1342434 - PK11_Derive*() doesn't set the failure reason

Status: UNCONFIRMED

(NSS :: Libraries, defect, P3)

Description

[Andrew Cagney]

User Agent: Mozilla/5.0 (X11; Fedora; Linux x86_64; rv:51.0) Gecko/20100101 Firefox/51.0
Build ID: 20170126153126

Steps to reproduce:

Call PK11_Derive() with bad arguments.

(I'm guessing that hg's "default" means "trunk")


Actual results:

As expected PK11_Derive() returns NULL.
Unfortunately, PORT_GetError() returns 0 so there's no hint as to why the arguments are bad.


Expected results:

PK11_Derive() and co should be calling:

   PORT_SetError(PK11_MapError(crv));

so that the underlying error reason is saved.

Comment 1

[Franziskus Kiefer [:franziskus]]

PK11_Derive calls PORT_SetError (sometimes). Do you have an example that doesn't set the error?

Comment 2

[Andrew Cagney]

Anything that causes the underlying ->C_DeriveKey() call to fail should have this problem (the status isn't set after an error return).  The case I encountered was:

--- Subject: PK11_Derive CKM_CONCATENATE_BASE_AND_KEY fails when KEY is from from CKM_ECDH1_DERIVE

I'm trying to understand why the PK11_Derive call:

    CK_OBJECT_HANDLE key_handle = PK11_GetSymKeyHandle(key);
    SECItem key_param = {
        .data = (unsigned char*)&key_handle,
        .len = sizeof(key_handle)
    };
    CK_ATTRIBUTE_TYPE operation = CKA_DERIVE;
    int key_size = 0;
    PK11SymKey *result = PK11_Derive(base_key,
CKM_CONCATENATE_BASE_AND_KEY, &key_param, CKM_MD5_KEY_DERIVATION,
                     operation, key_size);

works when "key" came from CKM_DH_PKCS_DERIVE but fails when the key
came from (CKM_ECDH1_DERIVE,CKD_NULL).  Debugging suggests that the
operation is rejected in lib/softoken/pkcs11c.c by the code:

            att2 = sftk_FindAttribute(newKey, CKA_VALUE);
            if (att2 == NULL) {
                sftk_FreeObject(newKey);
                crv = CKR_KEY_HANDLE_INVALID;
                break;
            }

I've found a work-around is to first CKM_EXTRACT_KEY_FROM_KEY and then
use that :-/

Is this expected?  And is there a clean way to detect a key like ECDH
that will cause problems.