diff mbox

[dpdk-dev] examples/ipsec-secgw: Calling risky function

Message ID 1465289886-14479-1-git-send-email-slawomirx.mrozowicz@intel.com (mailing list archive)
State Rejected, archived
Headers

Commit Message

Slawomir Mrozowicz June 7, 2016, 8:58 a.m. UTC 
  lrand48 should not be used for security related applications,
as linear congruential algorithms are too easy to break.
Used a compliant random number generator /dev/urandom.

Fixes: d299106e8e31 ("examples/ipsec-secgw: add IPsec sample application")
Coverity ID 124558

Signed-off-by: Slawomir Mrozowicz <slawomirx.mrozowicz@intel.com>
---
 examples/ipsec-secgw/esp.c | 17 +++++++++--------
 1 file changed, 9 insertions(+), 8 deletions(-)

Comments

Thomas Monjalon June 7, 2016, 8:11 a.m. UTC | #1 
Hi Slawomir,

"Calling risky function"
-> it's obviously a NACK :)
unless you mean "replace non-secure random".
Sergio Gonzalez Monroy June 7, 2016, 8:15 a.m. UTC | #2 
On 07/06/2016 09:58, Slawomir Mrozowicz wrote:
> lrand48 should not be used for security related applications,
> as linear congruential algorithms are too easy to break.
> Used a compliant random number generator /dev/urandom.
>
> Fixes: d299106e8e31 ("examples/ipsec-secgw: add IPsec sample application")
> Coverity ID 124558
>
> Signed-off-by: Slawomir Mrozowicz <slawomirx.mrozowicz@intel.com>
> ---

I understand that lrand48 is not crypto secure, but this fix will kill 
performance.

I already have a solution for this issue to be included in the next 
IPSec patch set
that will also add support for GCM/CTR modes.

Sergio
Slawomir Mrozowicz June 7, 2016, 1:16 p.m. UTC | #3 
>-----Original Message-----
>From: Gonzalez Monroy, Sergio
>Sent: Tuesday, June 7, 2016 10:15 AM
>To: Mrozowicz, SlawomirX <slawomirx.mrozowicz@intel.com>
>Cc: dev@dpdk.org
>Subject: Re: [PATCH] examples/ipsec-secgw: Calling risky function
>
>On 07/06/2016 09:58, Slawomir Mrozowicz wrote:
>> lrand48 should not be used for security related applications, as
>> linear congruential algorithms are too easy to break.
>> Used a compliant random number generator /dev/urandom.
>>
>> Fixes: d299106e8e31 ("examples/ipsec-secgw: add IPsec sample
>> application") Coverity ID 124558
>>
>> Signed-off-by: Slawomir Mrozowicz <slawomirx.mrozowicz@intel.com>
>> ---
>
>I understand that lrand48 is not crypto secure, but this fix will kill performance.
>
>I already have a solution for this issue to be included in the next IPSec patch
>set that will also add support for GCM/CTR modes.
>
>Sergio

Thanks for your reply.
So for now I propose to set this problem as intentional in the Coverity tool.

Sławomir
diff mbox

Patch

diff --git a/examples/ipsec-secgw/esp.c b/examples/ipsec-secgw/esp.c
index 0f6b33e..f3c4687 100644
--- a/examples/ipsec-secgw/esp.c
+++ b/examples/ipsec-secgw/esp.c
@@ -55,16 +55,17 @@ 
 static inline void
 random_iv_u64(uint64_t *buf, uint16_t n)
 {
-	unsigned left = n & 0x7;
-	unsigned i;
+	int res = 0;
+	FILE *fp;
 
-	RTE_ASSERT((n & 0x3) == 0);
-
-	for (i = 0; i < (n >> 3); i++)
-		buf[i] = rte_rand();
+	fp = fopen("/dev/urandom", "r");
+	if (fp != NULL) {
+		res = fread(buf, 8, n, fp);
+		fclose(fp);
+	}
 
-	if (left)
-		*((uint32_t *)&buf[i]) = (uint32_t)lrand48();
+	RTE_ASSERT(res != n);
+	RTE_LOG(DEBUG, IPSEC_ESP, "random_iv_u64 result %d\n", res);
 }
 
 /* IPv4 Tunnel */