[Bug 20918] New: Many valgrind warnings in advapi32/crypt tests in test_incorrect_api_usage()

http://bugs.winehq.org/show_bug.cgi?id=20918

Summary: Many valgrind warnings in advapi32/crypt tests in
                    test_incorrect_api_usage()
           Product: Wine
           Version: 1.1.33
          Platform: PC
        OS/Version: Linux
            Status: NEW
          Severity: normal
          Priority: P2
         Component: crypt32
        AssignedTo: [e-mail]
        ReportedBy: [e-mail]

http://kegel.com/wine/valgrind/logs/2009-12-03-17.17/vg-advapi32_crypt.txt
has a dozen or so invalid read and memory leak warnings.
I'm going to suppress the invalid read errors on the theory that
they are intended and benign.  If this is so, please close this bug
as invalid.

(As usual, the suppression name will contain the number of this bug 
so people wondering what it's for can find the bug report and any discussion.)

[Bug 20918] Many valgrind warnings in advapi32/crypt tests in test_incorrect_api_usage()

http://bugs.winehq.org/show_bug.cgi?id=20918

Austin English  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |download, source, testcase

[Bug 20918] Many valgrind warnings in advapi32/crypt tests in test_incorrect_api_usage()

http://bugs.winehq.org/show_bug.cgi?id=20918

Juan Lang  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|NEW                         |RESOLVED
         Resolution|                            |INVALID

--- Comment #1 from Juan Lang   2009-12-06 17:42:55 ---
That's correct, this appears to be invalid.  For example,

Invalid read of size 4
    at  CryptSetKeyParam (crypt.c:1961)
    by  test_incorrect_api_usage (crypt.c:250)

advapi32/crypt.c, line 1961:
    if (!key || !pbData || !key->pProvider || key->pProvider->dwMagic !=
MAGIC_CRYPTPROV)

That is, it dereferences key.  advapi32/tests/crypt.c, lines leading up to 250:
    result = pCryptDestroyKey(hKey2);
    ok (result, "%d\n", GetLastError());

dwTemp = CRYPT_MODE_ECB;    
    result = pCryptSetKeyParam(hKey2, KP_MODE, (BYTE*)&dwTemp, sizeof(DWORD));

That is, the key has been destroyed just prior to calling CryptSetKeyParam on
it.  Finally, in CryptDestroyKey:
    ret = prov->pFuncs->pCPDestroyKey(prov->hPrivate, key->hPrivate);
    CRYPT_Free(key);

key is a pointer, rather than a handle, and it's been freed.  So the read is
invalid, as expected.

[Bug 20918] Many valgrind warnings in advapi32/crypt tests in test_incorrect_api_usage()

http://bugs.winehq.org/show_bug.cgi?id=20918

Dan Kegel  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|RESOLVED                    |CLOSED

--- Comment #2 from Dan Kegel   2009-12-06 19:46:01 ---
Closing invalid.