Review Request: TMS ECC Key Recovery
https://fedorahosted.org/pki/ticket/252 - TMS - ECC Key Recovery patch for review: https://fedorahosted.org/pki/attachment/ticket/252/TMS-ECC-Recovery.patchFor... This patch provides code to allow ECC key recovery in the TMS environment. It was tested to work with tpsclient. The key injection part of implementation for the actual smart card tokens is scheduled to be done in #235 at a later time. I can do a demo tomorrow in office. thanks, Christina
Couple Questions: In the following piece of the patch: + SECITEM_FreeItem(&der, PR_FALSE); 81 SECKEY_DestroySubjectPublicKeyInfo(spki); 82 It looks like both are executed , even when the call to spki = SECKEY_DecodeDERSubjectPublicKeyInfo(&der); Fails and spki is NULL. Does the DestroySubjectPublicKeyInfo handle this? Also Will "der" be in a state to harm the call to SECITEM_FreeItem? Index: tps/src/engine/RA.cpp 187 =================================================================== 188 --- tps/src/engine/RA.cpp (revision 2497) 189 +++ tps/src/engine/RA.cpp (working copy) 190 @@ -1238,7 +1238,10 @@ 191 goto loser; 192 } else { 193 RA::Debug(LL_PER_PDU, "RecoverKey", "got public key =%s", tmp); 194 - *publicKey_s = PL_strdup(tmp); 195 + char *tmp_publicKey_s = PL_strdup(tmp); 196 + Buffer *decodePubKey = Util::URLDecode(tmp_publicKey_s); 197 + *publicKey_s = 198 + BTOA_DataToAscii(decodePubKey->getBuf(), decodePubKey->getLen()); 199 } 200 201 tmp = NULL; 202 @@ -1256,7 +1259,7 @@ 203 RA::Error(LL_PER_PDU, "RecoverKey", 204 "did not get iv_param for recovered key in DRM response"); 205 } else { 206 - RA::Debug(LL_PER_PDU, "ServerSideKeyGen", "got iv_param for recovered key =%s", tmp); 207 + RA::Debug(LL_PER_PDU, "RecoverKey", "got iv_param for recovered key =%s", tmp); 208 *ivParam_s = PL_strdup(tmp); 209 } In the above code we are doing a strdup giving publicKey_s. Are we freeing that string anywhere? ----- Original Message ----- From: "Christina Fu" <cfu@redhat.com> To: "pki-devel" <pki-devel@redhat.com> Sent: Monday, September 24, 2012 4:06:32 PM Subject: [Pki-devel] Review Request: TMS ECC Key Recovery https://fedorahosted.org/pki/ticket/252 - TMS - ECC Key Recovery patch for review: https://fedorahosted.org/pki/attachment/ticket/252/TMS-ECC-Recovery.patchFor... This patch provides code to allow ECC key recovery in the TMS environment. It was tested to work with tpsclient. The key injection part of implementation for the actual smart card tokens is scheduled to be done in #235 at a later time. I can do a demo tomorrow in office. thanks, Christina _______________________________________________ Pki-devel mailing list Pki-devel@redhat.com https://www.redhat.com/mailman/listinfo/pki-devel
Hi Jack, Thanks for the review. Please find my response in-line below. thanks, Christina On 09/26/2012 06:39 PM, John Magne wrote:
Couple Questions:
In the following piece of the patch:
+ SECITEM_FreeItem(&der, PR_FALSE); 81 SECKEY_DestroySubjectPublicKeyInfo(spki); 82
It looks like both are executed , even when the call to spki = SECKEY_DecodeDERSubjectPublicKeyInfo(&der);
Fails and spki is NULL. Does the DestroySubjectPublicKeyInfo handle this? Also Will "der" be in a state to harm the call to SECITEM_FreeItem?
DestroySubjectPublicKeyInfo checks before it frees so it is okay if spki happens to be null: void SECKEY_DestroySubjectPublicKeyInfo(CERTSubjectPublicKeyInfo *spki) { if (spki && spki->arena) { PORT_FreeArena(spki->arena, PR_FALSE); } } Also, the content of der is copied over in SECKEY_DecodeDERSubjectPublicKeyInfo(), so once the call is done, the code has no use for der any more, hence the SECITEM_FreeItem after. Also, SECITEM_FreeItem checks for null before it frees as well, so it's safe to call.
Index: tps/src/engine/RA.cpp 187 =================================================================== 188 --- tps/src/engine/RA.cpp (revision 2497) 189 +++ tps/src/engine/RA.cpp (working copy) 190 @@ -1238,7 +1238,10 @@ 191 goto loser; 192 } else { 193 RA::Debug(LL_PER_PDU, "RecoverKey", "got public key =%s", tmp); 194 - *publicKey_s = PL_strdup(tmp); 195 + char *tmp_publicKey_s = PL_strdup(tmp); 196 + Buffer *decodePubKey = Util::URLDecode(tmp_publicKey_s); 197 + *publicKey_s = 198 + BTOA_DataToAscii(decodePubKey->getBuf(), decodePubKey->getLen()); 199 } 200 201 tmp = NULL; 202 @@ -1256,7 +1259,7 @@ 203 RA::Error(LL_PER_PDU, "RecoverKey", 204 "did not get iv_param for recovered key in DRM response"); 205 } else { 206 - RA::Debug(LL_PER_PDU, "ServerSideKeyGen", "got iv_param for recovered key =%s", tmp); 207 + RA::Debug(LL_PER_PDU, "RecoverKey", "got iv_param for recovered key =%s", tmp); 208 *ivParam_s = PL_strdup(tmp); 209 }
In the above code we are doing a strdup giving publicKey_s.
Are we freeing that string anywhere?
I introduced two variables there. Yes, I should free them.
----- Original Message ----- From: "Christina Fu"<cfu@redhat.com> To: "pki-devel"<pki-devel@redhat.com> Sent: Monday, September 24, 2012 4:06:32 PM Subject: [Pki-devel] Review Request: TMS ECC Key Recovery
https://fedorahosted.org/pki/ticket/252 - TMS - ECC Key Recovery
patch for review: https://fedorahosted.org/pki/attachment/ticket/252/TMS-ECC-Recovery.patchFor...
This patch provides code to allow ECC key recovery in the TMS environment. It was tested to work with tpsclient. The key injection part of implementation for the actual smart card tokens is scheduled to be done in #235 at a later time.
I can do a demo tomorrow in office.
thanks, Christina
_______________________________________________ Pki-devel mailing list Pki-devel@redhat.com https://www.redhat.com/mailman/listinfo/pki-devel
participants (2)
-
Christina Fu -
John Magne