Hi all,
I've been asked by Mark to report my testings here.
Tested with qemu-for-ppc-2.8 commit 8b0262d0bcb7c697f2db8d6cd731cb478937902a.
Command line: qemu-system-ppc -M mac99 -m 256 -prom-env 'boot-args=-v' -bios foo/openbios-qemu-vga.elf -boot d -prom-env 'auto-boot?=false' -hda foo.qc2 -cdrom RhapsodyDR2.iso -nographic
Welcome to OpenBIOS v1.1 built on Jul 31 2016 05:52
0 > boot cd:%BOOT >> Not a bootable ELF image
switching to new context:
4+:interpret: exception -13 caught
interpret : update-key-map " get-key-map" " kbd" open-dev
$call-method drop gKeyMap 4 0 do swap dup @ >r 4+ swap dup dup @ r> or swap ! 4+ loop ; update-key-map gKeyMap 1+ c@ not h# 40 and if 0 stdout ! then failed with error ffffffed
Rhapsody Kernel Loader <16:33:44 980413> interpret: exception ffffffdf caught
interpret to DIH value &CL DIH to my-self 1 encode-int "
AAPL,boot-display" property &CL 0 100 " set-colors" DIH $CM " width" GMP ?dup 0= if 280 then " height" GMP ?dup 0= if 1E0 then 2dup to Dheight to Dwidth " linebytes" GMP ?dup 0= if Dwidth then " depth" GMP ?dup 0= if 8 then over value rowbytes active-package DIH ihandle>phandle to active-package : rect-setup >r >r rowbytes * + frame-buffer-adr + r> -rot r> ; : DRAW-RECTANGLE rect-setup 0 ?do 2dup 4 pick move 2 pick rowbytes d+ loop 3drop ; : FILL-RECTANGLE rect-setup 0 ?do dup 3 pick 3 pick fill rowbytes + loop 3drop ; : READ-RECTANGLE rect-setup >r swap r> 0 ?do 2dup 4 pick move rowbytes 3 pick d+ loop 3drop ; to active-package failed with error 00000100 call-method fill-rectangle: exception ffffffdf
call-method fill-rectangle failed with error ffffffdf
interpret: exception ffffffdf � caught
interpret drawImage failed with error 00000108
interpret: exception ffffffde caught
interpret dev /packages/obp-tftp failed with error 00000003
load isn't unique. ?no dev call-method fill-rectangle: exception ffffffdf
call-method fill-rectangle failed with error ffffffdf
call-method draw-rectangle: exception ffffffdf
call-method draw-rectangle failed with error ffffffdf
EXIT 0 >
On 04/08/16 15:27, Natalia Portillo wrote:
Hi all,
I've been asked by Mark to report my testings here.
Tested with qemu-for-ppc-2.8 commit 8b0262d0bcb7c697f2db8d6cd731cb478937902a.
Command line: qemu-system-ppc -M mac99 -m 256 -prom-env 'boot-args=-v' -bios foo/openbios-qemu-vga.elf -boot d -prom-env 'auto-boot?=false' -hda foo.qc2 -cdrom RhapsodyDR2.iso -nographic
Welcome to OpenBIOS v1.1 built on Jul 31 2016 05:52
0 > boot cd:%BOOT >> Not a bootable ELF image
switching to new context:
4+:interpret: exception -13 caught
interpret : update-key-map " get-key-map" " kbd" open-dev
$call-method drop gKeyMap 4 0 do swap dup @ >r 4+ swap dup dup @ r> or swap ! 4+ loop ; update-key-map gKeyMap 1+ c@ not h# 40 and if 0 stdout ! then failed with error ffffffed
Rhapsody Kernel Loader <16:33:44 980413> interpret: exception ffffffdf caught
interpret to DIH value &CL DIH to my-self 1 encode-int "
AAPL,boot-display" property &CL 0 100 " set-colors" DIH $CM " width" GMP ?dup 0= if 280 then " height" GMP ?dup 0= if 1E0 then 2dup to Dheight to Dwidth " linebytes" GMP ?dup 0= if Dwidth then " depth" GMP ?dup 0= if 8 then over value rowbytes active-package DIH ihandle>phandle to active-package : rect-setup >r >r rowbytes * + frame-buffer-adr + r> -rot r> ; : DRAW-RECTANGLE rect-setup 0 ?do 2dup 4 pick move 2 pick rowbytes d+ loop 3drop ; : FILL-RECTANGLE rect-setup 0 ?do dup 3 pick 3 pick fill rowbytes + loop 3drop ; : READ-RECTANGLE rect-setup >r swap r> 0 ?do 2dup 4 pick move rowbytes 3 pick d+ loop 3drop ; to active-package failed with error 00000100 call-method fill-rectangle: exception ffffffdf
call-method fill-rectangle failed with error ffffffdf
interpret: exception ffffffdf � caught
interpret drawImage failed with error 00000108
interpret: exception ffffffde caught
interpret dev /packages/obp-tftp failed with error 00000003
load isn't unique. ?no dev call-method fill-rectangle: exception ffffffdf
call-method fill-rectangle failed with error ffffffdf
call-method draw-rectangle: exception ffffffdf
call-method draw-rectangle failed with error ffffffdf
EXIT 0 >
Hmmm it seems like this and the other Rhapsody/OS X server boots are blowing up because the keyboard alias being looked for is "kbd" rather than "keyboard" which is what OpenBIOS uses. Fixing this should be fairly easy - I'll spin up a patch shortly.
ATB,
Mark.
On Thu, 4 Aug 2016, Mark Cave-Ayland wrote:
On 04/08/16 15:27, Natalia Portillo wrote:
0 > boot cd:%BOOT >> Not a bootable ELF image
switching to new context:
4+:interpret: exception -13 caught
interpret : update-key-map " get-key-map" " kbd" open-dev
$call-method drop gKeyMap 4 0 do swap dup @ >r 4+ swap dup dup @ r> or swap ! 4+ loop ; update-key-map gKeyMap 1+ c@ not h# 40 and if 0 stdout ! then failed with error ffffffed
Rhapsody Kernel Loader <16:33:44 980413> interpret: exception ffffffdf caught
interpret to DIH value &CL DIH to my-self 1 encode-int "
AAPL,boot-display" property &CL 0 100 " set-colors" DIH $CM " width" GMP ?dup 0= if 280 then " height" GMP ?dup 0= if 1E0 then 2dup to Dheight to Dwidth " linebytes" GMP ?dup 0= if Dwidth then " depth" GMP ?dup 0= if 8 then over value rowbytes active-package DIH ihandle>phandle to active-package : rect-setup >r >r rowbytes * + frame-buffer-adr + r> -rot r> ; : DRAW-RECTANGLE rect-setup 0 ?do 2dup 4 pick move 2 pick rowbytes d+ loop 3drop ; : FILL-RECTANGLE rect-setup 0 ?do dup 3 pick 3 pick fill rowbytes + loop 3drop ; : READ-RECTANGLE rect-setup >r swap r> 0 ?do 2dup 4 pick move rowbytes 3 pick d+ loop 3drop ; to active-package failed with error 00000100 call-method fill-rectangle: exception ffffffdf
call-method fill-rectangle failed with error ffffffdf
interpret: exception ffffffdf ??? caught
interpret drawImage failed with error 00000108
interpret: exception ffffffde caught
interpret dev /packages/obp-tftp failed with error 00000003
load isn't unique. ?no dev call-method fill-rectangle: exception ffffffdf
call-method fill-rectangle failed with error ffffffdf
call-method draw-rectangle: exception ffffffdf
call-method draw-rectangle failed with error ffffffdf
EXIT 0 >
Hmmm it seems like this and the other Rhapsody/OS X server boots are blowing up because the keyboard alias being looked for is "kbd" rather than "keyboard" which is what OpenBIOS uses. Fixing this should be fairly easy - I'll spin up a patch shortly.
That does not seem to match real hardware where it seems to be "keyboard" as well: http://nandra.segv.jp/NetBSD/G4.dump-device-tree.txt So it may be something else.
Regards, BALATON Zoltan
I think it is better to compare with a more appropiate device tree. Rhapsody DR did not even support the beige G3, but 8x00 and 9x00, and those OFs are archaic (and if I recall correctly, marked faulty by all Linux/ppc and NetBSD guys).
I can't find a while dump.
Also according to NetBSD/macppc installation guide for 9500: "NetBSD may not boot if screen/kbd are used (use ttya for your console)" *sic*
On 04/08/16 20:43, BALATON Zoltan wrote:
On Thu, 4 Aug 2016, Mark Cave-Ayland wrote:
On 04/08/16 15:27, Natalia Portillo wrote:
0 > boot cd:%BOOT >> Not a bootable ELF image
switching to new context:
4+:interpret: exception -13 caught
interpret : update-key-map " get-key-map" " kbd" open-dev
$call-method drop gKeyMap 4 0 do swap dup @ >r 4+ swap dup dup @ r> or swap ! 4+ loop ; update-key-map gKeyMap 1+ c@ not h# 40 and if 0 stdout ! then failed with error ffffffed
Rhapsody Kernel Loader <16:33:44 980413> interpret: exception ffffffdf caught
interpret to DIH value &CL DIH to my-self 1 encode-int "
AAPL,boot-display" property &CL 0 100 " set-colors" DIH $CM " width" GMP ?dup 0= if 280 then " height" GMP ?dup 0= if 1E0 then 2dup to Dheight to Dwidth " linebytes" GMP ?dup 0= if Dwidth then " depth" GMP ?dup 0= if 8 then over value rowbytes active-package DIH ihandle>phandle to active-package : rect-setup >r >r rowbytes * + frame-buffer-adr + r> -rot r> ; : DRAW-RECTANGLE rect-setup 0 ?do 2dup 4 pick move 2 pick rowbytes d+ loop 3drop ; : FILL-RECTANGLE rect-setup 0 ?do dup 3 pick 3 pick fill rowbytes + loop 3drop ; : READ-RECTANGLE rect-setup >r swap r> 0 ?do 2dup 4 pick move rowbytes 3 pick d+ loop 3drop ; to active-package failed with error 00000100 call-method fill-rectangle: exception ffffffdf
call-method fill-rectangle failed with error ffffffdf
interpret: exception ffffffdf ??? caught
interpret drawImage failed with error 00000108
interpret: exception ffffffde caught
interpret dev /packages/obp-tftp failed with error 00000003
load isn't unique. ?no dev call-method fill-rectangle: exception ffffffdf
call-method fill-rectangle failed with error ffffffdf
call-method draw-rectangle: exception ffffffdf
call-method draw-rectangle failed with error ffffffdf
EXIT 0 >
Hmmm it seems like this and the other Rhapsody/OS X server boots are blowing up because the keyboard alias being looked for is "kbd" rather than "keyboard" which is what OpenBIOS uses. Fixing this should be fairly easy - I'll spin up a patch shortly.
That does not seem to match real hardware where it seems to be "keyboard" as well: http://nandra.segv.jp/NetBSD/G4.dump-device-tree.txt So it may be something else.
Regards, BALATON Zoltan
On Thu, 4 Aug 2016, Natalia Portillo wrote:
I think it is better to compare with a more appropiate device tree.
Your command line said: "qemu-system-ppc -M mac99" and mac99 currently tries to match the machine with that device tree I've quoted. So in a way it is appropriate...
Rhapsody DR did not even support the beige G3, but 8x00 and 9x00, and those OFs are archaic (and if I recall correctly, marked faulty by all Linux/ppc and NetBSD guys).
...but -M mac99 may not be appropriate to run Rhapsody DR then.
Regards, BALATON Zoltan
Ideally, -M tsunami, -M gossamer, so on :p
I got once Rhapsody to boot on my G3 Beige. Long ago.
On 04/08/16 21:02, BALATON Zoltan wrote:
On Thu, 4 Aug 2016, Natalia Portillo wrote:
I think it is better to compare with a more appropiate device tree.
Your command line said: "qemu-system-ppc -M mac99" and mac99 currently tries to match the machine with that device tree I've quoted. So in a way it is appropriate...
Rhapsody DR did not even support the beige G3, but 8x00 and 9x00, and those OFs are archaic (and if I recall correctly, marked faulty by all Linux/ppc and NetBSD guys).
...but -M mac99 may not be appropriate to run Rhapsody DR then.
Regards, BALATON Zoltan
On Thu, 4 Aug 2016, Natalia Portillo wrote:
Ideally, -M tsunami, -M gossamer, so on :p
Yes, but that may need some time to implement all these.
I got once Rhapsody to boot on my G3 Beige. Long ago.
Then maybe try without -M mac99 as the default is -M g3beige but I'm not sure if there's any difference in OpenBIOS for those. It will have some different hardware emulated at least but at the point it failed that may not matter.
Regards, BALATON Zoltan
I've been waiting for this since me and Jocelyn tried to boot MorphOS 1.4 on OHW. There we discovered it tried to do some things on the RS232 registers and abandoned the idea because the peripherals emulation was simply far away from working with any guest.
That was eons ago.
I can wait :p
On 04/08/16 21:24, BALATON Zoltan wrote:
On Thu, 4 Aug 2016, Natalia Portillo wrote:
Ideally, -M tsunami, -M gossamer, so on :p
Yes, but that may need some time to implement all these.
I got once Rhapsody to boot on my G3 Beige. Long ago.
Then maybe try without -M mac99 as the default is -M g3beige but I'm not sure if there's any difference in OpenBIOS for those. It will have some different hardware emulated at least but at the point it failed that may not matter.
Regards, BALATON Zoltan
http://www.rhapsodyos.org/hardware/system_requirements/system_requirements_1... Seems this guy tested it on G3 beige.
Apple even provided a patch to enable G3's cache.
On 04/08/16 21:29, Natalia Portillo wrote:
I've been waiting for this since me and Jocelyn tried to boot MorphOS 1.4 on OHW. There we discovered it tried to do some things on the RS232 registers and abandoned the idea because the peripherals emulation was simply far away from working with any guest.
That was eons ago.
I can wait :p
On 04/08/16 21:24, BALATON Zoltan wrote:
On Thu, 4 Aug 2016, Natalia Portillo wrote:
Ideally, -M tsunami, -M gossamer, so on :p
Yes, but that may need some time to implement all these.
I got once Rhapsody to boot on my G3 Beige. Long ago.
Then maybe try without -M mac99 as the default is -M g3beige but I'm not sure if there's any difference in OpenBIOS for those. It will have some different hardware emulated at least but at the point it failed that may not matter.
Regards, BALATON Zoltan
Ok Mark with 4+ patch:
Rhapsody DR1 goes further, draws spinning ball before exiting to OF. Rhapsody DR2 goes further, stays some seconds in splash before exiting. OS X 1.0 Server goes further, instead of instant-exit to OF, shows different error message, changes color palette and hangs there (without modifying framebuffer). OS X 1.1 Server behaves same as 1.0. OS X 1.2 Server shows splash, then exits.
None arrives kernel but, now, we are in the road :p
Logs uploaded as *_4+_serial.log.
On 04/08/16 23:21, Natalia Portillo wrote:
http://www.rhapsodyos.org/hardware/system_requirements/system_requirements_1... Seems this guy tested it on G3 beige.
Apple even provided a patch to enable G3's cache.
On 04/08/16 21:29, Natalia Portillo wrote:
I've been waiting for this since me and Jocelyn tried to boot MorphOS 1.4 on OHW. There we discovered it tried to do some things on the RS232 registers and abandoned the idea because the peripherals emulation was simply far away from working with any guest.
That was eons ago.
I can wait :p
On 04/08/16 21:24, BALATON Zoltan wrote:
On Thu, 4 Aug 2016, Natalia Portillo wrote:
Ideally, -M tsunami, -M gossamer, so on :p
Yes, but that may need some time to implement all these.
I got once Rhapsody to boot on my G3 Beige. Long ago.
Then maybe try without -M mac99 as the default is -M g3beige but I'm not sure if there's any difference in OpenBIOS for those. It will have some different hardware emulated at least but at the point it failed that may not matter.
Regards, BALATON Zoltan
On 05/08/16 15:38, Natalia Portillo wrote:
Ok Mark with 4+ patch:
Rhapsody DR1 goes further, draws spinning ball before exiting to OF. Rhapsody DR2 goes further, stays some seconds in splash before exiting. OS X 1.0 Server goes further, instead of instant-exit to OF, shows different error message, changes color palette and hangs there (without modifying framebuffer). OS X 1.1 Server behaves same as 1.0. OS X 1.2 Server shows splash, then exits.
None arrives kernel but, now, we are in the road :p
Logs uploaded as *_4+_serial.log.
Great, thanks for testing! I'm away for most of this weekend so I'll probably take a look at this again next week :)
ATB,
Mark.
On 05/08/16 15:38, Natalia Portillo wrote:
Ok Mark with 4+ patch:
Rhapsody DR1 goes further, draws spinning ball before exiting to OF. Rhapsody DR2 goes further, stays some seconds in splash before exiting. OS X 1.0 Server goes further, instead of instant-exit to OF, shows different error message, changes color palette and hangs there (without modifying framebuffer). OS X 1.1 Server behaves same as 1.0. OS X 1.2 Server shows splash, then exits.
None arrives kernel but, now, we are in the road :p
Logs uploaded as *_4+_serial.log.
Just had a look and that looks almost clean now from the Forth side - will post the patches shortly.
ATB,
Mark.
On 04/08/16 20:43, BALATON Zoltan wrote:
On Thu, 4 Aug 2016, Mark Cave-Ayland wrote:
On 04/08/16 15:27, Natalia Portillo wrote:
0 > boot cd:%BOOT >> Not a bootable ELF image
switching to new context:
4+:interpret: exception -13 caught
interpret : update-key-map " get-key-map" " kbd" open-dev
$call-method drop gKeyMap 4 0 do swap dup @ >r 4+ swap dup dup @ r> or swap ! 4+ loop ; update-key-map gKeyMap 1+ c@ not h# 40 and if 0 stdout ! then failed with error ffffffed
Rhapsody Kernel Loader <16:33:44 980413> interpret: exception ffffffdf caught
interpret to DIH value &CL DIH to my-self 1 encode-int "
AAPL,boot-display" property &CL 0 100 " set-colors" DIH $CM " width" GMP ?dup 0= if 280 then " height" GMP ?dup 0= if 1E0 then 2dup to Dheight to Dwidth " linebytes" GMP ?dup 0= if Dwidth then " depth" GMP ?dup 0= if 8 then over value rowbytes active-package DIH ihandle>phandle to active-package : rect-setup >r >r rowbytes * + frame-buffer-adr + r> -rot r> ; : DRAW-RECTANGLE rect-setup 0 ?do 2dup 4 pick move 2 pick rowbytes d+ loop 3drop ; : FILL-RECTANGLE rect-setup 0 ?do dup 3 pick 3 pick fill rowbytes + loop 3drop ; : READ-RECTANGLE rect-setup >r swap r> 0 ?do 2dup 4 pick move rowbytes 3 pick d+ loop 3drop ; to active-package failed with error 00000100 call-method fill-rectangle: exception ffffffdf
call-method fill-rectangle failed with error ffffffdf
interpret: exception ffffffdf ??? caught
interpret drawImage failed with error 00000108
interpret: exception ffffffde caught
interpret dev /packages/obp-tftp failed with error 00000003
load isn't unique. ?no dev call-method fill-rectangle: exception ffffffdf
call-method fill-rectangle failed with error ffffffdf
call-method draw-rectangle: exception ffffffdf
call-method draw-rectangle failed with error ffffffdf
EXIT 0 >
Hmmm it seems like this and the other Rhapsody/OS X server boots are blowing up because the keyboard alias being looked for is "kbd" rather than "keyboard" which is what OpenBIOS uses. Fixing this should be fairly easy - I'll spin up a patch shortly.
That does not seem to match real hardware where it seems to be "keyboard" as well: http://nandra.segv.jp/NetBSD/G4.dump-device-tree.txt So it may be something else.
I'd say there are definitely multiple failures here :)
diff --git a/arch/ppc/qemu/qemu.fs b/arch/ppc/qemu/qemu.fs index 3d99a34..cc3784b 100644 --- a/arch/ppc/qemu/qemu.fs +++ b/arch/ppc/qemu/qemu.fs @@ -74,7 +74,7 @@ variable keyboard-phandle 0 keyboard-phandle ! active-package " /aliases" find-device keyboard-phandle @ get-package-path - encode-string " keyboard" property + encode-string " kbd" property active-package! then ; SYSTEM-initializer
According to the device trees I have, it's definitely "kbd" for the g3beige. And look at this:
" kbd" open-dev $call-method
Here $call-method has to throw if the named node doesn't exist. Perhaps the change is only required for g3beige or maybe non-USB keyboards?
There's also the 4+ which doesn't seem to be recognised. As a quick test boot QEMU with -prom-env 'auto-boot?=false' and then type the following:
: 4+ 4 + ; boot
ATB,
Mark.
File rhapsodydr2_4+.log
On 04/08/16 21:03, Mark Cave-Ayland wrote:
On 04/08/16 20:43, BALATON Zoltan wrote:
On Thu, 4 Aug 2016, Mark Cave-Ayland wrote:
On 04/08/16 15:27, Natalia Portillo wrote:
0 > boot cd:%BOOT >> Not a bootable ELF image
switching to new context:
4+:interpret: exception -13 caught
interpret : update-key-map " get-key-map" " kbd" open-dev
$call-method drop gKeyMap 4 0 do swap dup @ >r 4+ swap dup dup @ r> or swap ! 4+ loop ; update-key-map gKeyMap 1+ c@ not h# 40 and if 0 stdout ! then failed with error ffffffed
Rhapsody Kernel Loader <16:33:44 980413> interpret: exception ffffffdf caught
interpret to DIH value &CL DIH to my-self 1 encode-int "
AAPL,boot-display" property &CL 0 100 " set-colors" DIH $CM " width" GMP ?dup 0= if 280 then " height" GMP ?dup 0= if 1E0 then 2dup to Dheight to Dwidth " linebytes" GMP ?dup 0= if Dwidth then " depth" GMP ?dup 0= if 8 then over value rowbytes active-package DIH ihandle>phandle to active-package : rect-setup >r >r rowbytes * + frame-buffer-adr + r> -rot r> ; : DRAW-RECTANGLE rect-setup 0 ?do 2dup 4 pick move 2 pick rowbytes d+ loop 3drop ; : FILL-RECTANGLE rect-setup 0 ?do dup 3 pick 3 pick fill rowbytes + loop 3drop ; : READ-RECTANGLE rect-setup >r swap r> 0 ?do 2dup 4 pick move rowbytes 3 pick d+ loop 3drop ; to active-package failed with error 00000100 call-method fill-rectangle: exception ffffffdf
call-method fill-rectangle failed with error ffffffdf
interpret: exception ffffffdf ??? caught
interpret drawImage failed with error 00000108
interpret: exception ffffffde caught
interpret dev /packages/obp-tftp failed with error 00000003
load isn't unique. ?no dev call-method fill-rectangle: exception ffffffdf
call-method fill-rectangle failed with error ffffffdf
call-method draw-rectangle: exception ffffffdf
call-method draw-rectangle failed with error ffffffdf
EXIT 0 >
Hmmm it seems like this and the other Rhapsody/OS X server boots are blowing up because the keyboard alias being looked for is "kbd" rather than "keyboard" which is what OpenBIOS uses. Fixing this should be fairly easy - I'll spin up a patch shortly.
That does not seem to match real hardware where it seems to be "keyboard" as well: http://nandra.segv.jp/NetBSD/G4.dump-device-tree.txt So it may be something else.
I'd say there are definitely multiple failures here :)
diff --git a/arch/ppc/qemu/qemu.fs b/arch/ppc/qemu/qemu.fs index 3d99a34..cc3784b 100644 --- a/arch/ppc/qemu/qemu.fs +++ b/arch/ppc/qemu/qemu.fs @@ -74,7 +74,7 @@ variable keyboard-phandle 0 keyboard-phandle ! active-package " /aliases" find-device keyboard-phandle @ get-package-path
- encode-string " keyboard" property
- encode-string " kbd" property active-package! then
; SYSTEM-initializer
According to the device trees I have, it's definitely "kbd" for the g3beige. And look at this:
" kbd" open-dev $call-method
Here $call-method has to throw if the named node doesn't exist. Perhaps the change is only required for g3beige or maybe non-USB keyboards?
There's also the 4+ which doesn't seem to be recognised. As a quick test boot QEMU with -prom-env 'auto-boot?=false' and then type the following:
: 4+ 4 + ; boot
ATB,
Mark.
-
BALATON Zoltan -
Mark Cave-Ayland -
Natalia Portillo