From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:49530) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1cot6y-0003Uy-0E for qemu-devel@nongnu.org; Fri, 17 Mar 2017 10:44:13 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1cot6u-0008Mn-46 for qemu-devel@nongnu.org; Fri, 17 Mar 2017 10:44:12 -0400 Received: from mx1.redhat.com ([209.132.183.28]:57624) by eggs.gnu.org with esmtps (TLS1.0:DHE_RSA_AES_256_CBC_SHA1:32) (Exim 4.71) (envelope-from ) id 1cot6t-0008M8-Rt for qemu-devel@nongnu.org; Fri, 17 Mar 2017 10:44:08 -0400 Date: Fri, 17 Mar 2017 10:44:03 -0400 (EDT) From: Frediano Ziglio Message-ID: <1138905803.3616945.1489761843215.JavaMail.zimbra@redhat.com> In-Reply-To: <6a76ca9d-10cf-1de9-7430-5b3eb5e15aa2@amsat.org> References: <1489738282-28383-1-git-send-email-kraxel@redhat.com> <6a76ca9d-10cf-1de9-7430-5b3eb5e15aa2@amsat.org> MIME-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Transfer-Encoding: quoted-printable Subject: Re: [Qemu-devel] [PATCH] ui/egl-helpers: fix egl 1.5 display init List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Philippe =?utf-8?Q?Mathieu-Daud=C3=A9?= Cc: =?utf-8?Q?Marc-Andr=C3=A9?= Lureau , Gerd Hoffmann , qemu-devel@nongnu.org, hdegoede@redhat.com >=20 > On 03/17/2017 05:53 AM, Marc-Andr=C3=A9 Lureau wrote: > > On Fri, Mar 17, 2017 at 12:12 PM Gerd Hoffmann wrot= e: > > > >> Unfortunaly switching to getPlatformDisplayEXT isn't as easy as > >> implemented by 0ea1523fb6703aa0dcd65e66b59e96fec028e60a. See the > >> longish comment for the complete story. > >> > >> Cc: Frediano Ziglio > >> Suggested-by: Hans de Goede > >> Signed-off-by: Gerd Hoffmann > >> --- > > > > > > > > Reviewed-by: Marc-Andr=C3=A9 Lureau > > >=20 > Reviewed-by: Philippe Mathieu-Daud=C3=A9 >=20 > >> ui/egl-helpers.c | 53 > >> +++++++++++++++++++++++++++++++++++++++++++++++------ > >> 1 file changed, 47 insertions(+), 6 deletions(-) > >> > >> diff --git a/ui/egl-helpers.c b/ui/egl-helpers.c > >> index 584dd1b..48686cf 100644 > >> --- a/ui/egl-helpers.c > >> +++ b/ui/egl-helpers.c > >> @@ -192,6 +192,51 @@ EGLSurface qemu_egl_init_surface_x11(EGLContext e= ctx, > >> Window win) > >> > >> /* ------------------------------------------------------------------= ---- > >> */ > >> > >> +/* > >> + * Taken from glamor_egl.h from the Xorg xserver, which is MIT licens= ed > >> + * > >> + * Create an EGLDisplay from a native display type. This is a little > >> quirky > >> + * for a few reasons. > >> + * > >> + * 1: GetPlatformDisplayEXT and GetPlatformDisplay are the API you wa= nt > >> to > >> + * use, but have different function signatures in the third argument; > >> this > >> + * happens not to matter for us, at the moment, but it means epoxy wo= n't > >> alias > >> + * them together. > >> + * > >> + * 2: epoxy 1.3 and earlier don't understand EGL client extensions, w= hich > >> + * means you can't call "eglGetPlatformDisplayEXT" directly, as the > >> resolver > >> + * will crash. > >> + * > >> + * 3: You can't tell whether you have EGL 1.5 at this point, because > >> + * eglQueryString(EGL_VERSION) is a property of the display, which we > >> don't > >> + * have yet. So you have to query for extensions no matter what. > >> Fortunately > >> + * epoxy_has_egl_extension _does_ let you query for client extensions= , so > >> + * we don't have to write our own extension string parsing. > >> + * > >> + * 4. There is no EGL_KHR_platform_base to complement the EXT one, th= us > >> one > >> + * needs to know EGL 1.5 is supported in order to use the > >> eglGetPlatformDisplay > >> + * function pointer. > >> + * We can workaround this (circular dependency) by probing for the EG= L > >> 1.5 > >> + * platform extensions (EGL_KHR_platform_gbm and friends) yet it does= n't > >> seem > >> + * like mesa will be able to adverise these (even though it can do EG= L > >> 1.5). > >> + */ > >> +static EGLDisplay qemu_egl_get_display(void *native) > >> +{ > >> +#ifdef EGL_PLATFORM_GBM_MESA > >> + /* In practise any EGL 1.5 implementation would support the EXT > >> extension */ > >> + if (epoxy_has_egl_extension(NULL, "EGL_EXT_platform_base")) { Should not check EGL_MESA_platform_gbm instead? What happens if package is compiled against Mesa but in the system Nvidia OpenGL is installed? In this case EGL_PLATFORM_GBM_MESA is defines, possibly EGL_EXT_platform_base is supported but EGL_PLATFORM_GBM_MESA is not as Nvidia does not support it (so EGL_PLATFORM_GBM_MESA won't work). Maybe Xorg code support also Nvidia case (that does not have gbm) ? > >> + PFNEGLGETPLATFORMDISPLAYEXTPROC getPlatformDisplayEXT =3D > >> + (void *) eglGetProcAddress("eglGetPlatformDisplayEXT"); > >> + if (getPlatformDisplayEXT) { > >> + return getPlatformDisplayEXT(EGL_PLATFORM_GBM_MESA, nativ= e, > >> NULL); > >> + } > >> + } > >> +#endif > >> + > >> + /* fallback */ > >> + return eglGetDisplay(native); > >> +} > >> + > >> int qemu_egl_init_dpy(EGLNativeDisplayType dpy, bool gles, bool debug= ) > >> { > >> static const EGLint conf_att_gl[] =3D { > >> @@ -222,12 +267,8 @@ int qemu_egl_init_dpy(EGLNativeDisplayType dpy, b= ool > >> gles, bool debug) > >> setenv("LIBGL_DEBUG", "verbose", true); > >> } > >> > >> - egl_dbg("eglGetDisplay (dpy %p) ...\n", dpy); > >> -#ifdef EGL_MESA_platform_gbm > >> - qemu_egl_display =3D eglGetPlatformDisplayEXT(EGL_PLATFORM_GBM_ME= SA, > >> dpy, NULL); > >> -#else > >> - qemu_egl_display =3D eglGetDisplay(dpy); > >> -#endif > >> + egl_dbg("qemu_egl_get_display (dpy %p) ...\n", dpy); > >> + qemu_egl_display =3D qemu_egl_get_display(dpy); > >> if (qemu_egl_display =3D=3D EGL_NO_DISPLAY) { > >> error_report("egl: eglGetDisplay failed"); > >> return -1;