82227 - Signed scripts use new PSM interfaces.

Status: VERIFIED FIXED

(Core Graveyard :: Security: UI, defect)

Description

[David P. Drinan]

Change the mozilla signed scripts components to use the new PSM interfaces e.g. 
nsIX509Cert etc. 

Patch to follow soon.

Comment 1

[Mitchell Stoltz (not reading bugmail)]

This should be in 0.9.1 so that users of the beta will be able to use signed
scripts.

Comment 2

[David P. Drinan]

Attachment: Patch. Javi and mstoltz, please review.

Comment 3

[Mitchell Stoltz (not reading bugmail)]

r=mstoltz. David, can you check this in? I'll help you get approval.

Comment 4

[David P. Drinan]

Adding blizzard to cc-list.

Chris,

Please sr this bug.

Thanks!

Comment 5

[Christopher Blizzard (:blizzard)]

+    //-- Create a certificate principal with id and organization data
+    PRUnichar* fingerprint;
+    rv2 = pCert->GetSha1Fingerprint(&fingerprint);
+    nsCAutoString fingerprintStr;
+    fingerprintStr.AssignWithConversion(fingerprint);
+    PR_FREEIF(fingerprint);

Are you sure that the fingerprint can always be represented as a C string?  You
could have some data lossage here.

Other than that sr=blizzard

Comment 6

[Brendan Eich [:brendan]]

Please do not commit tabs in files:

+static void ContentCallback(void *arg, 
+                                           const char *buf,
+                                           unsigned long len)
+{
+}
+
+static PK11SymKey * GetDecryptKeyCallback(void *arg, 
+                                                 SECAlgorithmID *algid)
+{
+       return nsnull;
+}
+

etc.

/be

Comment 7

[David P. Drinan]

Fixed checked in.

Comment 8

[Mitchell Stoltz (not reading bugmail)]

I checked with javi and confirmed that we're not losing any data by converting
the fingerprint back to a one-byte string. The fingerprint is a hex string like
"38:A3:b6:93" and so on; no two-byte characters. 

Comment 9

[Mike Kaply [:mkaply]]

I believe Mac is broke on this change (although bonsai 
doesn't show it going into that build)

nsJar.cpp uses these hash interfaces, and now is passing 
PRUint32s instead of the new HashContext stuff.

Comment 10

[Christopher Blizzard (:blizzard)]

Crap, I can't believe that I missed those tabs.

Comment 11

[David P. Drinan]

I forgot to commit one file during checkin. It's checked in now.

Comment 12

[John Unruh]

Verified.