Re: Shell32: Reimplement SHGetFolder functions

[Alexandre Julliard]

Juan Lang  writes:
> +static void loadShell32(void)
> +{
> - if (hShell32)
> - {
> -
>    if (pMalloc)
> -
>    {
> -
>        pMalloc->lpVtbl->Release(pMalloc);
> -
>        pMalloc = NULL;
> -
>    }
> -
>    /* until it's really unloaded: */
> -
>    while (FreeLibrary(hShell32))
> -
>        ;

I'm not sure what you are trying to do here, but this causes an
infinite loop on XP (and it probably should under Wine too, since the
test binary imports shell32). Also some tests fail on XP:

shellpath.c:433: Test failed: SHGetFolderPath returned 0x00000000, expected 0x80070002
shellpath.c:438: Test failed: SHGetFolderLocation returned 0x00000000, expected E_FAIL
shellpath.c:442: Test failed: SHGetSpecialFolderPath succeeded, expected failure
shellpath.c:447: Test failed: SHGetFolderLocation returned 0x00000000, expected E_FAIL

Shell32: Reimplement SHGetFolder functions

[Juan Lang]

--- Alexandre Julliard  wrote:
> > +static void loadShell32(void)
> > +{
> > - if (hShell32)
> > - {
> > -
> >    if (pMalloc)
> > -
> >    {
> > -
> >        pMalloc->lpVtbl->Release(pMalloc);
> > -
> >        pMalloc = NULL;
> > -
> >    }
> > -
> >    /* until it's really unloaded: */
> > -
> >    while (FreeLibrary(hShell32))
> > -
> >        ;
>
> I'm not sure what you are trying to do here, but
> this causes an
> infinite loop on XP (and it probably should under
> Wine too, since the
> test binary imports shell32).

Really?  Odd, I first wrote this under XP.  I'm trying
to force shell32 to unload; native shell32 caches
values you've already queried and won't reinitialize
them till you reload it.

But thanks, I'll test some more.
--Juan

_______________________________
Do you Yahoo!?
Declare Yourself - Register online to vote today!
http://vote.yahoo.com

Shell32: Reimplement SHGetFolder functions

[Juan Lang]

--- Alexandre Julliard  wrote:
> > +static void loadShell32(void)
> > +{
> > - if (hShell32)
> > - {
> > -
> >    if (pMalloc)
> > -
> >    {
> > -
> >        pMalloc->lpVtbl->Release(pMalloc);
> > -
> >        pMalloc = NULL;
> > -
> >    }
> > -
> >    /* until it's really unloaded: */
> > -
> >    while (FreeLibrary(hShell32))
> > -
> >        ;
>
> I'm not sure what you are trying to do here, but
> this causes an
> infinite loop on XP (and it probably should under
> Wine too, since the
> test binary imports shell32).

I'm unable to reproduce the infinite loop under XP. 
As far as I can tell this shouldn't produce one,
either: since I don't modify the value of hShell32,
FreeLibrary should eventually fail when the refcount
on the library reaches 0 and the library is actually
unloaded.

There are other problems in WinXP that I'm working on
(ILFindLastID doesn't appear to be exported by name,
for example).  If you're sure an infinite loop is the
symptom, though, I'll resubmit without this test. 
(Pity, it was probably the most interesting test.)

--Juan

__________________________________
Do you Yahoo!?
Yahoo! Mail Address AutoComplete - You start. We finish.
http://promotions.yahoo.com/new_mail

Shell32: Reimplement SHGetFolder functions

[Alexandre Julliard]

Juan Lang  writes:
> I'm unable to reproduce the infinite loop under XP. 
> As far as I can tell this shouldn't produce one,
> either: since I don't modify the value of hShell32,
> FreeLibrary should eventually fail when the refcount
> on the library reaches 0 and the library is actually
> unloaded.

There is no guarantee of that, there are many cases where a dll cannot
be unloaded, for instance if it contains TLS data. You'll need to
spawn a child process if you want to reliably test the shell32 init
code.