Could need some help please. Can you post screenshots from "Show EDMAC" in debug menu while in Liveview mode with global draw on? The first overview page would suffice my needs. We can use this thread for collecting actual screenshots for all models. There are some old screenshots available here and there but not really much of 'em after merging-in FRSP branch. I would be interested in 5D3 (113 / 123) screenshots but any other DIGIC5 cam should do the trick for me, too.
Note: Some cameras have EDMAC_RAW_SLURP" enabled and the screenshots of those may vary...
70D.111A looked like this
before implementing some changes.
Note: added comments to the picture in yellow color (
click to resize):
(http://i60.tinypic.com/29pyvbo.png) (http://i60.tinypic.com/29pyvbo.png)
70D.111A looks like this
after using edmac raw_slurp to channel1:
(http://i59.tinypic.com/2s5zgi9.png) (http://i59.tinypic.com/2s5zgi9.png)
http://i59.tinypic.com/2s5zgi9.png
To be updated later by me for all 3 cases listed below.
Please add yours with model and FW version for the following cases like instructed (no more, no less):
- Case1: Turn on camera. Switch to LiveView. Take the screenshot from EDMAC's 1st overview page.
- Case2: Turn on camera. Switch to LiveView. Take a picture and afterwards take the screenshot from EDMAC's 1st overview page (no other actions in between).
- Case3: Turn on camera. Switch to MovieMode. Take a H264 video and while taking the video make a screenshot from EDMAC's 1st overview page.
- Case4: Turn on camera. Switch to MovieMode. Take a MLV video and while taking the video make a screenshot from EDMAC's 1st overview page.
EOSM.202 Case1: Turn on camera. Switch to LiveView. Take the screenshot from EDMAC's 1st overview page.
(https://i.imgur.com/lBKIECB.png)
EOSM.202 Case2: Turn on camera. Switch to LiveView. Take a picture and afterwards take the screenshot from EDMAC's 1st overview page (no other actions in between).
(https://i.imgur.com/fkHftBk.png)
EOSM.202 Case3: Turn on camera. Switch to MovieMode. Take a H264 video and while taking the video make a screenshot from EDMAC's 1st overview page.
(https://i.imgur.com/avvl8Nq.png)
[/size]EOSM.202[/size] Case4.1: Disable all modules. Turn on camera. [/size]Make a screenshot from EDMAC's 1st overview page.
(https://i.imgur.com/XQrGxTb.png)
[/size]
[/size]EOSM.202[/size] Case4.2: Enable MLV_REC modules. Turn on camera. Switch to MovieMode. Make a screenshot from EDMAC's 1st overview page.
(https://i.imgur.com/2q5blD7.png)
[/size]EOSM.202 Case4.3: Turn on camera. Switch to MovieMode. Take a MLV video and while taking the video make a screenshot from EDMAC's 1st overview page.
(https://i.imgur.com/gwZl2Wd.png)
/LE: Updated, again[/font]
5D3 1.2.3
Case 1:
(http://www.redyeti.net/extra/case1.png)
Case 2:
(http://www.redyeti.net/extra/case2.png)
Case 3:
(http://www.redyeti.net/extra/case3.png)
Thanx for sharing. Are yor screenshots done after a coldboot? İ did mine like that. Turn on cam in photo mode, switch to LV and take the screenshot w/o any other actıon in between. Can you check that and update your screenshot if result differs plz
I only did screeshots, the ML menu timeout bug makes it trickier.
BTW, after shooting,1 picture and few seconds of H.264 video, fields 3, 4, 10, 13, 21, 22, 25, 26 and 28 get populated with values ( buffer sizes? )
Ok can you please update your post with an additional screenshot after a fresh power-on and take a picture in LiveView. Afterwards take the same EDMAC screenshot please.
done
@Licaon_Kter. I appreciate all your efforts. See below...
@everybody: I updated the 1st post. Could you please update once again your post with the desired screenshots plz?
Canon 7D
------------
Immediately after the camera is turned on.
https://www.dropbox.com/s/12p8ijguoy6big0/VRAM0.PPM?dl=0
Live View turned on
https://www.dropbox.com/s/fa8exuqqsi4fehw/VRAM1.PPM?dl=0
Live View on, after picture taken
https://www.dropbox.com/s/jfpwszk3ho2ztax/VRAM2.PPM?dl=0
Whilst capturing video
https://www.dropbox.com/s/d9uhgw57205m3gu/VRAM3.PPM?dl=0
Cheers.... Otto
Updated, again
:P do share what you find so fascinating with them
550D-1.0.9. However do note that a lot of these values fluctuate, and they don't go stable, some of the ones that are green go grey and some grey go green.
In order: https://drive.google.com/folderview?id=0B_kjh0L1etrofmJhYlg5SENlVUhCaGVmYng4NEZ4ZTNBTTJvVG1WWlRXZEt3S1g5b3BQaXM&usp=sharing
Click the link, turns out the generated image link in google drive is temporary :P
Please everybody, can you add an additional screenshot or verify if mlv_rec was loaded and raw-recording option turned on in LV? see also my posted screenshot from 1st post with yellow colored comments. Also seen this discussion: http://magiclantern.fm/forum/index.php?topic=9741.msg148146#msg148146
I am especially interested what the EOS-M will show :P and I think chris_oversea's screenshots already have MLV_REC module enabled (plz confirm).
Note: while researching 70D I now got the feeling that EOS-M EDMAC's posted by Licaon show that there's something wrong on EOS-M EDMAC channel #8, too (just a guess but on 5D3 / 70D the correspondig edmac EDM#11 is green instead of red)
Updated yet again
Quote from: nikfreak on May 29, 2015, 03:16:04 PM
I think chris_oversea's screenshots already have MLV_REC module enabled (plz confirm).
No, MLV_REC, MLV_SND and RAW_REC were all disabled - I'm not a video guy so don't have any use for those modules.
Note that many of the EDMAC values and red/green/white colours change every second or so, which does make me wonder how much useful information you can obtain from a single screenshot for each of the three cases.
@chris: the values shown for edmac channel are worth it for me. It all depends on the mode you use (e.g. digital zoom or 5x etc).
For e.g. comparing your screenshots with those from my 1st post I immediately notice that something's not probably wrong on 70D but it differs. To be concrete: See pic2 of my 1st post EDM#1, #10, #28 differ in size. Compared to 5D3 screenshots from your case2 EDM#0, #10, #28, #21 all have the same size. I know, like you said, that the values change fast on the EDMAC overview pages. But that doesn't mind. Take the LV_buffers. Add 0x4000000 to your REG_EDMAC_WRITE_LV_ADDR and you will get the LV_Buffer addresses (consts.h).
Those channels that are being red marked seem to cause a problem for me on 70D. Again, taking the 2nd pic of my 1st post you will see that I miss a write channel on #21 or at least I can't seem to get it green colored so I guess there's some redirection problemor the channel gets disturbed by a feature / service. Additionally some might have noticed that 70D has also a problem with image review (zebras in QR mode look weird). I guess that's due to the size mismatches I am referring to above. I would appreciate comments from g3gg0 or a1ex on this. I may be proven wrong but before (=pic1) using edmac_raw_slurp #b32800 was being marked red for me and logs were full of "AbortCbr" messages. They are almost gone but still something's wrong (see EDM#21) and my logs still contain some garbage. I also feel that EOS-M got a problem with EDM#8 as per Licaon's screenshots. I can't say which feature / service would use that channel but a dm.log from EOS-M could help if my theory with green / red colors is correct.
Quote from: nikfreak on May 30, 2015, 01:10:00 PM
I also feel that EOS-M got a problem with EDM#8 as per Licaon's screenshots. I can't say which feature / service would use that channel but a dm.log from EOS-M could help if my theory with green / red colors is correct.
A what?
lol i will prepare a build for you to run on eos-m from dmspy branch. that contains a Debugmsg-Log entry in debug menu. it will start to create a dm.log file with 1MB once you chose that option. chose it again after some ime and it will stop creation and save dm.log to sdcard. i am not @home atm but if anyone else can share the build for eos-m we could speed things up a lil bit...
I can build, just plain dm-spy-experiments branch, right?
Yes
Only CONFIG_DEBUG_INTERCEPT = y is needed, right? ( it won't compile otherwise anyway )
Yes and check gettaskname stub
Quote from: nikfreak on May 30, 2015, 08:11:09 PM
Yes and check gettaskname stub
I'm not that advanced, do detail wtf that is. :P
NSTUB(0xFFA707A8, GetTaskName
already enabled and in there!
Here's the compiled zip for you. Post the dm.log so we can take a look. If my theory with EDM #8 is right then it shouldn't matter when you take the log (case 1-4-> all red)
https://drive.google.com/file/d/0B9Mu66yg5QzRZGQzNU04UlBRTGs/view?usp=sharing
I expect some interesting lines around LVx1_StartPreproPath.
Using your build:
Cold boot: https://bitbucket.org/snippets/Licaon_Kter/z8rb (https://bitbucket.org/snippets/Licaon_Kter/z8rb)
During RAW REC: https://bitbucket.org/snippets/Licaon_Kter/RXKG (https://bitbucket.org/snippets/Licaon_Kter/RXKG)
BTW, my build did not have the actual DebugLog menu item, strange.
Keep those logs online plz so I can compare with 70D!!! First look proves that my theory with #8 should be wrong but I would like to repeat the procedure at a later time by specifiying SetEdmac and StartEdmac for EOSM and share a new zip for you. Really appreciate all your efforts and feedbacks so far.
70D_111B
(http://i.imgur.com/jbxXmqj.jpg)
(http://i.imgur.com/A3LDJjs.jpg)
(http://i.imgur.com/0ED6llL.jpg)
(http://i.imgur.com/i4KXwAT.jpg)
digitalhouse you may have loaded MLV module but not turned it onso you see the raw resolution in ML menu, right? Do that and add the screenshot, too. 70D.111B looks same as 70D.111A.
@a1ex, @g3gg0:
I got a question and I am referring to the edmac screenshots from chris_overseas posted on 1st page here http://www.magiclantern.fm/forum/index.php?topic=15233.msg148199#msg148199.
On chris' screenshots EDMAC channel #5 and #21 seem to have the same write connection. At least both channels have (w1b) shown:
- 1440x779 <w1b>
- 3240x1079 <w1b>
from consts.h I see that 5D3.123 uses REG_EDMAC_WRITE_LV_ADDR 0xc0f04508. On the other side my screenshots from 1st post (and also dbgmsg's show that
70D uses EDM#0 REG_EDMAC_WRITE_LV_ADDR).
After using EDMAC_RAW_SLURP on 70D the debugmsg-logs got cleaned up from CRAW related messages (see 1st post). Now I
assume that #21 needs some kind of slurping, too. I would have another free write channel #33 for testing this. I can't understand why it is red colored and showing <
1b, 19>. Is this some kind of indication that there's something wrong or maybe the color means something like dual-use? On 5D3.123 it shows a simpple <
w1b> and is green colored. I assume that seeing REG_EDMAC_WRITE_LV_ADDR used on EDM #0 is not really a normal "behaviour" when comparing what's used on all other devices in consts.h.
Maybe I worry too much about EDMAC channels but I have some feeling there's something wrong after EDMAC_RAW_SLURP cleared up craw erros in debugmsg log's. So some words from you would be very welcome. Initially (..got a Q...) I thought I could maybe grab an instruction where (edmac-memcpy + raw.c) and how to redirect #21 to #33 or at least use #33 as some transfer/visibilty/connection channel for #0 or maybe move the EDMAC_WRITE_LV_ADDR from #0 to #33.
The red color (debug.c:2604) means something doesn't match our assumptions. You could try to find out which of the 3 branches is used (maybe assign 3 different colors).
REG_EDMAC_WRITE_LV_ADDR appears to be, indeed, on channel #0. It's the only one that matches the LCD resolution.
Ok redefined colors and for the suspicious EDM #21 it is this:
Quoteconn_r != 0xFF && dir != EDMAC_DIR_READ ? COLOR_RED : /* seems used for read, but dir is not read? */
From 1st post screenshot:Quote#21 f000000 3616x721 <1b,19>
I guess this one is also responsible for the broken zebras in QR. Btw I also undefined EDMAc_MEMCPY and EDMAC_RAW_SLURP but no change for EDM#21
1200D 1.0.1
(https://s31.postimg.org/6pxlqm7ob/VRAM0.jpg)
In case this is needed here.