Wow, 2003 is almost all green... only rpcrt4:server failing

[Dan Kegel]

The wine test suite is making great progress towards
passing on all platforms.

http://test.winehq.org/data/tests/rpcrt4:server.html
seems to be the sore thumb at the moment; it passes
on XP and Wine, but fails everywhere else.
I think this is the test that fails on the most platforms.
- Dan

[Paul Vriens]

Dan Kegel wrote:
> The wine test suite is making great progress towards
> passing on all platforms.
>
> http://test.winehq.org/data/tests/rpcrt4:server.html
> seems to be the sore thumb at the moment; it passes
> on XP and Wine, but fails everywhere else.
> I think this is the test that fails on the most platforms.
> - Dan

These failures on W2K3 started happening (or at least are more prominent) after 
the upgrade of winehq.org. Must be something with uninitialized stuff I presume.

[Austin English]

On Sun, Mar 8, 2009 at 9:39 AM, Dan Kegel  wrote:
> The wine test suite is making great progress towards
> passing on all platforms.
>
> http://test.winehq.org/data/tests/rpcrt4:server.html
> seems to be the sore thumb at the moment; it passes
> on XP and Wine, but fails everywhere else.
> I think this is the test that fails on the most platforms.
> - Dan

Also fails on OpenSolaris:
server.c:1302: Test failed: RpcServerUseProtseqEp(ncacn_ip_tcp) failed
with status 1703
server.c:1311: Using RpcServerRegisterIfEx
server.c:1327: Tests skipped: tcp_basic tests skipped due to earlier failure
server: 89 tests executed (0 marked as todo, 0 failures), 0 skipped.
server.c:1346: Test marked todo: RpcMgmtWaitServerListening failed
with status 1715
server: 37 tests executed (1 marked as todo, 1 failure), 1 skipped.

[Paul Vriens]

Dan Kegel wrote:
> The wine test suite is making great progress towards
> passing on all platforms.
>
> http://test.winehq.org/data/tests/rpcrt4:server.html
> seems to be the sore thumb at the moment; it passes
> on XP and Wine, but fails everywhere else.
> I think this is the test that fails on the most platforms.
> - Dan

Just some thing I've noticed. W2K3 had 13 failures before yesterday's commit 
round. Now it's down to 3. This most likely has to do with Rob's changes to WIDL 
as that's the only relevant area touched with yesterday's round. So maybe a few 
more of those will get things to zero ;)

The current w2k3 logs show two crashes in the child processes with 
RPC_X_NULL_REF_POINTER.

First 2 crashes are when get_filename() is called in basic_tests():

#if 0
   str = get_filename();
   ok(!strcmp(str, __FILE__), "get_filename() returned %s instead of %s\n", str, 
__FILE__);
   midl_user_free(str);
#endif

After commenting out that test block I get 2 new ones when get_s123() is called 
in pointer_tests():

#if 0
   s123 = get_s123();
   ok(s123->f1 == 1 && s123->f2 == 2 && s123->f3 == 3, "RPC get_s123\n");
   MIDL_user_free(s123);
#endif

[Paul Vriens]

Paul Vriens wrote:
> Dan Kegel wrote:
> > The wine test suite is making great progress towards
> > passing on all platforms.
> >
> > http://test.winehq.org/data/tests/rpcrt4:server.html
> > seems to be the sore thumb at the moment; it passes
> > on XP and Wine, but fails everywhere else.
> > I think this is the test that fails on the most platforms.
> > - Dan
>
> Just some thing I've noticed. W2K3 had 13 failures before yesterday's 
> commit round. Now it's down to 3. This most likely has to do with Rob's 
> changes to WIDL as that's the only relevant area touched with 
> yesterday's round. So maybe a few more of those will get things to zero ;)
>
> The current w2k3 logs show two crashes in the child processes with 
> RPC_X_NULL_REF_POINTER.
>
> First 2 crashes are when get_filename() is called in basic_tests():
>
> #if 0
>   str = get_filename();
>   ok(!strcmp(str, __FILE__), "get_filename() returned %s instead of 
> %s\n", str, __FILE__);
>   midl_user_free(str);
> #endif
>
> After commenting out that test block I get 2 new ones when get_s123() is 
> called in pointer_tests():
>
> #if 0
>   s123 = get_s123();
>   ok(s123->f1 == 1 && s123->f2 == 2 && s123->f3 == 3, "RPC get_s123\n");
>   MIDL_user_free(s123);
> #endif

Oh, and these crashes now seem to happen on all NT4+ platforms.