procfs task exe symlink

The kernel implements readlink of /proc/pid/exe by getting the file from
the first executable VMA.  Then the path to the file is reconstructed and
reported as the result.

Because of the VMA walk the code is slightly different on nommu systems.
This patch avoids separate /proc/pid/exe code on nommu systems.  Instead of
walking the VMAs to find the first executable file-backed VMA we store a
reference to the exec'd file in the mm_struct.

That reference would prevent the filesystem holding the executable file
from being unmounted even after unmapping the VMAs.  So we track the number
of VM_EXECUTABLE VMAs and drop the new reference when the last one is
unmapped.  This avoids pinning the mounted filesystem.

[akpm@linux-foundation.org: improve comments]
[yamamoto@valinux.co.jp: fix dup_mmap]
Signed-off-by: Matt Helsley <matthltc@us.ibm.com>
Cc: Oleg Nesterov <oleg@tv-sign.ru>
Cc: David Howells <dhowells@redhat.com>
Cc:"Eric W. Biederman" <ebiederm@xmission.com>
Cc: Christoph Hellwig <hch@lst.de>
Cc: Al Viro <viro@zeniv.linux.org.uk>
Cc: Hugh Dickins <hugh@veritas.com>
Signed-off-by: YAMAMOTO Takashi <yamamoto@valinux.co.jp>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>

12 files changed, 157 insertions, 81 deletions

diff --git a/fs/binfmt_flat.c b/fs/binfmt_flat.c
index c12cc36..3b40d45 100644
--- a/fs/binfmt_flat.c
+++ b/fs/binfmt_flat.c
@@ -531,7 +531,8 @@ 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, MAP_PRIVATE, 0);
+		textpos = do_mmap(bprm->file, 0, text_len, PROT_READ|PROT_EXEC,
+				  MAP_PRIVATE|MAP_EXECUTABLE, 0);
 		up_write(&current->mm->mmap_sem);
 		if (!textpos  || textpos >= (unsigned long) -4096) {
 			if (!textpos)
diff --git a/fs/exec.c b/fs/exec.c
index 711bc45..a138839 100644
--- a/fs/exec.c
+++ b/fs/exec.c
@@ -964,6 +964,8 @@ int flush_old_exec(struct linux_binprm * bprm)
 	if (retval)
 		goto out;
 
+	set_mm_exe_file(bprm->mm, bprm->file);
+
 	/*
 	 * Release all of the old mmap stuff
 	 */
diff --git a/fs/proc/base.c b/fs/proc/base.c
index c5e412a..b48ddb1 100644
--- a/fs/proc/base.c
+++ b/fs/proc/base.c
@@ -1181,6 +1181,81 @@ static const struct file_operations proc_pid_sched_operations = {
 
 #endif
 
+/*
+ * We added or removed a vma mapping the executable. The vmas are only mapped
+ * during exec and are not mapped with the mmap system call.
+ * Callers must hold down_write() on the mm's mmap_sem for these
+ */
+void added_exe_file_vma(struct mm_struct *mm)
+{
+	mm->num_exe_file_vmas++;
+}
+
+void removed_exe_file_vma(struct mm_struct *mm)
+{
+	mm->num_exe_file_vmas--;
+	if ((mm->num_exe_file_vmas == 0) && mm->exe_file){
+		fput(mm->exe_file);
+		mm->exe_file = NULL;
+	}
+
+}
+
+void set_mm_exe_file(struct mm_struct *mm, struct file *new_exe_file)
+{
+	if (new_exe_file)
+		get_file(new_exe_file);
+	if (mm->exe_file)
+		fput(mm->exe_file);
+	mm->exe_file = new_exe_file;
+	mm->num_exe_file_vmas = 0;
+}
+
+struct file *get_mm_exe_file(struct mm_struct *mm)
+{
+	struct file *exe_file;
+
+	/* We need mmap_sem to protect against races with removal of
+	 * VM_EXECUTABLE vmas */
+	down_read(&mm->mmap_sem);
+	exe_file = mm->exe_file;
+	if (exe_file)
+		get_file(exe_file);
+	up_read(&mm->mmap_sem);
+	return exe_file;
+}
+
+void dup_mm_exe_file(struct mm_struct *oldmm, struct mm_struct *newmm)
+{
+	/* It's safe to write the exe_file pointer without exe_file_lock because
+	 * this is called during fork when the task is not yet in /proc */
+	newmm->exe_file = get_mm_exe_file(oldmm);
+}
+
+static int proc_exe_link(struct inode *inode, struct path *exe_path)
+{
+	struct task_struct *task;
+	struct mm_struct *mm;
+	struct file *exe_file;
+
+	task = get_proc_task(inode);
+	if (!task)
+		return -ENOENT;
+	mm = get_task_mm(task);
+	put_task_struct(task);
+	if (!mm)
+		return -ENOENT;
+	exe_file = get_mm_exe_file(mm);
+	mmput(mm);
+	if (exe_file) {
+		*exe_path = exe_file->f_path;
+		path_get(&exe_file->f_path);
+		fput(exe_file);
+		return 0;
+	} else
+		return -ENOENT;
+}
+
 static void *proc_pid_follow_link(struct dentry *dentry, struct nameidata *nd)
 {
 	struct inode *inode = dentry->d_inode;
diff --git a/fs/proc/internal.h b/fs/proc/internal.h
index bc72f5c..45abb98 100644
--- a/fs/proc/internal.h
+++ b/fs/proc/internal.h
@@ -48,7 +48,6 @@ extern int maps_protect;
 
 extern void create_seq_entry(char *name, mode_t mode,
 				const struct file_operations *f);
-extern int proc_exe_link(struct inode *, struct path *);
 extern int proc_tid_stat(struct seq_file *m, struct pid_namespace *ns,
 				struct pid *pid, struct task_struct *task);
 extern int proc_tgid_stat(struct seq_file *m, struct pid_namespace *ns,
diff --git a/fs/proc/task_mmu.c b/fs/proc/task_mmu.c
index 7415eeb..e2b8e76 100644
--- a/fs/proc/task_mmu.c
+++ b/fs/proc/task_mmu.c
@@ -75,40 +75,6 @@ int task_statm(struct mm_struct *mm, int *shared, int *text,
 	return mm->total_vm;
 }
 
-int proc_exe_link(struct inode *inode, struct path *path)
-{
-	struct vm_area_struct * vma;
-	int result = -ENOENT;
-	struct task_struct *task = get_proc_task(inode);
-	struct mm_struct * mm = NULL;
-
-	if (task) {
-		mm = get_task_mm(task);
-		put_task_struct(task);
-	}
-	if (!mm)
-		goto out;
-	down_read(&mm->mmap_sem);
-
-	vma = mm->mmap;
-	while (vma) {
-		if ((vma->vm_flags & VM_EXECUTABLE) && vma->vm_file)
-			break;
-		vma = vma->vm_next;
-	}
-
-	if (vma) {
-		*path = vma->vm_file->f_path;
-		path_get(&vma->vm_file->f_path);
-		result = 0;
-	}
-
-	up_read(&mm->mmap_sem);
-	mmput(mm);
-out:
-	return result;
-}
-
 static void pad_len_spaces(struct seq_file *m, int len)
 {
 	len = 25 + sizeof(void*) * 6 - len;
diff --git a/fs/proc/task_nommu.c b/fs/proc/task_nommu.c
index 8011528..4b733f1 100644
--- a/fs/proc/task_nommu.c
+++ b/fs/proc/task_nommu.c
@@ -103,40 +103,6 @@ int task_statm(struct mm_struct *mm, int *shared, int *text,
 	return size;
 }
 
-int proc_exe_link(struct inode *inode, struct path *path)
-{
-	struct vm_list_struct *vml;
-	struct vm_area_struct *vma;
-	struct task_struct *task = get_proc_task(inode);
-	struct mm_struct *mm = get_task_mm(task);
-	int result = -ENOENT;
-
-	if (!mm)
-		goto out;
-	down_read(&mm->mmap_sem);
-
-	vml = mm->context.vmlist;
-	vma = NULL;
-	while (vml) {
-		if ((vml->vma->vm_flags & VM_EXECUTABLE) && vml->vma->vm_file) {
-			vma = vml->vma;
-			break;
-		}
-		vml = vml->next;
-	}
-
-	if (vma) {
-		*path = vma->vm_file->f_path;
-		path_get(&vma->vm_file->f_path);
-		result = 0;
-	}
-
-	up_read(&mm->mmap_sem);
-	mmput(mm);
-out:
-	return result;
-}
-
 /*
  * display mapping lines for a particular process's /proc/pid/maps
  */
diff --git a/include/linux/mm.h b/include/linux/mm.h
index fef602d..c31a9cd 100644
--- a/include/linux/mm.h
+++ b/include/linux/mm.h
@@ -1066,6 +1066,19 @@ extern void unlink_file_vma(struct vm_area_struct *);
 extern struct vm_area_struct *copy_vma(struct vm_area_struct **,
 	unsigned long addr, unsigned long len, pgoff_t pgoff);
 extern void exit_mmap(struct mm_struct *);
+
+#ifdef CONFIG_PROC_FS
+/* From fs/proc/base.c. callers must _not_ hold the mm's exe_file_lock */
+extern void added_exe_file_vma(struct mm_struct *mm);
+extern void removed_exe_file_vma(struct mm_struct *mm);
+#else
+static inline void added_exe_file_vma(struct mm_struct *mm)
+{}
+
+static inline void removed_exe_file_vma(struct mm_struct *mm)
+{}
+#endif /* CONFIG_PROC_FS */
+
 extern int may_expand_vm(struct mm_struct *mm, unsigned long npages);
 extern int install_special_mapping(struct mm_struct *mm,
 				   unsigned long addr, unsigned long len,
diff --git a/include/linux/mm_types.h b/include/linux/mm_types.h
index bc97bd5..eb7c16c 100644
--- a/include/linux/mm_types.h
+++ b/include/linux/mm_types.h
@@ -229,6 +229,12 @@ struct mm_struct {
 	struct task_struct *owner;	/* The thread group leader that */
 					/* owns the mm_struct.		*/
 #endif
+
+#ifdef CONFIG_PROC_FS
+	/* store ref to file /proc/<pid>/exe symlink points to */
+	struct file *exe_file;
+	unsigned long num_exe_file_vmas;
+#endif
 };
 
 #endif /* _LINUX_MM_TYPES_H */
diff --git a/include/linux/proc_fs.h b/include/linux/proc_fs.h
index 9b6c935..65f2299 100644
--- a/include/linux/proc_fs.h
+++ b/include/linux/proc_fs.h
@@ -9,7 +9,6 @@
 
 struct net;
 struct completion;
-
 /*
  * The proc filesystem constants/structures
  */
@@ -206,6 +205,12 @@ extern void proc_net_remove(struct net *net, const char *name);
 extern struct proc_dir_entry *proc_net_mkdir(struct net *net, const char *name,
 	struct proc_dir_entry *parent);
 
+/* While the {get|set|dup}_mm_exe_file functions are for mm_structs, they are
+ * only needed to implement /proc/<pid>|self/exe so we define them here. */
+extern void set_mm_exe_file(struct mm_struct *mm, struct file *new_exe_file);
+extern struct file *get_mm_exe_file(struct mm_struct *mm);
+extern void dup_mm_exe_file(struct mm_struct *oldmm, struct mm_struct *newmm);
+
 #else
 
 #define proc_root_driver NULL
@@ -255,6 +260,19 @@ static inline void pid_ns_release_proc(struct pid_namespace *ns)
 {
 }
 
+static inline void set_mm_exe_file(struct mm_struct *mm,
+				   struct file *new_exe_file)
+{}
+
+static inline struct file *get_mm_exe_file(struct mm_struct *mm)
+{
+	return NULL;
+}
+
+static inline void dup_mm_exe_file(struct mm_struct *oldmm,
+	       			   struct mm_struct *newmm)
+{}
+
 #endif /* CONFIG_PROC_FS */
 
 #if !defined(CONFIG_PROC_KCORE)
diff --git a/kernel/fork.c b/kernel/fork.c
index de5c16c..068ffe0 100644
--- a/kernel/fork.c
+++ b/kernel/fork.c
@@ -431,6 +431,7 @@ void mmput(struct mm_struct *mm)
 	if (atomic_dec_and_test(&mm->mm_users)) {
 		exit_aio(mm);
 		exit_mmap(mm);
+		set_mm_exe_file(mm, NULL);
 		if (!list_empty(&mm->mmlist)) {
 			spin_lock(&mmlist_lock);
 			list_del(&mm->mmlist);
@@ -543,6 +544,8 @@ struct mm_struct *dup_mm(struct task_struct *tsk)
 	if (init_new_context(tsk, mm))
 		goto fail_nocontext;
 
+	dup_mm_exe_file(oldmm, mm);
+
 	err = dup_mmap(mm, oldmm);
 	if (err)
 		goto free_pt;
diff --git a/mm/mmap.c b/mm/mmap.c
index 677d184..fac6633 100644
--- a/mm/mmap.c
+++ b/mm/mmap.c
@@ -230,8 +230,11 @@ static struct vm_area_struct *remove_vma(struct vm_area_struct *vma)
 	might_sleep();
 	if (vma->vm_ops && vma->vm_ops->close)
 		vma->vm_ops->close(vma);
-	if (vma->vm_file)
+	if (vma->vm_file) {
 		fput(vma->vm_file);
+		if (vma->vm_flags & VM_EXECUTABLE)
+			removed_exe_file_vma(vma->vm_mm);
+	}
 	mpol_put(vma_policy(vma));
 	kmem_cache_free(vm_area_cachep, vma);
 	return next;
@@ -623,8 +626,11 @@ again:			remove_next = 1 + (end > next->vm_end);
 		spin_unlock(&mapping->i_mmap_lock);
 
 	if (remove_next) {
-		if (file)
+		if (file) {
 			fput(file);
+			if (next->vm_flags & VM_EXECUTABLE)
+				removed_exe_file_vma(mm);
+		}
 		mm->map_count--;
 		mpol_put(vma_policy(next));
 		kmem_cache_free(vm_area_cachep, next);
@@ -1154,6 +1160,8 @@ munmap_back:
 		error = file->f_op->mmap(file, vma);
 		if (error)
 			goto unmap_and_free_vma;
+		if (vm_flags & VM_EXECUTABLE)
+			added_exe_file_vma(mm);
 	} else if (vm_flags & VM_SHARED) {
 		error = shmem_zero_setup(vma);
 		if (error)
@@ -1185,6 +1193,8 @@ munmap_back:
 		mpol_put(vma_policy(vma));
 		kmem_cache_free(vm_area_cachep, vma);
 		fput(file);
+		if (vm_flags & VM_EXECUTABLE)
+			removed_exe_file_vma(mm);
 	} else {
 		vma_link(mm, vma, prev, rb_link, rb_parent);
 		file = vma->vm_file;
@@ -1817,8 +1827,11 @@ int split_vma(struct mm_struct * mm, struct vm_area_struct * vma,
 	}
 	vma_set_policy(new, pol);
 
-	if (new->vm_file)
+	if (new->vm_file) {
 		get_file(new->vm_file);
+		if (vma->vm_flags & VM_EXECUTABLE)
+			added_exe_file_vma(mm);
+	}
 
 	if (new->vm_ops && new->vm_ops->open)
 		new->vm_ops->open(new);
@@ -2135,8 +2148,11 @@ struct vm_area_struct *copy_vma(struct vm_area_struct **vmap,
 			new_vma->vm_start = addr;
 			new_vma->vm_end = addr + len;
 			new_vma->vm_pgoff = pgoff;
-			if (new_vma->vm_file)
+			if (new_vma->vm_file) {
 				get_file(new_vma->vm_file);
+				if (vma->vm_flags & VM_EXECUTABLE)
+					added_exe_file_vma(mm);
+			}
 			if (new_vma->vm_ops && new_vma->vm_ops->open)
 				new_vma->vm_ops->open(new_vma);
 			vma_link(mm, new_vma, prev, rb_link, rb_parent);
diff --git a/mm/nommu.c b/mm/nommu.c
index 1d32fe8..ef8c62c 100644
--- a/mm/nommu.c
+++ b/mm/nommu.c
@@ -966,8 +966,13 @@ unsigned long do_mmap_pgoff(struct file *file,
 
 	INIT_LIST_HEAD(&vma->anon_vma_node);
 	atomic_set(&vma->vm_usage, 1);
-	if (file)
+	if (file) {
 		get_file(file);
+		if (vm_flags & VM_EXECUTABLE) {
+			added_exe_file_vma(current->mm);
+			vma->vm_mm = current->mm;
+		}
+	}
 	vma->vm_file	= file;
 	vma->vm_flags	= vm_flags;
 	vma->vm_start	= addr;
@@ -1022,8 +1027,11 @@ unsigned long do_mmap_pgoff(struct file *file,
 	up_write(&nommu_vma_sem);
 	kfree(vml);
 	if (vma) {
-		if (vma->vm_file)
+		if (vma->vm_file) {
 			fput(vma->vm_file);
+			if (vma->vm_flags & VM_EXECUTABLE)
+				removed_exe_file_vma(vma->vm_mm);
+		}
 		kfree(vma);
 	}
 	return ret;
@@ -1053,7 +1061,7 @@ EXPORT_SYMBOL(do_mmap_pgoff);
 /*
  * handle mapping disposal for uClinux
  */
-static void put_vma(struct vm_area_struct *vma)
+static void put_vma(struct mm_struct *mm, struct vm_area_struct *vma)
 {
 	if (vma) {
 		down_write(&nommu_vma_sem);
@@ -1075,8 +1083,11 @@ static void put_vma(struct vm_area_struct *vma)
 			realalloc -= kobjsize(vma);
 			askedalloc -= sizeof(*vma);
 
-			if (vma->vm_file)
+			if (vma->vm_file) {
 				fput(vma->vm_file);
+				if (vma->vm_flags & VM_EXECUTABLE)
+					removed_exe_file_vma(mm);
+			}
 			kfree(vma);
 		}
 
@@ -1113,7 +1124,7 @@ int do_munmap(struct mm_struct *mm, unsigned long addr, size_t len)
  found:
 	vml = *parent;
 
-	put_vma(vml->vma);
+	put_vma(mm, vml->vma);
 
 	*parent = vml->next;
 	realalloc -= kobjsize(vml);
@@ -1158,7 +1169,7 @@ void exit_mmap(struct mm_struct * mm)
 
 		while ((tmp = mm->context.vmlist)) {
 			mm->context.vmlist = tmp->next;
-			put_vma(tmp->vma);
+			put_vma(mm, tmp->vma);
 
 			realalloc -= kobjsize(tmp);
 			askedalloc -= sizeof(*tmp);