Skip to contentSkip to conversation

Revert "kmod: fix race in usermodehelper code"

PATCH In review Sebastian Andrzej Siewior 3 participants 3 messages Sep 23, 2009 – Sep 24, 2009

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 Revert "kmod: fix race in usermodehelper code"

This reverts commit c02e3f361:

|Author: Neil Horman <[email protected]>
|Date:   Tue Sep 22 16:43:36 2009 -0700
|
|    kmod: fix race in usermodehelper code
|
|    The user mode helper code has a race in it.  call_usermodehelper_exec()
|    takes an allocated subprocess_info structure, which it passes to a
|    workqueue, and then passes it to a kernel thread which it creates, after
|    which it calls complete to signal to the caller of
|    call_usermodehelper_exec() that it can free the subprocess_info struct.
|
|    But since we use that structure in the created thread, we can't call
|    complete from __call_usermodehelper(), which is where we create the kernel
|    thread.  We need to call complete() from within the kernel thread and then
|    not use subprocess_info afterward in the case of UMH_WAIT_EXEC.  Tested
|    successfully by me.
|
|    Signed-off-by: Neil Horman <[email protected]>
|    Cc: Rusty Russell <[email protected]>
|    Signed-off-by: Andrew Morton <[email protected]>
|    Signed-off-by: Linus Torvalds <[email protected]>

The pach is wrong IMHO. UMH_WAIT_EXEC is called with VFORK what ensures that the child finishes prior returing back to the parent. No race. The patch makes it even worse because it does the thing it claims not do:
- It calls ->complete() on UMH_WAIT_EXEC
- the complete() callback may de-allocated subinfo
As seen in the following call chain:

|[<c009f904>] (__link_path_walk+0x20/0xeb4) from [<c00a094c>] (path_walk+0x48/0x94)
|[<c00a094c>] (path_walk+0x48/0x94) from [<c00a0a34>] (do_path_lookup+0x24/0x4c)
|[<c00a0a34>] (do_path_lookup+0x24/0x4c) from [<c00a158c>] (do_filp_open+0xa4/0x83c)
|[<c00a158c>] (do_filp_open+0xa4/0x83c) from [<c009ba90>] (open_exec+0x24/0xe0)
|[<c009ba90>] (open_exec+0x24/0xe0) from [<c009bfa8>] (do_execve+0x7c/0x2e4)
|[<c009bfa8>] (do_execve+0x7c/0x2e4) from [<c0026a80>] (kernel_execve+0x34/0x80)
|[<c0026a80>] (kernel_execve+0x34/0x80) from [<c004b514>] (____call_usermodehelper+0x130/0x148)
|[<c004b514>] (____call_usermodehelper+0x130/0x148) from [<c0024858>] (kernel_thread_exit+0x0/0x8)

And the path pointer was NULL. Good that ARM's kernel_execve() doesn't check the pointer for NULL or else I wouldn't notice it.

The only race there might be is with UMH_NO_WAIT but it is too late for me to investigate it now. UMH_WAIT_PROC could probably also use VFORK and we could save one exec. So the only race I see is with UMH_NO_WAIT and recent scheduler changes where the child does not always run first might have trigger here something but as I said, it is late....

---
 kernel/kmod.c |   13 +++++--------
 1 files changed, 5 insertions(+), 8 deletions(-)
kernel/kmod.c +5 -8
index 689d20f..9fcb53a 100644
@@ -143,7 +143,6 @@ struct subprocess_info {
 static int ____call_usermodehelper(void *data)
 {
 	struct subprocess_info *sub_info = data;
-	enum umh_wait wait = sub_info->wait;
 	int retval;
 
 	BUG_ON(atomic_read(&sub_info->cred->usage) != 1);
@@ -185,14 +184,10 @@ static int ____call_usermodehelper(void *data)
 	 */
 	set_user_nice(current, 0);
 
-	if (wait == UMH_WAIT_EXEC)
-		complete(sub_info->complete);
-
 	retval = kernel_execve(sub_info->path, sub_info->argv, sub_info->envp);
 
 	/* Exec failed? */
-	if (wait != UMH_WAIT_EXEC)
-		sub_info->retval = retval;
+	sub_info->retval = retval;
 	do_exit(0);
 }
 
@@ -271,14 +266,16 @@ static void __call_usermodehelper(struct work_struct *work)
 
 	switch (wait) {
 	case UMH_NO_WAIT:
-	case UMH_WAIT_EXEC:
 		break;
 
 	case UMH_WAIT_PROC:
 		if (pid > 0)
 			break;
 		sub_info->retval = pid;
-		break;
+		/* FALLTHROUGH */
+
+	case UMH_WAIT_EXEC:
+		complete(sub_info->complete);
 	}
 }
 
Signature · 1 line
1.6.4.GIT
#1 lore raw
Quoted 6 lines · Sebastian Andrzej Siewior

The pach is wrong IMHO. UMH_WAIT_EXEC is called with VFORK what ensures that the child finishes prior returing back to the parent. No race. The patch makes it even worse because it does the thing it claims not do:
- It calls ->complete() on UMH_WAIT_EXEC
- the complete() callback may de-allocated subinfo

Yeah, I think you're right. That commit is bogus. Thanks for noticing.

Linus

#2 lore raw
In reply to Sebastian Andrzej Siewior
Quoted 23 lines · Sebastian Andrzej Siewior

This reverts commit c02e3f361:

|Author: Neil Horman <[email protected]>
|Date:   Tue Sep 22 16:43:36 2009 -0700
|
|    kmod: fix race in usermodehelper code
|
|    The user mode helper code has a race in it.  call_usermodehelper_exec()
|    takes an allocated subprocess_info structure, which it passes to a
|    workqueue, and then passes it to a kernel thread which it creates, after
|    which it calls complete to signal to the caller of
|    call_usermodehelper_exec() that it can free the subprocess_info struct.
|
|    But since we use that structure in the created thread, we can't call
|    complete from __call_usermodehelper(), which is where we create the kernel
|    thread.  We need to call complete() from within the kernel thread and then
|    not use subprocess_info afterward in the case of UMH_WAIT_EXEC.  Tested
|    successfully by me.
|
|    Signed-off-by: Neil Horman <[email protected]>
|    Cc: Rusty Russell <[email protected]>
|    Signed-off-by: Andrew Morton <[email protected]>
|    Signed-off-by: Linus Torvalds <[email protected]>
Quoted 3 lines

The pach is wrong IMHO. UMH_WAIT_EXEC is called with VFORK what ensures that the child finishes prior returing back to the parent. No race. The patch makes it even worse because it does the thing it claims not do:

Dang, you're right. I completely missed the use of CLONE_VFORK as a mechanism to prevent the parent from racing ahead of the child. Sorry for the noise. Neil

Quoted 1 line