Re: rsaenh: Revert part of d3c482250a8232e9657ecefb3e61fff0194e3fbe
Juan Lang <juan.lang(a)gmail.com> writes:
Commit d3c482250a8232e9657ecefb3e61fff0194e3fbe added a test that showed that the default provider should accept RC2 keys of up to 128 bits in length. The trouble is the fix was wrong: the default provider in Wine was the base cryptographic provider, but it has been the enhanced cryptographic provider for some time in Windows. The enhanced provider in Wine already supported 128 bit RC2 keys, and the test also used the default provider. Commit 7a6381e852a4daf0bbc77ffcb9da3bd85002e1d8 made the enhanced provider the default, so the test succeeds without this portion of the patch applied.
It fails for me: ../../../tools/runtest -q -P wine -M advapi32.dll -T ../../.. -p advapi32_test.exe.so crypt.c && touch crypt.ok crypt.c:972: Test failed: CryptImportKey error 2148073481 make: *** [crypt.ok] Error 1 -- Alexandre Julliard julliard(a)winehq.org
It fails for me:
../../../tools/runtest -q -P wine -M advapi32.dll -T ../../.. -p advapi32_test.exe.so crypt.c && touch crypt.ok crypt.c:972: Test failed: CryptImportKey error 2148073481 make: *** [crypt.ok] Error 1
Odd. On a clean wineprefix, I presume? --Juan
Juan Lang <juan.lang(a)gmail.com> writes:
It fails for me:
../../../tools/runtest -q -P wine -M advapi32.dll -T ../../.. -p advapi32_test.exe.so crypt.c && touch crypt.ok crypt.c:972: Test failed: CryptImportKey error 2148073481 make: *** [crypt.ok] Error 1
Odd. On a clean wineprefix, I presume?
Yes. On a brand new prefix it succeeds the first time, then fails. -- Alexandre Julliard julliard(a)winehq.org
It fails for me:
The trouble is, the test is invalid. It imports a key using CRYPT_IPSEC_HMAC_KEY, which Wine currently ignores. rsaenh currently doesn't implement HMAC. The chromium tests, on which this test case was built, also checks the resulting output from using this key, and Wine currently fails these tests too (bug 20340.) So the Wine test case that's currently succeeding is misleading: you can import a key 16 bytes in length, but if you use it expecting to use HMAC, you'll get the wrong result. I'll submit tests that show the appropriate key length, but I still think this part of the test should be reverted. --Juan
participants (2)
-
Alexandre Julliard -
Juan Lang