memblock_reserve(dt_phys, size);
We use section maps with 4K page size to create the swapper/idmaps. So far we have used !64K or 4K checks to handle the case where we use the section maps. This patch adds a new symbol, ARM64_SWAPPER_USES_SECTION_MAPS, to handle cases where we use section maps, instead of using the page size symbols. Cc: Ard Biesheuvel <ard.biesheuvel@linaro.org> Cc: Mark Rutland <mark.rutland@arm.com> Cc: Catalin Marinas <catalin.marinas@arm.com> Cc: Will Deacon <will.deacon@arm.com> Signed-off-by: Suzuki K. Poulose <suzuki.poulose@arm.com> --- arch/arm64/include/asm/kernel-pgtable.h | 31 ++++++++----- arch/arm64/mm/mmu.c | 72 ++++++++++++++----------------- 2 files changed, 52 insertions(+), 51 deletions(-)