Re: mapi32: initialize session pointer to zero in MAPILogonEx (2nd resend)

[Louis. Lenders]

diff --git a/dlls/mapi32/mapi32_main.c b/dlls/mapi32/mapi32_main.c
index 936d435..5aace9a 100644
--- a/dlls/mapi32/mapi32_main.c
+++ b/dlls/mapi32/mapi32_main.c
@@ -109,6 +109,7 @@ HRESULT WINAPI MAPILogonEx(ULONG_PTR uiparam,
LPWSTR profile,
{
     FIXME("(0x%08lx %s %p 0x%08x %p) Stub\n", uiparam,
           debugstr_w(profile), password, flags, session);
+    *session=0;
     return SUCCESS_SUCCESS;
}
> Are you guaranteed that session is not NULL?

No, no garantueed, but i looked up some code-samples by google, and as
far as i can tell, the current stub for MAPILogonEx will pass back garbage anyway to the application, so my thought was this patch would be rather harmless anyway.
> A test case would help.

I guess you're right about that, thanks anyway for the comments.
--Juan

mapi32: initialize session pointer to zero in MAPILogonEx (2nd resend)

[Juan Lang]

> > Are you guaranteed that session is not NULL?
>
> No, no garantueed, but i looked up some code-samples by google, and as far
> as i can tell, the current stub for MAPILogonEx will pass back garbage
> anyway to the application, so my thought was this patch would be rather
> harmless anyway.

Harmless unless some app passes NULL for session.  A test case that
shows that the function fails when session is NULL, or crashes, would
help.  If it crashes, your patch is fine.  If it fails more
gracefully, you'll need to do so as well.
--Juan

mapi32: initialize session pointer to zero in MAPILogonEx (2nd resend)

[James Hawkins]

On Wed, Oct 8, 2008 at 3:07 PM, Louis. Lenders
 wrote:
> diff --git a/dlls/mapi32/mapi32_main.c b/dlls/mapi32/mapi32_main.c
> index 936d435..5aace9a 100644
> --- a/dlls/mapi32/mapi32_main.c
> +++ b/dlls/mapi32/mapi32_main.c
> @@ -109,6 +109,7 @@ HRESULT WINAPI MAPILogonEx(ULONG_PTR uiparam,
> LPWSTR profile,
> {
>     FIXME("(0x%08lx %s %p 0x%08x %p) Stub\n", uiparam,
>           debugstr_w(profile), password, flags, session);
> - *session=0;
> return SUCCESS_SUCCESS;
>
> }
> > Are you guaranteed that session is not NULL?
>
> No, no garantueed, but i looked up some code-samples by google, and as far
> as i can tell, the current stub for MAPILogonEx will pass back garbage
> anyway to the application, so my thought was this patch would be rather
> harmless anyway.

The current stub doesn't pass back anything to the caller, and it
certainly doesn't crash.  Your change will crash an app that passes a
NULL session pointer to the function.