[Bug 27177] New: advapi32: CRYPT_UnicodeToANSI use signed int for string size

http://bugs.winehq.org/show_bug.cgi?id=27177

Summary: advapi32: CRYPT_UnicodeToANSI use signed int for
                    string size
           Product: Wine
           Version: unspecified
          Platform: x86
        OS/Version: All
            Status: UNCONFIRMED
          Severity: normal
          Priority: P2
         Component: advapi32
        AssignedTo: [e-mail]
        ReportedBy: [e-mail]

CRYPT_UnicodeToANSI use signed int for string size. It problem, if size is
negative. WideCharToMultiByte does not always work out well if count < 0.
example /dlls/advapi32/tests/crypt.c:560

[Bug 27177] advapi32: CRYPT_UnicodeToANSI use signed int for string size

http://bugs.winehq.org/show_bug.cgi?id=27177

Nikolay Sivov  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
           Severity|normal                      |enhancement

--- Comment #1 from Nikolay Sivov  2011-08-03 17:29:17 CDT ---
What's a problem exactly? WideCharToMultiByte takes signed int for buffer size.

[Bug 27177] advapi32: CRYPT_UnicodeToANSI use signed int for string size

http://bugs.winehq.org/show_bug.cgi?id=27177

--- Comment #2 from Alexandex Yastrebov  2011-08-03 20:21:49 CDT ---
The problem is that some functions pass DWORD(unsigned int) as buffer size,
accepted from the application, but  should correctly handle values ​​like
0xdeadbeaf. Example CryptEnumProvidersA. WideCharToMultiByte however not
obliged to correctly handle a negative buffer size.

[Bug 27177] advapi32: CRYPT_UnicodeToANSI use signed int for string size

http://bugs.winehq.org/show_bug.cgi?id=27177

--- Comment #3 from Dmitry Timoshkov  2011-08-04 00:58:27 CDT ---
Please provide a simple use case to demonstrate the problem.

[Bug 27177] advapi32: CRYPT_UnicodeToANSI use signed int for string size

http://bugs.winehq.org/show_bug.cgi?id=27177

--- Comment #4 from Alexandex Yastrebov  2011-08-04 04:50:15 CDT ---
I'm found bugs in ReactOS with wine tests
http://www.reactos.org/testman/detail.php?id=2537184.  If buffer size is a
0xdeadbeaf, ReactOS WideCharToMultiByte fail. Also Windows realization of
WideCharToMultiByte. If Wine realization not fail, it a problem of Wine.

[Bug 27177] advapi32: CRYPT_UnicodeToANSI use signed int for string size

http://bugs.winehq.org/show_bug.cgi?id=27177

--- Comment #5 from Dmitry Timoshkov  2011-08-04 09:03:00 CDT ---
(In reply to comment #4)
> I'm found bugs in ReactOS with wine tests
> http://www.reactos.org/testman/detail.php?id=2537184.  If buffer size is a
> 0xdeadbeaf, ReactOS WideCharToMultiByte fail. Also Windows realization of
> WideCharToMultiByte. If Wine realization not fail, it a problem of Wine.

Please provide a test case (preferably as a part of Wine test suite)
to show that's a real Wine problem and not something not related (like
a bug in ReactOS).

[Bug 27177] advapi32: CRYPT_UnicodeToANSI use signed int for string size

http://bugs.winehq.org/show_bug.cgi?id=27177

--- Comment #6 from Alexandex Yastrebov  2011-08-04 11:17:04 CDT ---
http://test.winehq.org/data/89dcf229d05d20ee17aff5712d59ba30efe0d1a9/linux_a...
When these bugs will be fixed, errors will be occur the same as that of
ReactOS. Wait.

[Bug 27177] advapi32: CRYPT_UnicodeToANSI use signed int for string size

http://bugs.winehq.org/show_bug.cgi?id=27177

Dmitry Timoshkov  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
          Component|advapi32                    |-unknown
         OS/Version|All                         |other

--- Comment #7 from Dmitry Timoshkov  2011-08-04 11:38:57 CDT ---
(In reply to comment #6)
> http://test.winehq.org/data/89dcf229d05d20ee17aff5712d59ba30efe0d1a9/linux_a...
> When these bugs will be fixed, errors will be occur the same as that of
> ReactOS. Wait.

This sounds like a pure speculation.

[Bug 27177] advapi32: CRYPT_UnicodeToANSI use signed int for string size

http://bugs.winehq.org/show_bug.cgi?id=27177

--- Comment #8 from Dmitry Timoshkov  2011-08-04 11:43:22 CDT ---
(In reply to comment #7)
> (In reply to comment #6)
> > http://test.winehq.org/data/89dcf229d05d20ee17aff5712d59ba30efe0d1a9/linux_a...
> > When these bugs will be fixed, errors will be occur the same as that of
> > ReactOS. Wait.
>
> This sounds like a pure speculation.

Moreover that claim could be interpreted that there is no any bug in Wine.

[Bug 27177] advapi32: CRYPT_UnicodeToANSI use signed int for string size

http://bugs.winehq.org/show_bug.cgi?id=27177

--- Comment #9 from Alexandex Yastrebov  2011-08-04 20:35:13 CDT ---
This code should work incorrectly even in Wine, but it's the same bug, signed
int != DWORD.

char* prov = "Microsoft Base Cryptographic Provider v1.0";
DWORD cbProvName = 0xFFFFFFFF; /* -1 
See http://source.winehq.org/source/dlls/advapi32/crypt.c#L130
*/
DWORD dwProvType;
char szProvName[128];
CryptEnumProvidersA(0, NULL, 0, &dwProvType, szProvName, &cbProvName);

ok(!strcmp(szProvName, prov), "expected %s, got %s\n", prov, szProvName);

[Bug 27177] advapi32: CRYPT_UnicodeToANSI use signed int for string size

http://bugs.winehq.org/show_bug.cgi?id=27177

Dan Kegel  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |RESOLVED
                 CC|                            |[e-mail]
         Resolution|                            |INVALID

--- Comment #10 from Dan Kegel  2011-08-05 01:33:37 CDT ---
Does this affect any real apps?
The test case fragment you gave seems buggy, as you're advertising a longer
buffer than is really present.

Marking invalid for now.  We can reopen if there's evidence that real apps
depend on this.

[Bug 27177] advapi32: CRYPT_UnicodeToANSI use signed int for string size

http://bugs.winehq.org/show_bug.cgi?id=27177

Austin English  changed:

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

--- Comment #11 from Austin English  2011-08-08 15:19:53 CDT ---
Closing.