From d0be89f6c2570a63ac44ccdd12473a54243cd296 Mon Sep 17 00:00:00 2001 From: Jian Peng Date: Tue, 17 May 2011 12:27:49 -0700 Subject: [PATCH] MIPS: topdown mmap support This patch introduced topdown mmap support in user process address space allocation policy. Recently, we ran some large applications that use mmap heavily and lead to OOM due to inflexible mmap allocation policy on MIPS32. Since most other major archs supported it for years, it is reasonable to follow the trend and reduce the pain of porting applications. Due to cache aliasing concern, arch_get_unmapped_area_topdown() and other helper functions are implemented in arch/mips/kernel/syscall.c. Signed-off-by: Jian Peng Cc: David Daney Cc: linux-mips@linux-mips.org Patchwork: https://patchwork.linux-mips.org/patch/2389/ Signed-off-by: Ralf Baechle --- arch/mips/include/asm/pgtable.h | 1 + arch/mips/mm/mmap.c | 193 ++++++++++++++++++++++++++++---- 2 files changed, 171 insertions(+), 23 deletions(-) diff --git a/arch/mips/include/asm/pgtable.h b/arch/mips/include/asm/pgtable.h index 7e40f3778179..b2202a68cf0f 100644 --- a/arch/mips/include/asm/pgtable.h +++ b/arch/mips/include/asm/pgtable.h @@ -414,6 +414,7 @@ int phys_mem_access_prot_allowed(struct file *file, unsigned long pfn, * constraints placed on us by the cache architecture. */ #define HAVE_ARCH_UNMAPPED_AREA +#define HAVE_ARCH_UNMAPPED_AREA_TOPDOWN /* * No page table caches to initialise diff --git a/arch/mips/mm/mmap.c b/arch/mips/mm/mmap.c index ae3c20a9556e..9ff5d0fac556 100644 --- a/arch/mips/mm/mmap.c +++ b/arch/mips/mm/mmap.c @@ -10,6 +10,7 @@ #include #include #include +#include #include #include @@ -17,21 +18,65 @@ unsigned long shm_align_mask = PAGE_SIZE - 1; /* Sane caches */ EXPORT_SYMBOL(shm_align_mask); +/* gap between mmap and stack */ +#define MIN_GAP (128*1024*1024UL) +#define MAX_GAP ((TASK_SIZE)/6*5) + +static int mmap_is_legacy(void) +{ + if (current->personality & ADDR_COMPAT_LAYOUT) + return 1; + + if (rlimit(RLIMIT_STACK) == RLIM_INFINITY) + return 1; + + return sysctl_legacy_va_layout; +} + +static unsigned long mmap_base(unsigned long rnd) +{ + unsigned long gap = rlimit(RLIMIT_STACK); + + if (gap < MIN_GAP) + gap = MIN_GAP; + else if (gap > MAX_GAP) + gap = MAX_GAP; + + return PAGE_ALIGN(TASK_SIZE - gap - rnd); +} + +static inline unsigned long COLOUR_ALIGN_DOWN(unsigned long addr, + unsigned long pgoff) +{ + unsigned long base = addr & ~shm_align_mask; + unsigned long off = (pgoff << PAGE_SHIFT) & shm_align_mask; + + if (base + off <= addr) + return base + off; + + return base - off; +} + #define COLOUR_ALIGN(addr,pgoff) \ ((((addr) + shm_align_mask) & ~shm_align_mask) + \ (((pgoff) << PAGE_SHIFT) & shm_align_mask)) -unsigned long arch_get_unmapped_area(struct file *filp, unsigned long addr, - unsigned long len, unsigned long pgoff, unsigned long flags) +enum mmap_allocation_direction {UP, DOWN}; + +static unsigned long arch_get_unmapped_area_foo(struct file *filp, + unsigned long addr0, unsigned long len, unsigned long pgoff, + unsigned long flags, enum mmap_allocation_direction dir) { - struct vm_area_struct * vmm; + struct mm_struct *mm = current->mm; + struct vm_area_struct *vma; + unsigned long addr = addr0; int do_color_align; - if (len > TASK_SIZE) + if (unlikely(len > TASK_SIZE)) return -ENOMEM; if (flags & MAP_FIXED) { - /* Even MAP_FIXED mappings must reside within TASK_SIZE. */ + /* Even MAP_FIXED mappings must reside within TASK_SIZE */ if (TASK_SIZE - len < addr) return -EINVAL; @@ -48,34 +93,130 @@ unsigned long arch_get_unmapped_area(struct file *filp, unsigned long addr, do_color_align = 0; if (filp || (flags & MAP_SHARED)) do_color_align = 1; + + /* requesting a specific address */ if (addr) { if (do_color_align) addr = COLOUR_ALIGN(addr, pgoff); else addr = PAGE_ALIGN(addr); - vmm = find_vma(current->mm, addr); + + vma = find_vma(mm, addr); if (TASK_SIZE - len >= addr && - (!vmm || addr + len <= vmm->vm_start)) + (!vma || addr + len <= vma->vm_start)) return addr; } - addr = current->mm->mmap_base; - if (do_color_align) - addr = COLOUR_ALIGN(addr, pgoff); - else - addr = PAGE_ALIGN(addr); - for (vmm = find_vma(current->mm, addr); ; vmm = vmm->vm_next) { - /* At this point: (!vmm || addr < vmm->vm_end). */ - if (TASK_SIZE - len < addr) - return -ENOMEM; - if (!vmm || addr + len <= vmm->vm_start) - return addr; - addr = vmm->vm_end; + if (dir == UP) { + addr = mm->mmap_base; + if (do_color_align) + addr = COLOUR_ALIGN(addr, pgoff); + else + addr = PAGE_ALIGN(addr); + + for (vma = find_vma(current->mm, addr); ; vma = vma->vm_next) { + /* At this point: (!vma || addr < vma->vm_end). */ + if (TASK_SIZE - len < addr) + return -ENOMEM; + if (!vma || addr + len <= vma->vm_start) + return addr; + addr = vma->vm_end; + if (do_color_align) + addr = COLOUR_ALIGN(addr, pgoff); + } + } else { + /* check if free_area_cache is useful for us */ + if (len <= mm->cached_hole_size) { + mm->cached_hole_size = 0; + mm->free_area_cache = mm->mmap_base; + } + + /* either no address requested or can't fit in requested address hole */ + addr = mm->free_area_cache; + if (do_color_align) { + unsigned long base = + COLOUR_ALIGN_DOWN(addr - len, pgoff); + + addr = base + len; + } + + /* make sure it can fit in the remaining address space */ + if (likely(addr > len)) { + vma = find_vma(mm, addr - len); + if (!vma || addr <= vma->vm_start) { + /* remember the address as a hint for next time */ + return mm->free_area_cache = addr-len; + } + } + + if (unlikely(mm->mmap_base < len)) + goto bottomup; + + addr = mm->mmap_base-len; if (do_color_align) - addr = COLOUR_ALIGN(addr, pgoff); + addr = COLOUR_ALIGN_DOWN(addr, pgoff); + + do { + /* + * Lookup failure means no vma is above this address, + * else if new region fits below vma->vm_start, + * return with success: + */ + vma = find_vma(mm, addr); + if (likely(!vma || addr+len <= vma->vm_start)) { + /* remember the address as a hint for next time */ + return mm->free_area_cache = addr; + } + + /* remember the largest hole we saw so far */ + if (addr + mm->cached_hole_size < vma->vm_start) + mm->cached_hole_size = vma->vm_start - addr; + + /* try just below the current vma->vm_start */ + addr = vma->vm_start-len; + if (do_color_align) + addr = COLOUR_ALIGN_DOWN(addr, pgoff); + } while (likely(len < vma->vm_start)); + +bottomup: + /* + * A failed mmap() very likely causes application failure, + * so fall back to the bottom-up function here. This scenario + * can happen with large stack limits and large mmap() + * allocations. + */ + mm->cached_hole_size = ~0UL; + mm->free_area_cache = TASK_UNMAPPED_BASE; + addr = arch_get_unmapped_area(filp, addr0, len, pgoff, flags); + /* + * Restore the topdown base: + */ + mm->free_area_cache = mm->mmap_base; + mm->cached_hole_size = ~0UL; + + return addr; } } +unsigned long arch_get_unmapped_area(struct file *filp, unsigned long addr0, + unsigned long len, unsigned long pgoff, unsigned long flags) +{ + return arch_get_unmapped_area_foo(filp, + addr0, len, pgoff, flags, UP); +} + +/* + * There is no need to export this but sched.h declares the function as + * extern so making it static here results in an error. + */ +unsigned long arch_get_unmapped_area_topdown(struct file *filp, + unsigned long addr0, unsigned long len, unsigned long pgoff, + unsigned long flags) +{ + return arch_get_unmapped_area_foo(filp, + addr0, len, pgoff, flags, DOWN); +} + void arch_pick_mmap_layout(struct mm_struct *mm) { unsigned long random_factor = 0UL; @@ -89,9 +230,15 @@ void arch_pick_mmap_layout(struct mm_struct *mm) random_factor &= 0xffffffful; } - mm->mmap_base = TASK_UNMAPPED_BASE + random_factor; - mm->get_unmapped_area = arch_get_unmapped_area; - mm->unmap_area = arch_unmap_area; + if (mmap_is_legacy()) { + mm->mmap_base = TASK_UNMAPPED_BASE + random_factor; + mm->get_unmapped_area = arch_get_unmapped_area; + mm->unmap_area = arch_unmap_area; + } else { + mm->mmap_base = mmap_base(random_factor); + mm->get_unmapped_area = arch_get_unmapped_area_topdown; + mm->unmap_area = arch_unmap_area_topdown; + } } static inline unsigned long brk_rnd(void) -- GitLab