Re: ntdll: Initialize APC call and result variables to suppress valgrind warnings

[Alexandre Julliard]

Dmitry Timoshkov <dmitry(a)codeweavers.com> writes:
> Alexandre said that it's OK to use memset for server requests because of
> backwards compatibility.

It's OK for requests, but why do you want to initialize the result
structure too?

-- 
Alexandre Julliard
julliard(a)winehq.org

ntdll: Initialize APC call and result variables to suppressvalgrind warnings

[Dmitry Timoshkov]

"Alexandre Julliard" <julliard(a)winehq.org> wrote:
> It's OK for requests, but why do you want to initialize the result
> structure too?

NTDLL_queue_process_apc for a local process calls invoke_apc() which
initializes the result to 0 before proceeding. Perhaps memset() should
be added for a not local case as well, or memset() should be removed
from invoke_apc() and replaced by explicit initialization as done in
my patch. What's your opinion?

-- 
Dmitry.

ntdll: Initialize APC call and result variables to suppressvalgrind warnings

[Alexandre Julliard]

"Dmitry Timoshkov" <dmitry(a)codeweavers.com> writes:
> "Alexandre Julliard" <julliard(a)winehq.org> wrote:
> > It's OK for requests, but why do you want to initialize the result
> > structure too?
>
> NTDLL_queue_process_apc for a local process calls invoke_apc() which
> initializes the result to 0 before proceeding. Perhaps memset() should
> be added for a not local case as well, or memset() should be removed
> from invoke_apc() and replaced by explicit initialization as done in
> my patch. What's your opinion?

In the local case there's no server call, so there's no need for a
memset. In the remote case there's a need for a memset in the target
process, i.e. in invoke_apc. A memset in the source process is always
useless since it never sends the result anywhere.

-- 
Alexandre Julliard
julliard(a)winehq.org