Resent-Date: Mon, 2 Aug 1999 10:30:54 +0200 (MET DST) From: zippel@fh-brandenburg.de (Roman Zippel) Subject: Re: mm fix To: zippel@fh-brandenburg.de (Roman Zippel) Date: Mon, 2 Aug 1999 10:30:51 +0200 (MET DST) Cc: linux-m68k@lists.linux-m68k.org In-Reply-To: from "Roman Zippel" at Jul 28, 99 06:53:40 pm Resent-From: linux-m68k@phil.uni-sb.de Hi, > Anyway, I wonder that this bug wasn't triggered before, for the 020/030 we > forgot to protect the pointer table, that we used to unmap the zero page. > The patch is needed for 2.2 and 2.3. Argh, when fixing this, I shouldn't break other cpus of course... bye, Roman --- arch/m68k/mm/init.c 1999/05/12 23:57:02 1.1.1.2 +++ arch/m68k/mm/init.c 1999/08/01 22:55:14 @@ -139,6 +139,7 @@ } static pmd_t *last_pgtable __initdata = NULL; +static pmd_t *zero_pgtable __initdata = NULL; __initfunc(static pmd_t * kernel_ptr_table(unsigned long *memavailp)) { @@ -236,7 +237,8 @@ #ifdef DEBUG printk ("[zero map]"); #endif - pte_dir = (pte_t *)kernel_ptr_table(memavailp); + zero_pgtable = kernel_ptr_table(memavailp); + pte_dir = (pte_t *)zero_pgtable; pmd_dir->pmd[0] = virt_to_phys(pte_dir) | _PAGE_TABLE | _PAGE_ACCESSED; pte_val(*pte_dir++) = 0; @@ -450,6 +452,9 @@ if (pgd_present(kernel_pg_dir[i])) init_pointer_table(pgd_page(kernel_pg_dir[i])); } + /* insert also pointer table that we used to unmap the zero page */ + if (zero_pgtable) + init_pointer_table((unsigned long)zero_pgtable); printk("Memory: %luk/%luk available (%dk kernel code, %dk data, %dk init)\n", (unsigned long) nr_free_pages << (PAGE_SHIFT-10),