[Patch resend v3 2/3] binfmt_misc: add persistent opened binary handler for containers

[James Bottomley]

>From a119fa7e0d4eb96a6290eb63a5bd597f5844fab4 Mon Sep 17 00:00:00 2001
From: James Bottomley <James.Bottomley at HansenPartnership.com>
Date: Wed, 17 Feb 2016 16:51:16 -0800
Subject: [PATCH 2/3] binfmt_misc: add persistent opened binary handler for
 containers

This patch adds a new flag 'F' to the binfmt handlers.  If you pass in
'F' the binary that runs the emulation will be opened immediately and
in future, will be cloned from the open file.

The net effect is that the handler survives both changeroots and mount
namespace changes, making it easy to work with foreign architecture
containers without contaminating the container image with the
emulator.

Signed-off-by: James Bottomley <James.Bottomley at HansenPartnership.com>
Acked-by: Serge Hallyn <serge.hallyn at canonical.com>
---
 fs/binfmt_misc.c | 41 +++++++++++++++++++++++++++++++++++++++--
 1 file changed, 39 insertions(+), 2 deletions(-)

diff --git a/fs/binfmt_misc.c b/fs/binfmt_misc.c
index 3a3ced7..8a108c4 100644
--- a/fs/binfmt_misc.c
+++ b/fs/binfmt_misc.c
@@ -26,6 +26,8 @@
 #include <linux/fs.h>
 #include <linux/uaccess.h>
 
+#include "internal.h"
+
 #ifdef DEBUG
 # define USE_DEBUG 1
 #else
@@ -43,6 +45,7 @@ enum {Enabled, Magic};
 #define MISC_FMT_PRESERVE_ARGV0 (1 << 31)
 #define MISC_FMT_OPEN_BINARY (1 << 30)
 #define MISC_FMT_CREDENTIALS (1 << 29)
+#define MISC_FMT_OPEN_FILE (1 << 28)
 
 typedef struct {
 	struct list_head list;
@@ -54,6 +57,7 @@ typedef struct {
 	char *interpreter;		/* filename of interpreter */
 	char *name;
 	struct dentry *dentry;
+	struct file *interp_file;
 } Node;
 
 static DEFINE_RWLOCK(entries_lock);
@@ -201,7 +205,13 @@ static int load_misc_binary(struct linux_binprm *bprm)
 	if (retval < 0)
 		goto error;
 
-	interp_file = open_exec(iname);
+	if (fmt->flags & MISC_FMT_OPEN_FILE && fmt->interp_file) {
+		interp_file = filp_clone_open(fmt->interp_file);
+		if (!IS_ERR(interp_file))
+			deny_write_access(interp_file);
+	} else {
+		interp_file = open_exec(iname);
+	}
 	retval = PTR_ERR(interp_file);
 	if (IS_ERR(interp_file))
 		goto error;
@@ -285,6 +295,11 @@ static char *check_special_flags(char *sfs, Node *e)
 			e->flags |= (MISC_FMT_CREDENTIALS |
 					MISC_FMT_OPEN_BINARY);
 			break;
+		case 'F':
+			pr_debug("register: flag: F: open interpreter file now\n");
+			p++;
+			e->flags |= MISC_FMT_OPEN_FILE;
+			break;
 		default:
 			cont = 0;
 		}
@@ -543,6 +558,8 @@ static void entry_status(Node *e, char *page)
 		*dp++ = 'O';
 	if (e->flags & MISC_FMT_CREDENTIALS)
 		*dp++ = 'C';
+	if (e->flags & MISC_FMT_OPEN_FILE)
+		*dp++ = 'F';
 	*dp++ = '\n';
 
 	if (!test_bit(Magic, &e->flags)) {
@@ -590,6 +607,11 @@ static void kill_node(Node *e)
 	}
 	write_unlock(&entries_lock);
 
+	if ((e->flags & MISC_FMT_OPEN_FILE) && e->interp_file) {
+		filp_close(e->interp_file, NULL);
+		e->interp_file = NULL;
+	}
+
 	if (dentry) {
 		drop_nlink(d_inode(dentry));
 		d_drop(dentry);
@@ -698,6 +720,21 @@ static ssize_t bm_register_write(struct file *file, const char __user *buffer,
 		goto out2;
 	}
 
+	if (e->flags & MISC_FMT_OPEN_FILE) {
+		struct file *f;
+
+		f = open_exec(e->interpreter);
+		if (IS_ERR(f)) {
+			err = PTR_ERR(f);
+			pr_notice("register: failed to install interpreter file %s\n", e->interpreter);
+			simple_release_fs(&bm_mnt, &entry_count);
+			iput(inode);
+			inode = NULL;
+			goto out2;
+		}
+		e->interp_file = f;
+	}
+
 	e->dentry = dget(dentry);
 	inode->i_private = e;
 	inode->i_fop = &bm_entry_operations;
@@ -716,7 +753,7 @@ out:
 
 	if (err) {
 		kfree(e);
-		return -EINVAL;
+		return err;
 	}
 	return count;
 }
-- 
2.6.6