Another call for testers

NiLuJe · 08-15-2018, 04:19 PM

Okay, now that FBInk 1.3.0 is out the door, I can focus on polishing one of the features it introduced: button_scan.

The idea, which came up during discussions around @sherman's kobo-rclone, is to automagically tap on the "Connect" button of the "USB Plugged In" popup.

After a few iterations, we've got something working, but there's a caveat:
While I'm pretty confident the "looking for the button" part is mostly okay everywhere (there may be a minor hitch on 16bpp FW (i.e., < 4.2)), I *know* the "pressing the button" part will NOT work everywhere (as-is).

Unfortunately, @sherman and I both have access only to a single model (H2O), so that's where you come in!

The idea is to install the prerelease version attached to this post (barring that, the *latest* FBInk version), and test the button_scan command (which means, again, yes, you'll need some kind of shell access to your device).

In particular, I'm looking for testers:

* On FW < 4.2
* With a H2O²r1
* With a Mini (where it *should* just work)
* With an Aura One (where it *should* just work)
* With an Aura (where it *should* just work)
* With a H2O²r2 (where it *should* just work)
* With an Aura SE (where it *should* just work)
* With an Aura SE r2
* With a Touch 2.0 (where it *should* just work)
* With a Libra

(I'll update that list as time goes on).

Tested:

* H2O [OK]
* Glo HD [OK]
* Glo [OK]
* Aura HD [OK]
* Clara HD [OK]
* Forma [OK]

If you have an old Touch, don't bother, those are going to be practically impossible to always handle right, sorry

.

I'll probably ask you to dump a proper screen capture of the USB popup, as well as a manually generated tap event in the general vicinity of the Connect button.

As for what to do, before running button_scan:

* Actually generate the popup, without needing to plug anything:
echo "usb plug add" > /tmp/nickel-hardware-status

* Capture a snapshot:
fbgrab usb.png
(You'll need a recent-ish FBGrab binary, either from KoboStuff or the standalone release).

* Capture an input event:
hexdump -x /dev/input/event1 (then a single tap on the connect button, and then ^C to interrupt hexdump).

* Test button_scan:
button_scan

If it succeeds in finding a button, you can add -p to make it attempt a touch, but see the next post before you cross that bridge

.

NOTE: On devices with native rotation support, please do this with Nickel in its native Portrait orientation (i.e., buttons on the right on the Libra/Forma).

NiLuJe · 08-15-2018, 04:23 PM

To avoid actually entering the "USB Connected" state for testing, you might want to simply: trigger the popup, capture a screenshot, *close* the popup, then display the screenshot with fbink (fbink -g file=usb.png), this ensures that a tap on that "fake" connect button will at worst only open a book or enter the Library

.

Because if you do want to really enter the "USB Connected" screen and still keep a shell session alive, that means USBNet, and you'll *have* to reboot your device after that, because trying to fake an unplug event will just crash the device when USBNet is enabled (... possibly with prejudice, on my H2O, it pretty much always wiped the stored WiFi passwords).

I do insist on the snapshot coming from FBGrab, to avoid any and all magical processing Nickel itself might do. The only exception to that rule *may* be for people on FW < 4.2, and even then, that's *in addition* to an FBGrab capture.

NiLuJe · 08-15-2018, 04:26 PM

As you may guess from the output of button_scan, I may need some more detailed info out of the poor guinea pig with a H2O²r1, because this one's extra weird

.

(So having a vague idea of what you're doing would be a plus ^^).

sherman · 08-15-2018, 05:29 PM

Hi, note that you can force the wifi to stay enabled when the USB is "connected" if you have enabled the "Force Wifi ON" setting in the developer options.

The developer options can be enabled with the following in the config file:

Quote:

[DeveloperSettings]
EnableDebugServices=true

NiLuJe · 08-15-2018, 07:16 PM

Ah, that's slightly less terrible than my "USBNet & reboot" approach, hadn't thought it'd persist through USBMS, good to know

.

EDIT: On my end, for WiFi to stick, I just needed

Code:

[DeveloperSettings]
ForceWifiOn=true

EnableDebugServices can stay false or missing (as I don't need the logging nor the telnet support

).

jsonora · 08-17-2018, 11:41 AM

I have tested it on my Glo HD and Glo.

Glo HD:

Code:

~ # hexdump -x /dev/input/event1
0000000    0148    0000    3ee4    0000    0003    0039    0001    0000
0000010    0148    0000    3ee7    0000    0003    0030    0001    0000
0000020    0148    0000    3ee9    0000    0003    0032    0001    0000
0000030    0148    0000    3eed    0000    0003    0035    03eb    0000
0000040    0148    0000    3eef    0000    0003    0036    0237    0000
0000050    0148    0000    3ef3    0000    0003    0018    0400    0000
0000060    0148    0000    3ef6    0000    0001    014a    0001    0000
0000070    0148    0000    3ef9    0000    0000    0002    0000    0000
0000080    0148    0000    3efb    0000    0000    0000    0000    0000
0000090    0148    0000    4357    0003    0003    0039    0001    0000
00000a0    0148    0000    435c    0003    0003    0030    0000    0000
00000b0    0148    0000    435e    0003    0003    0032    0000    0000
00000c0    0148    0000    4361    0003    0003    0035    03eb    0000
00000d0    0148    0000    4364    0003    0003    0036    0237    0000
00000e0    0148    0000    4367    0003    0003    0018    0000    0000
00000f0    0148    0000    4369    0003    0001    014a    0000    0000
0000100    0148    0000    436b    0003    0000    0002    0000    0000
0000110    0148    0000    436d    0003    0000    0000    0000    0000

~ # button_scan -p
[FBInk] Variable fb info: 1072x1448, 32bpp @ rotation: 3 (Counter Clockwise, 270°)
[FBInk] Fontsize set to 24x24 (IBM base glyph size: 8x8)
[FBInk] Line length: 44 cols, Page size: 60 rows
[FBInk] Fixed fb info: ID is "mxc_epdc_fb", length of fb mem: 6782976 bytes & line length: 4352 bytes
[FBInk] Kernel's HZ value appears to be 100
Button color is expected to be #D9D9D9
We need to match two buttons each between 134 and 268 pixels wide!
Looking for buttons in a 804x434 rectangle, from (53, 796) to (857, 1230)
Failed end of match after 1 consecutive matches @ (233, 835)
Failed end of match after 1 consecutive matches @ (499, 837)
Failed end of match after 1 consecutive matches @ (529, 840)
Failed end of match after 1 consecutive matches @ (530, 845)
Failed end of match after 1 consecutive matches @ (509, 847)
Failed end of match after 1 consecutive matches @ (408, 857)
Failed end of match after 1 consecutive matches @ (515, 858)
Failed end of match after 1 consecutive matches @ (504, 859)
Failed end of match after 1 consecutive matches @ (317, 911)
Failed end of match after 1 consecutive matches @ (478, 916)
Failed end of match after 1 consecutive matches @ (777, 918)
Failed end of match after 1 consecutive matches @ (686, 919)
Failed end of match after 1 consecutive matches @ (483, 925)
Failed end of match after 1 consecutive matches @ (722, 925)
Failed end of match after 1 consecutive matches @ (309, 926)
Failed end of match after 1 consecutive matches @ (788, 929)
Failed end of match after 1 consecutive matches @ (488, 934)
Failed end of match after 1 consecutive matches @ (342, 935)
Failed end of match after 1 consecutive matches @ (788, 935)
End of match 1 after 205 consecutive matches @ (403, 969)
End of match 2 after 205 consecutive matches @ (618, 969)
Matched on a 205x70 button! :)
x=1004, y=556
H2O²r1: x=443, y=515
Pressing the Connect button . . .
On iteration nr. 0, pixel (536, 1447) was #FFFFFF
On iteration nr. 1, pixel (536, 1447) was #FFFFFF
On iteration nr. 2, pixel (536, 1447) was #FFFFFF
On iteration nr. 3, pixel (536, 1447) was #000000
. . . appears to have been a success!

Glo:

Code:

~ # hexdump -x /dev/input/event1
0000000    d68b    5b76    d7b5    000c    0003    0001    019b    0000
0000010    d68b    5b76    dc74    000c    0003    0000    02cc    0000
0000020    d68b    5b76    de15    000c    0003    0018    0064    0000
0000030    d68b    5b76    df7e    000c    0001    014a    0001    0000
0000040    d68b    5b76    e0b9    000c    0000    0000    0000    0000
0000050    d68b    5b76    11ea    000e    0003    0001    019c    0000
0000060    d68b    5b76    146c    000e    0003    0000    02ce    0000
0000070    d68b    5b76    148a    000e    0003    0018    0065    0000
0000080    d68b    5b76    149d    000e    0000    0000    0000    0000
0000090    d68b    5b76    367a    000e    0003    0018    0000    0000
00000a0    d68b    5b76    36dd    000e    0001    014a    0000    0000
00000b0    d68b    5b76    3708    000e    0000    0000    0000    0000

~ # button_scan -p
[FBInk] Variable fb info: 758x1024, 32bpp @ rotation: 3 (Counter Clockwise, 270 °)
[FBInk] Fontsize set to 16x16 (IBM base glyph size: 8x8)
[FBInk] Line length: 47 cols, Page size: 64 rows
[FBInk] Fixed fb info: ID is "mxc_epdc_fb", length of fb mem: 3145728 bytes & line length: 3072 bytes
[FBInk] Kernel's HZ value appears to be 100
Button color is expected to be #D9D9D9
We need to match two buttons each between 94 and 189 pixels wide!
Looking for buttons in a 569x307 rectangle, from (37, 563) to (606, 870)
Failed end of match after 1 consecutive matches @ (134, 601)
Failed end of match after 1 consecutive matches @ (404, 609)
Failed end of match after 1 consecutive matches @ (206, 617)
Failed end of match after 1 consecutive matches @ (299, 620)
Failed end of match after 1 consecutive matches @ (371, 620)
Failed end of match after 1 consecutive matches @ (467, 662)
Failed end of match after 1 consecutive matches @ (558, 662)
Failed end of match after 1 consecutive matches @ (356, 669)
Failed end of match after 1 consecutive matches @ (306, 670)
Failed end of match after 1 consecutive matches @ (326, 672)
Failed end of match after 1 consecutive matches @ (461, 677)
End of match 1 after 148 consecutive matches @ (261, 700)
End of match 2 after 148 consecutive matches @ (419, 700)
Matched on a 148x56 button! :)
x=728, y=413
H2O²r1: x=295, y=344
Pressing the Connect button . . .
On iteration nr. 0, pixel (379, 1023) was #FFFFFF
On iteration nr. 1, pixel (379, 1023) was #FFFFFF
On iteration nr. 2, pixel (379, 1023) was #FFFFFF
On iteration nr. 3, pixel (379, 1023) was #FFFFFF
On iteration nr. 4, pixel (379, 1023) was #FFFFFF
On iteration nr. 5, pixel (379, 1023) was #FFFFFF
On iteration nr. 6, pixel (379, 1023) was #FFFFFF
On iteration nr. 7, pixel (379, 1023) was #FFFFFF
On iteration nr. 8, pixel (379, 1023) was #FFFFFF
On iteration nr. 9, pixel (379, 1023) was #FFFFFF
On iteration nr. 10, pixel (379, 1023) was #FFFFFF
On iteration nr. 11, pixel (379, 1023) was #FFFFFF
On iteration nr. 12, pixel (379, 1023) was #FFFFFF
On iteration nr. 13, pixel (379, 1023) was #FFFFFF
On iteration nr. 14, pixel (379, 1023) was #FFFFFF
On iteration nr. 15, pixel (379, 1023) was #FFFFFF
On iteration nr. 16, pixel (379, 1023) was #FFFFFF
On iteration nr. 17, pixel (379, 1023) was #FFFFFF
On iteration nr. 18, pixel (379, 1023) was #FFFFFF
On iteration nr. 19, pixel (379, 1023) was #FFFFFF
[FBInk] Failed to press the Connect button!

NiLuJe · 08-17-2018, 01:43 PM

@jsonora: That's exactly what I needed, thank you

.

Just to confirm: That worked properly on the Glo HD, right ^^ ?

And I assume the Glo did not support multi-touch gestures?

NiLuJe · 08-17-2018, 05:34 PM

I attached test binaries at the top, and updated the list.

If I didn't mess up, it adds support for older, non-MT devices (Touch A/B/C, Mini, Glo, Aura HD)

.

jsonora · 08-17-2018, 10:41 PM

Yes, it worked properly on the Glo HD.
And pinch to zoom doesn't work on the Glo. Multi-touch gestures don't seem to be supported.

FBInk-v1.4.0-15-g883243a worked on the Glo:

Code:

~ # hexdump -x /dev/input/event1
0000000    7936    5b77    6f68    000e    0003    0001    01a6    0000
0000010    7936    5b77    74aa    000e    0003    0000    02d1    0000
0000020    7936    5b77    74d1    000e    0003    0018    0064    0000
0000030    7936    5b77    74e0    000e    0001    014a    0001    0000
0000040    7936    5b77    74eb    000e    0000    0000    0000    0000
0000050    7936    5b77    e520    000e    0003    0000    02d0    0000
0000060    7936    5b77    e54d    000e    0003    0018    0065    0000
0000070    7936    5b77    e560    000e    0000    0000    0000    0000
0000080    7936    5b77    0c0d    000f    0003    0001    01a5    0000
0000090    7936    5b77    0c3e    000f    0003    0018    0064    0000
00000a0    7936    5b77    0c51    000f    0000    0000    0000    0000
00000b0    7936    5b77    3847    000f    0003    0000    02cf    0000
00000c0    7936    5b77    386f    000f    0003    0018    0065    0000
00000d0    7936    5b77    3882    000f    0000    0000    0000    0000
00000e0    7937    5b77    18e2    0000    0003    0001    01a4    0000
00000f0    7937    5b77    1908    0000    0003    0000    02ce    0000
0000100    7937    5b77    1913    0000    0003    0018    0064    0000
0000110    7937    5b77    1923    0000    0000    0000    0000    0000
0000120    7937    5b77    dee8    0002    0003    0018    0000    0000
0000130    7937    5b77    e176    0002    0001    014a    0000    0000
0000140    7937    5b77    e2ae    0002    0000    0000    0000    0000

~ # button_scan -p
[FBInk] Variable fb info: 758x1024, 32bpp @ rotation: 3 (Counter Clockwise, 270 °)
[FBInk] Fontsize set to 16x16 (IBM base glyph size: 8x8)
[FBInk] Line length: 47 cols, Page size: 64 rows
[FBInk] Fixed fb info: ID is "mxc_epdc_fb", length of fb mem: 3145728 bytes & line length: 3072 bytes
[FBInk] Enabled Kobo w/o Multi-Touch quirks
[FBInk] Kernel's HZ value appears to be 100
Button color is expected to be #D9D9D9
We need to match two buttons each between 94 and 189 pixels wide!
Looking for buttons in a 569x307 rectangle, from (37, 563) to (606, 870)
Failed end of match after 1 consecutive matches @ (134, 601)
Failed end of match after 1 consecutive matches @ (404, 609)
Failed end of match after 1 consecutive matches @ (206, 617)
Failed end of match after 1 consecutive matches @ (299, 620)
Failed end of match after 1 consecutive matches @ (371, 620)
Failed end of match after 1 consecutive matches @ (467, 662)
Failed end of match after 1 consecutive matches @ (558, 662)
Failed end of match after 1 consecutive matches @ (356, 669)
Failed end of match after 1 consecutive matches @ (306, 670)
Failed end of match after 1 consecutive matches @ (326, 672)
Failed end of match after 1 consecutive matches @ (461, 677)
End of match 1 after 148 consecutive matches @ (261, 700)
End of match 2 after 148 consecutive matches @ (419, 700)
Matched on a 148x56 button! :)
x=728, y=413
H2O²r1: x=295, y=344
Pressing the Connect button . . .
On iteration nr. 0, pixel (379, 1023) was #FFFFFF
On iteration nr. 1, pixel (379, 1023) was #FFFFFF
On iteration nr. 2, pixel (379, 1023) was #FFFFFF
On iteration nr. 3, pixel (379, 1023) was #FFFFFF
On iteration nr. 4, pixel (379, 1023) was #FFFFFF
On iteration nr. 5, pixel (379, 1023) was #000000
. . . appears to have been a success!

NiLuJe · 08-18-2018, 02:18 AM

@jsonora: Fantastic, many thanks!

For those following along, that leaves Mk7 devices, and the dreaded H2O²...

jsonora · 08-20-2018, 09:48 AM

I got my Clara HD yesterday.

"button_scan -p" failed.

Clara HD:

Code:

~ # hexdump -x /dev/input/event1
0000000    74e6    5b79    1d78    0000    0001    0145    0001    0000
0000010    74e6    5b79    1d78    0000    0001    014a    0001    0000
0000020    74e6    5b79    1d78    0000    0003    0039    0000    0000
0000030    74e6    5b79    1d78    0000    0003    003b    0000    0000
0000040    74e6    5b79    1d78    0000    0003    0035    03f2    0000
0000050    74e6    5b79    1d78    0000    0003    0036    0230    0000
0000060    74e6    5b79    1d78    0000    0003    003a    0013    0000
0000070    74e6    5b79    1d78    0000    0003    0030    0000    0000
0000080    74e6    5b79    1d78    0000    0003    0031    0000    0000
0000090    74e6    5b79    1d78    0000    0003    0034    0000    0000
00000a0    74e6    5b79    1d78    0000    0000    0002    0000    0000
00000b0    74e6    5b79    1d78    0000    0000    0000    0000    0000
00000c0    74e6    5b79    491d    0000    0003    0039    0000    0000
00000d0    74e6    5b79    491d    0000    0003    003b    0000    0000
00000e0    74e6    5b79    491d    0000    0003    0035    03f2    0000
00000f0    74e6    5b79    491d    0000    0003    0036    0230    0000
0000100    74e6    5b79    491d    0000    0003    003a    0013    0000
0000110    74e6    5b79    491d    0000    0003    0030    0000    0000
0000120    74e6    5b79    491d    0000    0003    0031    0000    0000
0000130    74e6    5b79    491d    0000    0003    0034    0000    0000
0000140    74e6    5b79    491d    0000    0000    0002    0000    0000
0000150    74e6    5b79    491d    0000    0000    0000    0000    0000
0000160    74e6    5b79    74bc    0000    0003    0039    0000    0000
0000170    74e6    5b79    74bc    0000    0003    003b    0000    0000
0000180    74e6    5b79    74bc    0000    0003    0035    03f2    0000
0000190    74e6    5b79    74bc    0000    0003    0036    0231    0000
00001a0    74e6    5b79    74bc    0000    0003    003a    0013    0000
00001b0    74e6    5b79    74bc    0000    0003    0030    0000    0000
00001c0    74e6    5b79    74bc    0000    0003    0031    0000    0000
00001d0    74e6    5b79    74bc    0000    0003    0034    0000    0000
00001e0    74e6    5b79    74bc    0000    0000    0002    0000    0000
00001f0    74e6    5b79    74bc    0000    0000    0000    0000    0000
0000200    74e6    5b79    a208    0000    0003    0039    0000    0000
0000210    74e6    5b79    a208    0000    0003    003b    0000    0000
0000220    74e6    5b79    a208    0000    0003    0035    03f3    0000
0000230    74e6    5b79    a208    0000    0003    0036    0231    0000
0000240    74e6    5b79    a208    0000    0003    003a    0013    0000
0000250    74e6    5b79    a208    0000    0003    0030    0000    0000
0000260    74e6    5b79    a208    0000    0003    0031    0000    0000
0000270    74e6    5b79    a208    0000    0003    0034    0000    0000
0000280    74e6    5b79    a208    0000    0000    0002    0000    0000
0000290    74e6    5b79    a208    0000    0000    0000    0000    0000
00002a0    74e6    5b79    c840    0000    0003    0039    0000    0000
00002b0    74e6    5b79    c840    0000    0003    003b    0000    0000
00002c0    74e6    5b79    c840    0000    0003    0035    03f3    0000
00002d0    74e6    5b79    c840    0000    0003    0036    0231    0000
00002e0    74e6    5b79    c840    0000    0003    003a    0013    0000
00002f0    74e6    5b79    c840    0000    0003    0030    0000    0000
0000300    74e6    5b79    c840    0000    0003    0031    0000    0000
0000310    74e6    5b79    c840    0000    0003    0034    0000    0000
0000320    74e6    5b79    c840    0000    0000    0002    0000    0000
0000330    74e6    5b79    c840    0000    0000    0000    0000    0000
0000340    74e6    5b79    f32b    0000    0003    0039    0000    0000
0000350    74e6    5b79    f32b    0000    0003    003b    0000    0000
0000360    74e6    5b79    f32b    0000    0003    0035    03f3    0000
0000370    74e6    5b79    f32b    0000    0003    0036    0231    0000
0000380    74e6    5b79    f32b    0000    0003    003a    0014    0000
0000390    74e6    5b79    f32b    0000    0003    0030    0000    0000
00003a0    74e6    5b79    f32b    0000    0003    0031    0000    0000
00003b0    74e6    5b79    f32b    0000    0003    0034    0000    0000
00003c0    74e6    5b79    f32b    0000    0000    0002    0000    0000
00003d0    74e6    5b79    f32b    0000    0000    0000    0000    0000
00003e0    74e6    5b79    1f95    0001    0003    0039    0000    0000
00003f0    74e6    5b79    1f95    0001    0003    003b    0000    0000
0000400    74e6    5b79    1f95    0001    0003    0035    03f4    0000
0000410    74e6    5b79    1f95    0001    0003    0036    0231    0000
0000420    74e6    5b79    1f95    0001    0003    003a    0014    0000
0000430    74e6    5b79    1f95    0001    0003    0030    0000    0000
0000440    74e6    5b79    1f95    0001    0003    0031    0000    0000
0000450    74e6    5b79    1f95    0001    0003    0034    0000    0000
0000460    74e6    5b79    1f95    0001    0000    0002    0000    0000
0000470    74e6    5b79    1f95    0001    0000    0000    0000    0000
0000480    74e6    5b79    49ad    0001    0003    0039    0000    0000
0000490    74e6    5b79    49ad    0001    0003    003b    0000    0000
00004a0    74e6    5b79    49ad    0001    0003    0035    03f4    0000
00004b0    74e6    5b79    49ad    0001    0003    0036    022f    0000
00004c0    74e6    5b79    49ad    0001    0003    003a    0014    0000
00004d0    74e6    5b79    49ad    0001    0003    0030    0000    0000
00004e0    74e6    5b79    49ad    0001    0003    0031    0000    0000
00004f0    74e6    5b79    49ad    0001    0003    0034    0000    0000
0000500    74e6    5b79    49ad    0001    0000    0002    0000    0000
0000510    74e6    5b79    49ad    0001    0000    0000    0000    0000
0000520    74e6    5b79    9ca1    0001    0001    014a    0000    0000
0000530    74e6    5b79    9ca1    0001    0001    0145    0000    0000
0000540    74e6    5b79    9ca1    0001    0000    0000    0000    0000

~ # button_scan -p
[FBInk] Variable fb info: 1072x1448, 32bpp @ rotation: 3 (Counter Clockwise, 270°)
[FBInk] Fontsize set to 24x24 (IBM base glyph size: 8x8)
[FBInk] Line length: 44 cols, Page size: 60 rows
[FBInk] Fixed fb info: ID is "mxc_epdc_fb", length of fb mem: 6782976 bytes & line length: 4352 bytes
[FBInk] Enabled Kobo Mark 7 quirks
[FBInk] Kernel's HZ value appears to be 100
Button color is expected to be #D9D9D9
We need to match two buttons each between 134 and 268 pixels wide!
Looking for buttons in a 804x434 rectangle, from (53, 796) to (857, 1230)
Failed end of match after 1 consecutive matches @ (233, 835)
Failed end of match after 1 consecutive matches @ (499, 837)
Failed end of match after 1 consecutive matches @ (529, 840)
Failed end of match after 1 consecutive matches @ (530, 845)
Failed end of match after 1 consecutive matches @ (509, 847)
Failed end of match after 1 consecutive matches @ (408, 857)
Failed end of match after 1 consecutive matches @ (515, 858)
Failed end of match after 1 consecutive matches @ (504, 859)
Failed end of match after 1 consecutive matches @ (317, 911)
Failed end of match after 1 consecutive matches @ (478, 916)
Failed end of match after 1 consecutive matches @ (777, 918)
Failed end of match after 1 consecutive matches @ (686, 919)
Failed end of match after 1 consecutive matches @ (483, 925)
Failed end of match after 1 consecutive matches @ (722, 925)
Failed end of match after 1 consecutive matches @ (309, 926)
Failed end of match after 1 consecutive matches @ (788, 929)
Failed end of match after 1 consecutive matches @ (488, 934)
Failed end of match after 1 consecutive matches @ (342, 935)
Failed end of match after 1 consecutive matches @ (788, 935)
End of match 1 after 205 consecutive matches @ (403, 969)
End of match 2 after 205 consecutive matches @ (618, 969)
Matched on a 205x70 button! :)
x=1004, y=556
H2O²r1: x=443, y=515
Pressing the Connect button . . .
On iteration nr. 0, pixel (536, 1447) was #FFFFFF
On iteration nr. 1, pixel (536, 1447) was #FFFFFF
On iteration nr. 2, pixel (536, 1447) was #FFFFFF
On iteration nr. 3, pixel (536, 1447) was #FFFFFF
On iteration nr. 4, pixel (536, 1447) was #FFFFFF
On iteration nr. 5, pixel (536, 1447) was #FFFFFF
On iteration nr. 6, pixel (536, 1447) was #FFFFFF
On iteration nr. 7, pixel (536, 1447) was #FFFFFF
On iteration nr. 8, pixel (536, 1447) was #FFFFFF
On iteration nr. 9, pixel (536, 1447) was #FFFFFF
On iteration nr. 10, pixel (536, 1447) was #FFFFFF
On iteration nr. 11, pixel (536, 1447) was #FFFFFF
On iteration nr. 12, pixel (536, 1447) was #FFFFFF
On iteration nr. 13, pixel (536, 1447) was #FFFFFF
On iteration nr. 14, pixel (536, 1447) was #FFFFFF
On iteration nr. 15, pixel (536, 1447) was #FFFFFF
On iteration nr. 16, pixel (536, 1447) was #FFFFFF
On iteration nr. 17, pixel (536, 1447) was #FFFFFF
On iteration nr. 18, pixel (536, 1447) was #FFFFFF
On iteration nr. 19, pixel (536, 1447) was #FFFFFF
[FBInk] Failed to press the Connect button!

patrik · 08-20-2018, 11:46 AM

Kobo Aura HD

Code:

[root@(none) ~]# hexdump -x /dev/input/event1
0000000    e0b5    5b7a    6fb0    000a    0003    0001    024e    0000
0000010    e0b5    5b7a    7473    000a    0003    0000    0403    0000
0000020    e0b5    5b7a    7498    000a    0003    0018    0064    0000
0000030    e0b5    5b7a    74a5    000a    0001    014a    0001    0000
0000040    e0b5    5b7a    74b1    000a    0000    0000    0000    0000
0000050    e0b5    5b7a    106b    000b    0003    0000    0402    0000
0000060    e0b5    5b7a    1094    000b    0003    0018    0065    0000
0000070    e0b5    5b7a    10a6    000b    0000    0000    0000    0000
0000080    e0b5    5b7a    27c9    000b    0003    0018    0000    0000
0000090    e0b5    5b7a    27ec    000b    0001    014a    0000    0000
00000a0    e0b5    5b7a    27f4    000b    0000    0000    0000    0000

Code:

[root@(none) ~]# button_scan
[FBInk] Variable fb info: 1080x1440, 32bpp @ rotation: 1 (Clockwise, 90°)
[FBInk] Fontsize set to 24x24 (IBM base glyph size: 8x8)
[FBInk] Line length: 45 cols, Page size: 60 rows
[FBInk] It's a perfect fit!
[FBInk] Fixed fb info: ID is "mxc_epdc_fb", length of fb mem: 6684672 bytes & line length: 4352 bytes
[FBInk] Kernel's HZ value appears to be 100
Button color is expected to be #D9D9D9
We need to match two buttons each between 135 and 270 pixels wide!
Looking for buttons in a 810x432 rectangle, from (54, 792) to (864, 1224)
Failed end of match after 1 consecutive matches @ (522, 825)
Failed end of match after 1 consecutive matches @ (313, 826)
Failed end of match after 1 consecutive matches @ (385, 827)
Failed end of match after 1 consecutive matches @ (463, 827)
Failed end of match after 1 consecutive matches @ (501, 827)
Failed end of match after 1 consecutive matches @ (560, 827)
Failed end of match after 1 consecutive matches @ (120, 886)
Failed end of match after 1 consecutive matches @ (217, 902)
Failed end of match after 1 consecutive matches @ (110, 910)
Failed end of match after 1 consecutive matches @ (151, 914)
End of match 1 after 184 consecutive matches @ (390, 989)
End of match 2 after 184 consecutive matches @ (584, 989)
Matched on a 184x70 button! :)
x=1024, y=588
H2O²r1: x=415, y=491

Code:

[root@(none) ~]# button_scan -p
[FBInk] Variable fb info: 1080x1440, 32bpp @ rotation: 1 (Clockwise, 90°)
[FBInk] Fontsize set to 24x24 (IBM base glyph size: 8x8)
[FBInk] Line length: 45 cols, Page size: 60 rows
[FBInk] It's a perfect fit!
[FBInk] Fixed fb info: ID is "mxc_epdc_fb", length of fb mem: 6684672 bytes & line length: 4352 bytes
[FBInk] Kernel's HZ value appears to be 100
Button color is expected to be #D9D9D9
We need to match two buttons each between 135 and 270 pixels wide!
Looking for buttons in a 810x432 rectangle, from (54, 792) to (864, 1224)
Failed end of match after 1 consecutive matches @ (522, 825)
Failed end of match after 1 consecutive matches @ (313, 826)
Failed end of match after 1 consecutive matches @ (385, 827)
Failed end of match after 1 consecutive matches @ (463, 827)
Failed end of match after 1 consecutive matches @ (501, 827)
Failed end of match after 1 consecutive matches @ (560, 827)
Failed end of match after 1 consecutive matches @ (120, 886)
Failed end of match after 1 consecutive matches @ (217, 902)
Failed end of match after 1 consecutive matches @ (110, 910)
Failed end of match after 1 consecutive matches @ (151, 914)
End of match 1 after 184 consecutive matches @ (390, 989)
End of match 2 after 184 consecutive matches @ (584, 989)
Matched on a 184x70 button! :)
x=1024, y=588
H2O²r1: x=415, y=491
Pressing the Connect button . . .
On iteration nr. 0, pixel (540, 1439) was #FFFFFF
On iteration nr. 1, pixel (540, 1439) was #FFFFFF
On iteration nr. 2, pixel (540, 1439) was #FFFFFF
On iteration nr. 3, pixel (540, 1439) was #FFFFFF
On iteration nr. 4, pixel (540, 1439) was #FFFFFF
On iteration nr. 5, pixel (540, 1439) was #FFFFFF
On iteration nr. 6, pixel (540, 1439) was #FFFFFF
On iteration nr. 7, pixel (540, 1439) was #FFFFFF
On iteration nr. 8, pixel (540, 1439) was #FFFFFF
On iteration nr. 9, pixel (540, 1439) was #FFFFFF
On iteration nr. 10, pixel (540, 1439) was #FFFFFF
On iteration nr. 11, pixel (540, 1439) was #FFFFFF
On iteration nr. 12, pixel (540, 1439) was #FFFFFF
On iteration nr. 13, pixel (540, 1439) was #FFFFFF
On iteration nr. 14, pixel (540, 1439) was #FFFFFF
On iteration nr. 15, pixel (540, 1439) was #FFFFFF
On iteration nr. 16, pixel (540, 1439) was #FFFFFF
On iteration nr. 17, pixel (540, 1439) was #FFFFFF
On iteration nr. 18, pixel (540, 1439) was #FFFFFF
On iteration nr. 19, pixel (540, 1439) was #FFFFFF
[FBInk] Failed to press the Connect button!

NiLuJe · 08-20-2018, 12:11 PM

@jsonora: Thank you (again

). I updated the package in the first post (it *might* take a few tries, the pressure events are all over the place, so I took some liberties to implement something that made sense vs. what I caught from your log

).

@patrik: Make sure you're using the package from the first post of *this* thread, instead of the latest FBInk release from its own thread

. ( And/or double-check your model: tail -c 3 /mnt/onboard/.kobo/version )

patrik · 08-20-2018, 12:18 PM

I installed fbink from _this_ thread, did install kbstuff after in case that overwrote something.

Check my model? Don't you think I know what model I have?!? Or I am missing something.

Output: 350

NiLuJe · 08-20-2018, 12:25 PM

@patrik: There should be an extra line of setup log for that model, which I'm not seeing in your log, which leads me to believe you're using an outdated binary (or that I fucked up the detection of the Aura HD in particular, because it's in a different switch block than the other devices like it).

That's the reasoning behind me asking for the exact model number

.

The bad news is, I can't see anything wrong in the code, and the check looks okay for the Aura HD (whose id is, indeed, 350

), so my next question will be *another* stupid one, sorry: if you run fbink with no argument, it'll print the help screen, and at the top, the version should be FBInk v1.4.0-56-g2aff05a for Kobo.

EDIT: Whoop, stop the presses, yes, it's the version from KoboStuff that overwrote it (and that one is indeed a tiny bit older right now)

.

08-15-2018, 04:19 PM	#1
NiLuJe BLAM! Posts: 13,477 Karma: 26012492 Join Date: Jun 2010 Location: Paris, France Device: Kindle 2i, 3g, 4, 5w, PW, PW2, PW5; Kobo H2O, Forma, Elipsa, Sage, C2E	Another call for testers Okay, now that FBInk 1.3.0 is out the door, I can focus on polishing one of the features it introduced: button_scan. The idea, which came up during discussions around @sherman's kobo-rclone, is to automagically tap on the "Connect" button of the "USB Plugged In" popup. After a few iterations, we've got something working, but there's a caveat: While I'm pretty confident the "looking for the button" part is mostly okay everywhere (there may be a minor hitch on 16bpp FW (i.e., < 4.2)), I know the "pressing the button" part will NOT work everywhere (as-is). Unfortunately, @sherman and I both have access only to a single model (H2O), so that's where you come in! The idea is to install the prerelease version attached to this post (barring that, the *latest* FBInk version), and test the button_scan command (which means, again, yes, you'll need some kind of shell access to your device). In particular, I'm looking for testers: * On FW < 4.2 * With a H2O²r1 * With a Mini (where it should just work) * With an Aura One (where it should just work) * With an Aura (where it should just work) * With a H2O²r2 (where it should just work) * With an Aura SE (where it should just work) * With an Aura SE r2 * With a Touch 2.0 (where it should just work) * With a Libra (I'll update that list as time goes on). Tested: * H2O [OK] * Glo HD [OK] * Glo [OK] * Aura HD [OK] * Clara HD [OK] * Forma [OK] If you have an old Touch, don't bother, those are going to be practically impossible to always handle right, sorry . I'll probably ask you to dump a proper screen capture of the USB popup, as well as a manually generated tap event in the general vicinity of the Connect button. As for what to do, before running button_scan: * Actually generate the popup, without needing to plug anything: echo "usb plug add" > /tmp/nickel-hardware-status * Capture a snapshot: fbgrab usb.png (You'll need a recent-ish FBGrab binary, either from KoboStuff or the standalone release). * Capture an input event: hexdump -x /dev/input/event1 (then a single tap on the connect button, and then ^C to interrupt hexdump). * Test button_scan: button_scan If it succeeds in finding a button, you can add -p to make it attempt a touch, but see the next post before you cross that bridge . NOTE: On devices with native rotation support, please do this with Nickel in its native Portrait orientation (i.e., buttons on the right on the Libra/Forma). Last edited by NiLuJe; 09-30-2019 at 09:52 PM. Reason: No rc currently needed

08-15-2018, 04:23 PM	#2
NiLuJe BLAM! Posts: 13,477 Karma: 26012492 Join Date: Jun 2010 Location: Paris, France Device: Kindle 2i, 3g, 4, 5w, PW, PW2, PW5; Kobo H2O, Forma, Elipsa, Sage, C2E	To avoid actually entering the "USB Connected" state for testing, you might want to simply: trigger the popup, capture a screenshot, close the popup, then display the screenshot with fbink (fbink -g file=usb.png), this ensures that a tap on that "fake" connect button will at worst only open a book or enter the Library . Because if you do want to really enter the "USB Connected" screen and still keep a shell session alive, that means USBNet, and you'll have to reboot your device after that, because trying to fake an unplug event will just crash the device when USBNet is enabled (... possibly with prejudice, on my H2O, it pretty much always wiped the stored WiFi passwords). I do insist on the snapshot coming from FBGrab, to avoid any and all magical processing Nickel itself might do. The only exception to that rule may be for people on FW < 4.2, and even then, that's in addition to an FBGrab capture. Last edited by NiLuJe; 08-15-2018 at 04:42 PM.

08-15-2018, 07:16 PM	#5
NiLuJe BLAM! Posts: 13,477 Karma: 26012492 Join Date: Jun 2010 Location: Paris, France Device: Kindle 2i, 3g, 4, 5w, PW, PW2, PW5; Kobo H2O, Forma, Elipsa, Sage, C2E	Ah, that's slightly less terrible than my "USBNet & reboot" approach, hadn't thought it'd persist through USBMS, good to know . EDIT: On my end, for WiFi to stick, I just needed Code: [DeveloperSettings] ForceWifiOn=true EnableDebugServices can stay false or missing (as I don't need the logging nor the telnet support ). Last edited by NiLuJe; 09-12-2018 at 09:09 PM.

08-17-2018, 01:43 PM	#7
NiLuJe BLAM! Posts: 13,477 Karma: 26012492 Join Date: Jun 2010 Location: Paris, France Device: Kindle 2i, 3g, 4, 5w, PW, PW2, PW5; Kobo H2O, Forma, Elipsa, Sage, C2E	@jsonora: That's exactly what I needed, thank you . Just to confirm: That worked properly on the Glo HD, right ^^ ? And I assume the Glo did not support multi-touch gestures? Last edited by NiLuJe; 08-17-2018 at 05:33 PM.

08-20-2018, 12:11 PM	#13
NiLuJe BLAM! Posts: 13,477 Karma: 26012492 Join Date: Jun 2010 Location: Paris, France Device: Kindle 2i, 3g, 4, 5w, PW, PW2, PW5; Kobo H2O, Forma, Elipsa, Sage, C2E	@jsonora: Thank you (again ). I updated the package in the first post (it might take a few tries, the pressure events are all over the place, so I took some liberties to implement something that made sense vs. what I caught from your log ). @patrik: Make sure you're using the package from the first post of this thread, instead of the latest FBInk release from its own thread . ( And/or double-check your model: tail -c 3 /mnt/onboard/.kobo/version ) Last edited by NiLuJe; 08-20-2018 at 12:14 PM.

08-15-2018, 04:26 PM	#3
NiLuJe BLAM! Posts: 13,477 Karma: 26012492 Join Date: Jun 2010 Location: Paris, France Device: Kindle 2i, 3g, 4, 5w, PW, PW2, PW5; Kobo H2O, Forma, Elipsa, Sage, C2E	As you may guess from the output of button_scan, I may need some more detailed info out of the poor guinea pig with a H2O²r1, because this one's extra weird . (So having a vague idea of what you're doing would be a plus ^^).

08-17-2018, 05:34 PM	#8
NiLuJe BLAM! Posts: 13,477 Karma: 26012492 Join Date: Jun 2010 Location: Paris, France Device: Kindle 2i, 3g, 4, 5w, PW, PW2, PW5; Kobo H2O, Forma, Elipsa, Sage, C2E	I attached test binaries at the top, and updated the list. If I didn't mess up, it adds support for older, non-MT devices (Touch A/B/C, Mini, Glo, Aura HD) .

08-18-2018, 02:18 AM	#10
NiLuJe BLAM! Posts: 13,477 Karma: 26012492 Join Date: Jun 2010 Location: Paris, France Device: Kindle 2i, 3g, 4, 5w, PW, PW2, PW5; Kobo H2O, Forma, Elipsa, Sage, C2E	@jsonora: Fantastic, many thanks! For those following along, that leaves Mk7 devices, and the dreaded H2O²...

08-20-2018, 12:18 PM	#14
patrik Guru Posts: 653 Karma: 4568205 Join Date: Jan 2010 Location: Sweden Device: Kobo Forma	I installed fbink from _this_ thread, did install kbstuff after in case that overwrote something. Check my model? Don't you think I know what model I have?!? Or I am missing something. Output: 350

08-20-2018, 12:25 PM	#15
NiLuJe BLAM! Posts: 13,477 Karma: 26012492 Join Date: Jun 2010 Location: Paris, France Device: Kindle 2i, 3g, 4, 5w, PW, PW2, PW5; Kobo H2O, Forma, Elipsa, Sage, C2E	@patrik: There should be an extra line of setup log for that model, which I'm not seeing in your log, which leads me to believe you're using an outdated binary (or that I fucked up the detection of the Aura HD in particular, because it's in a different switch block than the other devices like it). That's the reasoning behind me asking for the exact model number . The bad news is, I can't see anything wrong in the code, and the check looks okay for the Aura HD (whose id is, indeed, 350 ), so my next question will be another stupid one, sorry: if you run fbink with no argument, it'll print the help screen, and at the top, the version should be FBInk v1.4.0-56-g2aff05a for Kobo. EDIT: Whoop, stop the presses, yes, it's the version from KoboStuff that overwrote it (and that one is indeed a tiny bit older right now) . Last edited by NiLuJe; 08-20-2018 at 12:28 PM.

Similar Threads
Thread	Thread Starter	Forum	Replies	Last Post
Call for testers with a Mark 7 device	NiLuJe	Kobo Developer's Corner	9	07-13-2018 12:04 PM
call for testers from source: dramatic performance improvement	chaley	Calibre	3	12-15-2010 09:45 AM
OS X testers needed	kovidgoyal	Calibre	23	11-18-2010 12:11 PM
Call for testers	kovidgoyal	Calibre	289	09-19-2008 01:45 PM