net: af_packet: Validate parameter size for PACKET_HDRLEN control message
Subsystems (3)
PATCH 1/1 net: af_packet: Validate parameter size for PACKET_HDRLEN control message
Building af_packet may fail with
In function ‘copy_from_user’,
inlined from ‘packet_getsockopt’ at net/packet/af_packet.c:3215:21:arch/x86/include/asm/uaccess_32.h:211:26: error: call to
‘copy_from_user_overflow’ declared with attribute error: copy_from_user() buffer size is not provably correct
if built with W=1 due to a missing parameter size validation.
--- net/packet/af_packet.c | 2 ++ 1 file changed, 2 insertions(+)
net/packet/af_packet.c +2 -0
@@ -3210,6 +3210,8 @@ static int packet_getsockopt(struct socket *sock, int level, int optname, val = po->tp_version;break ;case PACKET_HDRLEN : +if (len <sizeof (int )) +return -EINVAL ;if (len >sizeof (int )) len =sizeof (int );if (copy_from_user (&val, optval, len))
Signature · 1 line
1.7.9.7
David Miller
maintainerThe first thing this function does is test whether len < 0, therefore your change is unnecessary.
If the user gives us something between 0 and sizeof(int), that's their problem, and they'll get a partial int copied back into userspace as a result instead of the complete integer.
Please don't blindly silence warnings like this, thanks.
Quoted 21 lines · Guenter Roeck
Building
af_packetmay fail withIn function ‘copy_from_user’,
inlined from ‘packet_getsockopt’ atnet/packet/af_packet.c:3215:21:arch/x86/include/asm/uaccess_32.h:211:26: error: call to
‘copy_from_user_overflow’ declared with attribute error:copy_from_user()buffer size is not provably correctif built with W=1 due to a missing parameter size validation.
Signed-off-by Guenter Roeck <[email protected]>--- net/packet/af_packet.c | 2 ++ 1 file changed, 2 insertions(+)net/packet/af_packet.c +0
-0
net/packet/af_packet.c static int packet_getsockopt(struct socket *sock, int level, int optname, · line 3214
Quoted 5 lines
net/packet/af_packet.c +2
-0@@ -3210,6 +3210,8 @@ static int packet_getsockopt(struct socket *sock, int level, int optname, val = po->tp_version;break ;case PACKET_HDRLEN : +if (len <sizeof (int )) +return -EINVAL ;
I think this could break some user space applications here, those who e.g. only pass an uint16_t to packet_getsockopt with PACKET_HDRLEN.
Quoted 3 lines
net/packet/af_packet.c +0
-0if (len >sizeof (int )) len =sizeof (int );if (copy_from_user (&val, optval, len))
David Miller
maintainerFrom: Daniel Borkmann <[email protected]>
Date: Wed, 27 Feb 2013 21:22:17 +0100
Quoted 18 lines · Guenter Roeck
Quoted 17 lines
Building
af_packetmay fail withIn function ‘copy_from_user’,
inlined from ‘packet_getsockopt’ atnet/packet/af_packet.c:3215:21:arch/x86/include/asm/uaccess_32.h:211:26: error: call to
‘copy_from_user_overflow’ declared with attribute error:
copy_from_user()
buffer size is not provably correctif built with W=1 due to a missing parameter size validation.
Signed-off-by Guenter Roeck <[email protected]>--- net/packet/af_packet.c | 2 ++ 1 file changed, 2 insertions(+)
Quoted 15 lines
net/packet/af_packet.c static int packet_getsockopt(struct socket · line 3215
Quoted 11 lines
net/packet/af_packet.c +2
-0@@ -3210,6 +3210,8 @@ static int packet_getsockopt(struct socket *sock, int level, int optname, val = po->tp_version;break ;case PACKET_HDRLEN : +if (len <sizeof (int )) +return -EINVAL ;I think this could break some user space applications here, those who e.g. only pass
anuint16_ttopacket_getsockoptwithPACKET_HDRLEN.
Well, their shit is broken on big endian then.
Quoted 37 lines · David Miller
From: Daniel Borkmann <[email protected]>
Date: Wed, 27 Feb 2013 21:22:17 +0100Quoted 18 lines · Guenter Roeck
Quoted 17 lines
Building
af_packetmay fail withIn function ‘copy_from_user’,
inlined from ‘packet_getsockopt’ atnet/packet/af_packet.c:3215:21:arch/x86/include/asm/uaccess_32.h:211:26: error: call to
‘copy_from_user_overflow’ declared with attribute error:
copy_from_user()
buffer size is not provably correctif built with W=1 due to a missing parameter size validation.
Signed-off-by Guenter Roeck <[email protected]>--- net/packet/af_packet.c | 2 ++ 1 file changed, 2 insertions(+)Quoted 15 lines
net/packet/af_packet.c static int packet_getsockopt(struct socket · line 3215
Quoted 11 lines
net/packet/af_packet.c +2
-0@@ -3210,6 +3210,8 @@ static int packet_getsockopt(struct socket *sock, int level, int optname, val = po->tp_version;break ;case PACKET_HDRLEN : +if (len <sizeof (int )) +return -EINVAL ;I think this could break some user space applications here, those who e.g. only pass
anuint16_ttopacket_getsockoptwithPACKET_HDRLEN.
Quoted 1 line
Well, their shit is broken on big endian then.
There must be something else going on anyway ... yes, my patch fixes the warning/error, but copy_from_user should only bail out if the copy size can be larger than the provided buffer (unless I misunderstand the code in copy_from_user). And the second check should take care of that.
Anyway, point taken. I'll waste my time elsewhere :).
Thanks,
Guenter
Quoted 38 lines · Guenter Roeck
Quoted 37 lines · David Miller
From: Daniel Borkmann <[email protected]>
Date: Wed, 27 Feb 2013 21:22:17 +0100Quoted 33 lines · Guenter Roeck
Quoted 22 lines
Building
af_packetmay fail withIn function ‘copy_from_user’,
inlined from ‘packet_getsockopt’ atnet/packet/af_packet.c:3215:21:arch/x86/include/asm/uaccess_32.h:211:26: error: call to
‘copy_from_user_overflow’ declared with attribute error:
copy_from_user()
buffer size is not provably correctif built with W=1 due to a missing parameter size validation.
Signed-off-by Guenter Roeck <[email protected]>--- net/packet/af_packet.c | 2 ++ 1 file changed, 2 insertions(+)net/packet/af_packet.c +0
-0net/packet/af_packet.c static int packet_getsockopt(struct socket · line 3215
Quoted 6 lines
net/packet/af_packet.c +2
-0@@ -3210,6 +3210,8 @@ static int packet_getsockopt(struct socket *sock, int level, int optname, val = po->tp_version;break ;case PACKET_HDRLEN : +if (len <sizeof (int )) +return -EINVAL ;I think this could break some user space applications here, those who e.g. only pass
anuint16_ttopacket_getsockoptwithPACKET_HDRLEN.
Quoted 6 lines
Quoted 1 line
Well, their shit is broken on big endian then.
There must be something else going on anyway ... yes, my patch fixes the warning/error, but
copy_from_usershould only bail out if the copy size can be larger than the provided buffer (unless I misunderstand the code incopy_from_user). And the second check should take care of that.
Fair enough, from what I read the implementation on x86_64 uses gcc's __builtin_object_size(<X>, 0) [1]. Since the <to> (<X>) argument is known at compile time (val:int), __builtin_object_size() will return sizeof(int)-1, the number of bytes from val start to the end of the object val pointer points to. Since our length that we pass can be [0, sizeof(int)] the compiler cannot prove it, if the copy_from_user() buffer size is correct. Thus, "buffer size is not provably correct". Applications not passing int to this getsockopt(2) are screwed up then anyway.
[1] http://gcc.gnu.org/onlinedocs/gcc/Object-Size-Checking.html