qemu-devel.nongnu.org archive mirror
 help / color / mirror / Atom feed
From: Max Reitz <mreitz@redhat.com>
To: Gerd Hoffmann <kraxel@redhat.com>, qemu-devel@nongnu.org
Subject: Re: [Qemu-devel] [PATCH 1/9] shaders: initialize vertexes once
Date: Mon, 14 Sep 2015 18:19:16 +0200	[thread overview]
Message-ID: <55F6F384.1040107@redhat.com> (raw)
In-Reply-To: <55F6EF34.4010703@redhat.com>

[-- Attachment #1: Type: text/plain, Size: 5662 bytes --]

On 14.09.2015 18:00, Max Reitz wrote:
> On 09.09.2015 13:20, Gerd Hoffmann wrote:
>> Create a buffer for the vertex data and place vertexes
>> there at initialization time.  Then just use the buffer
>> for each texture blit.
>>
>> Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
>> ---
>>  include/ui/shader.h |  4 +++-
>>  ui/console-gl.c     |  7 ++++++-
>>  ui/shader.c         | 32 +++++++++++++++++++++++++++-----
>>  3 files changed, 36 insertions(+), 7 deletions(-)
>>
>> diff --git a/include/ui/shader.h b/include/ui/shader.h
>> index 8509596..f7d8618 100644
>> --- a/include/ui/shader.h
>> +++ b/include/ui/shader.h
>> @@ -3,7 +3,9 @@
>>  
>>  #include <epoxy/gl.h>
>>  
>> -void qemu_gl_run_texture_blit(GLint texture_blit_prog);
>> +GLuint qemu_gl_init_texture_blit(GLint texture_blit_prog);
>> +void qemu_gl_run_texture_blit(GLint texture_blit_prog,
>> +                              GLint texture_blit_vao);
>>  
>>  GLuint qemu_gl_create_compile_shader(GLenum type, const GLchar *src);
>>  GLuint qemu_gl_create_link_program(GLuint vert, GLuint frag);
>> diff --git a/ui/console-gl.c b/ui/console-gl.c
>> index cb45cf8..baf397b 100644
>> --- a/ui/console-gl.c
>> +++ b/ui/console-gl.c
>> @@ -33,6 +33,7 @@
>>  
>>  struct ConsoleGLState {
>>      GLint texture_blit_prog;
>> +    GLint texture_blit_vao;
>>  };
>>  
>>  /* ---------------------------------------------------------------------- */
>> @@ -47,6 +48,9 @@ ConsoleGLState *console_gl_init_context(void)
>>          exit(1);
>>      }
>>  
>> +    gls->texture_blit_vao =
>> +        qemu_gl_init_texture_blit(gls->texture_blit_prog);
>> +
>>      return gls;
>>  }
>>  
>> @@ -131,7 +135,8 @@ void surface_gl_render_texture(ConsoleGLState *gls,
>>      glClearColor(0.1f, 0.1f, 0.1f, 0.0f);
>>      glClear(GL_COLOR_BUFFER_BIT);
>>  
>> -    qemu_gl_run_texture_blit(gls->texture_blit_prog);
>> +    qemu_gl_run_texture_blit(gls->texture_blit_prog,
>> +                             gls->texture_blit_vao);
>>  }
>>  
>>  void surface_gl_destroy_texture(ConsoleGLState *gls,
>> diff --git a/ui/shader.c b/ui/shader.c
>> index 52a4632..ada1c4c 100644
>> --- a/ui/shader.c
>> +++ b/ui/shader.c
>> @@ -29,21 +29,43 @@
>>  
>>  /* ---------------------------------------------------------------------- */
>>  
>> -void qemu_gl_run_texture_blit(GLint texture_blit_prog)
>> +GLuint qemu_gl_init_texture_blit(GLint texture_blit_prog)
>>  {
>> -    GLfloat in_position[] = {
>> +    static const GLfloat in_position[] = {
>>          -1, -1,
>>          1,  -1,
>>          -1,  1,
>>          1,   1,
>>      };
>>      GLint l_position;
>> +    GLuint vao, buffer;
>> +
>> +    glGenVertexArrays(1, &vao);
>> +    glBindVertexArray(vao);
>> +
>> +    /* this is the VBO that holds the vertex data */
>> +    glGenBuffers(1, &buffer);
>> +    glBindBuffer(GL_ARRAY_BUFFER, buffer);
>> +    glBufferData(GL_ARRAY_BUFFER, sizeof(in_position), in_position,
>> +                 GL_STATIC_DRAW);
>>  
>> -    glUseProgram(texture_blit_prog);
>>      l_position = glGetAttribLocation(texture_blit_prog, "in_position");
>> -    glVertexAttribPointer(l_position, 2, GL_FLOAT, GL_FALSE, 0, in_position);
>> +    glVertexAttribPointer(l_position, 2, GL_FLOAT, GL_FALSE, 0, 0);
>>      glEnableVertexAttribArray(l_position);
>> -    glDrawArrays(GL_TRIANGLE_STRIP, l_position, 4);
>> +
>> +    glBindBuffer (GL_ARRAY_BUFFER, 0);
>> +    glBindVertexArray (0);
>> +    glDeleteBuffers (1, &buffer);
> 
> Although I may be wrong, I don't think glVertexAttribPointer() loads the
> buffer data into the given vertex attribute.
> 
> As far as I can see from the specification regarding vertex array
> objects, they only represent which data is to be used for vertex
> attributes ("All state related to the definition of data used by the
> vertex processor is encapsulated in a vertex array object.").
> 
> glVertexAttribPointer():
> (1) determines the attribute's format,
> (2) binds a vertex attribute index to a shader attribute index,
> (3) binds the data to the vertex attribute.
> 
> Step (3) (glBindVertexBuffer()) does not appear to load the data, but
> only define its origin ("The source of data for a generic vertex
> attribute may be determined by attaching a buffer object to a vertex
> array object with [glBindVertexBuffer()]").
> 
> Therefore, I'm not sure whether deleting the buffer is right here. Maybe
> OpenGL uses reference counting here, too, so it remembers that the
> buffer is still in use by the VAO, and so the glDeleteBuffers()
> operation will only decrease its refcount, but not actually end up
> deleting it. But I don't think so (compare the documentation of
> glDeleteBuffers() with glDeleteShader(); the former says it will delete
> the buffer, whereas the latter explicitly mentions that the shader will
> not be deleted as long as it is attached to a program).
> 
> All in all I don't think we should delete the buffer as long as it is in
> use by the VAO.

Oh, and I just noticed: With glDeleteBuffers(), I get a segmentation
fault when qemu exits (somewhere deep in fglrx_dri.so). Without, the
segfault disappears.

Max

>> +
>> +    return vao;
>> +}
>> +
>> +void qemu_gl_run_texture_blit(GLint texture_blit_prog,
>> +                              GLint texture_blit_vao)
>> +{
>> +    glUseProgram(texture_blit_prog);
>> +    glBindVertexArray(texture_blit_vao);
>> +    glDrawArrays(GL_TRIANGLE_STRIP, 0, 4);
>>  }
>>  
>>  /* ---------------------------------------------------------------------- */
>>
> 
> 



[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 473 bytes --]

  reply	other threads:[~2015-09-14 16:19 UTC|newest]

Thread overview: 33+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2015-09-09 11:20 [Qemu-devel] [PATCH 0/9] add virgl rendering support Gerd Hoffmann
2015-09-09 11:20 ` [Qemu-devel] [PATCH 1/9] shaders: initialize vertexes once Gerd Hoffmann
2015-09-10 12:40   ` Marc-André Lureau
2015-09-10 13:22     ` Gerd Hoffmann
2015-09-10 13:42       ` Marc-André Lureau
2015-09-10 14:59         ` Gerd Hoffmann
2015-09-14 16:00   ` Max Reitz
2015-09-14 16:19     ` Max Reitz [this message]
2015-09-15  7:10       ` Gerd Hoffmann
2015-09-09 11:20 ` [Qemu-devel] [PATCH 2/9] sdl2: quick & dirty flicker workaround Gerd Hoffmann
2015-09-10 12:40   ` Marc-André Lureau
2015-09-14 16:16   ` Max Reitz
2015-09-09 11:20 ` [Qemu-devel] [PATCH 3/9] ui/console: add opengl context and scanout support interfaces Gerd Hoffmann
2015-09-10 12:40   ` Marc-André Lureau
2015-09-15  8:30   ` Paolo Bonzini
2015-09-09 11:20 ` [Qemu-devel] [PATCH 4/9] virtio-gpu: update headers for virgl/3d Gerd Hoffmann
2015-09-10 12:41   ` Marc-André Lureau
2015-09-09 11:20 ` [Qemu-devel] [PATCH 5/9] virtio-gpu: add 3d mode and virgl rendering support Gerd Hoffmann
2015-09-14 18:14   ` Max Reitz
2015-09-15  7:33     ` Gerd Hoffmann
2015-09-15  8:33   ` Paolo Bonzini
2015-09-09 11:20 ` [Qemu-devel] [PATCH 6/9] sdl2/opengl: add opengl context and scanout support Gerd Hoffmann
2015-09-14 18:49   ` Max Reitz
2015-09-15  7:54     ` Gerd Hoffmann
2015-09-16 13:44       ` Max Reitz
2015-09-09 11:20 ` [Qemu-devel] [PATCH 7/9] opengl: add egl-context.[ch] helpers Gerd Hoffmann
2015-09-11 14:13   ` Marc-André Lureau
2015-09-09 11:20 ` [Qemu-devel] [PATCH 8/9] gtk/opengl: add opengl context and scanout support (egl) Gerd Hoffmann
2015-09-11 14:36   ` Marc-André Lureau
2015-09-09 11:20 ` [Qemu-devel] [PATCH 9/9] gtk/opengl: add opengl context and scanout support (GtkGLArea) Gerd Hoffmann
2015-09-11 14:44   ` Marc-André Lureau
2015-09-14 13:50     ` Gerd Hoffmann
2015-09-11 16:10 ` [Qemu-devel] [PATCH 0/9] add virgl rendering support Marc-André Lureau

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=55F6F384.1040107@redhat.com \
    --to=mreitz@redhat.com \
    --cc=kraxel@redhat.com \
    --cc=qemu-devel@nongnu.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).