qemu-devel.nongnu.org archive mirror
 help / color / mirror / Atom feed
* [Qemu-devel] [PATCH] sdl: Avoid unnecessary resizing of the display surface
@ 2012-05-01  5:53 Stefan Weil
  2012-05-01 14:00 ` Paolo Bonzini
  2012-05-04 19:49 ` Stefan Weil
  0 siblings, 2 replies; 3+ messages in thread
From: Stefan Weil @ 2012-05-01  5:53 UTC (permalink / raw)
  To: qemu-devel; +Cc: Paolo Bonzini, Anthony Liguori, Stefan Weil

If neither width nor height changes, nothing has to be done.

Cc: Anthony Liguori <aliguori@us.ibm.com>
Signed-off-by: Stefan Weil <sw@weilnetz.de>
---

This patch improves SDL for any host (for example with remote X displays),
but the main reason why I wrote it was another problem:

On w32 / w64 hosts, qemu-system-arm has a deadlock when it calls
sdl_resize_displaysurface during Linux boot. One thread waits for
a critical region, another thread waits inside SDL_SetVideoMode.

The patch avoids this problem. Paolo, maybe you have an idea what
could cause the deadlock. Debugging on wxx is terrible - up to now,
I did not succeed in analysing the lock situation with gdb.

Regards,
Stefan W.

 ui/sdl.c |    7 +++++--
 1 files changed, 5 insertions(+), 2 deletions(-)

diff --git a/ui/sdl.c b/ui/sdl.c
index 8700b7a..8c1c71c 100644
--- a/ui/sdl.c
+++ b/ui/sdl.c
@@ -224,8 +224,11 @@ static void sdl_free_displaysurface(DisplaySurface *surface)
 
 static DisplaySurface* sdl_resize_displaysurface(DisplaySurface *surface, int width, int height)
 {
-    sdl_free_displaysurface(surface);
-    return sdl_create_displaysurface(width, height);
+    if (surface->width != width || surface->height != height) {
+        sdl_free_displaysurface(surface);
+        surface = sdl_create_displaysurface(width, height);
+    }
+    return surface;
 }
 
 /* generic keyboard conversion */
-- 
1.7.9

^ permalink raw reply related	[flat|nested] 3+ messages in thread

* Re: [Qemu-devel] [PATCH] sdl: Avoid unnecessary resizing of the display surface
  2012-05-01  5:53 [Qemu-devel] [PATCH] sdl: Avoid unnecessary resizing of the display surface Stefan Weil
@ 2012-05-01 14:00 ` Paolo Bonzini
  2012-05-04 19:49 ` Stefan Weil
  1 sibling, 0 replies; 3+ messages in thread
From: Paolo Bonzini @ 2012-05-01 14:00 UTC (permalink / raw)
  To: Stefan Weil; +Cc: Anthony Liguori, qemu-devel

Il 01/05/2012 07:53, Stefan Weil ha scritto:
> If neither width nor height changes, nothing has to be done.
> 
> Cc: Anthony Liguori <aliguori@us.ibm.com>
> Signed-off-by: Stefan Weil <sw@weilnetz.de>
> ---
> 
> This patch improves SDL for any host (for example with remote X displays),
> but the main reason why I wrote it was another problem:
> 
> On w32 / w64 hosts, qemu-system-arm has a deadlock when it calls
> sdl_resize_displaysurface during Linux boot. One thread waits for
> a critical region, another thread waits inside SDL_SetVideoMode.
> 
> The patch avoids this problem. Paolo, maybe you have an idea what
> could cause the deadlock.

No idea---honestly, my w32 interest is limited to replacing hacks with
proper solutions. :)  It turns out that in many cases you can fix bugs
along the way and that helps having people look at your patches...

Paolo

^ permalink raw reply	[flat|nested] 3+ messages in thread

* Re: [Qemu-devel] [PATCH] sdl: Avoid unnecessary resizing of the display surface
  2012-05-01  5:53 [Qemu-devel] [PATCH] sdl: Avoid unnecessary resizing of the display surface Stefan Weil
  2012-05-01 14:00 ` Paolo Bonzini
@ 2012-05-04 19:49 ` Stefan Weil
  1 sibling, 0 replies; 3+ messages in thread
From: Stefan Weil @ 2012-05-04 19:49 UTC (permalink / raw)
  To: Anthony Liguori; +Cc: qemu-devel

Am 01.05.2012 07:53, schrieb Stefan Weil:
> If neither width nor height changes, nothing has to be done.
>
> Cc: Anthony Liguori<aliguori@us.ibm.com>
> Signed-off-by: Stefan Weil<sw@weilnetz.de>
> ---
>
> This patch improves SDL for any host (for example with remote X displays),
> but the main reason why I wrote it was another problem:
>
> On w32 / w64 hosts, qemu-system-arm has a deadlock when it calls
> sdl_resize_displaysurface during Linux boot. One thread waits for
> a critical region, another thread waits inside SDL_SetVideoMode.
>
> The patch avoids this problem. Paolo, maybe you have an idea what
> could cause the deadlock. Debugging on wxx is terrible - up to now,
> I did not succeed in analysing the lock situation with gdb.
>
> Regards,
> Stefan W.
>
>   ui/sdl.c |    7 +++++--
>   1 files changed, 5 insertions(+), 2 deletions(-)
>
> diff --git a/ui/sdl.c b/ui/sdl.c
> index 8700b7a..8c1c71c 100644
> --- a/ui/sdl.c
> +++ b/ui/sdl.c
> @@ -224,8 +224,11 @@ static void sdl_free_displaysurface(DisplaySurface *surface)
>
>   static DisplaySurface* sdl_resize_displaysurface(DisplaySurface *surface, int width, int height)
>   {
> -    sdl_free_displaysurface(surface);
> -    return sdl_create_displaysurface(width, height);
> +    if (surface->width != width || surface->height != height) {
> +        sdl_free_displaysurface(surface);
> +        surface = sdl_create_displaysurface(width, height);
> +    }
> +    return surface;
>   }
>
>   /* generic keyboard conversion */
>    

Please don't apply this patch. It is not complete and can cause a
crash (seen on w32 when resizing the SDL window).

I'll send an update.

Regards,
Stefan W.

^ permalink raw reply	[flat|nested] 3+ messages in thread

end of thread, other threads:[~2012-05-04 19:49 UTC | newest]

Thread overview: 3+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2012-05-01  5:53 [Qemu-devel] [PATCH] sdl: Avoid unnecessary resizing of the display surface Stefan Weil
2012-05-01 14:00 ` Paolo Bonzini
2012-05-04 19:49 ` Stefan Weil

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).