[PATCH V6] ARM: LPAE: Fix mapping in alloc_init_section for unaligned addresses
Russell King - ARM Linux
linux at arm.linux.org.uk
Mon Mar 18 06:45:43 EDT 2013
On Mon, Mar 18, 2013 at 11:20:47AM +0530, Sricharan R wrote:
> Hi Russell,
>
> On Monday 18 March 2013 01:22 AM, Christoffer Dall wrote:
> > On Sat, Mar 16, 2013 at 10:05 PM, Sricharan R <r.sricharan at ti.com> wrote:
> >> From: R Sricharan <r.sricharan at ti.com>
> >>
> >> With LPAE enabled, alloc_init_section() does not map the entire
> >> address space for unaligned addresses.
> >>
> >> The issue also reproduced with CMA + LPAE. CMA tries to map 16MB
> >> with page granularity mappings during boot. alloc_init_pte()
> >> is called and out of 16MB, only 2MB gets mapped and rest remains
> >> unaccessible.
> >>
> >> Because of this OMAP5 boot is broken with CMA + LPAE enabled.
> >> Fix the issue by ensuring that the entire addresses are
> >> mapped.
> >>
> >> Signed-off-by: R Sricharan <r.sricharan at ti.com>
> >> Cc: Catalin Marinas <catalin.marinas at arm.com>
> >> Cc: Christoffer Dall <chris at cloudcar.com>
> >> Cc: Russell King <linux at arm.linux.org.uk>
> >> Cc: Santosh Shilimkar <santosh.shilimkar at ti.com>
> >> Tested-by: Laura Abbott <lauraa at codeaurora.org>
> >> Acked-by: Catalin Marinas <catalin.marinas at arm.com>
> >> ---
> >> [V2] Moved the loop to alloc_init_pte as per Russell's
> >> feedback and changed the subject accordingly.
> >> Using PMD_XXX instead of SECTION_XXX to avoid
> >> different loop increments with/without LPAE.
> >>
> >> [v3] Removed the dummy variable phys and updated
> >> the commit log for CMA case.
> >>
> >> [v4] Resending with updated change log and
> >> updating the tags.
> >>
> >> [v5] Renamed alloc_init_section to alloc_init_pmd
> >> and moved the loop back there. Also introduced
> >> map_init_section as per Catalin's comments.
> >>
> >> [v6] Corrected tags and updated the comments for code.
> >>
> >> arch/arm/mm/mmu.c | 73 ++++++++++++++++++++++++++++++++++-------------------
> >> 1 file changed, 47 insertions(+), 26 deletions(-)
> >>
> >> diff --git a/arch/arm/mm/mmu.c b/arch/arm/mm/mmu.c
> >> index e95a996..7897894 100644
> >> --- a/arch/arm/mm/mmu.c
> >> +++ b/arch/arm/mm/mmu.c
> >> @@ -598,39 +598,60 @@ static void __init alloc_init_pte(pmd_t *pmd, unsigned long addr,
> >> } while (pte++, addr += PAGE_SIZE, addr != end);
> >> }
> >>
> >> -static void __init alloc_init_section(pud_t *pud, unsigned long addr,
> >> - unsigned long end, phys_addr_t phys,
> >> - const struct mem_type *type)
> >> +static void __init map_init_section(pmd_t *pmd, unsigned long addr,
> >> + unsigned long end, phys_addr_t phys,
> >> + const struct mem_type *type)
> >> {
> >> - pmd_t *pmd = pmd_offset(pud, addr);
> >> -
> >> +#ifndef CONFIG_ARM_LPAE
> >> /*
> >> - * Try a section mapping - end, addr and phys must all be aligned
> >> - * to a section boundary. Note that PMDs refer to the individual
> >> - * L1 entries, whereas PGDs refer to a group of L1 entries making
> >> - * up one logical pointer to an L2 table.
> >> + * In classic MMU format, puds and pmds are folded in to
> >> + * the pgds. pmd_offset gives the PGD entry. PGDs refer to a
> >> + * group of L1 entries making up one logical pointer to
> >> + * an L2 table (2MB), where as PMDs refer to the individual
> >> + * L1 entries (1MB). Hence increment to get the correct
> >> + * offset for odd 1MB sections.
> >> + * (See arch/arm/include/asm/pgtable-2level.h)
> >> */
> >> - if (type->prot_sect && ((addr | end | phys) & ~SECTION_MASK) == 0) {
> >> - pmd_t *p = pmd;
> >> -
> >> -#ifndef CONFIG_ARM_LPAE
> >> - if (addr & SECTION_SIZE)
> >> - pmd++;
> >> + if (addr & SECTION_SIZE)
> >> + pmd++;
> >> #endif
> >> + do {
> >> + *pmd = __pmd(phys | type->prot_sect);
> >> + phys += SECTION_SIZE;
> >> + } while (pmd++, addr += SECTION_SIZE, addr != end);
> >>
> >> - do {
> >> - *pmd = __pmd(phys | type->prot_sect);
> >> - phys += SECTION_SIZE;
> >> - } while (pmd++, addr += SECTION_SIZE, addr != end);
> >> + flush_pmd_entry(pmd);
> >> +}
> >>
> >> - flush_pmd_entry(p);
> >> - } else {
> >> +static void __init alloc_init_pmd(pud_t *pud, unsigned long addr,
> >> + unsigned long end, phys_addr_t phys,
> >> + const struct mem_type *type)
> >> +{
> >> + pmd_t *pmd = pmd_offset(pud, addr);
> >> + unsigned long next;
> >> +
> >> + do {
> >> /*
> >> - * No need to loop; pte's aren't interested in the
> >> - * individual L1 entries.
> >> + * With LPAE, we must loop over to map
> >> + * all the pmds for the given range.
> >> */
> >> - alloc_init_pte(pmd, addr, end, __phys_to_pfn(phys), type);
> >> - }
> >> + next = pmd_addr_end(addr, end);
> >> +
> >> + /*
> >> + * Try a section mapping - addr, next and phys must all be
> >> + * aligned to a section boundary.
> >> + */
> >> + if (type->prot_sect &&
> >> + ((addr | next | phys) & ~SECTION_MASK) == 0) {
> >> + map_init_section(pmd, addr, next, phys, type);
> >> + } else {
> >> + alloc_init_pte(pmd, addr, next,
> >> + __phys_to_pfn(phys), type);
> >> + }
> >> +
> >> + phys += next - addr;
> >> +
> >> + } while (pmd++, addr = next, addr != end);
> >> }
> >>
> >> static void __init alloc_init_pud(pgd_t *pgd, unsigned long addr,
> >> @@ -641,7 +662,7 @@ static void __init alloc_init_pud(pgd_t *pgd, unsigned long addr,
> >>
> >> do {
> >> next = pud_addr_end(addr, end);
> >> - alloc_init_section(pud, addr, next, phys, type);
> >> + alloc_init_pmd(pud, addr, next, phys, type);
> >> phys += next - addr;
> >> } while (pud++, addr = next, addr != end);
> >> }
> >> --
> >> 1.7.9.5
> >>
> > Acked-by: Christoffer Dall <chris at cloudcar.com>
>
> I am not able to add this in to the patch system because my login fails.
> I was trying using the credentials registered with linux-arm-kernel
> mailing list. Can you please help me here ?
The mailing list and my site are two entirely separate and independent
sites. In fact now, the mailing list is hosted by a separate
individual.
More information about the linux-arm-kernel
mailing list