azlig and Milan Broz reported that when the tty layer is done with a
buffer, the data can hang around in it for a very long time. That
sometimes can "leak" to userspace under some conditions.
Because of this, just zero out the data after the tty layer is finished
with it, for buffers that we "think" should be zeroed out.
Greg Kroah-Hartman (1):
tty: wipe buffer if not echoing data
Linus Torvalds (1):
tty: wipe buffer.
drivers/tty/n_tty.c | 20 +++++++++++++++++---
drivers/tty/tty_buffer.c | 6 +++++-
2 files changed, 22 insertions(+), 4 deletions(-)
--
2.19.0
From: Greg KH <[email protected]>
If we are not echoing the data to userspace, then perhaps it is a
"secret" so we should wipe it once we are done with it.
This mirrors the logic that the audit code has.
Reported-by: aszlig <[email protected]>
Tested-by: Milan Broz <[email protected]>
Tested-by: aszlig <[email protected]>
Cc: Willy Tarreau <[email protected]>
Signed-off-by: Greg Kroah-Hartman <[email protected]>
---
drivers/tty/n_tty.c | 20 +++++++++++++++++---
1 file changed, 17 insertions(+), 3 deletions(-)
diff --git a/drivers/tty/n_tty.c b/drivers/tty/n_tty.c
index 431742201709..d259a6585dbe 100644
--- a/drivers/tty/n_tty.c
+++ b/drivers/tty/n_tty.c
@@ -152,17 +152,28 @@ static inline unsigned char *echo_buf_addr(struct n_tty_data *ldata, size_t i)
return &ldata->echo_buf[i & (N_TTY_BUF_SIZE - 1)];
}
+/* If we are not echoing the data, perhaps this is a secret so erase it */
+static void zero_buffer(struct tty_struct *tty, u8 *buffer, int size)
+{
+ bool icanon = !!L_ICANON(tty);
+ bool no_echo = !L_ECHO(tty);
+
+ if (icanon && no_echo)
+ memset(buffer, 0x00, size);
+}
+
static int tty_copy_to_user(struct tty_struct *tty, void __user *to,
size_t tail, size_t n)
{
struct n_tty_data *ldata = tty->disc_data;
size_t size = N_TTY_BUF_SIZE - tail;
- const void *from = read_buf_addr(ldata, tail);
+ void *from = read_buf_addr(ldata, tail);
int uncopied;
if (n > size) {
tty_audit_add_data(tty, from, size);
uncopied = copy_to_user(to, from, size);
+ zero_buffer(tty, from, size);
if (uncopied)
return uncopied;
to += size;
@@ -171,7 +182,9 @@ static int tty_copy_to_user(struct tty_struct *tty, void __user *to,
}
tty_audit_add_data(tty, from, n);
- return copy_to_user(to, from, n);
+ uncopied = copy_to_user(to, from, n);
+ zero_buffer(tty, from, n);
+ return uncopied;
}
/**
@@ -1960,11 +1973,12 @@ static int copy_from_read_buf(struct tty_struct *tty,
n = min(head - ldata->read_tail, N_TTY_BUF_SIZE - tail);
n = min(*nr, n);
if (n) {
- const unsigned char *from = read_buf_addr(ldata, tail);
+ unsigned char *from = read_buf_addr(ldata, tail);
retval = copy_to_user(*b, from, n);
n -= retval;
is_eof = n == 1 && *from == EOF_CHAR(tty);
tty_audit_add_data(tty, from, n);
+ zero_buffer(tty, from, n);
smp_store_release(&ldata->read_tail, ldata->read_tail + n);
/* Turn single EOF into zero-length read */
if (L_EXTPROC(tty) && ldata->icanon && is_eof &&
--
2.19.0
From: Linus Torvalds <[email protected]>
After we are done with the tty buffer, zero it out.
Reported-by: aszlig <[email protected]>
Tested-by: Milan Broz <[email protected]>
Tested-by: aszlig <[email protected]>
Cc: Willy Tarreau <[email protected]>
Signed-off-by: Greg Kroah-Hartman <[email protected]>
---
drivers/tty/tty_buffer.c | 6 +++++-
1 file changed, 5 insertions(+), 1 deletion(-)
diff --git a/drivers/tty/tty_buffer.c b/drivers/tty/tty_buffer.c
index c996b6859c5e..ae3ce330200e 100644
--- a/drivers/tty/tty_buffer.c
+++ b/drivers/tty/tty_buffer.c
@@ -468,11 +468,15 @@ receive_buf(struct tty_port *port, struct tty_buffer *head, int count)
{
unsigned char *p = char_buf_ptr(head, head->read);
char *f = NULL;
+ int n;
if (~head->flags & TTYB_NORMAL)
f = flag_buf_ptr(head, head->read);
- return port->client_ops->receive_buf(port, p, f, count);
+ n = port->client_ops->receive_buf(port, p, f, count);
+ if (n > 0)
+ memset(p, 0, n);
+ return n;
}
/**
--
2.19.0
On 02/10/2018 19:17, Greg Kroah-Hartman wrote:
> From: Greg KH <[email protected]>
>
> If we are not echoing the data to userspace, then perhaps it is a
> "secret" so we should wipe it once we are done with it.
Just to explain our test case for cryptsetup, where aszlig initially reported it:
cryptsetup reads a passphrase from terminal, derives a candidate key for keyslot
decryption and immediately wipes the passphrase from memory, because it is
not needed anymore.
And here is the problem - the kernel keeps this passphrase in a tty buffer memory
that is no longer available to userspace. It can be retrieved from memory dump
later, even after the crypt mapping is destroyed.
Almost all userspace tools working with passphrases through tty access have
the same problem here.
> This mirrors the logic that the audit code has.
>
> Reported-by: aszlig <[email protected]>
> Tested-by: Milan Broz <[email protected]>
> Tested-by: aszlig <[email protected]>
> Cc: Willy Tarreau <[email protected]>
> Signed-off-by: Greg Kroah-Hartman <[email protected]>
> ---
> drivers/tty/n_tty.c | 20 +++++++++++++++++---
...
> static int tty_copy_to_user(struct tty_struct *tty, void __user *to,
> size_t tail, size_t n)
> {
> struct n_tty_data *ldata = tty->disc_data;
> size_t size = N_TTY_BUF_SIZE - tail;
> - const void *from = read_buf_addr(ldata, tail);
> + void *from = read_buf_addr(ldata, tail);
> int uncopied;
>
> if (n > size) {
> tty_audit_add_data(tty, from, size);
> uncopied = copy_to_user(to, from, size);
> + zero_buffer(tty, from, size);
I think Linus mentioned in some previous mail that there should be
zero_buffer(tty, from, size - uncopied);
to avoid wiping of yet uncopied data.
I tested the unmodified version though...
> if (uncopied)
> return uncopied;
> to += size;
> @@ -171,7 +182,9 @@ static int tty_copy_to_user(struct tty_struct *tty, void __user *to,
> }
>
> tty_audit_add_data(tty, from, n);
> - return copy_to_user(to, from, n);
> + uncopied = copy_to_user(to, from, n);
> + zero_buffer(tty, from, n);
> + return uncopied;
> }
Milan