Re[2]: [PATCH v3] comctl32/imagelist: fix ImageList_Read

[Denis Malikov]

>Вторник, 10 июля 2018, 14:55 +07:00 от Nikolay Sivov <nsivov at codeweavers.com>:
>
>
>
>On 07/10/2018 10:52 AM, Denis Malikov wrote:
>>
>>
>>> Четверг,  5 июля 2018, 22:48 +07:00 от Nikolay Sivov < nsivov at codeweavers.com >:
>>>
>>> On 06/25/2018 04:27 PM, Denis Malikov wrote:
>>>
>>>> Fix for versions x600 and x620 and pointer calculation for mixing image and mask bits.
>>>>
>>>> Tested on *.reg files extracted from:
>>>>    - XP/2003 key HKCU\Software\Microsoft\Windows\CurrentVersion\Explorer\TrayNotify;
>>>>    - Vista/7 key HKEY_CLASSES_ROOT\Local Settings\Software\Microsoft\Windows\CurrentVersion\TrayNotify
>>>>
>>>> Signed-off-by: Denis Malikov <  mdn40000 at mail.ru >
>>>> ---
>>>>    dlls/comctl32/imagelist.c | 16 ++++++++++------
>>>>    1 file changed, 10 insertions(+), 6 deletions(-)
>>>>
>>>> diff --git a/dlls/comctl32/imagelist.c b/dlls/comctl32/imagelist.c
>>>> index a08d60752e..be901462a0 100644
>>>> --- a/dlls/comctl32/imagelist.c
>>>> +++ b/dlls/comctl32/imagelist.c
>>>> @@ -2267,7 +2267,9 @@ HIMAGELIST WINAPI ImageList_Read(IStream *pstm)
>>>>    return NULL;
>>>>        if (ilHead.usMagic != (('L' << 8) | 'I'))
>>>>    return NULL;
>>>> -    if (ilHead.usVersion != 0x101) /* probably version? */
>>>> +    if (ilHead.usVersion != 0x101 &&
>>>> +        ilHead.usVersion != 0x600 && /* XP/2003 version */
>>>> +        ilHead.usVersion != 0x620)   /* Vista/7 version */
>>>>    return NULL;
>>>>
>>>>        TRACE("cx %u, cy %u, flags 0x%04x, cCurImage %u, cMaxImage %u\n",
>>>> @@ -2296,23 +2298,25 @@ HIMAGELIST WINAPI ImageList_Read(IStream *pstm)
>>>>        {
>>>>            DWORD *ptr = image_bits;
>>>>            BYTE *mask_ptr = mask_bits;
>>>> -        int stride = himl->cy * image_info->bmiHeader.biWidth;
>>>> +        int stride = himl->cy * (ilHead.usVersion != 0x101 ? himl->cx : image_info->bmiHeader.biWidth);
>>>> +        int image_step = ilHead.usVersion != 0x101 ? 1 : TILE_COUNT;
>>>> +        int mask_step = ilHead.usVersion != 0x101 ? 4 : 8;
>>>>
>>>>            if (image_info->bmiHeader.biHeight > 0)  /* bottom-up */
>>>>            {
>>>>                ptr += image_info->bmiHeader.biHeight * image_info->bmiHeader.biWidth - stride;
>>>> -            mask_ptr += (image_info->bmiHeader.biHeight * image_info->bmiHeader.biWidth - stride) / 8;
>>>> +            mask_ptr += (image_info->bmiHeader.biHeight * image_info->bmiHeader.biWidth - stride) / mask_step;
>>>>                stride = -stride;
>>>>                image_info->bmiHeader.biHeight = himl->cy;
>>>>            }
>>>>            else image_info->bmiHeader.biHeight = -himl->cy;
>>>>
>>>> -        for (i = 0; i < ilHead.cCurImage; i += TILE_COUNT)
>>>> +        for (i = 0; i < ilHead.cCurImage; i += image_step)
>>>>            {
>>>> -            add_dib_bits( himl, i, min( ilHead.cCurImage - i, TILE_COUNT ),
>>>> +            add_dib_bits( himl, i, min( ilHead.cCurImage - i, image_step ),
>>>>                              himl->cx, himl->cy, image_info, mask_info, ptr, mask_ptr );
>>>>                ptr += stride;
>>>> -            mask_ptr += stride / 8;
>>>> +            mask_ptr += stride / mask_step;
>>>>            }
>>>>        }
>>>>        else
>>> I think that it's possible 'else' branch also needs adjusting, when
>>> something else than ILC_COLOR32 is used. I'm not sure if we have a test
>>> for it, something like "Create -> Save -> check stream output".
>> Yes, actually you right but i have no any data source for testing/debugging non-ILC_COLOR32 branch.
>> --
>It should be easy to create, you can even use wine tests for that - 
>create v6 version object, and do Write() with appropriate format.

But may be it is possible to accept this patch as is? I can not even supposed how long it takes, but this fix i need today for consequent work
--