Re: [Qemu-devel] Re: [PATCH] VMware SVGA II emulation
Hi, On 02/04/07, Thiemo Seufer <[EMAIL PROTECTED]> wrote: [..] I left the sdl_copy bit out because it broke scrolling of a guest Linux framebuffer in the moment it initialized its virtual consoles. Maybe that's a bug in libsdl 1.2. Thanks for checking this. I didn't have any ideas as to what could be affecting the Cirrus VGA in the patch, but yes, the .dpy_copy is used in it. This is to remove two obsolete variables in vl.c that I have accidentally left in the patch and only noticed it in the commitdiff: --- a/vl.c +++ b/vl.c @@ -543,10 +543,6 @@ int kbd_mouse_is_absolute(void) return qemu_put_mouse_event_current->qemu_put_mouse_event_absolute; } -void (*kbd_mouse_set)(int x, int y, int on) = NULL; -void (*kbd_cursor_define)(int width, int height, int bpp, int hot_x, int hot_y, - uint8_t *image, uint8_t *mask) = NULL; - void do_info_mice(void) { QEMUPutMouseEntry *cursor; Cheers, Andrzej
Re: [Qemu-devel] Re: [PATCH] VMware SVGA II emulation
andrzej zaborowski wrote: > Hi, > I looked at how the embedding of a standard VGA in the VMware SVGA > could be done (mainly out of curiosity) and it wasn't difficult. I had > to make small changes in hw/vga.c but I think it's made more flexible > now. Attached is a second version of the VMware SVGA patch. This time > including the "Host-accelerated mouse cursor support in SDL" patch in > the same file. I added a FIFO for buffering screen update requests so > as to avoid touching the SDL buffer between scheduled updates, (which > could lead to drawing over qemu monitor or even segfaults if the > resolution was changed in meantime). I also found what was wrong with > the accelerated rectangle filling and fixed a couple of other issues. > The palette size was also wrong in the older patch - credits to > Anthony Liguori who spotted it. > > I have not tested switching between std VGA graphic modes and VMware > SVGA mode, but the switching between VGA text mode and VMware SVGA > mode worked fine (although the black console background was becoming > not exactly black). Savevm/loadvm should also work, including the VGA > part. > In this version VMware SVGA is a standalone VGA so we're using only > one qemu console, like with "-cirrusvga". I left the sdl_copy bit out because it broke scrolling of a guest Linux framebuffer in the moment it initialized its virtual consoles. Maybe that's a bug in libsdl 1.2. Thiemo