[Bug 38327] New: Uninitialized matrix

https://bugs.winehq.org/show_bug.cgi?id=38327

Bug ID: 38327
           Summary: Uninitialized matrix
           Product: Wine
           Version: 1.7.39
          Hardware: x86
                OS: Linux
            Status: UNCONFIRMED
          Severity: normal
          Priority: P2
         Component: directx-d3d
          Assignee: [e-mail]
          Reporter: [e-mail]
      Distribution: ---

I got warning
-----
glsl_shader.c: In function ‘shader_glsl_load_constants’:
glsl_shader.c:799: warning: ‘temp._44’ is used uninitialized in this function
glsl_shader.c:744: note: ‘temp._44’ was declared here
glsl_shader.c:799: warning: ‘temp._43’ is used uninitialized in this function
glsl_shader.c:744: note: ‘temp._43’ was declared here
glsl_shader.c:799: warning: ‘temp._42’ is used uninitialized in this function
glsl_shader.c:744: note: ‘temp._42’ was declared here
glsl_shader.c:799: warning: ‘temp._41’ is used uninitialized in this function
glsl_shader.c:744: note: ‘temp._41’ was declared here
glsl_shader.c:799: warning: ‘temp._34’ is used uninitialized in this function
glsl_shader.c:744: note: ‘temp._34’ was declared here
glsl_shader.c:799: warning: ‘temp._24’ is used uninitialized in this function
glsl_shader.c:744: note: ‘temp._24’ was declared here
glsl_shader.c:799: warning: ‘temp._14’ is used uninitialized in this function
glsl_shader.c:744: note: ‘temp._14’ was declared here
----
this is a function
---
/* Taken and adapted from Mesa. */
static BOOL invert_matrix_3d(struct wined3d_matrix *out, const struct
wined3d_matrix *in)
{
    float pos, neg, t, det;
    struct wined3d_matrix temp;
...
    *out = temp;
---
The function is for matrix 3x3 while wined3d_matrix is 4x4.
But local variable (temp) is not initialized and may contain arbitrary values.
Should the matrix to be zeroed explicitly or there must be function to invert
4x4 matrix?

[Bug 38327] Uninitialized matrix

https://bugs.winehq.org/show_bug.cgi?id=38327

Matteo Bruni  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |[e-mail]

--- Comment #1 from Matteo Bruni  ---
Those values are not really used so it doesn't matter in practice. I don't get
those warnings, what compiler / version are you using?

[Bug 38327] Uninitialized matrix

https://bugs.winehq.org/show_bug.cgi?id=38327

--- Comment #2 from Sergey Isakov  ---
(In reply to Matteo Bruni from comment #1)
> Those values are not really used so it doesn't matter in practice. I don't
> get those warnings, what compiler / version are you using?

Apple's gcc-4.2.1 with -Os optimization. Without optimization there are no such
warning. (?)
Anyway those values are influences on inverted matrix. If you want I can write
invert_matrix_4x4.

[Bug 38327] Uninitialized matrix

https://bugs.winehq.org/show_bug.cgi?id=38327

Matteo Bruni  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
                 OS|Linux                       |Mac OS X

--- Comment #3 from Matteo Bruni  ---
(In reply to Sergey Isakov from comment #2)
> (In reply to Matteo Bruni from comment #1)
> > Those values are not really used so it doesn't matter in practice. I don't
> > get those warnings, what compiler / version are you using?
>
> Apple's gcc-4.2.1 with -Os optimization. Without optimization there are no
> such warning. (?)

What does happen with the default -O2?
> Anyway those values are influences on inverted matrix. If you want I can
> write invert_matrix_4x4.

They are ignored on purpose, I'm following the GL behavior. I just thought
about one testcase which should confirm that D3D does the same, I'm going to
test that and send a fix if I'm wrong.

[Bug 38327] Uninitialized matrix

https://bugs.winehq.org/show_bug.cgi?id=38327

--- Comment #4 from Sergey Isakov  ---
Yes, I see really 3x3 used.

[Bug 38327] Uninitialized matrix

https://bugs.winehq.org/show_bug.cgi?id=38327

--- Comment #5 from Sergey Isakov  ---
(In reply to Matteo Bruni from comment #3)
> (In reply to Sergey Isakov from comment #2)
> > (In reply to Matteo Bruni from comment #1)
> > > Those values are not really used so it doesn't matter in practice. I don't
> > > get those warnings, what compiler / version are you using?
> >
> > Apple's gcc-4.2.1 with -Os optimization. Without optimization there are no
> > such warning. (?)
>
> What does happen with the default -O2?

I see in WIKI
---
J. Huk: [feb 10] "-O2" is consider safe for most software "-O3" is not - from
my experience, wine will [crash] with most apps, when compiled with "-O3"
---

But this is very old advice.
For me -Os works fine in all cases.

[Bug 38327] Uninitialized matrix

https://bugs.winehq.org/show_bug.cgi?id=38327

--- Comment #6 from Matteo Bruni  ---
I'm just asking if you still get warnings with the default -O2.

[Bug 38327] Uninitialized matrix

https://bugs.winehq.org/show_bug.cgi?id=38327

--- Comment #7 from Sergey Isakov  ---
Never saw before

[Bug 38327] Uninitialized matrix

https://bugs.winehq.org/show_bug.cgi?id=38327

Sergey Isakov  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |RESOLVED
         Resolution|---                         |WONTFIX

--- Comment #8 from Sergey Isakov  ---
Not important

[Bug 38327] Uninitialized matrix

https://bugs.winehq.org/show_bug.cgi?id=38327

Nikolay Sivov  changed:

What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|RESOLVED                    |CLOSED
         Resolution|WONTFIX                     |INVALID

--- Comment #9 from Nikolay Sivov  ---
Closing.

[Bug 38327] Uninitialized matrix

https://bugs.winehq.org/show_bug.cgi?id=38327

--- Comment #10 from Sergey Isakov  ---
I think status of this bug must be WONT FIX because compiler warnings present.