bootmem: factor out the marking of a PFN range
Johannes Weiner [Thu, 24 Jul 2008 04:28:06 +0000 (21:28 -0700)]
Introduce new helpers that mark a range that resides completely on a node
or node-agnostic ranges that might also span node boundaries.

The free/reserve API functions will then directly use these helpers.

Note that the free/reserve semantics become more strict: while the prior
code took basically arbitrary range arguments and marked the PFNs that
happen to fall into that range, the new code requires node-specific ranges
to be completely on the node.  The node-agnostic requests might span node
boundaries as long as the nodes are contiguous.

Passing ranges that do not satisfy these criteria is a bug.

[akpm@linux-foundation.org: fix printk warnings]
Signed-off-by: Johannes Weiner <hannes@saeurebad.de>
Cc: Ingo Molnar <mingo@elte.hu>
Cc: Yinghai Lu <yhlu.kernel@gmail.com>
Cc: Andi Kleen <andi@firstfloor.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

mm/bootmem.c

index 9d03ff6..e5415a5 100644 (file)
@@ -234,6 +234,9 @@ static void __init __free(bootmem_data_t *bdata,
                sidx + PFN_DOWN(bdata->node_boot_start),
                eidx + PFN_DOWN(bdata->node_boot_start));
 
+       if (bdata->hint_idx > sidx)
+               bdata->hint_idx = sidx;
+
        for (idx = sidx; idx < eidx; idx++)
                if (!test_and_clear_bit(idx, bdata->node_bootmem_map))
                        BUG();
@@ -263,40 +266,57 @@ static int __init __reserve(bootmem_data_t *bdata, unsigned long sidx,
        return 0;
 }
 
-static void __init free_bootmem_core(bootmem_data_t *bdata, unsigned long addr,
-                                    unsigned long size)
+static int __init mark_bootmem_node(bootmem_data_t *bdata,
+                               unsigned long start, unsigned long end,
+                               int reserve, int flags)
 {
        unsigned long sidx, eidx;
-       unsigned long i;
 
-       BUG_ON(!size);
+       bdebug("nid=%td start=%lx end=%lx reserve=%d flags=%x\n",
+               bdata - bootmem_node_data, start, end, reserve, flags);
 
-       /* out range */
-       if (addr + size < bdata->node_boot_start ||
-               PFN_DOWN(addr) > bdata->node_low_pfn)
-               return;
-       /*
-        * round down end of usable mem, partially free pages are
-        * considered reserved.
-        */
+       BUG_ON(start < PFN_DOWN(bdata->node_boot_start));
+       BUG_ON(end > bdata->node_low_pfn);
 
-       if (addr >= bdata->node_boot_start &&
-                       PFN_DOWN(addr - bdata->node_boot_start) < bdata->hint_idx)
-               bdata->hint_idx = PFN_DOWN(addr - bdata->node_boot_start);
+       sidx = start - PFN_DOWN(bdata->node_boot_start);
+       eidx = end - PFN_DOWN(bdata->node_boot_start);
 
-       /*
-        * Round up to index to the range.
-        */
-       if (PFN_UP(addr) > PFN_DOWN(bdata->node_boot_start))
-               sidx = PFN_UP(addr) - PFN_DOWN(bdata->node_boot_start);
+       if (reserve)
+               return __reserve(bdata, sidx, eidx, flags);
        else
-               sidx = 0;
+               __free(bdata, sidx, eidx);
+       return 0;
+}
+
+static int __init mark_bootmem(unsigned long start, unsigned long end,
+                               int reserve, int flags)
+{
+       unsigned long pos;
+       bootmem_data_t *bdata;
+
+       pos = start;
+       list_for_each_entry(bdata, &bdata_list, list) {
+               int err;
+               unsigned long max;
+
+               if (pos < PFN_DOWN(bdata->node_boot_start)) {
+                       BUG_ON(pos != start);
+                       continue;
+               }
+
+               max = min(bdata->node_low_pfn, end);
 
-       eidx = PFN_DOWN(addr + size - bdata->node_boot_start);
-       if (eidx > bdata->node_low_pfn - PFN_DOWN(bdata->node_boot_start))
-               eidx = bdata->node_low_pfn - PFN_DOWN(bdata->node_boot_start);
+               err = mark_bootmem_node(bdata, pos, max, reserve, flags);
+               if (reserve && err) {
+                       mark_bootmem(start, pos, 0, 0);
+                       return err;
+               }
 
-       __free(bdata, sidx, eidx);
+               if (max == end)
+                       return 0;
+               pos = bdata->node_low_pfn;
+       }
+       BUG();
 }
 
 /**
@@ -307,12 +327,17 @@ static void __init free_bootmem_core(bootmem_data_t *bdata, unsigned long addr,
  *
  * Partial pages will be considered reserved and left as they are.
  *
- * Only physical pages that actually reside on @pgdat are marked.
+ * The range must reside completely on the specified node.
  */
 void __init free_bootmem_node(pg_data_t *pgdat, unsigned long physaddr,
                              unsigned long size)
 {
-       free_bootmem_core(pgdat->bdata, physaddr, size);
+       unsigned long start, end;
+
+       start = PFN_UP(physaddr);
+       end = PFN_DOWN(physaddr + size);
+
+       mark_bootmem_node(pgdat->bdata, start, end, 0, 0);
 }
 
 /**
@@ -322,83 +347,16 @@ void __init free_bootmem_node(pg_data_t *pgdat, unsigned long physaddr,
  *
  * Partial pages will be considered reserved and left as they are.
  *
- * All physical pages within the range are marked, no matter what
- * node they reside on.
+ * The range must be contiguous but may span node boundaries.
  */
 void __init free_bootmem(unsigned long addr, unsigned long size)
 {
-       bootmem_data_t *bdata;
-       list_for_each_entry(bdata, &bdata_list, list)
-               free_bootmem_core(bdata, addr, size);
-}
-
-/*
- * Marks a particular physical memory range as unallocatable. Usable RAM
- * might be used for boot-time allocations - or it might get added
- * to the free page pool later on.
- */
-static int __init can_reserve_bootmem_core(bootmem_data_t *bdata,
-                       unsigned long addr, unsigned long size, int flags)
-{
-       unsigned long sidx, eidx;
-       unsigned long i;
-
-       BUG_ON(!size);
-
-       /* out of range, don't hold other */
-       if (addr + size < bdata->node_boot_start ||
-               PFN_DOWN(addr) > bdata->node_low_pfn)
-               return 0;
-
-       /*
-        * Round up to index to the range.
-        */
-       if (addr > bdata->node_boot_start)
-               sidx= PFN_DOWN(addr - bdata->node_boot_start);
-       else
-               sidx = 0;
-
-       eidx = PFN_UP(addr + size - bdata->node_boot_start);
-       if (eidx > bdata->node_low_pfn - PFN_DOWN(bdata->node_boot_start))
-               eidx = bdata->node_low_pfn - PFN_DOWN(bdata->node_boot_start);
-
-       for (i = sidx; i < eidx; i++) {
-               if (test_bit(i, bdata->node_bootmem_map)) {
-                       if (flags & BOOTMEM_EXCLUSIVE)
-                               return -EBUSY;
-               }
-       }
-
-       return 0;
-
-}
-
-static void __init reserve_bootmem_core(bootmem_data_t *bdata,
-                       unsigned long addr, unsigned long size, int flags)
-{
-       unsigned long sidx, eidx;
-       unsigned long i;
-
-       BUG_ON(!size);
-
-       /* out of range */
-       if (addr + size < bdata->node_boot_start ||
-               PFN_DOWN(addr) > bdata->node_low_pfn)
-               return;
-
-       /*
-        * Round up to index to the range.
-        */
-       if (addr > bdata->node_boot_start)
-               sidx= PFN_DOWN(addr - bdata->node_boot_start);
-       else
-               sidx = 0;
+       unsigned long start, end;
 
-       eidx = PFN_UP(addr + size - bdata->node_boot_start);
-       if (eidx > bdata->node_low_pfn - PFN_DOWN(bdata->node_boot_start))
-               eidx = bdata->node_low_pfn - PFN_DOWN(bdata->node_boot_start);
+       start = PFN_UP(addr);
+       end = PFN_DOWN(addr + size);
 
-       return __reserve(bdata, sidx, eidx, flags);
+       mark_bootmem(start, end, 0, 0);
 }
 
 /**
@@ -410,18 +368,17 @@ static void __init reserve_bootmem_core(bootmem_data_t *bdata,
  *
  * Partial pages will be reserved.
  *
- * Only physical pages that actually reside on @pgdat are marked.
+ * The range must reside completely on the specified node.
  */
 int __init reserve_bootmem_node(pg_data_t *pgdat, unsigned long physaddr,
                                 unsigned long size, int flags)
 {
-       int ret;
+       unsigned long start, end;
 
-       ret = can_reserve_bootmem_core(pgdat->bdata, physaddr, size, flags);
-       if (ret < 0)
-               return -ENOMEM;
-       reserve_bootmem_core(pgdat->bdata, physaddr, size, flags);
-       return 0;
+       start = PFN_DOWN(physaddr);
+       end = PFN_UP(physaddr + size);
+
+       return mark_bootmem_node(pgdat->bdata, start, end, 1, flags);
 }
 
 #ifndef CONFIG_HAVE_ARCH_BOOTMEM_NODE
@@ -433,24 +390,17 @@ int __init reserve_bootmem_node(pg_data_t *pgdat, unsigned long physaddr,
  *
  * Partial pages will be reserved.
  *
- * All physical pages within the range are marked, no matter what
- * node they reside on.
+ * The range must be contiguous but may span node boundaries.
  */
 int __init reserve_bootmem(unsigned long addr, unsigned long size,
                            int flags)
 {
-       bootmem_data_t *bdata;
-       int ret;
+       unsigned long start, end;
 
-       list_for_each_entry(bdata, &bdata_list, list) {
-               ret = can_reserve_bootmem_core(bdata, addr, size, flags);
-               if (ret < 0)
-                       return ret;
-       }
-       list_for_each_entry(bdata, &bdata_list, list)
-               reserve_bootmem_core(bdata, addr, size, flags);
+       start = PFN_DOWN(addr);
+       end = PFN_UP(addr + size);
 
-       return 0;
+       return mark_bootmem(start, end, 1, flags);
 }
 #endif /* !CONFIG_HAVE_ARCH_BOOTMEM_NODE */
 
@@ -663,7 +613,7 @@ void * __init alloc_bootmem_section(unsigned long size,
        if (start_nr != section_nr || end_nr != section_nr) {
                printk(KERN_WARNING "alloc_bootmem failed on section %ld.\n",
                       section_nr);
-               free_bootmem_core(pgdat->bdata, __pa(ptr), size);
+               free_bootmem_node(pgdat, __pa(ptr), size);
                ptr = NULL;
        }