[PATCH v2] gdi32: Match second_name during font search.

[Huw Davies]

On Fri, May 14, 2021 at 07:17:41PM +0800, Jiangyi Chen wrote:

> From 570f2a9cf61f74978ee0da5915028f6c9a457dfb Mon Sep 17 00:00:00 2001
> From: Jiangyi Chen <chenjiangyi at uniontech.com>
> Date: Fri, 7 May 2021 14:01:13 +0800
> Subject: [PATCH v2] gdi32: Match second_name during font search.
> 
> For example, there is a substituting MS Shell Dlg -> SimSun.the first parameter of GdipCreateFontFamilyFromName is MS Shell Dlg,then second_name is needed to find SimSun.Otherwise, it may cause GdipDrawString to crash indirectly.
> 
> Signed-off-by: Jiangyi Chen <chenjiangyi at uniontech.com>
> ---
>  dlls/gdi32/font.c | 3 ++-
>  1 file changed, 2 insertions(+), 1 deletion(-)
> 
> diff --git a/dlls/gdi32/font.c b/dlls/gdi32/font.c
> index 2d2831ec725..24e6d82c100 100644
> --- a/dlls/gdi32/font.c
> +++ b/dlls/gdi32/font.c
> @@ -2772,6 +2772,7 @@ static BOOL family_matches( struct gdi_font_family *family, const WCHAR *face_na
>      struct gdi_font_face *face;
>  
>      if (!facename_compare( face_name, family->family_name, LF_FACESIZE - 1 )) return TRUE;
> +    if (!facename_compare( face_name, family->second_name, LF_FACESIZE - 1 )) return TRUE;
>      LIST_FOR_EACH_ENTRY( face, get_family_face_list(family), struct gdi_font_face, entry )
>          if (!facename_compare( face_name, face->full_name, LF_FACESIZE - 1 )) return TRUE;
>      return FALSE;
> @@ -2879,7 +2880,7 @@ static BOOL CDECL font_EnumFonts( PHYSDEV dev, LOGFONTW *lf, FONTENUMPROCW proc,
>              if (!family_matches(family, face_name)) continue;
>              LIST_FOR_EACH_ENTRY( face, get_family_face_list(family), struct gdi_font_face, entry )
>              {
> -                if (!face_matches( family->family_name, face, face_name )) continue;
> +                if (!face_matches( family->family_name, face, face_name ) && !face_matches( family->second_name, face, face_name )) continue;
>                  if (!enum_face_charsets( family, face, enum_charsets, count, proc, lparam, orig_name ))
>                      return FALSE;
>  	    }

A test for this would be nice (based on the gdi32 api, not gdiplus).

Also, if gdiplus is crashing, this suggests there is a bug there too.

Huw.