s390/vdso: put vdso datapage in a separate vma

Add a separate "[vvar]" mapping for the vdso datapage, since it
doesn't need to be executable or COW-able.

This is actually the s390 implementation of commit 87154938
("arm64: vdso: put vdso datapage in a separate vma")
Reviewed-by: Alexander Gordeev <agordeev@linux.ibm.com>
Signed-off-by: Heiko Carstens <hca@linux.ibm.com>
Signed-off-by: Vasily Gorbik <gor@linux.ibm.com>

arch/s390/kernel/vdso.c

@@ -40,6 +40,14 @@ static int __init vdso_setup(char *str)
 }
 __setup("vdso=", vdso_setup);
 
+static vm_fault_t vvar_fault(const struct vm_special_mapping *sm,
+			     struct vm_area_struct *vma, struct vm_fault *vmf)
+{
+	if (vmf->pgoff == 0)
+		return vmf_insert_pfn(vma, vmf->address, virt_to_pfn(vdso_data));
+	return VM_FAULT_SIGBUS;
+}
+
 static int vdso_mremap(const struct vm_special_mapping *sm,
 		       struct vm_area_struct *vma)
 {
@@ -47,6 +55,11 @@ static int vdso_mremap(const struct vm_special_mapping *sm,
 	return 0;
 }
 
+static struct vm_special_mapping vvar_mapping = {
+	.name = "[vvar]",
+	.fault = vvar_fault,
+};
+
 static struct vm_special_mapping vdso_mapping = {
 	.name = "[vdso]",
 	.mremap = vdso_mremap,
@@ -61,38 +74,41 @@ early_initcall(vdso_getcpu_init); /* Must be called before SMP init */
 
 int arch_setup_additional_pages(struct linux_binprm *bprm, int uses_interp)
 {
+	unsigned long addr, vdso_text_start, vdso_text_len, vdso_mapping_len;
 	struct mm_struct *mm = current->mm;
 	struct vm_area_struct *vma;
-	unsigned long vdso_base;
 	int rc;
 
 	if (!vdso_enabled || is_compat_task())
 		return 0;
 	if (mmap_write_lock_killable(mm))
 		return -EINTR;
-	vdso_base = get_unmapped_area(NULL, 0, vdso_pages << PAGE_SHIFT, 0, 0);
-	rc = vdso_base;
-	if (IS_ERR_VALUE(vdso_base))
+	vdso_text_len = vdso_pages << PAGE_SHIFT;
+	vdso_mapping_len = vdso_text_len + PAGE_SIZE;
+	addr = get_unmapped_area(NULL, 0, vdso_mapping_len, 0, 0);
+	rc = addr;
+	if (IS_ERR_VALUE(addr))
 		goto out;
-	/*
-	 * our vma flags don't have VM_WRITE so by default, the process
-	 * isn't allowed to write those pages.
-	 * gdb can break that with ptrace interface, and thus trigger COW
-	 * on those pages but it's then your responsibility to never do that
-	 * on the "data" page of the vDSO or you'll stop getting kernel
-	 * updates and your nice userland gettimeofday will be totally dead.
-	 * It's fine to use that for setting breakpoints in the vDSO code
-	 * pages though.
-	 */
-	vma = _install_special_mapping(mm, vdso_base, vdso_pages << PAGE_SHIFT,
+	/* VM_MAYWRITE for COW so gdb can set breakpoints */
+	vdso_text_start = addr;
+	vma = _install_special_mapping(mm, addr, vdso_text_len,
 				       VM_READ|VM_EXEC|
 				       VM_MAYREAD|VM_MAYWRITE|VM_MAYEXEC,
 				       &vdso_mapping);
 	rc = PTR_ERR(vma);
 	if (IS_ERR(vma))
 		goto out;
-	current->mm->context.vdso_base = vdso_base;
-	rc = 0;
+	addr += vdso_text_len;
+	vma = _install_special_mapping(mm, addr, PAGE_SIZE,
+				       VM_READ|VM_MAYREAD|VM_PFNMAP,
+				       &vvar_mapping);
+	if (IS_ERR(vma)) {
+		do_munmap(mm, vdso_text_start, vdso_text_len, NULL);
+		rc = PTR_ERR(vma);
+	} else {
+		current->mm->context.vdso_base = vdso_text_start;
+		rc = 0;
+	}
 out:
 	mmap_write_unlock(mm);
 	return rc;
@@ -103,15 +119,14 @@ static int __init vdso_init(void)
 	struct page **pages;
 	int i;
 
-	vdso_pages = ((vdso64_end - vdso64_start) >> PAGE_SHIFT) + 1;
+	vdso_pages = (vdso64_end - vdso64_start) >> PAGE_SHIFT;
 	pages = kcalloc(vdso_pages + 1, sizeof(struct page *), GFP_KERNEL);
 	if (!pages) {
 		vdso_enabled = 0;
 		return -ENOMEM;
 	}
-	for (i = 0; i < vdso_pages - 1; i++)
+	for (i = 0; i < vdso_pages; i++)
 		pages[i] = virt_to_page(vdso64_start + i * PAGE_SIZE);
-	pages[vdso_pages - 1] = virt_to_page(vdso_data);
 	pages[vdso_pages] = NULL;
 	vdso_mapping.pages = pages;
 	return 0;