[PATCH] i386 __verify_write fixes

This patch does a few cleanups/fixes with __verify_write:
 - Only compile it when needed.
 - Move test for KERNEL_DS out of line.
 - The mmap semaphore is needed to access the vma list.
 - Use fixmap for the WP test.
 - Removes an obsolete comment in fixmap.h

arch/i386/kernel/i386_ksyms.c

@@ -72,7 +72,9 @@ EXPORT_SYMBOL(pfn_to_nid);
 #ifdef CONFIG_X86_NUMAQ
 EXPORT_SYMBOL(xquad_portio);
 #endif
+#ifndef CONFIG_X86_WP_WORKS_OK
 EXPORT_SYMBOL(__verify_write);
+#endif
 EXPORT_SYMBOL(dump_thread);
 EXPORT_SYMBOL(dump_fpu);
 EXPORT_SYMBOL(dump_extended_fpu);

arch/i386/mm/fault.c

@@ -30,17 +30,20 @@ extern void die(const char *,struct pt_regs *,long);
 
 extern int console_loglevel;
 
+#ifndef CONFIG_X86_WP_WORKS_OK
 /*
  * Ugly, ugly, but the goto's result in better assembly..
  */
 int __verify_write(const void * addr, unsigned long size)
 {
+	struct mm_struct *mm = current->mm;
 	struct vm_area_struct * vma;
 	unsigned long start = (unsigned long) addr;
 
-	if (!size)
+	if (!size || segment_eq(get_fs(),KERNEL_DS))
 		return 1;
 
+	down_read(&mm->mmap_sem);
 	vma = find_vma(current->mm, start);
 	if (!vma)
 		goto bad_area;
@@ -80,6 +83,13 @@ int __verify_write(const void * addr, unsigned long size)
 		if (!(vma->vm_flags & VM_WRITE))
 			goto bad_area;;
 	}
+	/*
+	 * We really need to hold mmap_sem over the whole access to
+	 * userspace, else another thread could change permissions.
+	 * This is unfixable, so don't use i386-class machines for
+	 * critical servers.
+	 */
+	up_read(&mm->mmap_sem);
 	return 1;
 
 check_stack:
@@ -89,6 +99,7 @@ int __verify_write(const void * addr, unsigned long size)
 		goto good_area;
 
 bad_area:
+	up_read(&mm->mmap_sem);
 	return 0;
 
 out_of_memory:
@@ -98,6 +109,7 @@ int __verify_write(const void * addr, unsigned long size)
 	}
 	goto bad_area;
 }
+#endif
 
 /*
  * Unlock any spinlocks which will prevent us from getting the

arch/i386/mm/init.c

@@ -378,15 +378,10 @@ void __init paging_init(void)
  * This function cannot be __init, since exceptions don't work in that
  * section.
  */
-static int do_test_wp_bit(unsigned long vaddr);
+static int do_test_wp_bit(void);
 
 void __init test_wp_bit(void)
 {
-	const unsigned long vaddr = PAGE_OFFSET;
-	pgd_t *pgd;
-	pmd_t *pmd;
-	pte_t *pte, old_pte;
-
 	if (cpu_has_pse) {
 		/* Ok, all PSE-capable CPUs are definitely handling the WP bit right. */
 		boot_cpu_data.wp_works_ok = 1;
@@ -395,17 +390,10 @@ void __init test_wp_bit(void)
 
 	printk("Checking if this processor honours the WP bit even in supervisor mode... ");
 
-	pgd = swapper_pg_dir + __pgd_offset(vaddr);
-	pmd = pmd_offset(pgd, vaddr);
-	pte = pte_offset_kernel(pmd, vaddr);
-	old_pte = *pte;
-	*pte = pfn_pte(0, PAGE_READONLY);
-	local_flush_tlb();
-
-	boot_cpu_data.wp_works_ok = do_test_wp_bit(vaddr);
-
-	*pte = old_pte;
-	local_flush_tlb();
+	/* Any page-aligned address will do, the test is non-destructive */
+	__set_fixmap(FIX_WP_TEST, __pa(&swapper_pg_dir), PAGE_READONLY);
+	boot_cpu_data.wp_works_ok = do_test_wp_bit();
+	clear_fixmap(FIX_WP_TEST);
 
 	if (!boot_cpu_data.wp_works_ok) {
 		printk("No.\n");
@@ -550,7 +538,7 @@ void __init pgtable_cache_init(void)
 #endif
 
 /* Put this after the callers, so that it cannot be inlined */
-static int do_test_wp_bit(unsigned long vaddr)
+static int do_test_wp_bit(void)
 {
 	char tmp_reg;
 	int flag;
@@ -564,7 +552,7 @@ static int do_test_wp_bit(unsigned long vaddr)
 		"	.align 4	\n"
 		"	.long 1b,2b	\n"
 		".previous		\n"
-		:"=m" (*(char *) vaddr),
+		:"=m" (*(char *)fix_to_virt(FIX_WP_TEST)),
 		 "=q" (tmp_reg),
 		 "=r" (flag)
 		:"2" (1)

arch/i386/mm/ioremap.c

@@ -289,7 +289,7 @@ void __init bt_iounmap(void *addr, unsigned long size)
 
 	idx = FIX_BTMAP_BEGIN;
 	while (nrpages > 0) {
-		__set_fixmap(idx, 0, __pgprot(0));
+		clear_fixmap(idx);
 		--idx;
 		--nrpages;
 	}

include/asm-i386/fixmap.h

@@ -41,13 +41,6 @@
  * TLB entries of such buffers will not be flushed across
  * task switches.
  */
-
-/*
- * on UP currently we will have no trace of the fixmap mechanizm,
- * no page table allocations, etc. This might change in the
- * future, say framebuffers for the console driver(s) could be
- * fix-mapped?
- */
 enum fixed_addresses {
 #ifdef CONFIG_X86_LOCAL_APIC
 	FIX_APIC_BASE,	/* local (CPU) APIC) -- required for SMP or not */
@@ -81,6 +74,7 @@ enum fixed_addresses {
 #define NR_FIX_BTMAPS	16
 	FIX_BTMAP_END = __end_of_permanent_fixed_addresses,
 	FIX_BTMAP_BEGIN = FIX_BTMAP_END + NR_FIX_BTMAPS - 1,
+	FIX_WP_TEST,
 	__end_of_fixed_addresses
 };
 
@@ -94,6 +88,10 @@ extern void __set_fixmap (enum fixed_addresses idx,
  */
 #define set_fixmap_nocache(idx, phys) \
 		__set_fixmap(idx, phys, PAGE_KERNEL_NOCACHE)
+
+#define clear_fixmap(idx) \
+		__set_fixmap(idx, 0, __pgprot(0))
+
 /*
  * used by vmalloc.c.
  *

include/asm-i386/uaccess.h

@@ -64,7 +64,6 @@ int __verify_write(const void *, unsigned long);
 
 #define access_ok(type,addr,size) ( (__range_ok(addr,size) == 0) && \
 			 ((type) == VERIFY_READ || boot_cpu_data.wp_works_ok || \
-			 segment_eq(get_fs(),KERNEL_DS) || \
 			  __verify_write((void *)(addr),(size))))
 
 #endif