Skip to contentSkip to conversation

Avoid buffer overflows in get_user_pages()

PATCH In review Jonathan Corbet 5 participants 6 messages Feb 11, 2008 – Feb 14, 2008

Subsystems (2)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 Avoid buffer overflows in get_user_pages()

Avoid buffer overflows in get_user_pages()

So I spent a while pounding my head against my monitor trying to figure out the vmsplice() vulnerability - how could a failure to check for *read* access turn into a root exploit? It turns out that it's a buffer overflow problem which is made easy by the way get_user_pages() is coded.

In particular, "len" is a signed int, and it is only checked at the *end* of a do {} while() loop. So, if it is passed in as zero, the loop will execute once and decrement len to -1. At that point, the loop will proceed until the next invalid address is found; in the process, it will likely overflow the pages array passed in to get_user_pages().

I think that, if get_user_pages() has been asked to grab zero pages, that's what it should do. Thus this patch; it is, among other things, enough to block the (already fixed) root exploit and any others which might be lurking in similar code. I also think that the number of pages should be unsigned, but changing the prototype of this function probably requires some more careful review.

mm/memory.c +2 -0
index e5628a5..7f50fd8 100644
@@ -989,6 +989,8 @@ int get_user_pages(struct task_struct *tsk, struct mm_struct *mm,
 	int i;
 	unsigned int vm_flags;
 
+	if (len <= 0)
+		return 0;
 	/* 
 	 * Require read or write permissions.
 	 * If 'force' is set, we only require the "MAY" flags.
#1 lore raw
Quoted 28 lines · Jonathan Corbet

Avoid buffer overflows in get_user_pages()

So I spent a while pounding my head against my monitor trying to figure out the vmsplice() vulnerability - how could a failure to check for *read* access turn into a root exploit? It turns out that it's a buffer overflow problem which is made easy by the way get_user_pages() is coded.

In particular, "len" is a signed int, and it is only checked at the *end* of a do {} while() loop. So, if it is passed in as zero, the loop will execute once and decrement len to -1. At that point, the loop will proceed until the next invalid address is found; in the process, it will likely overflow the pages array passed in to get_user_pages().

I think that, if get_user_pages() has been asked to grab zero pages, that's what it should do. Thus this patch; it is, among other things, enough to block the (already fixed) root exploit and any others which might be lurking in similar code. I also think that the number of pages should be unsigned, but changing the prototype of this function probably requires some more careful review.

mm/memory.c +0 -0
index e5628a5..7f50fd8 100644

mm/memory.c int get_user_pages(struct task_struct *tsk, struct · line 993

Quoted 5 lines
mm/memory.c +2 -0
@@ -989,6 +989,8 @@ int get_user_pages(struct task_struct *tsk, struct
 mm_struct *mm, int i;
 	unsigned int vm_flags;
 
+	if (len <= 0)
+		return 0;

BUG_ON()?

maintainer
#2 lore raw
In reply to Jonathan Corbet
Quoted 8 lines · On Mon, 11 Feb 2008 16:17:33 -0700 Jonathan Corbet <[email protected]> wrote:

Avoid buffer overflows in get_user_pages()

So I spent a while pounding my head against my monitor trying to figure out the vmsplice() vulnerability - how could a failure to check for *read* access turn into a root exploit? It turns out that it's a buffer overflow problem which is made easy by the way get_user_pages() is coded.

Quoted 5 lines

In particular, "len" is a signed int, and it is only checked at the *end* of a do {} while() loop. So, if it is passed in as zero, the loop will execute once and decrement len to -1. At that point, the loop will proceed until the next invalid address is found; in the process, it will likely overflow the pages array passed in to get_user_pages().

Sounds convincing.

Quoted 14 lines

I think that, if get_user_pages() has been asked to grab zero pages, that's what it should do. Thus this patch; it is, among other things, enough to block the (already fixed) root exploit and any others which might be lurking in similar code. I also think that the number of pages should be unsigned, but changing the prototype of this function probably requires some more careful review.

mm/memory.c +0 -0
index e5628a5..7f50fd8 100644

mm/memory.c int get_user_pages(struct task_struct *tsk, struct mm_struct *mm, · line 996

Quoted 8 lines
mm/memory.c +2 -0
@@ -989,6 +989,8 @@ int get_user_pages(struct task_struct *tsk, struct mm_struct *mm,
 	int i;
 	unsigned int vm_flags;
 
+	if (len <= 0)
+		return 0;
 	/* 
 	 * Require read or write permissions.
 	 * If 'force' is set, we only require the "MAY" flags.

Can we just convert

	do {
		...
	} while (len);

into

	while (len) {
		...
	}

?

#3 lore raw
Quoted 6 lines · Andrew Morton

mm/memory.c int get_user_pages(struct task_struct *tsk, struct mm_struct *mm, · line 996

Quoted 5 lines
mm/memory.c +2 -0
+	if (len <= 0)
+		return 0;
 	/* 
 	 * Require read or write permissions.
 	 * If 'force' is set, we only require the "MAY" flags.

Can we just convert

Quoted 7 lines
	do {
		...
	} while (len);
into
	while (len) {
		...
	}

How would that help?

Rather

	while (len > 0) {
		...
	}

would do the trick.

Signature · 1 line
Jiri Kosina
#4 lore raw
In reply to Jonathan Corbet

for stable (.22 .23 .24) ?

git id in mainline: 900cf086fd2fbad07f72f4575449e0d0958f860f

--

tested in: http://repo.or.cz/w/linux-2.6.22.y-op.git testing

Quoted 36 lines · Jonathan Corbet

Avoid buffer overflows in get_user_pages()

So I spent a while pounding my head against my monitor trying to figure out the vmsplice() vulnerability - how could a failure to check for *read* access turn into a root exploit? It turns out that it's a buffer overflow problem which is made easy by the way get_user_pages() is coded.

In particular, "len" is a signed int, and it is only checked at the *end* of a do {} while() loop. So, if it is passed in as zero, the loop will execute once and decrement len to -1. At that point, the loop will proceed until the next invalid address is found; in the process, it will likely overflow the pages array passed in to get_user_pages().

I think that, if get_user_pages() has been asked to grab zero pages, that's what it should do. Thus this patch; it is, among other things, enough to block the (already fixed) root exploit and any others which might be lurking in similar code. I also think that the number of pages should be unsigned, but changing the prototype of this function probably requires some more careful review.

mm/memory.c +2 -0
index e5628a5..7f50fd8 100644
@@ -989,6 +989,8 @@ int get_user_pages(struct task_struct *tsk, struct
 mm_struct *mm,
 int i;
 unsigned int vm_flags;
 
+	if (len <= 0)
+	return 0;
 /*
 * Require read or write permissions.
Quoted 7 lines

* If 'force' is set, we only require the "MAY" flags.
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/

--
Thanks,
Oliver

#5 lore raw
Quoted 3 lines · Oliver Pinter

for stable (.22 .23 .24) ?

git id in mainline: 900cf086fd2fbad07f72f4575449e0d0958f860f

I sent it to the stable folks a couple days ago.

Thanks,

jon

Jonathan Corbet / LWN.net / [email protected]