| Message ID | CAHt6W4duQaPmjK-USDx7GpAk0VuFOY4+W1D0rPgZ52_Lx+Ug5w@mail.gmail.com |
|---|---|
| State | New |
| Headers | show |
Il 18/06/2013 12:17, Frediano Ziglio ha scritto: > Modern notebook support 1366x768 resolution. The resolution width is > not multiple of 16 causing some problems. > > QEMU VGA emulation requires width resolution to be multiple of 8. > > VNC implementation requires width resolution to be multiple of 16. > > Signed-off-by: Frediano Ziglio <frediano.ziglio@citrix.com > <mailto:frediano.ziglio@citrix.com>> Tested-by: Fabio Fantoni <fabio.fantoni@m2r.biz> I tested it for a long time with spice on xen (because qxl will be fully working only after adding SSE support on hvm domUs). It works, I think it is good to add this and the respective vgabios patch on upstream. > --- > hw/display/vga.c | 2 +- > ui/vnc.c | 34 +++++++++++++++++++--------------- > 2 files changed, 20 insertions(+), 16 deletions(-) > > Updates: > - rebased > - fixed style problems > - fixed typos in comment > > Attached patch for last vgabios in order to get some new resolutions. > Still had no time to test deeply. > > diff --git a/hw/display/vga.c b/hw/display/vga.c > index 21a108d..0053b0f 100644 > --- a/hw/display/vga.c > +++ b/hw/display/vga.c > @@ -648,7 +648,7 @@ void vbe_ioport_write_data(void *opaque, uint32_t > addr, uint32_t val) > } > break; > case VBE_DISPI_INDEX_XRES: > - if ((val <= VBE_DISPI_MAX_XRES) && ((val & 7) == 0)) { > + if ((val <= VBE_DISPI_MAX_XRES) && ((val & 1) == 0)) { > s->vbe_regs[s->vbe_index] = val; > } > break; > diff --git a/ui/vnc.c b/ui/vnc.c > index dfc7459..2a2bb90 100644 > --- a/ui/vnc.c > +++ b/ui/vnc.c > @@ -911,26 +911,30 @@ static int vnc_update_client(VncState *vs, int > has_dirty) > for (y = 0; y < height; y++) { > int x; > int last_x = -1; > - for (x = 0; x < width / 16; x++) { > - if (test_and_clear_bit(x, vs->dirty[y])) { > + for (x = 0; x < width; x += 16) { > + if (test_and_clear_bit(x/16, vs->dirty[y])) { > if (last_x == -1) { > last_x = x; > } > } else { > if (last_x != -1) { > - int h = find_and_clear_dirty_height(vs, y, > last_x, x, > - height); > + int h = find_and_clear_dirty_height(vs, y, > last_x/16, > + x/16, height); > > - n += vnc_job_add_rect(job, last_x * 16, y, > - (x - last_x) * 16, h); > + n += vnc_job_add_rect(job, last_x, y, > + (x - last_x), h); > } > last_x = -1; > } > } > if (last_x != -1) { > - int h = find_and_clear_dirty_height(vs, y, last_x, x, > height); > - n += vnc_job_add_rect(job, last_x * 16, y, > - (x - last_x) * 16, h); > + int h = find_and_clear_dirty_height(vs, y, last_x/16, > x/16, > + height); > + if (x > width) { > + x = width; > + } > + n += vnc_job_add_rect(job, last_x, y, > + (x - last_x), h); > } > } > > @@ -1861,7 +1865,7 @@ static void framebuffer_update_request(VncState > *vs, int incremental, > int w, int h) > { > int i; > - const size_t width = surface_width(vs->vd->ds) / 16; > + const size_t width = (surface_width(vs->vd->ds)+15) / 16; > const size_t height = surface_height(vs->vd->ds); > > if (y_position > height) { > @@ -2686,10 +2690,6 @@ static int > vnc_refresh_server_surface(VncDisplay *vd) > * Check and copy modified bits from guest to server surface. > * Update server dirty map. > */ > - cmp_bytes = 64; > - if (cmp_bytes > vnc_server_fb_stride(vd)) { > - cmp_bytes = vnc_server_fb_stride(vd); > - } > if (vd->guest.format != VNC_SERVER_FB_FORMAT) { > int width = pixman_image_get_width(vd->server); > tmpbuf = qemu_pixman_linebuf_create(VNC_SERVER_FB_FORMAT, width); > @@ -2710,8 +2710,12 @@ static int > vnc_refresh_server_surface(VncDisplay *vd) > } > server_ptr = server_row; > > - for (x = 0; x + 15 < width; > + cmp_bytes = 64; > + for (x = 0; x < width; > x += 16, guest_ptr += cmp_bytes, server_ptr += > cmp_bytes) { > + if (width - x < 16) { > + cmp_bytes = 4 * (width - x); > + } > if (!test_and_clear_bit((x / 16), vd->guest.dirty[y])) > continue; > if (memcmp(server_ptr, guest_ptr, cmp_bytes) == 0) > -- > 1.7.10.4 > > > > 2013/5/10 Andreas Färber <afaerber@suse.de <mailto:afaerber@suse.de>> > > Am 15.03.2013 19:14, schrieb Frediano Ziglio: > > Modern notebook support 136x768 resolution. The resolution width is > > 1366? > > > not multiple of 16 causing some problems. > > "a multiple"? (me not a native English speaker) > > > > > Qemu VGA emulation require width resolution to be multiple of 8. > > "QEMU" > > > > > VNC implementation require width resolution to be multiple of 16. > > "requires" or "implementations" > > > > > This patch remove these limits. Was tested with a Windows > machine with > > standard vga and 1366x768 as resolution. I had to update vgabios as > > version in qemu (pc-bios/vgabios-stdvga.bin) is quite old. I > also had > > to add some patches on top of VGABIOS 0.7a to add some new > > resolutions. > > > > I have some doubt about this patch > > - are other UI (sdl, cocoa, qxl) happy if resolution is not > multiple of 16 ? > > SDL and Gtk+ should be easily testable; if you CC Peter Maydell or > me we > can try to test Cocoa. CC'ing QXL guys. > > > - scanline is computed exactly without any alignment (so 1366 8 > bit is > > 1366 bytes) while getting vesa information from a laptop it seems to > > use some kind of alignment (if became 0x580 which is 1408 bytes). > > Perhaps should I change either VGABIOS and Qemu to make this > > alignment? > > Concerns and personal comments are better placed below ---. :) > > > > > Signed-off-by: Frediano Ziglio <frediano.ziglio@citrix.com > <mailto:frediano.ziglio@citrix.com>> > > > > --- > > hw/vga.c | 2 +- > > File has moved to hw/display/. > > > ui/vnc.c | 27 +++++++++++++-------------- > > 2 files changed, 14 insertions(+), 15 deletions(-) > > I don't see VGABIOS being updated here despite being mentioned > above? Is > that done in a different patch? Still needed? > > > diff --git a/hw/vga.c b/hw/vga.c > > index 1caf23d..d229f06 100644 > > --- a/hw/vga.c > > +++ b/hw/vga.c > > @@ -651,7 +651,7 @@ void vbe_ioport_write_data(void *opaque, > uint32_t > > addr, uint32_t val) > > } > > break; > > case VBE_DISPI_INDEX_XRES: > > - if ((val <= VBE_DISPI_MAX_XRES) && ((val & 7) == 0)) { > > + if ((val <= VBE_DISPI_MAX_XRES) && ((val & 1) == 0)) { > > s->vbe_regs[s->vbe_index] = val; > > } > > break; > > diff --git a/ui/vnc.c b/ui/vnc.c > > index ff4e2ae..328d14d 100644 > > --- a/ui/vnc.c > > +++ b/ui/vnc.c > > @@ -907,26 +907,27 @@ static int vnc_update_client(VncState *vs, > int has_dirty) > > for (y = 0; y < height; y++) { > > int x; > > int last_x = -1; > > - for (x = 0; x < width / 16; x++) { > > - if (test_and_clear_bit(x, vs->dirty[y])) { > > + for (x = 0; x < width; x += 16) { > > + if (test_and_clear_bit(x/16, vs->dirty[y])) { > [snip] > > Please check if scripts/checkpatch.pl <http://checkpatch.pl> > complains about missing spaces > around operators. > > Regards, > Andreas > > -- > SUSE LINUX Products GmbH, Maxfeldstr. 5, 90409 Nürnberg, Germany > GF: Jeff Hawn, Jennifer Guild, Felix Imendörffer; HRB 16746 AG > Nürnberg > > > > > Nessun virus nel messaggio. > Controllato da AVG - www.avg.com <http://www.avg.com> > Versione: 2013.0.3345 / Database dei virus: 3199/6420 - Data di > rilascio: 18/06/2013 >
Hi, > Tested-by: Fabio Fantoni <fabio.fantoni@m2r.biz> > > I tested it for a long time with spice on xen (because qxl will be fully > working only after adding SSE support on hvm domUs). It works, I think > it is good to add this and the respective vgabios patch on upstream. >> case VBE_DISPI_INDEX_XRES: >> - if ((val <= VBE_DISPI_MAX_XRES) && ((val & 7) == 0)) { >> + if ((val <= VBE_DISPI_MAX_XRES) && ((val & 1) == 0)) { >> s->vbe_regs[s->vbe_index] = val; >> } >> break; It's not that simple. With 32bit depths common today it will work fine, but for lower depths (especially those lower than 8bit) this will give you broken scanline alignment. cheers, Gerd
2013/7/23 Gerd Hoffmann <kraxel@redhat.com> > Hi, > > > Tested-by: Fabio Fantoni <fabio.fantoni@m2r.biz> > > > > I tested it for a long time with spice on xen (because qxl will be fully > > working only after adding SSE support on hvm domUs). It works, I think > > it is good to add this and the respective vgabios patch on upstream. > > >> case VBE_DISPI_INDEX_XRES: > >> - if ((val <= VBE_DISPI_MAX_XRES) && ((val & 7) == 0)) { > >> + if ((val <= VBE_DISPI_MAX_XRES) && ((val & 1) == 0)) { > >> s->vbe_regs[s->vbe_index] = val; > >> } > >> break; > > It's not that simple. With 32bit depths common today it will work fine, > but for lower depths (especially those lower than 8bit) this will give > you broken scanline alignment. > > cheers, > Gerd > > > In the card I tested the scanline is keep aligned but for this reason is not directly computed by maxx * bits but is something bigger. Frediano
On Sun, Jul 28, 2013 at 05:56:01PM +0100, Frediano Ziglio wrote: > 2013/7/23 Gerd Hoffmann <[1]kraxel@redhat.com> > > Â Hi, > > Tested-by: Fabio Fantoni <[2]fabio.fantoni@m2r.biz> > > > > I tested it for a long time with spice on xen (because qxl will be > fully > > working only after adding SSE support on hvm domUs). It works, I think > > it is good to add this and the respective vgabios patch on upstream. > > >> Â Â Â Â Â case VBE_DISPI_INDEX_XRES: > >> - Â Â Â Â Â Â if ((val <= VBE_DISPI_MAX_XRES) && ((val & 7) == > 0)) { > >> + Â Â Â Â Â Â if ((val <= VBE_DISPI_MAX_XRES) && ((val & 1) == > 0)) { > >> Â Â Â Â Â Â Â Â Â s->vbe_regs[s->vbe_index] = val; > >> Â Â Â Â Â Â Â } > >> Â Â Â Â Â Â Â break; > > It's not that simple. Â With 32bit depths common today it will work > fine, > but for lower depths (especially those lower than 8bit) this will give > you broken scanline alignment. > > cheers, > Â Gerd > > In the card I tested the scanline is keep aligned but for this reason is > not directly computed by maxx * bits but is something bigger. > Hmm.. so is an update needed for this patch? It'd be nice to get this patch in! -- Pasi > Frediano >
Il 25/09/2013 18:12, Pasi Kärkkäinen ha scritto: > On Sun, Jul 28, 2013 at 05:56:01PM +0100, Frediano Ziglio wrote: >> 2013/7/23 Gerd Hoffmann <[1]kraxel@redhat.com> >> >> Â Hi, >> > Tested-by: Fabio Fantoni <[2]fabio.fantoni@m2r.biz> >> > >> > I tested it for a long time with spice on xen (because qxl will be >> fully >> > working only after adding SSE support on hvm domUs). It works, I think >> > it is good to add this and the respective vgabios patch on upstream. >> >> >> Â Â Â Â Â case VBE_DISPI_INDEX_XRES: >> >> - Â Â Â Â Â Â if ((val <= VBE_DISPI_MAX_XRES) && ((val & 7) == >> 0)) { >> >> + Â Â Â Â Â Â if ((val <= VBE_DISPI_MAX_XRES) && ((val & 1) == >> 0)) { >> >> Â Â Â Â Â Â Â Â Â s->vbe_regs[s->vbe_index] = val; >> >> Â Â Â Â Â Â Â } >> >> Â Â Â Â Â Â Â break; >> >> It's not that simple. Â With 32bit depths common today it will work >> fine, >> but for lower depths (especially those lower than 8bit) this will give >> you broken scanline alignment. >> >> cheers, >> Â Gerd >> >> In the card I tested the scanline is keep aligned but for this reason is >> not directly computed by maxx * bits but is something bigger. >> > Hmm.. so is an update needed for this patch? It'd be nice to get this patch in! Ping Some statistics for evidence the importance to add support for 1366x768 and other resolutions: http://www.rapidtables.com/web/dev/screen-resolution-statistics.htm http://www.geek.com/chips/most-popular-screen-resolution-increases-to-1366-x-768-1482311/ > > -- Pasi > >> Frediano >> > > _______________________________________________ > Xen-devel mailing list > Xen-devel@lists.xen.org > http://lists.xen.org/xen-devel
diff --git a/hw/display/vga.c b/hw/display/vga.c index 21a108d..0053b0f 100644 --- a/hw/display/vga.c +++ b/hw/display/vga.c @@ -648,7 +648,7 @@ void vbe_ioport_write_data(void *opaque, uint32_t addr, uint32_t val) } break; case VBE_DISPI_INDEX_XRES: - if ((val <= VBE_DISPI_MAX_XRES) && ((val & 7) == 0)) { + if ((val <= VBE_DISPI_MAX_XRES) && ((val & 1) == 0)) { s->vbe_regs[s->vbe_index] = val; } break; diff --git a/ui/vnc.c b/ui/vnc.c index dfc7459..2a2bb90 100644 --- a/ui/vnc.c +++ b/ui/vnc.c @@ -911,26 +911,30 @@ static int vnc_update_client(VncState *vs, int has_dirty) for (y = 0; y < height; y++) { int x; int last_x = -1; - for (x = 0; x < width / 16; x++) { - if (test_and_clear_bit(x, vs->dirty[y])) { + for (x = 0; x < width; x += 16) { + if (test_and_clear_bit(x/16, vs->dirty[y])) { if (last_x == -1) { last_x = x; } } else { if (last_x != -1) { - int h = find_and_clear_dirty_height(vs, y, last_x, x, - height); + int h = find_and_clear_dirty_height(vs, y, last_x/16, + x/16, height); - n += vnc_job_add_rect(job, last_x * 16, y, - (x - last_x) * 16, h); + n += vnc_job_add_rect(job, last_x, y, + (x - last_x), h); } last_x = -1; } } if (last_x != -1) { - int h = find_and_clear_dirty_height(vs, y, last_x, x, height); - n += vnc_job_add_rect(job, last_x * 16, y, - (x - last_x) * 16, h); + int h = find_and_clear_dirty_height(vs, y, last_x/16, x/16, + height); + if (x > width) { + x = width; + } + n += vnc_job_add_rect(job, last_x, y,
Modern notebook support 1366x768 resolution. The resolution width is not multiple of 16 causing some problems. QEMU VGA emulation requires width resolution to be multiple of 8. VNC implementation requires width resolution to be multiple of 16. Signed-off-by: Frediano Ziglio <frediano.ziglio@citrix.com> --- hw/display/vga.c | 2 +- ui/vnc.c | 34 +++++++++++++++++++--------------- 2 files changed, 20 insertions(+), 16 deletions(-) Updates: - rebased - fixed style problems - fixed typos in comment Attached patch for last vgabios in order to get some new resolutions. Still had no time to test deeply. + (x - last_x), h); } } @@ -1861,7 +1865,7 @@ static void framebuffer_update_request(VncState *vs, int incremental, int w, int h) { int i; - const size_t width = surface_width(vs->vd->ds) / 16; + const size_t width = (surface_width(vs->vd->ds)+15) / 16; const size_t height = surface_height(vs->vd->ds); if (y_position > height) { @@ -2686,10 +2690,6 @@ static int vnc_refresh_server_surface(VncDisplay *vd) * Check and copy modified bits from guest to server surface. * Update server dirty map. */ - cmp_bytes = 64; - if (cmp_bytes > vnc_server_fb_stride(vd)) { - cmp_bytes = vnc_server_fb_stride(vd); - } if (vd->guest.format != VNC_SERVER_FB_FORMAT) { int width = pixman_image_get_width(vd->server); tmpbuf = qemu_pixman_linebuf_create(VNC_SERVER_FB_FORMAT, width); @@ -2710,8 +2710,12 @@ static int vnc_refresh_server_surface(VncDisplay *vd) } server_ptr = server_row; - for (x = 0; x + 15 < width; + cmp_bytes = 64; + for (x = 0; x < width; x += 16, guest_ptr += cmp_bytes, server_ptr += cmp_bytes) { + if (width - x < 16) { + cmp_bytes = 4 * (width - x); + } if (!test_and_clear_bit((x / 16), vd->guest.dirty[y])) continue; if (memcmp(server_ptr, guest_ptr, cmp_bytes) == 0)