x86, mm: Find early page table buffer together
authorYinghai Lu <yinghai@kernel.org>
Sat, 17 Nov 2012 03:38:44 +0000 (19:38 -0800)
committerH. Peter Anvin <hpa@linux.intel.com>
Sat, 17 Nov 2012 19:59:06 +0000 (11:59 -0800)
We should not do that in every calling of init_memory_mapping.

At the same time need to move down early_memtest, and could remove after_bootmem
checking.

-v2: fix one early_memtest with 32bit by passing max_pfn_mapped instead.

Signed-off-by: Yinghai Lu <yinghai@kernel.org>
Link: http://lkml.kernel.org/r/1353123563-3103-8-git-send-email-yinghai@kernel.org
Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>
arch/x86/mm/init.c

index 51f919febf6409581843aa9e5aa479908dc4f094..1ce0d033fafc7acb87a6e7957ce8edee35144e78 100644 (file)
@@ -274,16 +274,6 @@ unsigned long __init_refok init_memory_mapping(unsigned long start,
        memset(mr, 0, sizeof(mr));
        nr_range = split_mem_range(mr, 0, start, end);
 
-       /*
-        * Find space for the kernel direct mapping tables.
-        *
-        * Later we should allocate these tables in the local node of the
-        * memory mapped. Unfortunately this is done currently before the
-        * nodes are discovered.
-        */
-       if (!after_bootmem)
-               find_early_table_space(start, end);
-
        for (i = 0; i < nr_range; i++)
                ret = kernel_physical_mapping_init(mr[i].start, mr[i].end,
                                                   mr[i].page_size_mask);
@@ -296,6 +286,36 @@ unsigned long __init_refok init_memory_mapping(unsigned long start,
 
        __flush_tlb_all();
 
+       return ret >> PAGE_SHIFT;
+}
+
+void __init init_mem_mapping(void)
+{
+       probe_page_size_mask();
+
+       /*
+        * Find space for the kernel direct mapping tables.
+        *
+        * Later we should allocate these tables in the local node of the
+        * memory mapped. Unfortunately this is done currently before the
+        * nodes are discovered.
+        */
+#ifdef CONFIG_X86_64
+       find_early_table_space(0, max_pfn<<PAGE_SHIFT);
+#else
+       find_early_table_space(0, max_low_pfn<<PAGE_SHIFT);
+#endif
+       max_low_pfn_mapped = init_memory_mapping(0, max_low_pfn<<PAGE_SHIFT);
+       max_pfn_mapped = max_low_pfn_mapped;
+
+#ifdef CONFIG_X86_64
+       if (max_pfn > max_low_pfn) {
+               max_pfn_mapped = init_memory_mapping(1UL<<32,
+                                                    max_pfn<<PAGE_SHIFT);
+               /* can we preseve max_low_pfn ?*/
+               max_low_pfn = max_pfn;
+       }
+#endif
        /*
         * Reserve the kernel pagetable pages we used (pgt_buf_start -
         * pgt_buf_end) and free the other ones (pgt_buf_end - pgt_buf_top)
@@ -311,32 +331,14 @@ unsigned long __init_refok init_memory_mapping(unsigned long start,
         * RO all the pagetable pages, including the ones that are beyond
         * pgt_buf_end at that time.
         */
-       if (!after_bootmem && pgt_buf_end > pgt_buf_start)
+       if (pgt_buf_end > pgt_buf_start)
                x86_init.mapping.pagetable_reserve(PFN_PHYS(pgt_buf_start),
                                PFN_PHYS(pgt_buf_end));
 
-       if (!after_bootmem)
-               early_memtest(start, end);
+       /* stop the wrong using */
+       pgt_buf_top = 0;
 
-       return ret >> PAGE_SHIFT;
-}
-
-void __init init_mem_mapping(void)
-{
-       probe_page_size_mask();
-
-       /* max_pfn_mapped is updated here */
-       max_low_pfn_mapped = init_memory_mapping(0, max_low_pfn<<PAGE_SHIFT);
-       max_pfn_mapped = max_low_pfn_mapped;
-
-#ifdef CONFIG_X86_64
-       if (max_pfn > max_low_pfn) {
-               max_pfn_mapped = init_memory_mapping(1UL<<32,
-                                                    max_pfn<<PAGE_SHIFT);
-               /* can we preseve max_low_pfn ?*/
-               max_low_pfn = max_pfn;
-       }
-#endif
+       early_memtest(0, max_pfn_mapped << PAGE_SHIFT);
 }
 
 /*