VM: add "vm_mmap()" helper function

This continues the theme started with vm_brk() and vm_munmap():
vm_mmap() does the same thing as do_mmap(), but additionally does the
required VM locking.

This uninlines (and rewrites it to be clearer) do_mmap(), which sadly
duplicates it in mm/mmap.c and mm/nommu.c.  But that way we don't have
to export our internal do_mmap_pgoff() function.

Some day we hopefully don't have to export do_mmap() either, if all
modular users can become the simpler vm_mmap() instead.  We're actually
very close to that already, with the notable exception of the (broken)
use in i810, and a couple of stragglers in binfmt_elf.

Signed-off-by: Linus Torvalds <[email protected]>

arch/tile/kernel/single_step.c

@@ -346,12 +346,10 @@ void single_step_once(struct pt_regs *regs)
 		}
 
 		/* allocate a cache line of writable, executable memory */
-		down_write(&current->mm->mmap_sem);
-		buffer = (void __user *) do_mmap(NULL, 0, 64,
+		buffer = (void __user *) vm_mmap(NULL, 0, 64,
 					  PROT_EXEC | PROT_READ | PROT_WRITE,
 					  MAP_PRIVATE | MAP_ANONYMOUS,
 					  0);
-		up_write(&current->mm->mmap_sem);
 
 		if (IS_ERR((void __force *)buffer)) {
 			kfree(state);

arch/x86/ia32/ia32_aout.c

@@ -379,26 +379,22 @@ static int load_aout_binary(struct linux_binprm *bprm, struct pt_regs *regs)
 			goto beyond_if;
 		}
 
-		down_write(&current->mm->mmap_sem);
-		error = do_mmap(bprm->file, N_TXTADDR(ex), ex.a_text,
+		error = vm_mmap(bprm->file, N_TXTADDR(ex), ex.a_text,
 				PROT_READ | PROT_EXEC,
 				MAP_FIXED | MAP_PRIVATE | MAP_DENYWRITE |
 				MAP_EXECUTABLE | MAP_32BIT,
 				fd_offset);
-		up_write(&current->mm->mmap_sem);
 
 		if (error != N_TXTADDR(ex)) {
 			send_sig(SIGKILL, current, 0);
 			return error;
 		}
 
-		down_write(&current->mm->mmap_sem);
-		error = do_mmap(bprm->file, N_DATADDR(ex), ex.a_data,
+		error = vm_mmap(bprm->file, N_DATADDR(ex), ex.a_data,
 				PROT_READ | PROT_WRITE | PROT_EXEC,
 				MAP_FIXED | MAP_PRIVATE | MAP_DENYWRITE |
 				MAP_EXECUTABLE | MAP_32BIT,
 				fd_offset + ex.a_text);
-		up_write(&current->mm->mmap_sem);
 		if (error != N_DATADDR(ex)) {
 			send_sig(SIGKILL, current, 0);
 			return error;
@@ -482,12 +478,10 @@ static int load_aout_library(struct file *file)
 		goto out;
 	}
 	/* Now use mmap to map the library into memory. */
-	down_write(&current->mm->mmap_sem);
-	error = do_mmap(file, start_addr, ex.a_text + ex.a_data,
+	error = vm_mmap(file, start_addr, ex.a_text + ex.a_data,
 			PROT_READ | PROT_WRITE | PROT_EXEC,
 			MAP_FIXED | MAP_PRIVATE | MAP_DENYWRITE | MAP_32BIT,
 			N_TXTOFF(ex));
-	up_write(&current->mm->mmap_sem);
 	retval = error;
 	if (error != start_addr)
 		goto out;

arch/x86/kvm/x86.c

@@ -6336,13 +6336,11 @@ int kvm_arch_prepare_memory_region(struct kvm *kvm,
 		if (npages && !old.rmap) {
 			unsigned long userspace_addr;
 
-			down_write(&current->mm->mmap_sem);
-			userspace_addr = do_mmap(NULL, 0,
+			userspace_addr = vm_mmap(NULL, 0,
 						 npages * PAGE_SIZE,
 						 PROT_READ | PROT_WRITE,
 						 map_flags,
 						 0);
-			up_write(&current->mm->mmap_sem);
 
 			if (IS_ERR((void *)userspace_addr))
 				return PTR_ERR((void *)userspace_addr);

drivers/gpu/drm/drm_bufs.c

@@ -1510,8 +1510,8 @@ int drm_freebufs(struct drm_device *dev, void *data,
  * \param arg pointer to a drm_buf_map structure.
  * \return zero on success or a negative number on failure.
  *
- * Maps the AGP, SG or PCI buffer region with do_mmap(), and copies information
- * about each buffer into user space. For PCI buffers, it calls do_mmap() with
+ * Maps the AGP, SG or PCI buffer region with vm_mmap(), and copies information
+ * about each buffer into user space. For PCI buffers, it calls vm_mmap() with
  * offset equal to 0, which drm_mmap() interpretes as PCI buffers and calls
  * drm_mmap_dma().
  */
@@ -1553,18 +1553,14 @@ int drm_mapbufs(struct drm_device *dev, void *data,
 				retcode = -EINVAL;
 				goto done;
 			}
-			down_write(&current->mm->mmap_sem);
-			virtual = do_mmap(file_priv->filp, 0, map->size,
+			virtual = vm_mmap(file_priv->filp, 0, map->size,
 					  PROT_READ | PROT_WRITE,
 					  MAP_SHARED,
 					  token);
-			up_write(&current->mm->mmap_sem);
 		} else {
-			down_write(&current->mm->mmap_sem);
-			virtual = do_mmap(file_priv->filp, 0, dma->byte_count,
+			virtual = vm_mmap(file_priv->filp, 0, dma->byte_count,
 					  PROT_READ | PROT_WRITE,
 					  MAP_SHARED, 0);
-			up_write(&current->mm->mmap_sem);
 		}
 		if (virtual > -1024UL) {
 			/* Real error */

drivers/gpu/drm/exynos/exynos_drm_gem.c

@@ -581,10 +581,8 @@ int exynos_drm_gem_mmap_ioctl(struct drm_device *dev, void *data,
 	obj->filp->f_op = &exynos_drm_gem_fops;
 	obj->filp->private_data = obj;
 
-	down_write(&current->mm->mmap_sem);
-	addr = do_mmap(obj->filp, 0, args->size,
+	addr = vm_mmap(obj->filp, 0, args->size,
 			PROT_READ | PROT_WRITE, MAP_SHARED, 0);
-	up_write(&current->mm->mmap_sem);
 
 	drm_gem_object_unreference_unlocked(obj);
 

drivers/gpu/drm/i810/i810_dma.c

@@ -129,6 +129,7 @@ static int i810_map_buffer(struct drm_buf *buf, struct drm_file *file_priv)
 	if (buf_priv->currently_mapped == I810_BUF_MAPPED)
 		return -EINVAL;
 
+	/* This is all entirely broken */
 	down_write(&current->mm->mmap_sem);
 	old_fops = file_priv->filp->f_op;
 	file_priv->filp->f_op = &i810_buffer_fops;

drivers/gpu/drm/i915/i915_gem.c

@@ -1087,11 +1087,9 @@ i915_gem_mmap_ioctl(struct drm_device *dev, void *data,
 	if (obj == NULL)
 		return -ENOENT;
 
-	down_write(&current->mm->mmap_sem);
-	addr = do_mmap(obj->filp, 0, args->size,
+	addr = vm_mmap(obj->filp, 0, args->size,
 		       PROT_READ | PROT_WRITE, MAP_SHARED,
 		       args->offset);
-	up_write(&current->mm->mmap_sem);
 	drm_gem_object_unreference_unlocked(obj);
 	if (IS_ERR((void *)addr))
 		return addr;

fs/binfmt_aout.c

@@ -319,24 +319,20 @@ static int load_aout_binary(struct linux_binprm * bprm, struct pt_regs * regs)
 			goto beyond_if;
 		}
 
-		down_write(&current->mm->mmap_sem);
-		error = do_mmap(bprm->file, N_TXTADDR(ex), ex.a_text,
+		error = vm_mmap(bprm->file, N_TXTADDR(ex), ex.a_text,
 			PROT_READ | PROT_EXEC,
 			MAP_FIXED | MAP_PRIVATE | MAP_DENYWRITE | MAP_EXECUTABLE,
 			fd_offset);
-		up_write(&current->mm->mmap_sem);
 
 		if (error != N_TXTADDR(ex)) {
 			send_sig(SIGKILL, current, 0);
 			return error;
 		}
 
-		down_write(&current->mm->mmap_sem);
- 		error = do_mmap(bprm->file, N_DATADDR(ex), ex.a_data,
+		error = vm_mmap(bprm->file, N_DATADDR(ex), ex.a_data,
 				PROT_READ | PROT_WRITE | PROT_EXEC,
 				MAP_FIXED | MAP_PRIVATE | MAP_DENYWRITE | MAP_EXECUTABLE,
 				fd_offset + ex.a_text);
-		up_write(&current->mm->mmap_sem);
 		if (error != N_DATADDR(ex)) {
 			send_sig(SIGKILL, current, 0);
 			return error;
@@ -417,12 +413,10 @@ static int load_aout_library(struct file *file)
 		goto out;
 	}
 	/* Now use mmap to map the library into memory. */
-	down_write(&current->mm->mmap_sem);
-	error = do_mmap(file, start_addr, ex.a_text + ex.a_data,
+	error = vm_mmap(file, start_addr, ex.a_text + ex.a_data,
 			PROT_READ | PROT_WRITE | PROT_EXEC,
 			MAP_FIXED | MAP_PRIVATE | MAP_DENYWRITE,
 			N_TXTOFF(ex));
-	up_write(&current->mm->mmap_sem);
 	retval = error;
 	if (error != start_addr)
 		goto out;

fs/binfmt_elf.c

@@ -958,10 +958,8 @@ static int load_elf_binary(struct linux_binprm *bprm, struct pt_regs *regs)
 		   and some applications "depend" upon this behavior.
 		   Since we do not have the power to recompile these, we
 		   emulate the SVr4 behavior. Sigh. */
-		down_write(&current->mm->mmap_sem);
-		error = do_mmap(NULL, 0, PAGE_SIZE, PROT_READ | PROT_EXEC,
+		error = vm_mmap(NULL, 0, PAGE_SIZE, PROT_READ | PROT_EXEC,
 				MAP_FIXED | MAP_PRIVATE, 0);
-		up_write(&current->mm->mmap_sem);
 	}
 
 #ifdef ELF_PLAT_INIT
@@ -1046,16 +1044,14 @@ static int load_elf_library(struct file *file)
 		eppnt++;
 
 	/* Now use mmap to map the library into memory. */
-	down_write(&current->mm->mmap_sem);
-	error = do_mmap(file,
+	error = vm_mmap(file,
 			ELF_PAGESTART(eppnt->p_vaddr),
 			(eppnt->p_filesz +
 			 ELF_PAGEOFFSET(eppnt->p_vaddr)),
 			PROT_READ | PROT_WRITE | PROT_EXEC,
 			MAP_FIXED | MAP_PRIVATE | MAP_DENYWRITE,
 			(eppnt->p_offset -
 			 ELF_PAGEOFFSET(eppnt->p_vaddr)));
-	up_write(&current->mm->mmap_sem);
 	if (error != ELF_PAGESTART(eppnt->p_vaddr))
 		goto out_free_ph;
 

fs/binfmt_elf_fdpic.c

@@ -390,21 +390,17 @@ static int load_elf_fdpic_binary(struct linux_binprm *bprm,
 	    (executable_stack == EXSTACK_DEFAULT && VM_STACK_FLAGS & VM_EXEC))
 		stack_prot |= PROT_EXEC;
 
-	down_write(&current->mm->mmap_sem);
-	current->mm->start_brk = do_mmap(NULL, 0, stack_size, stack_prot,
+	current->mm->start_brk = vm_mmap(NULL, 0, stack_size, stack_prot,
 					 MAP_PRIVATE | MAP_ANONYMOUS |
 					 MAP_UNINITIALIZED | MAP_GROWSDOWN,
 					 0);
 
 	if (IS_ERR_VALUE(current->mm->start_brk)) {
-		up_write(&current->mm->mmap_sem);
 		retval = current->mm->start_brk;
 		current->mm->start_brk = 0;
 		goto error_kill;
 	}
 
-	up_write(&current->mm->mmap_sem);
-
 	current->mm->brk = current->mm->start_brk;
 	current->mm->context.end_brk = current->mm->start_brk;
 	current->mm->context.end_brk +=
@@ -955,10 +951,8 @@ static int elf_fdpic_map_file_constdisp_on_uclinux(
 	if (params->flags & ELF_FDPIC_FLAG_EXECUTABLE)
 		mflags |= MAP_EXECUTABLE;
 
-	down_write(&mm->mmap_sem);
-	maddr = do_mmap(NULL, load_addr, top - base,
+	maddr = vm_mmap(NULL, load_addr, top - base,
 			PROT_READ | PROT_WRITE | PROT_EXEC, mflags, 0);
-	up_write(&mm->mmap_sem);
 	if (IS_ERR_VALUE(maddr))
 		return (int) maddr;
 
@@ -1096,10 +1090,8 @@ static int elf_fdpic_map_file_by_direct_mmap(struct elf_fdpic_params *params,
 
 		/* create the mapping */
 		disp = phdr->p_vaddr & ~PAGE_MASK;
-		down_write(&mm->mmap_sem);
-		maddr = do_mmap(file, maddr, phdr->p_memsz + disp, prot, flags,
+		maddr = vm_mmap(file, maddr, phdr->p_memsz + disp, prot, flags,
 				phdr->p_offset - disp);
-		up_write(&mm->mmap_sem);
 
 		kdebug("mmap[%d] <file> sz=%lx pr=%x fl=%x of=%lx --> %08lx",
 		       loop, phdr->p_memsz + disp, prot, flags,
@@ -1143,10 +1135,8 @@ static int elf_fdpic_map_file_by_direct_mmap(struct elf_fdpic_params *params,
 			unsigned long xmaddr;
 
 			flags |= MAP_FIXED | MAP_ANONYMOUS;
-			down_write(&mm->mmap_sem);
-			xmaddr = do_mmap(NULL, xaddr, excess - excess1,
+			xmaddr = vm_mmap(NULL, xaddr, excess - excess1,
 					 prot, flags, 0);
-			up_write(&mm->mmap_sem);
 
 			kdebug("mmap[%d] <anon>"
 			       " ad=%lx sz=%lx pr=%x fl=%x of=0 --> %08lx",

fs/binfmt_flat.c

@@ -542,10 +542,8 @@ static int load_flat_file(struct linux_binprm * bprm,
 		 */
 		DBG_FLT("BINFMT_FLAT: ROM mapping of file (we hope)\n");
 
-		down_write(&current->mm->mmap_sem);
-		textpos = do_mmap(bprm->file, 0, text_len, PROT_READ|PROT_EXEC,
+		textpos = vm_mmap(bprm->file, 0, text_len, PROT_READ|PROT_EXEC,
 				  MAP_PRIVATE|MAP_EXECUTABLE, 0);
-		up_write(&current->mm->mmap_sem);
 		if (!textpos || IS_ERR_VALUE(textpos)) {
 			if (!textpos)
 				textpos = (unsigned long) -ENOMEM;
@@ -556,10 +554,8 @@ static int load_flat_file(struct linux_binprm * bprm,
 
 		len = data_len + extra + MAX_SHARED_LIBS * sizeof(unsigned long);
 		len = PAGE_ALIGN(len);
-		down_write(&current->mm->mmap_sem);
-		realdatastart = do_mmap(0, 0, len,
+		realdatastart = vm_mmap(0, 0, len,
 			PROT_READ|PROT_WRITE|PROT_EXEC, MAP_PRIVATE, 0);
-		up_write(&current->mm->mmap_sem);
 
 		if (realdatastart == 0 || IS_ERR_VALUE(realdatastart)) {
 			if (!realdatastart)
@@ -603,10 +599,8 @@ static int load_flat_file(struct linux_binprm * bprm,
 
 		len = text_len + data_len + extra + MAX_SHARED_LIBS * sizeof(unsigned long);
 		len = PAGE_ALIGN(len);
-		down_write(&current->mm->mmap_sem);
-		textpos = do_mmap(0, 0, len,
+		textpos = vm_mmap(0, 0, len,
 			PROT_READ | PROT_EXEC | PROT_WRITE, MAP_PRIVATE, 0);
-		up_write(&current->mm->mmap_sem);
 
 		if (!textpos || IS_ERR_VALUE(textpos)) {
 			if (!textpos)

fs/binfmt_som.c

@@ -147,31 +147,25 @@ static int map_som_binary(struct file *file,
 	code_size = SOM_PAGEALIGN(hpuxhdr->exec_tsize);
 	current->mm->start_code = code_start;
 	current->mm->end_code = code_start + code_size;
-	down_write(&current->mm->mmap_sem);
-	retval = do_mmap(file, code_start, code_size, prot,
+	retval = vm_mmap(file, code_start, code_size, prot,
 			flags, SOM_PAGESTART(hpuxhdr->exec_tfile));
-	up_write(&current->mm->mmap_sem);
 	if (retval < 0 && retval > -1024)
 		goto out;
 
 	data_start = SOM_PAGESTART(hpuxhdr->exec_dmem);
 	data_size = SOM_PAGEALIGN(hpuxhdr->exec_dsize);
 	current->mm->start_data = data_start;
 	current->mm->end_data = bss_start = data_start + data_size;
-	down_write(&current->mm->mmap_sem);
-	retval = do_mmap(file, data_start, data_size,
+	retval = vm_mmap(file, data_start, data_size,
 			prot | PROT_WRITE, flags,
 			SOM_PAGESTART(hpuxhdr->exec_dfile));
-	up_write(&current->mm->mmap_sem);
 	if (retval < 0 && retval > -1024)
 		goto out;
 
 	som_brk = bss_start + SOM_PAGEALIGN(hpuxhdr->exec_bsize);
 	current->mm->start_brk = current->mm->brk = som_brk;
-	down_write(&current->mm->mmap_sem);
-	retval = do_mmap(NULL, bss_start, som_brk - bss_start,
+	retval = vm_mmap(NULL, bss_start, som_brk - bss_start,
 			prot | PROT_WRITE, MAP_FIXED | MAP_PRIVATE, 0);
-	up_write(&current->mm->mmap_sem);
 	if (retval > 0 || retval < -1024)
 		retval = 0;
 out:

include/linux/mm.h

@@ -1393,31 +1393,20 @@ extern int install_special_mapping(struct mm_struct *mm,
 
 extern unsigned long get_unmapped_area(struct file *, unsigned long, unsigned long, unsigned long, unsigned long);
 
-extern unsigned long do_mmap_pgoff(struct file *file, unsigned long addr,
-	unsigned long len, unsigned long prot,
-	unsigned long flag, unsigned long pgoff);
 extern unsigned long mmap_region(struct file *file, unsigned long addr,
 	unsigned long len, unsigned long flags,
 	vm_flags_t vm_flags, unsigned long pgoff);
-
-static inline unsigned long do_mmap(struct file *file, unsigned long addr,
-	unsigned long len, unsigned long prot,
-	unsigned long flag, unsigned long offset)
-{
-	unsigned long ret = -EINVAL;
-	if ((offset + PAGE_ALIGN(len)) < offset)
-		goto out;
-	if (!(offset & ~PAGE_MASK))
-		ret = do_mmap_pgoff(file, addr, len, prot, flag, offset >> PAGE_SHIFT);
-out:
-	return ret;
-}
-
+extern unsigned long do_mmap(struct file *, unsigned long,
+        unsigned long, unsigned long,
+        unsigned long, unsigned long);
 extern int do_munmap(struct mm_struct *, unsigned long, size_t);
 
 /* These take the mm semaphore themselves */
 extern unsigned long vm_brk(unsigned long, unsigned long);
 extern int vm_munmap(struct mm_struct *, unsigned long, size_t);
+extern unsigned long vm_mmap(struct file *, unsigned long,
+        unsigned long, unsigned long,
+        unsigned long, unsigned long);
 
 /* truncate.c */
 extern void truncate_inode_pages(struct address_space *, loff_t);