[SPARC64]: Top-down address space allocation for 32-bit tasks.

Currently allocations are very constrained for 32-bit processes.
It grows down-up from 0x70000000 to 0xf0000000 which gives about
2GB of stack + dynamic mmap() space.

So support the top-down method, and we need to override the
generic helper function in order to deal with D-cache coloring.

With these changes I was able to squeeze out a mmap() just over
3.6GB in size in a 32-bit process.

Signed-off-by: David S. Miller <davem@davemloft.net>

4 files changed, 189 insertions, 10 deletions

diff --git a/arch/sparc64/kernel/binfmt_aout32.c b/arch/sparc64/kernel/binfmt_aout32.c
index cb9ecd0172c0..d7caa60a0074 100644
--- a/arch/sparc64/kernel/binfmt_aout32.c
+++ b/arch/sparc64/kernel/binfmt_aout32.c
@@ -239,6 +239,8 @@ static int load_aout32_binary(struct linux_binprm * bprm, struct pt_regs * regs)
                 (current->mm->start_data = N_DATADDR(ex));
         current->mm->brk = ex.a_bss +
                 (current->mm->start_brk = N_BSSADDR(ex));
+        current->mm->free_area_cache = current->mm->mmap_base;
+        current->mm->cached_hole_size = 0;
 
         current->mm->mmap = NULL;
         compute_creds(bprm);
diff --git a/arch/sparc64/kernel/sys_sparc.c b/arch/sparc64/kernel/sys_sparc.c
index 8840415408be..61dffb9349bd 100644
--- a/arch/sparc64/kernel/sys_sparc.c
+++ b/arch/sparc64/kernel/sys_sparc.c
@@ -82,9 +82,34 @@ static inline int straddles_64bit_va_hole(unsigned long start, unsigned long end
         return 1;
 }
 
-#define COLOUR_ALIGN(addr,pgoff)                \
-        ((((addr)+SHMLBA-1)&~(SHMLBA-1)) +      \
-         (((pgoff)<<PAGE_SHIFT) & (SHMLBA-1)))
+/* These functions differ from the default implementations in
+ * mm/mmap.c in two ways:
+ *
+ * 1) For file backed MAP_SHARED mmap()'s we D-cache color align,
+ *    for fixed such mappings we just validate what the user gave us.
+ * 2) For 64-bit tasks we avoid mapping anything within 4GB of
+ *    the spitfire/niagara VA-hole.
+ */
+
+static inline unsigned long COLOUR_ALIGN(unsigned long addr,
+                                         unsigned long pgoff)
+{
+        unsigned long base = (addr+SHMLBA-1)&~(SHMLBA-1);
+        unsigned long off = (pgoff<<PAGE_SHIFT) & (SHMLBA-1);
+
+        return base + off;
+}
+
+static inline unsigned long COLOUR_ALIGN_DOWN(unsigned long addr,
+                                              unsigned long pgoff)
+{
+        unsigned long base = addr & ~(SHMLBA-1);
+        unsigned long off = (pgoff<<PAGE_SHIFT) & (SHMLBA-1);
+
+        if (base + off <= addr)
+                return base + off;
+        return base - off;
+}
 
 unsigned long arch_get_unmapped_area(struct file *filp, unsigned long addr, unsigned long len, unsigned long pgoff, unsigned long flags)
 {
@@ -106,7 +131,7 @@ unsigned long arch_get_unmapped_area(struct file *filp, unsigned long addr, unsi
 
         if (test_thread_flag(TIF_32BIT))
                 task_size = 0xf0000000UL;
-        if (len > task_size || len >= VA_EXCLUDE_START)
+        if (unlikely(len > task_size || len >= VA_EXCLUDE_START))
                 return -ENOMEM;
 
         do_color_align = 0;
@@ -125,11 +150,12 @@ unsigned long arch_get_unmapped_area(struct file *filp, unsigned long addr, unsi
                         return addr;
         }
 
-        if (len <= mm->cached_hole_size) {
+        if (len > mm->cached_hole_size) {
+                start_addr = addr = mm->free_area_cache;
+        } else {
+                start_addr = addr = TASK_UNMAPPED_BASE;
                 mm->cached_hole_size = 0;
-                mm->free_area_cache = TASK_UNMAPPED_BASE;
         }
-        start_addr = addr = mm->free_area_cache;
 
         task_size -= len;
 
@@ -146,7 +172,7 @@ full_search:
                         addr = VA_EXCLUDE_END;
                         vma = find_vma(mm, VA_EXCLUDE_END);
                 }
-                if (task_size < addr) {
+                if (unlikely(task_size < addr)) {
                         if (start_addr != TASK_UNMAPPED_BASE) {
                                 start_addr = addr = TASK_UNMAPPED_BASE;
                                 mm->cached_hole_size = 0;
@@ -154,7 +180,7 @@ full_search:
                         }
                         return -ENOMEM;
                 }
-                if (!vma || addr + len <= vma->vm_start) {
+                if (likely(!vma || addr + len <= vma->vm_start)) {
                         /*
                          * Remember the place where we stopped the search:
                          */
@@ -170,6 +196,121 @@ full_search:
         }
 }
 
+unsigned long
+arch_get_unmapped_area_topdown(struct file *filp, const unsigned long addr0,
+                          const unsigned long len, const unsigned long pgoff,
+                          const unsigned long flags)
+{
+        struct vm_area_struct *vma;
+        struct mm_struct *mm = current->mm;
+        unsigned long task_size = 0xf0000000UL;
+        unsigned long addr = addr0;
+        int do_color_align;
+
+        /* This should only ever run for 32-bit processes.  */
+        BUG_ON(!test_thread_flag(TIF_32BIT));
+
+        if (flags & MAP_FIXED) {
+                /* We do not accept a shared mapping if it would violate
+                 * cache aliasing constraints.
+                 */
+                if ((flags & MAP_SHARED) &&
+                    ((addr - (pgoff << PAGE_SHIFT)) & (SHMLBA - 1)))
+                        return -EINVAL;
+                return addr;
+        }
+
+        if (unlikely(len > task_size))
+                return -ENOMEM;
+
+        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);
+
+                vma = find_vma(mm, addr);
+                if (task_size - len >= addr &&
+                    (!vma || addr + len <= vma->vm_start))
+                        return addr;
+        }
+
+        /* 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_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;
+}
+
 /* Try to align mapping such that we align it as much as possible. */
 unsigned long get_fb_unmapped_area(struct file *filp, unsigned long orig_addr, unsigned long len, unsigned long pgoff, unsigned long flags)
 {
@@ -213,6 +354,37 @@ unsigned long get_fb_unmapped_area(struct file *filp, unsigned long orig_addr, u
         return addr;
 }
 
+/* Essentially the same as PowerPC... */
+void arch_pick_mmap_layout(struct mm_struct *mm)
+{
+        /*
+         * Fall back to the standard layout if the personality
+         * bit is set, or if the expected stack growth is unlimited:
+         */
+        if (!test_thread_flag(TIF_32BIT) ||
+            (current->personality & ADDR_COMPAT_LAYOUT) ||
+            current->signal->rlim[RLIMIT_STACK].rlim_cur == RLIM_INFINITY ||
+            sysctl_legacy_va_layout) {
+                mm->mmap_base = TASK_UNMAPPED_BASE;
+                mm->get_unmapped_area = arch_get_unmapped_area;
+                mm->unmap_area = arch_unmap_area;
+        } else {
+                /* We know it's 32-bit */
+                unsigned long task_size = 0xf0000000UL;
+                unsigned long gap;
+
+                gap = current->signal->rlim[RLIMIT_STACK].rlim_cur;
+                if (gap < 128 * 1024 * 1024)
+                        gap = 128 * 1024 * 1024;
+                if (gap > (task_size / 6 * 5))
+                        gap = (task_size / 6 * 5);
+
+                mm->mmap_base = task_size - (gap & PAGE_MASK);
+                mm->get_unmapped_area = arch_get_unmapped_area_topdown;
+                mm->unmap_area = arch_unmap_area_topdown;
+        }
+}
+
 asmlinkage unsigned long sparc_brk(unsigned long brk)
 {
         /* People could try to be nasty and use ta 0x6d in 32bit programs */
diff --git a/include/asm-sparc64/pgtable.h b/include/asm-sparc64/pgtable.h
index d427ce649214..ed4124edf837 100644
--- a/include/asm-sparc64/pgtable.h
+++ b/include/asm-sparc64/pgtable.h
@@ -752,8 +752,11 @@ extern int io_remap_pfn_range(struct vm_area_struct *vma, unsigned long from,
 
 #include <asm-generic/pgtable.h>
 
-/* We provide our own get_unmapped_area to cope with VA holes for userland */
+/* We provide our own get_unmapped_area to cope with VA holes and
+ * SHM area cache aliasing for userland.
+ */
 #define HAVE_ARCH_UNMAPPED_AREA
+#define HAVE_ARCH_UNMAPPED_AREA_TOPDOWN
 
 /* We provide a special get_unmapped_area for framebuffer mmaps to try and use
  * the largest alignment possible such that larget PTEs can be used.
diff --git a/include/asm-sparc64/processor.h b/include/asm-sparc64/processor.h
index 685479fb4364..c6896b88283e 100644
--- a/include/asm-sparc64/processor.h
+++ b/include/asm-sparc64/processor.h
@@ -217,6 +217,8 @@ static inline void prefetchw(const void *x)
 
 #define spin_lock_prefetch(x)   prefetchw(x)
 
+#define HAVE_ARCH_PICK_MMAP_LAYOUT
+
 #endif /* !(__ASSEMBLY__) */
 
 #endif /* !(__ASM_SPARC64_PROCESSOR_H) */