Re: Undocumented behaviour in ReqQueryValueEx when 'count' and 'type'point to the same address

[Dmitry Timoshkov]

"Sam Dennis"  wrote:
> - size = 4;
> - ret = RegQueryValueExA(hkey_main, "BIN32", NULL, &size, buffer, &size);
> - ok(ret == ERROR_SUCCESS, "expected ERROR_SUCCESS, got %d\n", ret);

What this test is supposed to test?

Undocumented behaviour in ReqQueryValueEx when 'count' and 'type'point to the same address

[Juan Lang]

> > - size = 4;
> > - ret = RegQueryValueExA(hkey_main, "BIN32", NULL, &size, buffer, &size);
> > - ok(ret == ERROR_SUCCESS, "expected ERROR_SUCCESS, got %d\n", ret);
>
> What this test is supposed to test?

It tests the order in which type and count are checked.  When type and
count are (erroneously) passed the same pointer, RegQueryValueExA
still succeeds on Windows.  It fails on Wine, due to type being set to
0 early in the function, which caused a regression in Sam's (buggy)
program.
--Juan

Undocumented behaviour in ReqQueryValueEx when 'count' and'type'point to the same address

[Dmitry Timoshkov]

"Juan Lang"  wrote:
> > > - size = 4;
> > > - ret = RegQueryValueExA(hkey_main, "BIN32", NULL, &size, buffer, &size);
> > > - ok(ret == ERROR_SUCCESS, "expected ERROR_SUCCESS, got %d\n", ret);
> >
> > What this test is supposed to test?
>
> It tests the order in which type and count are checked.  When type and
> count are (erroneously) passed the same pointer, RegQueryValueExA
> still succeeds on Windows.  It fails on Wine, due to type being set to
> 0 early in the function, which caused a regression in Sam's (buggy)
> program.

Shouldn't the test also check for returned values in size and buffer?

Undocumented behaviour in ReqQueryValueEx when 'count' and'type'point to the same address

[Juan Lang]

> Shouldn't the test also check for returned values in size and buffer?

Yes, I expect so.
--Juan