Skip to contentSkip to conversation

support japanese JP106 keyboard on new console.

PATCH In review Hiroshi Miura 5 participants 25 messages Jan 24, 2003 – Feb 5, 2003 Input drivers

lore
Share & tools
lore
#0 lore raw

PATCH 1/1 support japanese JP106 keyboard on new console.

Hi,

After re-writting a console layer, a japanese keyboard is not supported (or degraded). This patch fixs it.

A USB keyboard driver may have same problem, but I don't have one.

linux-2.5.59/drivers/input/keyboard/atkbd.c renamed from drivers/input/keyboard/atkbd.c +13 -1
@@ -309,6 +309,12 @@
 	if (atkbd_command(atkbd, &atkbd->oldset, ATKBD_CMD_GSCANSET))
 		atkbd->oldset = 2;
 
+	if (atkbd->id == 0xab02) {
+		printk("atkbd: jp109(106) keyboard found\n");
+		param[0] = atkbd_set;
+		atkbd_command(atkbd, param, ATKBD_CMD_SSCANSET);
+		return 5;
+	}
 /*
  * For known special keyboards we can go ahead and set the correct set.
  * We check for NCD PS/2 Sun, NorthGate OmniKey 101 and
@@ -531,6 +537,12 @@
 	else
 		memcpy(atkbd->keycode, atkbd_set2_keycode, sizeof(atkbd->keycode));
 
+	if (atkbd->set == 5) {
+		atkbd->keycode[0x13] = 0x70;  /* Hiragana/Katakana */
+		atkbd->keycode[0x6a] = 0x7c;  /* Yen, pipe 124*/
+		atkbd->set = 2;
+	}
+
 	atkbd->dev.name = atkbd->name;
 	atkbd->dev.phys = atkbd->phys;
 	atkbd->dev.id.bustype = BUS_I8042;
@@ -544,7 +556,7 @@
 
 	input_register_device(&atkbd->dev);
 
-	printk(KERN_INFO "input: %s on %s\n", atkbd->name, serio->phys);
+	printk(KERN_INFO "input: %s (0x%x) on %s\n", atkbd->name, atkbd->id, serio->phys);
 }
 
 
Signature · 4 lines
Hiroshi Miura  --- http://www.da-cha.org/ 
NTTDATA Corp. Marketing & Business Strategy Planning Dept. --- [email protected] 
Key fingerprint = 9117 9407 5684 FBF1 4063  15B4 401D D077 04AB 8617
-- My hacking life is happy as the day is long
#1 lore raw
Quoted 2 lines · Hiroshi Miura

After re-writting a console layer, a japanese keyboard is not supported (or degraded). This patch fixs it.

This patch doesn't work, all normal keyboards - not just japanese ones have id of 0xab02.

Quoted 44 lines

A USB keyboard driver may have same problem, but I don't have one.

linux-2.5.59/drivers/input/keyboard/atkbd.c renamed from drivers/input/keyboard/atkbd.c +13 -1
@@ -309,6 +309,12 @@
 	if (atkbd_command(atkbd, &atkbd->oldset, ATKBD_CMD_GSCANSET))
 		atkbd->oldset = 2;
 
+	if (atkbd->id == 0xab02) {
+		printk("atkbd: jp109(106) keyboard found\n");
+		param[0] = atkbd_set;
+		atkbd_command(atkbd, param, ATKBD_CMD_SSCANSET);
+		return 5;
+	}
 /*
  * For known special keyboards we can go ahead and set the correct set.
  * We check for NCD PS/2 Sun, NorthGate OmniKey 101 and
@@ -531,6 +537,12 @@
 	else
 		memcpy(atkbd->keycode, atkbd_set2_keycode, sizeof(atkbd->keycode));
 
+	if (atkbd->set == 5) {
+		atkbd->keycode[0x13] = 0x70;  /* Hiragana/Katakana */
+		atkbd->keycode[0x6a] = 0x7c;  /* Yen, pipe 124*/
+		atkbd->set = 2;
+	}
+
 	atkbd->dev.name = atkbd->name;
 	atkbd->dev.phys = atkbd->phys;
 	atkbd->dev.id.bustype = BUS_I8042;
@@ -544,7 +556,7 @@
 
 	input_register_device(&atkbd->dev);
 
-	printk(KERN_INFO "input: %s on %s\n", atkbd->name, serio->phys);
+	printk(KERN_INFO "input: %s (0x%x) on %s\n", atkbd->name, atkbd->id, serio->phys);
 }
 
 
Signature · 4 lines
Hiroshi Miura  --- http://www.da-cha.org/ 
NTTDATA Corp. Marketing & Business Strategy Planning Dept. --- [email protected] 
Key fingerprint = 9117 9407 5684 FBF1 4063  15B4 401D D077 04AB 8617
-- My hacking life is happy as the day is long
Signature · 2 lines
Vojtech Pavlik
SuSE Labs
#2 lore raw

Please fix atkbd_set2_keycode table in atkbd.c for jp106 keyboard.

Quoted 7 lines · Vojtech Pavlik
Quoted 2 lines · Hiroshi Miura

After re-writting a console layer, a japanese keyboard is not supported (or degraded). This patch fixs it.

This patch doesn't work, all normal keyboards - not just japanese ones have id of 0xab02.

I agree this point. It's difficult to detect jp106 keyboard automatically. Many venders use common internal circuits with us keyborad.

Quoted 21 lines
Quoted 21 lines

A USB keyboard driver may have same problem, but I don't have one.

linux-2.5.59/drivers/input/keyboard/atkbd.c 2003-01-24 09:13:11.000000000 +0900 renamed from drivers/input/keyboard/atkbd.c 2002-12-03 07:59:41.000000000 +0900 +6 -0
@@ -309,6 +309,12 @@
      if (atkbd_command(atkbd, &atkbd->oldset, ATKBD_CMD_GSCANSET))
              atkbd->oldset = 2;
 
+     if (atkbd->id == 0xab02) {
+             printk("atkbd: jp109(106) keyboard found\n");
+             param[0] = atkbd_set;
+             atkbd_command(atkbd, param, ATKBD_CMD_SSCANSET);
+             return 5;
+     }
 /*
  * For known special keyboards we can go ahead and set the correct set.
  * We check for NCD PS/2 Sun, NorthGate OmniKey 101 and
@@ -531,6 +537,12 @@
      else
              memcpy(atkbd->keycode, atkbd_set2_keycode, sizeof(atkbd->keycode));
 
Quoted 3 lines

linux-2.5.59/drivers/input/keyboard/atkbd.c 2003-01-24 09:13:11.000000000 +0900 line 542

Quoted 3 lines
linux-2.5.59/drivers/input/keyboard/atkbd.c 2003-01-24 09:13:11.000000000 +0900 renamed from drivers/input/keyboard/atkbd.c 2002-12-03 07:59:41.000000000 +0900 +3 -0
+     if (atkbd->set == 5) {
+             atkbd->keycode[0x13] = 0x70;  /* Hiragana/Katakana */
+             atkbd->keycode[0x6a] = 0x7c;  /* Yen, pipe 124*/

I think he catches good point. Kernel 2.0-2.4 use keycode 124 (0x7c) for scancode 0x6a. 2.5 uses keycode 183. This breaks jp106 keymaps. We cannot type '\' and '|' from jp106 keyboard on 2.5 kernel.
I believe there is no impact by changing keycode 183 to 124.

Quoted 8 lines
Quoted 4 lines
linux-2.5.59/drivers/input/keyboard/atkbd.c 2003-01-24 09:13:11.000000000 +0900 renamed from drivers/input/keyboard/atkbd.c 2002-12-03 07:59:41.000000000 +0900 +0 -1
--

Hiroshi Miura --- http://www.da-cha.org/
NTTDATA Corp. Marketing & Business Strategy Planning Dept. --- [email protected]

--
Vojtech Pavlik
SuSE Labs

Regard,
Osamu Tomita

#3 lore raw
Quoted 34 lines · Osamu Tomita

Please fix atkbd_set2_keycode table in atkbd.c for jp106 keyboard.

Quoted 7 lines · Vojtech Pavlik
Quoted 2 lines · Hiroshi Miura

After re-writting a console layer, a japanese keyboard is not supported (or degraded). This patch fixs it.

This patch doesn't work, all normal keyboards - not just japanese ones have id of 0xab02.

I agree this point. It's difficult to detect jp106 keyboard automatically. Many venders use common internal circuits with us keyborad.

Quoted 21 lines
Quoted 21 lines

A USB keyboard driver may have same problem, but I don't have one.

linux-2.5.59/drivers/input/keyboard/atkbd.c 2003-01-24 09:13:11.000000000 +0900 renamed from drivers/input/keyboard/atkbd.c 2002-12-03 07:59:41.000000000 +0900 +6 -0
@@ -309,6 +309,12 @@
      if (atkbd_command(atkbd, &atkbd->oldset, ATKBD_CMD_GSCANSET))
              atkbd->oldset = 2;
 
+     if (atkbd->id == 0xab02) {
+             printk("atkbd: jp109(106) keyboard found\n");
+             param[0] = atkbd_set;
+             atkbd_command(atkbd, param, ATKBD_CMD_SSCANSET);
+             return 5;
+     }
 /*
  * For known special keyboards we can go ahead and set the correct set.
  * We check for NCD PS/2 Sun, NorthGate OmniKey 101 and
@@ -531,6 +537,12 @@
      else
              memcpy(atkbd->keycode, atkbd_set2_keycode, sizeof(atkbd->keycode));
 
Quoted 7 lines
Quoted 3 lines

linux-2.5.59/drivers/input/keyboard/atkbd.c 2003-01-24 09:13:11.000000000 +0900 line 542

Quoted 3 lines
linux-2.5.59/drivers/input/keyboard/atkbd.c 2003-01-24 09:13:11.000000000 +0900 renamed from drivers/input/keyboard/atkbd.c 2002-12-03 07:59:41.000000000 +0900 +3 -0
+     if (atkbd->set == 5) {
+             atkbd->keycode[0x13] = 0x70;  /* Hiragana/Katakana */
+             atkbd->keycode[0x6a] = 0x7c;  /* Yen, pipe 124*/

I think he catches good point. Kernel 2.0-2.4 use keycode 124 (0x7c) for scancode 0x6a. 2.5 uses keycode 183. This breaks jp106 keymaps. We cannot type '\' and '|' from jp106 keyboard on 2.5 kernel.
I believe there is no impact by changing keycode 183 to 124.

Well, it's not so easy. Fortunately KEY_KPCOMMA can be relatively easily moved elsewhere, however keys 181 to 198 are 'international and language keys', defined the same way as USB/HID spec (please take a look at it). Having a single one of them remapped elsewhere doesn't look so nice.

Signature · 2 lines
Vojtech Pavlik
SuSE Labs
#4 lore raw
Quoted 27 lines
Quoted 19 lines
Quoted 19 lines
Quoted 19 lines
linux-2.5.59/drivers/input/keyboard/atkbd.c 2003-01-24 09:13:11.000000000 +0900 renamed from drivers/input/keyboard/atkbd.c 2002-12-03 07:59:41.000000000 +0900 +6 -0
@@ -309,6 +309,12 @@
      if (atkbd_command(atkbd, &atkbd->oldset, ATKBD_CMD_GSCANSET))
              atkbd->oldset = 2;
 
+     if (atkbd->id == 0xab02) {
+             printk("atkbd: jp109(106) keyboard found\n");
+             param[0] = atkbd_set;
+             atkbd_command(atkbd, param, ATKBD_CMD_SSCANSET);
+             return 5;
+     }
 /*
  * For known special keyboards we can go ahead and set the correct set.
  * We check for NCD PS/2 Sun, NorthGate OmniKey 101 and
@@ -531,6 +537,12 @@
      else
              memcpy(atkbd->keycode, atkbd_set2_keycode, sizeof(atkbd->keycode));
 
Quoted 7 lines
Quoted 3 lines

linux-2.5.59/drivers/input/keyboard/atkbd.c 2003-01-24 09:13:11.000000000 +0900 line 542

Quoted 3 lines
linux-2.5.59/drivers/input/keyboard/atkbd.c 2003-01-24 09:13:11.000000000 +0900 renamed from drivers/input/keyboard/atkbd.c 2002-12-03 07:59:41.000000000 +0900 +3 -0
+     if (atkbd->set == 5) {
+             atkbd->keycode[0x13] = 0x70;  /* Hiragana/Katakana */
+             atkbd->keycode[0x6a] = 0x7c;  /* Yen, pipe 124*/

I think he catches good point. Kernel 2.0-2.4 use keycode 124 (0x7c) for scancode 0x6a. 2.5 uses keycode 183. This breaks jp106 keymaps. We cannot type '\' and '|' from jp106 keyboard on 2.5 kernel.
I believe there is no impact by changing keycode 183 to 124.

Quoted 4 lines

Well, it's not so easy. Fortunately KEY_KPCOMMA can be relatively easily moved elsewhere, however keys 181 to 198 are 'international and language keys', defined the same way as USB/HID spec (please take a look at it). Having a single one of them remapped elsewhere doesn't look so nice.

The keymapping on my Japanese keyboard changes quite a bit when it's set up to use set 3, instead of the default set 2. I've temporarily switched back to using set 2, until I've got time to set it up properly, but the language keys don't function in set 2, (they generate the same scancode as the space bar).

Let me know if I can provide any extra info from this keyboard - it's an IBM 5576 Keyboard-2, part number 94X1110.

John.

#5 lore raw
Quoted 32 lines · John Bradford
Quoted 27 lines
Quoted 19 lines
Quoted 19 lines
Quoted 19 lines
linux-2.5.59/drivers/input/keyboard/atkbd.c 2003-01-24 09:13:11.000000000 +0900 renamed from drivers/input/keyboard/atkbd.c 2002-12-03 07:59:41.000000000 +0900 +6 -0
@@ -309,6 +309,12 @@
      if (atkbd_command(atkbd, &atkbd->oldset, ATKBD_CMD_GSCANSET))
              atkbd->oldset = 2;
 
+     if (atkbd->id == 0xab02) {
+             printk("atkbd: jp109(106) keyboard found\n");
+             param[0] = atkbd_set;
+             atkbd_command(atkbd, param, ATKBD_CMD_SSCANSET);
+             return 5;
+     }
 /*
  * For known special keyboards we can go ahead and set the correct set.
  * We check for NCD PS/2 Sun, NorthGate OmniKey 101 and
@@ -531,6 +537,12 @@
      else
              memcpy(atkbd->keycode, atkbd_set2_keycode, sizeof(atkbd->keycode));
 
Quoted 7 lines
Quoted 3 lines

linux-2.5.59/drivers/input/keyboard/atkbd.c 2003-01-24 09:13:11.000000000 +0900 line 542

Quoted 3 lines
linux-2.5.59/drivers/input/keyboard/atkbd.c 2003-01-24 09:13:11.000000000 +0900 renamed from drivers/input/keyboard/atkbd.c 2002-12-03 07:59:41.000000000 +0900 +3 -0
+     if (atkbd->set == 5) {
+             atkbd->keycode[0x13] = 0x70;  /* Hiragana/Katakana */
+             atkbd->keycode[0x6a] = 0x7c;  /* Yen, pipe 124*/

I think he catches good point. Kernel 2.0-2.4 use keycode 124 (0x7c) for scancode 0x6a. 2.5 uses keycode 183. This breaks jp106 keymaps. We cannot type '\' and '|' from jp106 keyboard on 2.5 kernel.
I believe there is no impact by changing keycode 183 to 124.

Quoted 4 lines

Well, it's not so easy. Fortunately KEY_KPCOMMA can be relatively easily moved elsewhere, however keys 181 to 198 are 'international and language keys', defined the same way as USB/HID spec (please take a look at it). Having a single one of them remapped elsewhere doesn't look so nice.

Quoted 8 lines

The keymapping on my Japanese keyboard changes quite a bit when it's set up to use set 3, instead of the default set 2. I've temporarily switched back to using set 2, until I've got time to set it up properly, but the language keys don't function in set 2, (they generate the same scancode as the space bar).

Let me know if I can provide any extra info from this keyboard - it's an IBM 5576 Keyboard-2, part number 94X1110.

If you can provide the complete key -> scancode table for that keyboard, or at least differences against standard US keyboard for both set 2 and set 3, that'd be great.

Signature · 2 lines
Vojtech Pavlik
SuSE Labs
#6 lore raw
Quoted 6 lines
Quoted 6 lines

The keymapping on my Japanese keyboard changes quite a bit when it's set up to use set 3, instead of the default set 2. I've temporarily switched back to using set 2, until I've got time to set it up properly, but the language keys don't function in set 2, (they generate the same scancode as the space bar).

Quoted 6 lines
Quoted 2 lines

Let me know if I can provide any extra info from this keyboard - it's an IBM 5576 Keyboard-2, part number 94X1110.

If you can provide the complete key -> scancode table for that keyboard, or at least differences against standard US keyboard for both set 2 and set 3, that'd be great.

OK, half way through writing it down, I've realised what is happening
- in set 2, the keyboard seems to be emulating another keyboard layout, (but apparently not a US or UK one).

For example, ; and : are on different keys. Shift ; gives + and shift : gives *

The keycodes generated, though, show up as follows:

; - 39
shift ; (+) - 13
: - 42, 39, release 39, release 42
shift : (*) - 9

So, if I press : on it's own, the keyboard actually simulates pressing shift and ;

Likewise, if I press right-shift and @ to get ` the keyboard actually sends a release 54 as soon as I press @, even though I'm still holding down right-shift, then sends 41, release 41, then 54

I suspect set 3 will producde a more direct mapping of a single keycode for each key, but we shall see...

I'll send along the complete list when I've finished it, (it's taking ages :-) ).

John.

#7 lore raw
Quoted 27 lines · John Bradford
Quoted 6 lines
Quoted 6 lines

The keymapping on my Japanese keyboard changes quite a bit when it's set up to use set 3, instead of the default set 2. I've temporarily switched back to using set 2, until I've got time to set it up properly, but the language keys don't function in set 2, (they generate the same scancode as the space bar).

Quoted 6 lines
Quoted 2 lines

Let me know if I can provide any extra info from this keyboard - it's an IBM 5576 Keyboard-2, part number 94X1110.

If you can provide the complete key -> scancode table for that keyboard, or at least differences against standard US keyboard for both set 2 and set 3, that'd be great.

OK, half way through writing it down, I've realised what is happening
- in set 2, the keyboard seems to be emulating another keyboard layout, (but apparently not a US or UK one).

For example, ; and : are on different keys. Shift ; gives + and shift : gives *

The keycodes generated, though, show up as follows:

; - 39
shift ; (+) - 13
: - 42, 39, release 39, release 42
shift : (*) - 9

Quoted 6 lines

So, if I press : on it's own, the keyboard actually simulates pressing shift and ;

Likewise, if I press right-shift and @ to get ` the keyboard actually sends a release 54 as soon as I press @, even though I'm still holding down right-shift, then sends 41, release 41, then 54

Yeah, I've seen this before on some weird keyboards. We need to make sure we understand properly decode what the keyboard is pretending to be
- no need to really detect what keys were pressed in reality and what the keyboard just wants us to think are pressed.

Quoted 2 lines

I suspect set 3 will producde a more direct mapping of a single keycode for each key, but we shall see...

I hope so.

Quoted 2 lines

I'll send along the complete list when I've finished it, (it's taking ages :-) ).

Signature · 2 lines
Vojtech Pavlik
SuSE Labs
#8 lore raw

Set2 scancodes for Japanese keyboard

Quoted 3 lines
Quoted 3 lines
Quoted 3 lines

If you can provide the complete key -> scancode table for that keyboard, or at least differences against standard US keyboard for both set 2 and set 3, that'd be great.

Here are the set 2 ones, I've missed out most of the letter and number keys, because they all followed the obvious numerical sequence. I'll send the set 3 ones separately, (when I've typed them in :-) ).

Key
Keycode
Make scancode
Break scancode

Escape
1
0x01
0x81

F1
59
0x3b
0xbb

F10
68
0x44
0xc4

F11
87
0x57
0xd7

F12
88
0x58
0xd8

Printscreen?, (marked 'page', and something I can't read). 99
0xe0 0x2a 0xe0 0x37
0xe0 0xb7 0xe0 0xaa

Sys Rq, (alt and key above)
84
0x54
0xd4

Scroll lock
70
0x46
0xc6

Numlock, (shift scroll lock)
69
0x45
0xc5

Pause
119
0xe1 0x1d 0x45
0xe1 0x9d 0xc5

Break
101
0xe0 0x46
0xe0 0xc6

Hankaku/Zenkaku
None
0xff
none

1
2
0x02
0x83

0
11
0x0b
0x8b

-
12
0x0c
0x8c

shift -
13
0xb6 0x0d
0x8d 0x36

^
42,7
0x2a 0x07
0x8 0xaa

shift ^
none
0xff
none

yen
none
0xff
none

shift yen ( | )
43
0x2b
0xab

backspace
14
0x0e
0x8e

tab
15
0x0f
0x8f

q
16
0x10
0x90

p
25
0x19
0x99

@
42,3
0x2a 0x03
0x83 0xaa

shift @
release 54, 41, release 41, 54
0xb6 0x29
0xa9 0x36

[
26
0x1a
0x9a

enter
28
0x1c
0x9c

caps lock
58
0x3a
0xba

a
30
0x1e
0xae

l
38
0x26
0xa6

;
39
0x27
0xa7

shift ; ( + )
13
0x0d
0x8d

:
42 39
0x2a 0x27
0xa7 0xaa

shift : ( * )
9
0x09
0x89

]
27
0x1b
0x9b

left shift 42
0x2a
0xaa

z
44
0x2c
0xad

m
50
0x32
0xb2

,
51
0x33
0xb3

#9 lore raw

[Corrected] Set2 scancodes for Japanese keyboard

In reply to Vojtech Pavlik

Some got missed off the first time:

Key
Keycode
Make scancode
Break scancode

Escape
1
0x01
0x81

F1
59
0x3b
0xbb

F10
68
0x44
0xc4

F11
87
0x57
0xd7

F12
88
0x58
0xd8

Printscreen?, (marked 'page', and something I can't read). 99
0xe0 0x2a 0xe0 0x37
0xe0 0xb7 0xe0 0xaa

Sys Rq, (alt and key above)
84
0x54
0xd4

Scroll lock
70
0x46
0xc6

Numlock, (shift scroll lock)
69
0x45
0xc5

Pause
119
0xe1 0x1d 0x45
0xe1 0x9d 0xc5

Break
101
0xe0 0x46
0xe0 0xc6

Hankaku/Zenkaku
None
0xff
none

1
2
0x02
0x83

0
11
0x0b
0x8b

-
12
0x0c
0x8c

shift -
13
0xb6 0x0d
0x8d 0x36

^
42,7
0x2a 0x07
0x8 0xaa

shift ^
none
0xff
none

yen
none
0xff
none

shift yen ( | )
43
0x2b
0xab

backspace
14
0x0e
0x8e

tab
15
0x0f
0x8f

q
16
0x10
0x90

p
25
0x19
0x99

@
42,3
0x2a 0x03
0x83 0xaa

shift @
release 54, 41, release 41, 54
0xb6 0x29
0xa9 0x36

[
26
0x1a
0x9a

enter
28
0x1c
0x9c

caps lock
58
0x3a
0xba

a
30
0x1e
0xae

l
38
0x26
0xa6

;
39
0x27
0xa7

shift ; ( + )
13
0x0d
0x8d

:
42 39
0x2a 0x27
0xa7 0xaa

shift : ( * )
9
0x09
0x89

]
27
0x1b
0x9b

left shift 42
0x2a
0xaa

z
44
0x2c
0xad

m
50
0x32
0xb2

,
51
0x33
0xb3

#10 lore raw

Re: [Corrected] Set2 scancodes for Japanese keyboard

Quoted 1 line · John Bradford

Some got missed off the first time:

What kernel this is tested with? What method used? These don't look like Set2 codes AT ALL.

Quoted 204 lines

Key
Keycode
Make scancode
Break scancode

Escape
1
0x01
0x81

F1
59
0x3b
0xbb

F10
68
0x44
0xc4

F11
87
0x57
0xd7

F12
88
0x58
0xd8

Printscreen?, (marked 'page', and something I can't read). 99
0xe0 0x2a 0xe0 0x37
0xe0 0xb7 0xe0 0xaa

Sys Rq, (alt and key above)
84
0x54
0xd4

Scroll lock
70
0x46
0xc6

Numlock, (shift scroll lock)
69
0x45
0xc5

Pause
119
0xe1 0x1d 0x45
0xe1 0x9d 0xc5

Break
101
0xe0 0x46
0xe0 0xc6

Hankaku/Zenkaku
None
0xff
none

1
2
0x02
0x83

0
11
0x0b
0x8b

-
12
0x0c
0x8c

shift -
13
0xb6 0x0d
0x8d 0x36

^
42,7
0x2a 0x07
0x8 0xaa

shift ^
none
0xff
none

yen
none
0xff
none

shift yen ( | )
43
0x2b
0xab

backspace
14
0x0e
0x8e

tab
15
0x0f
0x8f

q
16
0x10
0x90

p
25
0x19
0x99

@
42,3
0x2a 0x03
0x83 0xaa

shift @
release 54, 41, release 41, 54
0xb6 0x29
0xa9 0x36

[
26
0x1a
0x9a

enter
28
0x1c
0x9c

caps lock
58
0x3a
0xba

a
30
0x1e
0xae

l
38
0x26
0xa6

;
39
0x27
0xa7

shift ; ( + )
13
0x0d
0x8d

:
42 39
0x2a 0x27
0xa7 0xaa

shift : ( * )
9
0x09
0x89

]
27
0x1b
0x9b

left shift 42
0x2a
0xaa

z
44
0x2c
0xad

m
50
0x32
0xb2

,
51
0x33
0xb3

Signature · 2 lines
Vojtech Pavlik
SuSE Labs
#11 lore raw

Re: [Corrected] Set2 scancodes for Japanese keyboard

Quoted 7 lines · John Bradford
Quoted 2 lines

Some got missed off the first time:

What kernel this is tested with? What method used? These don't look like Set2 codes AT ALL.

The kernel is 2.4.20. The keycode is the output from showkey, and the make and break codes are the output from showkey -s.

Should I have used I8042_DEBUG_IO instead? :-/

John.

#12 lore raw

Re: [Corrected] Set2 scancodes for Japanese keyboard

Quoted 5 lines · John Bradford
Quoted 5 lines · John Bradford
Quoted 2 lines

Some got missed off the first time:

Quoted 7 lines
Quoted 2 lines

What kernel this is tested with? What method used? These don't look like Set2 codes AT ALL.

The kernel is 2.4.20. The keycode is the output from showkey, and the make and break codes are the output from showkey -s.

Should I have used I8042_DEBUG_IO instead? :-/

With 2.5, yes, that'd be much better. And yet better it'd be if you'd have used the "i8042_direct=1" kernel option, and for set3 the "atkbd_set=3" option.

I'm sorry to tell you after you wrote it all down, but these are set1 scancodes you see.

Signature · 2 lines
Vojtech Pavlik
SuSE Labs
#13 lore raw

Re: [Corrected] Set2 scancodes for Japanese keyboard

Quoted 6 lines
Quoted 6 lines
Quoted 2 lines

What kernel this is tested with? What method used? These don't look like Set2 codes AT ALL.

The kernel is 2.4.20. The keycode is the output from showkey, and the make and break codes are the output from showkey -s.

Quoted 8 lines
Quoted 1 line

Should I have used I8042_DEBUG_IO instead? :-/

With 2.5, yes, that'd be much better. And yet better it'd be if you'd have used the "i8042_direct=1" kernel option, and for set3 the "atkbd_set=3" option.

I'm sorry to tell you after you wrote it all down, but these are set1 scancodes you see.

I should have realised anyway - I remember now it translates everything to set 1.

Oh dear, well, it might take a while - I packed my serial terminal away yesterday, because a wire needs to be soldered on it :-), (which explains why I was writing it all down :-) )

John.

#14 lore raw

Re: [Corrected] Set2 scancodes for Japanese keyboard

In reply to Vojtech Pavlik
Quoted 2 lines · Vojtech Pavlik

I'm sorry to tell you after you wrote it all down, but these are set1 scancodes you see.

What about "translated set2" ?

So far I have not yet seen cases where the translation was nonstandard. That is, for all keyboards I have looked at or received reports on the table given in
http://www.win.tue.nl/~aeb/linux/kbd/scancodes-4.html#kscancodes
described the conversion from set2 to set1.

So, given the translated set2 codes, also the original set2 codes are known with high probability. (The translation is almost 1-1.)

Andries

#15 lore raw

Re: [Corrected] Set2 scancodes for Japanese keyboard

Quoted 4 lines · Andries Brouwer
Quoted 2 lines

I'm sorry to tell you after you wrote it all down, but these are set1 scancodes you see.

What about "translated set2" ?

Correct.

Quoted 8 lines

So far I have not yet seen cases where the translation was nonstandard. That is, for all keyboards I have looked at or received reports on the table given in
http://www.win.tue.nl/~aeb/linux/kbd/scancodes-4.html#kscancodes
described the conversion from set2 to set1.

So, given the translated set2 codes, also the original set2 codes are known with high probability. (The translation is almost 1-1.)

Correct. Given set of not translated set2 codes, the translated set2 codes are created much easier, hence the not translated set2 codes offer better value.

And, translated set3 just plain doesn't make sense.

Signature · 2 lines
Vojtech Pavlik
SuSE Labs
#16 lore raw

Set3 scancodes for Japanese keyboard

In reply to Vojtech Pavlik
Quoted 6 lines
Quoted 6 lines
Quoted 2 lines

What kernel this is tested with? What method used? These don't look like Set2 codes AT ALL.

The kernel is 2.4.20. The keycode is the output from showkey, and the make and break codes are the output from showkey -s.

Quoted 8 lines
Quoted 1 line

Should I have used I8042_DEBUG_IO instead? :-/

With 2.5, yes, that'd be much better. And yet better it'd be if you'd have used the "i8042_direct=1" kernel option, and for set3 the "atkbd_set=3" option.

I'm sorry to tell you after you wrote it all down, but these are set1 scancodes you see.

Here are the set 3 scancodes for my Japanese keyboard.

If the set 2 codes are still of interest, I can do those as well, but set 3 seems to be a much more useful mode, because there is a direct mapping of keys to scancodes, instead of the more complicated similated shift key codes that set 2 produces.

ESC                             08 F0 08
F1                              07 F0 07
F2                              0F F0 0F
F3                              17 F0 17
F4                              1F F0 1F
F5                              27 F0 27
F6                              2F F0 2F
F7                              37 F0 37
F8                              3F F0 3F
F9                              47 F0 47
F10                             4F F0 4F
F11                             56 F0 56
F12                             5E F0 5E
HANKAKU/ZENKAKU                 0E F0 0E
1                               16 F0 16
2                               1E F0 1E
3                               26 F0 26
4                               25 F0 25
5                               2E F0 2E
6                               36 F0 36
7                               3D F0 3D
8                               3E F0 3E
9                               46 F0 46
0                               45 F0 45
-                               4E F0 4E
^                               55 F0 55
YEN                             13 F0 13
BACKSPACE                       66 F0 66
TAB                             0D F0 0D
Q                               15 F0 15
W                               1D F0 1D
E                               24 F0 24
R                               2D F0 2D
T                               2C F0 2C
Y                               35 F0 35
U                               3C F0 3C
I                               43 F0 43
O                               44 F0 44
P                               4D F0 4D
@                               54 F0 54
[                               5B F0 5B
A                               1C F0 1C
S                               1B F0 1B
D                               23 F0 23
F                               2B F0 2B
G                               34 F0 34
H                               33 F0 33
J                               3B F0 3B
K                               42 F0 42
L                               4B F0 4B
;                               4C F0 4C
:                               52 F0 52
]                               53 F0 53
ENTER                           5A F0 5A
LEFT SHIFT                      12 F0 12
Z                               1A F0 1A
X                               22 F0 22
C                               21 F0 21
V                               2A F0 2A
B                               32 F0 32
N                               31 F0 31
M                               3A F0 3A
,                               41 F0 41
.                               49 F0 49
/                               4A F0 4A
\                               5C F0 5C
RIGHT SHIFT                     59 F0 59
LEFT CONTROL                    11 F0 11
KANJI/KANAKANA                  19 F0 19
MUHENKAN                        85 F0 85
SPACE                           29 F0 29
ZENKOHO/HENKAN (JIKOHO)/ZENKOHO 86 F0 86
HIRAGANA                        87 F0 87
ZENMEN KI                       39 F0 39
RIGHT CONTROL                   58 F0 58

PAGE/SYSRQ 57 F0 57
SCROLL LOCK
5F <-
ED ->
FA <-
00 ->
FA <-
F0 <-
5F <-
SHIFT-SCROLL LOCK (NUMLOCK) 5F F0 5F
PAUSE/BREAK 62 F0 62

INSERT                          67 F0 67
HOME                            6E F0 6E
PAGE UP                         6F F0 6F
DELETE                          64 F0 64
END                             65 F0 65
PAGE DOWN                       6D F0 6D
CURSOR UP                       63 F0 63
CURSOR DOWN                     60 F0 60
CURSOR LEFT                     61 F0 61
CURSOR RIGHT                    69 F0 69

KEYPAD * 7E F0 7E
KEYPAD / 77 F0 77
KEYPAD ,
76 <-
ED ->
FA <-
00 ->
FA <-
F0 <-
76 <-

KEYPAD -                        84 F0 84
KEYPAD +                        7C F0 7C
KEYPAD ENTER                    79 F0 79
KEYPAD .                        71 F0 71
KEYPAD 0                        70 F0 70
KEYPAD 1                        69 F0 69
KEYPAD 2                        72 F0 72
KEYPAD 3                        7A F0 7A
KEYPAD 4                        6B F0 6B
KEYPAD 5                        73 F0 73
KEYPAD 6                        74 F0 74
KEYPAD 7                        6C F0 6C
KEYPAD 8                        75 F0 75
KEYPAD 9                        7D F0 7D

John.

#17 lore raw
In reply to Vojtech Pavlik
Quoted 44 lines · Vojtech Pavlik
Quoted 34 lines · Osamu Tomita

Please fix atkbd_set2_keycode table in atkbd.c for jp106 keyboard.

Quoted 7 lines · Vojtech Pavlik
Quoted 2 lines · Hiroshi Miura

After re-writting a console layer, a japanese keyboard is not supported (or degraded). This patch fixs it.

This patch doesn't work, all normal keyboards - not just japanese ones have id of 0xab02.

I agree this point. It's difficult to detect jp106 keyboard automatically. Many venders use common internal circuits with us keyborad.

Quoted 21 lines
Quoted 21 lines

A USB keyboard driver may have same problem, but I don't have one.

linux-2.5.59/drivers/input/keyboard/atkbd.c 2003-01-24 09:13:11.000000000 +0900 renamed from drivers/input/keyboard/atkbd.c 2002-12-03 07:59:41.000000000 +0900 +6 -0
@@ -309,6 +309,12 @@
      if (atkbd_command(atkbd, &atkbd->oldset, ATKBD_CMD_GSCANSET))
              atkbd->oldset = 2;
 
+     if (atkbd->id == 0xab02) {
+             printk("atkbd: jp109(106) keyboard found\n");
+             param[0] = atkbd_set;
+             atkbd_command(atkbd, param, ATKBD_CMD_SSCANSET);
+             return 5;
+     }
 /*
  * For known special keyboards we can go ahead and set the correct set.
  * We check for NCD PS/2 Sun, NorthGate OmniKey 101 and
@@ -531,6 +537,12 @@
      else
              memcpy(atkbd->keycode, atkbd_set2_keycode, sizeof(atkbd->keycode));
 
Quoted 7 lines
Quoted 3 lines

linux-2.5.59/drivers/input/keyboard/atkbd.c 2003-01-24 09:13:11.000000000 +0900 line 542

Quoted 3 lines
linux-2.5.59/drivers/input/keyboard/atkbd.c 2003-01-24 09:13:11.000000000 +0900 renamed from drivers/input/keyboard/atkbd.c 2002-12-03 07:59:41.000000000 +0900 +3 -0
+     if (atkbd->set == 5) {
+             atkbd->keycode[0x13] = 0x70;  /* Hiragana/Katakana */
+             atkbd->keycode[0x6a] = 0x7c;  /* Yen, pipe 124*/

I think he catches good point. Kernel 2.0-2.4 use keycode 124 (0x7c) for scancode 0x6a. 2.5 uses keycode 183. This breaks jp106 keymaps. We cannot type '\' and '|' from jp106 keyboard on 2.5 kernel.
I believe there is no impact by changing keycode 183 to 124.

Quoted 8 lines

Well, it's not so easy. Fortunately KEY_KPCOMMA can be relatively easily moved elsewhere, however keys 181 to 198 are 'international and language keys', defined the same way as USB/HID spec (please take a look at it). Having a single one of them remapped elsewhere doesn't look so nice.

--
Vojtech Pavlik
SuSE Labs

Thanks. I see. Keycode in the kernel should be unified. I'll use newer kbd utility (kbd-1.08) and rewrite keymap like below.
keycode 124 = backslash bar #Compatibility for 2.4 (for a while)
keycode 183 = backslash bar

BTW. I've extracted japanese keyboard specification from OADG's documents. About OADG, please visit http://www.oadg.org/.
I attach those files. Usefull?

-Regards,
Osamu Tomita

#18 lore raw
Quoted 10 lines · Osamu Tomita
Quoted 8 lines

Well, it's not so easy. Fortunately KEY_KPCOMMA can be relatively easily moved elsewhere, however keys 181 to 198 are 'international and language keys', defined the same way as USB/HID spec (please take a look at it). Having a single one of them remapped elsewhere doesn't look so nice.

--
Vojtech Pavlik
SuSE Labs

Thanks. I see. Keycode in the kernel should be unified.

Definitely. And cast in stone and never changed again.

Quoted 3 lines

I'll use newer kbd utility (kbd-1.08) and rewrite keymap like below.
keycode 124 = backslash bar #Compatibility for 2.4 (for a while)
keycode 183 = backslash bar

Yes, that'd be great.

Quoted 3 lines

BTW. I've extracted japanese keyboard specification from OADG's documents. About OADG, please visit http://www.oadg.org/.
I attach those files. Usefull?

Yes, I'm sure they'll be.

Signature · 2 lines
Vojtech Pavlik
SuSE Labs
#19 lore raw
In reply to Hiroshi Miura

I have a question.

Quoted 24 lines · Hiroshi Miura

Hi,

After re-writting a console layer, a japanese keyboard is not supported (or degraded). This patch fixs it.

A USB keyboard driver may have same problem, but I don't have one.

linux-2.5.59/drivers/input/keyboard/atkbd.c 2003-01-24 09:13:11.000000000 +0900 renamed from drivers/input/keyboard/atkbd.c 2002-12-03 07:59:41.000000000 +0900 +6 -0
@@ -309,6 +309,12 @@
        if (atkbd_command(atkbd, &atkbd->oldset, ATKBD_CMD_GSCANSET))
                atkbd->oldset = 2;
 
+       if (atkbd->id == 0xab02) {
+               printk("atkbd: jp109(106) keyboard found\n");
+               param[0] = atkbd_set;
+               atkbd_command(atkbd, param, ATKBD_CMD_SSCANSET);
+               return 5;
+       }
 /*
  * For known special keyboards we can go ahead and set the correct set.
  * We check for NCD PS/2 Sun, NorthGate OmniKey 101 and

linux-2.5.59/drivers/input/keyboard/atkbd.c 2003-01-24 09:13:11.000000000 +0900 line 541

Quoted 5 lines
linux-2.5.59/drivers/input/keyboard/atkbd.c 2003-01-24 09:13:11.000000000 +0900 renamed from drivers/input/keyboard/atkbd.c 2002-12-03 07:59:41.000000000 +0900 +2 -0
@@ -531,6 +537,12 @@
        else
                memcpy(atkbd->keycode, atkbd_set2_keycode, sizeof(atkbd->keycode));
 
+       if (atkbd->set == 5) {
+               atkbd->keycode[0x13] = 0x70;  /* Hiragana/Katakana */

I'm interesting in the reason to use keycode 0x70 for 'Hiragana/Katakana' key. Please clarify.

Regard,
Osamu Tomita

#20 lore raw

Hi, Mr. Tomita,

In message "Re: [PATCH 2.5.59] support japanese JP106 keyboard on new console."

    on 03/01/25, Osamu Tomita <[email protected]> writes:
Quoted 6 lines

I have a question.

linux-2.5.59/drivers/input/keyboard/atkbd.c line 541

Quoted 2 lines
linux-2.5.59/drivers/input/keyboard/atkbd.c renamed from drivers/input/keyboard/atkbd.c +2 -0
+       if (atkbd->set == 5) {
+               atkbd->keycode[0x13] = 0x70;  /* Hiragana/Katakana */

I'm interesting in the reason to use keycode 0x70 for 'Hiragana/Katakana' key. Please clarify.

This is answer for your question.

I have two point of view about this.
these are my try and error process.

First, I press Katakana_Hiragana key on console with 2.4.19, it warns

keyboard: unrecognized scancode (70) - ignored

Two, in a 2.4.20's pc_keyb.c, there is a comment,
/*
* The keycodes below are randomly located in 89-95,112-118,120-127.
* They could be thrown away (and all occurrences below replaced by 0),
* but that would force many users to use the `setkeycodes' utility, where
* they needed not before. It does not matter that there are duplicates, as
* long as no duplication occurs for any single keyboard. ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
*/

Hiragana_Katakana was not defined before and I want to define a keycode point. When I saw 2.4.20 pc_keyb.c source, I found all keycode below 127 was used, then there is no room. But the comment tell me I can use 120-123, 125-127 with Japanese keyboard because these are not used on JP89/109 keyboards. (124 is, as you know, Yen key) THese are defined for a latin keyboards. So I use 120.

How do you think about it?

--
Hiroshi Miura

#21 lore raw
Quoted 31 lines · Hiroshi Miura

Hi, Mr. Tomita,

In message "Re: [PATCH 2.5.59] support japanese JP106 keyboard on new console."

    on 03/01/25, Osamu Tomita <[email protected]> writes:
Quoted 6 lines

I have a question.

linux-2.5.59/drivers/input/keyboard/atkbd.c line 541

Quoted 2 lines
linux-2.5.59/drivers/input/keyboard/atkbd.c renamed from drivers/input/keyboard/atkbd.c +2 -0
+       if (atkbd->set == 5) {
+               atkbd->keycode[0x13] = 0x70;  /* Hiragana/Katakana */

I'm interesting in the reason to use keycode 0x70 for 'Hiragana/Katakana' key. Please clarify.

This is answer for your question.

I have two point of view about this.
these are my try and error process.

First, I press Katakana_Hiragana key on console with 2.4.19, it warns

keyboard: unrecognized scancode (70) - ignored

Two, in a 2.4.20's pc_keyb.c, there is a comment,
/*
* The keycodes below are randomly located in 89-95,112-118,120-127.
* They could be thrown away (and all occurrences below replaced by 0),
* but that would force many users to use the `setkeycodes' utility, where
* they needed not before. It does not matter that there are duplicates, as
* long as no duplication occurs for any single keyboard. ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
*/

Quoted 8 lines

Hiragana_Katakana was not defined before and I want to define a keycode point. When I saw 2.4.20 pc_keyb.c source, I found all keycode below 127 was used, then there is no room. But the comment tell me I can use 120-123, 125-127 with Japanese keyboard because these are not used on JP89/109 keyboards. (124 is, as you know, Yen key) THese are defined for a latin keyboards. So I use 120.

How do you think about it?

In 2.4 you can, in 2.5 the 'as long as no duplication occurs for any single keyboard' is not valid anymore, and the keycode for hiragana/katakana is defined to be 183 I think.

Signature · 2 lines
Vojtech Pavlik
SuSE Labs
#22 lore raw
Quoted 9 lines
Quoted 9 lines

Hiragana_Katakana was not defined before and I want to define a keycode point.
When I saw 2.4.20 pc_keyb.c source, I found all keycode below 127 was used, then there is no room. But the comment tell me I can use 120-123, 125-127 with Japanese keyboard because these are not used on JP89/109 keyboards.
(124 is, as you know, Yen key) THese are defined for a latin keyboards. So I use 120.

Quoted 5 lines
Quoted 1 line

How do you think about it?

In 2.4 you can, in 2.5 the 'as long as no duplication occurs for any single keyboard' is not valid anymore, and the keycode for hiragana/katakana is defined to be 183 I think.

We assigned 182 to hiragana/katakana for set 3 in 2.5, and left 183 undefined. Should we change the 2.5 keycode to 183?

John.

#23 lore raw
Quoted 9 lines · John Bradford
Quoted 9 lines
Quoted 9 lines

Hiragana_Katakana was not defined before and I want to define a keycode point.
When I saw 2.4.20 pc_keyb.c source, I found all keycode below 127 was used, then there is no room. But the comment tell me I can use 120-123, 125-127 with Japanese keyboard because these are not used on JP89/109 keyboards.
(124 is, as you know, Yen key) THese are defined for a latin keyboards. So I use 120.

Quoted 8 lines
Quoted 5 lines
Quoted 1 line

How do you think about it?

In 2.4 you can, in 2.5 the 'as long as no duplication occurs for any single keyboard' is not valid anymore, and the keycode for hiragana/katakana is defined to be 183 I think.

We assigned 182 to hiragana/katakana for set 3 in 2.5, and left 183 undefined. Should we change the 2.5 keycode to 183?

No - just bad memory on my side, it might as well be 183. ;)

Signature · 2 lines
Vojtech Pavlik
SuSE Labs
#24 lore raw
In reply to Vojtech Pavlik

In message "Re: [PATCH 2.5.59] support japanese JP106 keyboard on new console."

    on 03/02/02, Vojtech Pavlik <[email protected]> writes:
Quoted 3 lines

In 2.4 you can, in 2.5 the 'as long as no duplication occurs for any single keyboard' is not valid anymore, and the keycode for hiragana/katakana is defined to be 183 I think.

I can understand the rule of international and language keys in 2.5. I think hiragana/katakana is defined to be '182'.
The keycode 183 is a yen/pipe key.

linux-2.5.59/drivers/input/keyboard/atkbd.c 2003-02-05 12:32:43.000000000 +0900 renamed from drivers/input/keyboard/atkbd.c 2002-12-03 07:59:41.000000000 +0900 +1 -1
@@ -39,7 +39,7 @@
 
 static unsigned char atkbd_set2_keycode[512] = {
          0, 67, 65, 63, 61, 59, 60, 88,  0, 68, 66, 64, 62, 15, 41, 85,
-         0, 56, 42,  0, 29, 16,  2, 89,  0,  0, 44, 31, 30, 17,  3, 90,
+         0, 56, 42,182, 29, 16,  2, 89,  0,  0, 44, 31, 30, 17,  3, 90,
          0, 46, 45, 32, 18,  5,  4, 91,  0, 57, 47, 33, 20, 19,  6,  0,
          0, 49, 48, 35, 34, 21,  7,  0,  0,  0, 50, 36, 22,  8,  9,  0,
          0, 51, 37, 23, 24, 11, 10,  0,  0, 52, 53, 38, 39, 25, 12,  0,

--
Hiroshi Miura