Usage of BIO

[openssl.org #2090] [patch] NULL-pointer handling in asn1_multi()

Dear all,

another case where NULL should be handled is asn1_multi(). It looks like

   sk = sk_ASN1_TYPE_new_null();
   [...]
         sk_ASN1_TYPE_push(sk, typ);

sk_ASN1_TYPE_push() is defined as sk_push(), which dereferences sk. A
simple patch is attached.

Best regards,

   Martin

[openssl.org #2091] [patch] NULL-pointer check in OBJ_obj2txt()

Dear all,

I think the following snippet from OBJ_obj2txt() is missing a NULL-pointer
check.

                s=OBJ_nid2ln(nid);
                if (s == NULL)
                        s=OBJ_nid2sn(nid);
                if (buf)
                        BUF_strlcpy(buf,s,buf_len);

OBJ_nid2sn() may return NULL, in this case BUF_strlcpy() dereferences
the NULL pointer. This problem can be triggered from an external test
application.

If the parameter has neither ln nor sn, I suggest that -1 is returned.

The attached short patch against today's snapshot fixes this problem

Best regards,

   Martin

Re: [openssl.org #2091] [patch] NULL-pointer check in OBJ_obj2txt()

On Tue, Nov 03, 2009 at 10:09:04AM +0100, Martin Kaiser via RT wrote:

> If the parameter has neither ln nor sn, I suggest that -1 is returned.
> 
> The attached short patch against today's snapshot fixes this problem
> 

Thinking about this again, would it make sense to add an error to the
queue like the patch below?

Best regards,

   Martin

diff --git a/crypto/objects/obj_dat.c b/crypto/objects/obj_dat.c
index e999ef7..fd98769 100644
--- a/crypto/objects/obj_dat.c
+++ b/crypto/objects/obj_dat.c
 <at>  <at>  -466,7 +466,7  <at>  <at>  ASN1_OBJECT *OBJ_txt2obj(const char *s, int no_name)
 int OBJ_obj2txt(char *buf, int buf_len, const ASN1_OBJECT *a, int no_name)
 {
        int i,n=0,len,nid, first, use_bn;
-       BIGNUM *bl;
+       BIGNUM *bl=NULL;
        unsigned long l;
        const unsigned char *p;
        char tbuf[DECIMAL_SIZE(i)+DECIMAL_SIZE(l)+2];
 <at>  <at>  -483,6 +483,11  <at>  <at>  int OBJ_obj2txt(char *buf, int buf_len, const ASN1_OBJECT *
                s=OBJ_nid2ln(nid);
                if (s == NULL)
                        s=OBJ_nid2sn(nid);
+               if (s == NULL)
+                       {
+                       OBJerr(OBJ_F_OBJ_OBJ2TXT,OBJ_R_UNKNOWN_NID_NAME);
+                       goto err;
+                       }
                if (buf)
                        BUF_strlcpy(buf,s,buf_len);
                n=strlen(s);
 <at>  <at>  -494,7 +499,6  <at>  <at>  int OBJ_obj2txt(char *buf, int buf_len, const ASN1_OBJECT *a
        p=a->data;

        first = 1;
-       bl = NULL;

        while (len > 0)
                {
diff --git a/crypto/objects/objects.h b/crypto/objects/objects.h
index bd0ee52..ab43d9d 100644
--- a/crypto/objects/objects.h
+++ b/crypto/objects/objects.h
 <at>  <at>  -1127,10 +1127,12  <at>  <at>  void ERR_load_OBJ_strings(void);
 #define OBJ_F_OBJ_NID2LN                                102
 #define OBJ_F_OBJ_NID2OBJ                               103
 #define OBJ_F_OBJ_NID2SN                                104
+#define OBJ_F_OBJ_OBJ2TXT                               105

 /* Reason codes. */
 #define OBJ_R_MALLOC_FAILURE                            100
 #define OBJ_R_UNKNOWN_NID                               101
+#define OBJ_R_UNKNOWN_NID_NAME                          102

 #ifdef  __cplusplus
 }

______________________________________________________________________
OpenSSL Project                                 http://www.openssl.org
Development Mailing List                       openssl-dev <at> openssl.org
Automated List Manager                           majordomo <at> openssl.org

Re: [openssl.org #2089] [PATCH] DTLS Fragment size bug

On Fri, 2009-10-30 at 18:11 +0100, Robin Seggelmann via RT wrote:
> DTLS fragmentation doesn't consider the additional data required with  
> using encryption, so the packet size then exceeds the MTU when  
> fragmentation is performed. This patch subtracts the size of the mac  
> and the additional encryption bytes of the maximum possible length for  
> a fragment, if necessary.
> 
> Thanks to Daniel Mentz for finding this bug.
> 

... 

> +	if (s->write_hash)
> +		mac_size = EVP_MD_CTX_size(s->write_hash);
> +	else
> +		mac_size = 0;
> +

This was applied to the 0.9.8 branch too, where it causes a SEGV. It
should be EVP_MD_size() there.

Index: ssl/d1_both.c
===================================================================
RCS file: /home/dwmw2/openssl-cvs/openssl/ssl/d1_both.c,v
retrieving revision 1.4.2.21
diff -u -p -r1.4.2.21 d1_both.c
--- ssl/d1_both.c	2 Nov 2009 13:36:56 -0000	1.4.2.21
+++ ssl/d1_both.c	3 Nov 2009 20:41:54 -0000
 <at>  <at>  -226,7 +226,7  <at>  <at>  int dtls1_do_write(SSL *s, int type)
 			(int)s->d1->w_msg_hdr.msg_len + DTLS1_HM_HEADER_LENGTH);

 	if (s->write_hash)
-		mac_size = EVP_MD_CTX_size(s->write_hash);
+		mac_size = EVP_MD_size(s->write_hash);
 	else
 		mac_size = 0;

--

-- 
David Woodhouse                            Open Source Technology Centre
David.Woodhouse <at> intel.com                              Intel Corporation

______________________________________________________________________
OpenSSL Project                                 http://www.openssl.org
Development Mailing List                       openssl-dev <at> openssl.org
Automated List Manager                           majordomo <at> openssl.org

[openssl.org #2092] Updated AES-GCM/AES-CCM/CMAC patch

This is is an update to our previous contribution.

It should apply cleanly to openssl-1.0.0-beta3, that's been tested on a
couple of Unix'y platforms here.

There are some bug fixes which deal with corner cases in AES-GCM and
AES-CCM, and CMAC now works in DES modes.
Thanks to Paul Suhler for his testing of GCM which picked a couple of
subtle bugs there.

It should also build on Windows now - but I can't get beta3 to build
(unmodified) on Windows in my environment for some reason so I've only been
able to test that with a close cousin of this patch applied to older
OpenSSL variants..
Note that I had to drop some of the test cases to get this to build on
Windows, if you need the older/more comprehensive test cases, you'll have
to dig them out of earlier patches.

(See attached file: ibm4.patch)

Peter

SSL_write reurns SSL3_WRITE_PENDING

OpenSSL hardware acceleration exploration...

Zlib compression

RE: SSL_write reurns SSL3_WRITE_PENDING