Skip to contentSkip to conversation

9p/trans_fd: Check file mode at opening

PATCH In review Alexey Kardashevskiy 3 participants 6 messages Jul 28, 2020 – Jul 29, 2020

Subsystems (3)
lore
Share & tools
lore
#0 lore raw

PATCH 1/1 9p/trans_fd: Check file mode at opening

The "fd" transport layer uses 2 file descriptors passed externally and calls kernel_write()/kernel_read() on these. If files were opened without FMODE_WRITE/FMODE_READ, WARN_ON_ONCE() will fire.

This adds file mode checking in p9_fd_open; this returns -EBADF to preserve the original behavior.

Found by syzkaller.

---
 net/9p/trans_fd.c | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)
net/9p/trans_fd.c +6 -1
index 13cd683a658a..62cdfbd01f0a 100644
@@ -797,6 +797,7 @@ static int parse_opts(char *params, struct p9_fd_opts *opts)
 
 static int p9_fd_open(struct p9_client *client, int rfd, int wfd)
 {
+	bool perm;
 	struct p9_trans_fd *ts = kzalloc(sizeof(struct p9_trans_fd),
 					   GFP_KERNEL);
 	if (!ts)
@@ -804,12 +805,16 @@ static int p9_fd_open(struct p9_client *client, int rfd, int wfd)
 
 	ts->rd = fget(rfd);
 	ts->wr = fget(wfd);
-	if (!ts->rd || !ts->wr) {
+	perm = ts->rd && (ts->rd->f_mode & FMODE_READ) &&
+	       ts->wr && (ts->wr->f_mode & FMODE_WRITE);
+	if (!ts->rd || !ts->wr || !perm) {
 		if (ts->rd)
 			fput(ts->rd);
 		if (ts->wr)
 			fput(ts->wr);
 		kfree(ts);
+		if (!perm)
+			return -EBADF;
 		return -EIO;
 	}
 
Signature · 1 line
2.17.1
#1 lore raw

Hi Alexey,

Working on 9p now ?!? ;-)

Cc'ing Dominique Martinet who appears to be the person who takes care of 9p these days.

Quoted 7 lines · On Tue, 28 Jul 2020 22:41:29 +1000 Alexey Kardashevskiy <[email protected]> wrote:

The "fd" transport layer uses 2 file descriptors passed externally and calls kernel_write()/kernel_read() on these. If files were opened without FMODE_WRITE/FMODE_READ, WARN_ON_ONCE() will fire.

This adds file mode checking in p9_fd_open; this returns -EBADF to preserve the original behavior.

So this would cause open() to fail with EBADF, which might look a bit weird to userspace since it didn't pass an fd... Is this to have a different error than -EIO that is returned when either rfd or wfd doesn't point to an open file descriptor ? If yes, why do we care ?

Quoted 37 lines

Found by syzkaller.

---
 net/9p/trans_fd.c | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)
net/9p/trans_fd.c +6 -1
index 13cd683a658a..62cdfbd01f0a 100644
@@ -797,6 +797,7 @@ static int parse_opts(char *params, struct p9_fd_opts *opts)
 
 static int p9_fd_open(struct p9_client *client, int rfd, int wfd)
 {
+	bool perm;
 	struct p9_trans_fd *ts = kzalloc(sizeof(struct p9_trans_fd),
 					   GFP_KERNEL);
 	if (!ts)
@@ -804,12 +805,16 @@ static int p9_fd_open(struct p9_client *client, int rfd, int wfd)
 
 	ts->rd = fget(rfd);
 	ts->wr = fget(wfd);
-	if (!ts->rd || !ts->wr) {
+	perm = ts->rd && (ts->rd->f_mode & FMODE_READ) &&
+	       ts->wr && (ts->wr->f_mode & FMODE_WRITE);
+	if (!ts->rd || !ts->wr || !perm) {
 		if (ts->rd)
 			fput(ts->rd);
 		if (ts->wr)
 			fput(ts->wr);
 		kfree(ts);
+		if (!perm)
+			return -EBADF;
 		return -EIO;
 	}
 
#2 lore raw
Quoted 3 lines · Greg Kurz

Hi Alexey,

Working on 9p now ?!? ;-)

No, I am running syzkaller and seeing things :)

Quoted 10 lines

Cc'ing Dominique Martinet who appears to be the person who takes care of 9p these days.

Quoted 4 lines · On Tue, 28 Jul 2020 22:41:29 +1000 Alexey Kardashevskiy <[email protected]> wrote:

The "fd" transport layer uses 2 file descriptors passed externally and calls kernel_write()/kernel_read() on these. If files were opened without FMODE_WRITE/FMODE_READ, WARN_ON_ONCE() will fire.

Quoted 8 lines
Quoted 3 lines

This adds file mode checking in p9_fd_open; this returns -EBADF to preserve the original behavior.

So this would cause open() to fail with EBADF, which might look a bit weird to userspace since it didn't pass an fd... Is this to have a different error than -EIO that is returned when either rfd or wfd doesn't point to an open file descriptor ?

This is only to preserve the existing behavior.

Quoted 1 line

If yes, why do we care ?

Without the patch, p9_fd_open() produces a kernel warning which is not great by itself and becomes crash with panic_on_warn.

Quoted 39 lines
Quoted 38 lines

Found by syzkaller.

---
 net/9p/trans_fd.c | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)
net/9p/trans_fd.c +6 -1
index 13cd683a658a..62cdfbd01f0a 100644
@@ -797,6 +797,7 @@ static int parse_opts(char *params, struct p9_fd_opts *opts)
 
 static int p9_fd_open(struct p9_client *client, int rfd, int wfd)
 {
+	bool perm;
 	struct p9_trans_fd *ts = kzalloc(sizeof(struct p9_trans_fd),
 					   GFP_KERNEL);
 	if (!ts)
@@ -804,12 +805,16 @@ static int p9_fd_open(struct p9_client *client, int rfd, int wfd)
 
 	ts->rd = fget(rfd);
 	ts->wr = fget(wfd);
-	if (!ts->rd || !ts->wr) {
+	perm = ts->rd && (ts->rd->f_mode & FMODE_READ) &&
+	       ts->wr && (ts->wr->f_mode & FMODE_WRITE);
+	if (!ts->rd || !ts->wr || !perm) {
 		if (ts->rd)
 			fput(ts->rd);
 		if (ts->wr)
 			fput(ts->wr);
 		kfree(ts);
+		if (!perm)
+			return -EBADF;
 		return -EIO;
 	}
 
Signature · 1 line
Alexey
#3 lore raw
Quoted 9 lines · On Wed, 29 Jul 2020 09:50:21 +1000 Alexey Kardashevskiy <[email protected]> wrote:
Quoted 3 lines · Greg Kurz

Hi Alexey,

Working on 9p now ?!? ;-)

No, I am running syzkaller and seeing things :)

:)

Quoted 20 lines
Quoted 11 lines

Cc'ing Dominique Martinet who appears to be the person who takes care of 9p these days.

Quoted 4 lines · On Tue, 28 Jul 2020 22:41:29 +1000 Alexey Kardashevskiy <[email protected]> wrote:

The "fd" transport layer uses 2 file descriptors passed externally and calls kernel_write()/kernel_read() on these. If files were opened without FMODE_WRITE/FMODE_READ, WARN_ON_ONCE() will fire.

Quoted 8 lines
Quoted 3 lines

This adds file mode checking in p9_fd_open; this returns -EBADF to preserve the original behavior.

So this would cause open() to fail with EBADF, which might look a bit weird to userspace since it didn't pass an fd... Is this to have a different error than -EIO that is returned when either rfd or wfd doesn't point to an open file descriptor ?

Quoted 8 lines

This is only to preserve the existing behavior.

Quoted 1 line

If yes, why do we care ?

Without the patch, p9_fd_open() produces a kernel warning which is not great by itself and becomes crash with panic_on_warn.

I don't question the patch, just the errno. Why not returning -EIO ?

Quoted 42 lines
Quoted 41 lines
Quoted 40 lines

Found by syzkaller.

---
 net/9p/trans_fd.c | 7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)
net/9p/trans_fd.c +6 -1
index 13cd683a658a..62cdfbd01f0a 100644
@@ -797,6 +797,7 @@ static int parse_opts(char *params, struct p9_fd_opts *opts)
 
 static int p9_fd_open(struct p9_client *client, int rfd, int wfd)
 {
+	bool perm;
 	struct p9_trans_fd *ts = kzalloc(sizeof(struct p9_trans_fd),
 					   GFP_KERNEL);
 	if (!ts)
@@ -804,12 +805,16 @@ static int p9_fd_open(struct p9_client *client, int rfd, int wfd)
 
 	ts->rd = fget(rfd);
 	ts->wr = fget(wfd);
-	if (!ts->rd || !ts->wr) {
+	perm = ts->rd && (ts->rd->f_mode & FMODE_READ) &&
+	       ts->wr && (ts->wr->f_mode & FMODE_WRITE);
+	if (!ts->rd || !ts->wr || !perm) {
 		if (ts->rd)
 			fput(ts->rd);
 		if (ts->wr)
 			fput(ts->wr);
 		kfree(ts);
+		if (!perm)
+			return -EBADF;
 		return -EIO;
 	}
 
#4 lore raw
In reply to Greg Kurz

Greg Kurz wrote on Tue, Jul 28, 2020:

Quoted 3 lines
Quoted 3 lines

The "fd" transport layer uses 2 file descriptors passed externally and calls kernel_write()/kernel_read() on these. If files were opened without FMODE_WRITE/FMODE_READ, WARN_ON_ONCE() will fire.

There already is a fix in linux-next as a39c46067c84 ("net/9p: validate fds in p9_fd_open")

Quoted 7 lines
Quoted 2 lines

This adds file mode checking in p9_fd_open; this returns -EBADF to preserve the original behavior.

So this would cause open() to fail with EBADF, which might look a bit weird to userspace since it didn't pass an fd... Is this to have a different error than -EIO that is returned when either rfd or wfd doesn't point to an open file descriptor ? If yes, why do we care ?

FWIW the solution taken just returns EIO as it would if an invalid fd was given, but since it did pass an fd EBADF actually makes sense to me?

However to the second question I'm not sure I care :)

Quoted 1 line
Quoted 1 line

Found by syzkaller.

I'm starting to understand where David comment came from the other day, I guess it's still time to change my mind and submit to linus now I've had time to test it...

Signature · 1 line
Dominique
#5 lore raw
Quoted 12 lines · On Wed, 29 Jul 2020 08:14:49 +0200 Dominique Martinet <[email protected]>…

Greg Kurz wrote on Tue, Jul 28, 2020:

Quoted 3 lines
Quoted 3 lines

The "fd" transport layer uses 2 file descriptors passed externally and calls kernel_write()/kernel_read() on these. If files were opened without FMODE_WRITE/FMODE_READ, WARN_ON_ONCE() will fire.

There already is a fix in linux-next as a39c46067c84 ("net/9p: validate fds in p9_fd_open")

Quoted 4 lines
Quoted 2 lines

This adds file mode checking in p9_fd_open; this returns -EBADF to preserve the original behavior.

So this would cause open() to fail with EBADF, which might look a bit

Oops... this seems to rather end up in mount(). :)

Quoted 7 lines
Quoted 3 lines

weird to userspace since it didn't pass an fd... Is this to have a different error than -EIO that is returned when either rfd or wfd doesn't point to an open file descriptor ? If yes, why do we care ?

FWIW the solution taken just returns EIO as it would if an invalid fd was given, but since it did pass an fd EBADF actually makes sense to me?

POSIX says:

https://pubs.opengroup.org/onlinepubs/9699919799/functions/V2_chap02.html

[EBADF]
Bad file descriptor. A file descriptor argument is out of range, refers to no open file, or a read (write) request is made to a file that is only open for writing (reading).

It seems that EBADF would be appropriate for both the existing and the new error path.

Quoted 8 lines

However to the second question I'm not sure I care :)

Quoted 1 line
Quoted 1 line

Found by syzkaller.

I'm starting to understand where David comment came from the other day, I guess it's still time to change my mind and submit to linus now I've had time to test it...