EOS M100 porting

srsa_4c

Attached is another source snapshot. I'm publishing it because it uses "ROM" modifications (currently two 64kB pages of ROM are replaced with a modified RAM copy). The conversion of platform code is only partially done, so much of the usual copied firmware code is still there.
I'm currently using a suspected free space that I found next to the "Lime" code blob that loads to 0x1040000. That space is between 0x11A0440 ... 0x1340000. "Lime" is the Xtensa CPU used for networking.
The port hasn't improved much over the last published version, so I'm still not posting binaries.

A few words about remapping ROM.
This happens at boot time, in platform/m100/sub/100a/boot.c, in the function setup_mmu_for_chdk(). A bunch of firmware functions (with rather lame names) are also used in that function, their purpose is to update CPU caches and branch predictor. The function addresses can be found in stubs_entry_2.S. Anybody looking for them can find them in a disassembly - the 0xdffxxxxx routines are in a copied blob, the blob can be found by looking at the "Detected address ranges" info in stubs_entry.S. A file named platform/m100/sub/mmuutils.c is a library with MMU-related routines and helper macros.

The ARM CPU related library routines are identical in DIGIC 7 and 8 cameras (PowerShot and EOS). A copy of 100a ROM can be found here.

edit:
added a source files only archive (for those not wishing to deal with patches or compiling the source)

m100_100a_unfinished_201002.diff

336.25KB

m100_100a_201002_filesonly.7z

55.95KB

c_joerg

srsa_4c
This happens at boot time, in platform/m100/sub/100a/boot.c, in the function setup_mmu_for_chdk().

Where can i find boot.c for m100?
Here is no m100 platform
https://app.assembla.com/spaces/chdk/subversion/source/HEAD/trunk/platform/

srsa_4c

c_joerg
Where can i find boot.c for m100?
Here is no m100 platform
https://app.assembla.com/spaces/chdk/subversion/source/HEAD/trunk/platform/

Correct. I did not add the files to svn because the port is not "there" yet. I added an archive with loader and platform files to that post. If you do want to compile, you need to apply the patch to trunk source.

c_joerg

srsa_4c
If you do want to compile, you need to apply the patch to trunk source.

I tried...

shooting.c:138:3: error: 'MODE_HDR_BACKLIGHT_CTRL' undeclared here (not in a function)
 { MODE_HDR_BACKLIGHT_CTRL   ,32846 }, // hdr backlight control
   ^
gmake[1]: *** [shooting.o] Error 1
gmake: *** [all-recursive] Error 1

srsa_4c

c_joerg
I tried...

[code]
shooting.c:138:3: error: 'MODE_HDR_BACKLIGHT_CTRL' undeclared here (not in a function)
{ MODE_HDR_BACKLIGHT_CTRL ,32846 }, // hdr backlight control
^
gmake[1]: *** [shooting.o] Error 1
gmake: *** [all-recursive] Error 1[/code]

Patch == diff, the first file attached.
But you might be better off making the rest of changes by hand. Find the file indicated after "Index:", remove lines prefixed by -, add lines prefixed by +. Do not copy the prefix itself.

Index: core/conf.c
===================================================================
--- core/conf.c	(revision 5575)
+++ core/conf.c	(working copy)
@@ -1640,7 +1640,8 @@
        || (m == MODE_SPORTS)            // True if Sports mode (SX280HS, multiple issues with raw buffer; storing raw makes little sense in this mode anyway)
 #endif
 #if defined(CAM_DISABLE_RAW_IN_HDR)
-       || (m == MODE_HDR)            // True if HDR mode (G7X)
+       || (m == MODE_HDR)                // True if HDR mode (G7X)
+       || (m == MODE_HDR_BACKLIGHT_CTRL) // True if HDR backlight control mode (M100)
 #endif
     );
 }
Index: include/modelist.h
===================================================================
--- include/modelist.h	(revision 5575)
+++ include/modelist.h	(working copy)
@@ -123,6 +123,7 @@
     MODE_ZOOM_BLUR          , // ixus990_sd970 "The optical zoom operates automatically during shooting"
 
     MODE_PANNING            , // G7X2
+    MODE_HDR_BACKLIGHT_CTRL , // hdr backlight control, m100
 
 };

c_joerg

srsa_4cBut you might be better off making the rest of changes by hand.

Many Thanksâ€¦ 😀 😀 😀
I got CHDK working on my M100. It really works a lot. I haven't had a crash yet. The M100 with its many limited features will benefit greatly from CHDK.Would it be risky to modify and run M3M10_Focus.lua?

srsa_4cPatch == diff, the first file attached.

I understood that in principle. I'm just not that familiar with the tool chain. Is there somewhere a description of how to create and use the diff?

srsa_4c

c_joerg
I got CHDK working on my M100. It really works a lot. I haven't had a crash yet.

Cool. Let me know if that changes.

Would it be risky to modify and run M3M10_Focus.lua?

I made a modified one December last year, attached.

Is there somewhere a description of how to create and use the diff?

From command line, "svn diff" and "svn patch". Or you could install a GUI svn client (there's this for Windows). I don't have time for longer explanation, sorry.

I continued the transition to ROM-patch based source. At first, it crashed during boot. Then I realized I'm changing code/data inside CHDK core, so I put in an extra cache clean for that area too. Now it works, hopefully it will continue doing so. Combined archive attached.

m100_100a_201004.7z

93.39KB

M3M10M100_Focus.lua

13.03KB

c_joerg

srsa_4c
I made a modified one December last year, attached.

Thanks.
You made the changes on version 0.4.
I have already posted version 0.5 here. But I can merge it.
https://chdk.setepontos.com/d/13756/69

srsa_4c
From command line, "svn diff" and "svn patch". Or you could install a GUI svn client (there's this for Windows). I don't have time for longer explanation, sorry.

Itâ€™s fine. Something to start withâ€¦

I've tried rawopint.lua and MDFB2013.lua. Works very well on the usual. The only thing that I notice is that the optical temperature is higher than that of the sensor. I guess they are reversed.

Caefix

😀 😁 :-*

--- Modelist ---@ 0
PS12  PID:13009
0 0 IS0 ND0 (0)<1
m100-100a-CHDK_DE-1.5.0-5570
M5 SD5 Dm0 Mm2 F0 AF1L0
A/DCIM/204___10/Img_0972
:---=----=--+-=----=----
1 AUTO
2 P
3 TV
4 AV
5 M
12 CREATIVE_EFFECT
19 FISHEYE
21 HDR
26 LANDSCAPE
30 MINIATURE
31 MONOCHROME
33 NIGHT_SCENE
36 PORTRAIT
42 SOFTFOCUS
43 SPORTS
48 TOY_CAMERA
61 VIDEO_STD
63 VIDEO_TIME_LAPSE
64 HYBRID_AUTO
69 VIDEO_M
71 ART_BOLD
72 WATERPAINTING
73 CLOSEUP
74 FOOD
75 SELFIE
79 ??? Unknown ???
2020/10/6 # 17:35:20 ---
-26 ----- 2 Sec -----

---#Helperl ~ LOG_973.TXT
1@ PS12  PID:13009
m100-100a-CHDK_DE-1.5.0-5570
M5 SD5 Dm0 Mm2 F0 AF1L0
A/DCIM/204___10/Img_0972
0 IS0 ND0 (0)<1
2020/10/6 # 17:38:25 ---
Â°C Opt.37, CCD:31, Bat:30
700ms ((100003))1 5541mV$
972/ Bat:7650mV -74

With reloaded battery IÂ´ll look again. (Ubasic-Voodoodetector, tiny variation...)
...happens together with gk.log.

Exception!! [data abort]
 -> Write Access (0x00000000)
Occured Time  2020:10:06 16:52:41
ISR : vector=27
< Exc Registers >
R 0  : 0x40F0403D
R 1  : 0x00000000
R 2  : 0x00001000
R 3  : 0x00000000
R 4  : 0x40F0403D
R 5  : 0x00000000
R 6  : 0x40F0403D
R 7  : 0x00000000
R 8  : 0x00000000
R 9  : 0xE05B0AB4
R10  : 0x00000000
R11  : 0x00000000
R12  : 0x00000000
R13  : 0x002B07A0
R14  : 0xE0028277
PC   : 0x40F534AC
CPSR : 0x000001F3
DFSR : 0x00000807
IFSR : 0x0000000F
FAR  : 0x00000000
IFAR : 0x50FD6D1F
< StackDump >

M100-100a-CHDK_DE-1.5.0-5570-13009
 
Exception!! [data abort]
 -> Write Access (0x0000023D)
Occured Time  2020:10:06 20:56:33
ISR : vector=6
< Exc Registers >
R 0  : 0x40F0403D
R 1  : 0x00000000
R 2  : 0x00001000
R 3  : 0x0000023D
R 4  : 0x40F0403D
R 5  : 0x00000000
R 6  : 0x40F0403D
R 7  : 0x00000020
R 8  : 0x00000000
R 9  : 0xE05B0AB4
R10  : 0x00000000
R11  : 0x00000000
R12  : 0x00000000
R13  : 0x002B07A0
R14  : 0xE0028277
PC   : 0x40F53706
CPSR : 0x200001F3
DFSR : 0x00000807
IFSR : 0x0000040F
FAR  : 0x0000023D
IFAR : 0x54FD6D17
< StackDump >

... zoom_keys not defined 🤣 I enjoy the wheel...

c_joerg

Two things that I noticeâ€¦

1) The M100 changes its black level from 512 to 2048 at around ISO320. We know that from the G1x and S110

2) With the M100 there is a difference between ISO and sv96 value
The M100 has
ISO100 (sv) => sv96 = 480
All my other cameras have
ISO100 (sv) => sv96= 411

sv=sv96_to_iso(sv96_real_to_market(get_prop(props.SV)))

srsa_4c

c_joerg
1) The M100 changes its black level from 512 to 2048 at around ISO320. We know that from the G1x and S110

Thanks, I'll switch the port to calculate black level instead of assuming it being fixed.

2) With the M100 there is a difference between ISO and sv96 value
The M100 has
ISO100 (sv) => sv96 = 480
All my other cameras have
ISO100 (sv) => sv96= 411
[code]sv=sv96_to_iso(sv96_real_to_market(get_prop(props.SV)))[/code]

Both PROPCASE_SV_MARKET and PROPCASE_SV appear to have the same value, indeed 480 when setting ISO 100 with Canon controls.

@Caefix
I'd appreciate if you could tell how you crashed the camera. If it's a script, make that script as short as possible. And please don't waste my time with this sort of thing.

I recommend running a build from the latest source as I moved the CHDK core to a different memory area.

c_joerg

srsa_4c
Both PROPCASE_SV_MARKET and PROPCASE_SV appear to have the same value, indeed 480 when setting ISO 100 with Canon controls.

Right. It is the same with the M3. We already had this discussion here at the M3
https://chdk.setepontos.com/d/12542

Caefix

:😀 Without zoomkeys we have trouble to adjust some script settings.
So I modified my

// gui_menu.c
        case KEY_UP:
        case JOGDIAL_LEFT: //	M100 // 558 ~> Line 703
        case KEY_ZOOM_OUT:

        case JOGDIAL_RIGHT: //	M100 // 562 ~> Line 708
        case KEY_ZOOM_IN:

It affects all wheel cams, but it simplyfies cursor placement on cams with dual zoomlever, too.

// I dropped the 93kb zip content into my trunk, & it works really a lot... 😎
Here I listen to focus beeps in green frame, but the output is always 0 false.
How to get get_focus()? 🙁

--[[
@title FoTest.lua
--]]
repeat
press("shoot_half")
print("get_focus() ",get_focus(),get_shooting())
sleep (200)
print(" get_focus()",get_focus(),get_shooting())
release("shoot_half")
until get_wheel()~=0

srsa_4c

Caefix
:😀 Without zoomkeys we have trouble to adjust some script settings.

You can use half-shoot as replacement for moving the selected digit.

How to get get_focus()? 🙁

It's not implemented. It would rely on a firmware function named
[tt]GetFocusLensSubjectDistanceFromLens[/tt]
but that just returns zero. Same as on M3 and M10.

Caefix

😉 ...some other observations:
get_zoom()=0, dof.near=dof.hyp_dist, dof.far=-1, dof.dof=-1, dof.coc=19
Strange x=get_focus_state() values (bitmap?) in Auto (collected in Log 'growing' with script)
Cam ~'never' starts with first button press...

--[[
@title FoTestr.lua 
@param    M Mode
@default  M 2
@range    M 0 79
@param    s sleep
@default  s 400
@range    s 0 1000
--]]

capmode=require("capmode")
props=require("propcase")
print_screen(-125)
set_console_layout(0,0,45,6)
lineal=lineal or ":...,....1....,....2....,....3....,....4.."
print(lineal)

function m_info(a,b,X)
	dof=get_dofinfo()
	print(a,dof.focal_length,dof.eff_focal_length,capmode.get_canon(),capmode.get(),
	get_prop(65),-get_focus(),get_shooting() or b,get_focus_mode(),X,dof.focus_valid,get_focus_ok())
	print(get_zoom(),dof.min_stack_dist,dof.near,dof.focus,dof.far,dof.hyp_valid,dof.hyp_dist,dof.dof,dof.coc)
	print(get_mode(),-s,get_prop(props.AF_LOCK),get_prop(props.AF_FRAME),get_prop(props.METERING_MODE),get_prop(props.DRIVE_MODE),'--')
end

if is_capture_mode_valid(M) then set_capture_mode(M); play_sound(1) else M=2 end

repeat
	press("shoot_half")
	x=get_focus_state()
	if x~=0 then m_info(" ",7777,x) end

	sleep(s)
	x=get_focus_state()
	if x~=0 then m_info("-",8888,x) end
	release("shoot_half")

	sleep(2*s)
until get_wheel()~=0

set_console_layout(0,0,25,5)

LOG_0125.TXT

10.94KB

c_joerg

Caefix
😉 ...some other observations:
get_zoom()=0, dof.near=dof.hyp_dist, dof.far=-1, dof.dof=-1, dof.coc=19 Strange )

Did you read that and understand where the values â€‹â€‹come from?
https://chdk.setepontos.com/d/13756

Caefix
Cam ~'never' starts with first button press...

One Power on?

if I turn on the camera too quickly after inserting the SD, my M100 sometimes doesn't turn on either. I have to wait a bit when switching on (until the LED is off).

Caefix

c_joerg
One Power on?

I trained that ... & changed the card.
Do we need Boot to <play> after (short) ON on M100?
& Thx for nice links... 😎

srsa_4c

c_joerg
if I turn on the camera too quickly after inserting the SD, my M100 sometimes doesn't turn on either. I have to wait a bit when switching on (until the LED is off).

I see that, but it's not too bad (at least here). Waiting for a ~second is enough. Do any of you need to wait longer before the cam is willing to start?
There's a worse issue with startup. If I remove the battery and leave it out for some time, I have to wait a long time (15 seconds are enough*) before the camera can be successfully started. If I attempt to start sooner, I get the LED blink (sign of diskboot executing) and cam stays off. Unsuccessful attempts elongate the "dead" period.
I suspect these are connected to some startup check that my current code isn't handling correctly.

*edit:
corrected amount of "dead" time

c_joerg

srsa_4cI see that, but it's not too bad (at least here). Waiting for a ~second is enough. Do any of you need to wait longer before the cam is willing to start?

When I close the SD card door, the green LED goes on briefly twice. That takes maybe 2s. If I switch on after that, then everything is fine.
If I switch the camera on too early, the green LED stays on for about 10s and prevents it from being switched on again.

srsa_4cThere's a worse issue with startup. If I remove the battery and leave it out for some time, I have to wait a long time (15 seconds are enough*) before the camera can be successfully started. If I attempt to start sooner, I get the LED blink (sign of diskboot executing) and cam stays off. Unsuccessful attempts elongate the "dead" period.

I have not yet come into this situation.

I also had no other undefined crashes. Iâ€™m really happy right now

srsa_4c

c_joerg
When I close the SD card door, the green LED goes on briefly twice. That takes maybe 2s.

I've never seen it blinking twice. Maybe it depends on card...?

If I switch the camera on too early, the green LED stays on for about 10s and prevents it from being switched on again.

That's probably a lockup or crash (never seen that either). Do you get new romlog with matching timestamp after that (I suspect not)?

edit:
It looks like memory is getting corrupted, but only when booting with the above mentioned conditions (closing card door or battery out->in->closing door). Need to investigate...

edit2:
I also occurred to me that the CPU cache routines used in CHDK may not be adequate in a dual-core setup. We're probably only clearing the cache on one core...

« Previous Page Next Page »