Re: [PATCH v4 00/13] macfb: fixes for booting MacOS

[Laurent Vivier]

Le 08/10/2021 à 00:12, Mark Cave-Ayland a écrit :
> This is the next set of patches to allow users to boot MacOS in QEMU's
> q800 machine.
> 
> Patches 1 to 4 are fixes for existing bugs that I discovered whilst
> developing the remainder of the patchset whilst patch 5 simplifies the
> registration of the framebuffer RAM.
> 
> Patch 6 adds trace events to the framebuffer register accesses. The
> framebuffer registers are not officially documented, so the macfb
> device changes here are based upon reading of Linux/NetBSD source code,
> using gdbstub during the MacOS toolbox ROM initialisation, and changing
> the framebuffer size/depth within MacOS itself with these trace events
> enabled.
> 
> Patches 7 and 8 implement the mode sense logic documented in Apple
> Technical Note HW26 "Macintosh Quadra Built-In Video" and configure the
> default display type to be VGA.
> 
> Patch 9 implements the common monitor modes used for VGA at 640x480 and
> 800x600 for 1, 2, 4, 8 and 24-bit depths and also the Apple 21" color
> monitor at 1152x870 with 8-bit depth.
> 
> Patches 10 and 11 fix up errors in the 1-bit and 24-bit pixel encodings
> discovered when testing these color depths in MacOS.
> 
> Patch 12 adds a timer to implement the 60.15Hz VBL interrupt which is
> required for MacOS to process mouse movements, whilst patch 13 wires the
> same interrupt to a dedicated pin on VIA2 reserved for the video
> interrupt on the Quadra 800.
> 
> Signed-off-by: Mark Cave-Ayland <mark.cave-ayland@ilande.co.uk>
> 
> v4:
> - Rebase onto master
> - Add R-B tags from Laurent
> - Change %d to %u for the macfb_ctrl_* trace-events size parameter in patch 6
> 
> v3:
> - Rebase onto master
> - Add Laurent's code for displaying supported display modes as an error hint 
> in patch 9
>   and update the patch description
> - Fix typo in patch 8 description: 32-bit -> 24-bit
> - Split Error API changes from patch 1 into a new patch 2
> 
> v2:
> - Rebase onto master
> - Add R-B tags from Zoltan, Philippe and Laurent
> - Rework macfb_common_realize() to return a bool in patch 1
> - Add Fixes tag to patch 2
> - Use Laurent's suggested change for s->current_palette (slightly modified) 
> in patch 3
> - Change size trace-events parameter to unsigned int in patch 5
> - Add assert() as suggested by Philippe in patch 7
> - Move calculation of next VBL time into a separate macfb_next_vbl() function 
> in patch 11
> 
> Mark Cave-Ayland (13):
>   macfb: handle errors that occur during realize
>   macfb: update macfb.c to use the Error API best practices
>   macfb: fix invalid object reference in macfb_common_realize()
>   macfb: fix overflow of color_palette array
>   macfb: use memory_region_init_ram() in macfb_common_realize() for the
>     framebuffer
>   macfb: add trace events for reading and writing the control registers
>   macfb: implement mode sense to allow display type to be detected
>   macfb: add qdev property to specify display type
>   macfb: add common monitor modes supported by the MacOS toolbox ROM
>   macfb: fix up 1-bit pixel encoding
>   macfb: fix 24-bit RGB pixel encoding
>   macfb: add vertical blank interrupt
>   q800: wire macfb IRQ to separate video interrupt on VIA2
> 
>  hw/display/macfb.c         | 386 ++++++++++++++++++++++++++++++++++---
>  hw/display/trace-events    |   7 +
>  hw/m68k/q800.c             |  23 ++-
>  include/hw/display/macfb.h |  43 +++++
>  4 files changed, 429 insertions(+), 30 deletions(-)
> 

Applied to my q800 branch,

Thanks,
Laurent