Change in ...coreboot[master]: libpayload/drivers/i8042: ignore special keys on keyboard

[Matt Delco (Code Review)]

Matt Delco has uploaded this change for review. ( https://review.coreboot.org/c/coreboot/+/33303

Change subject: libpayload/drivers/i8042: ignore special keys on keyboard
......................................................................

libpayload/drivers/i8042: ignore special keys on keyboard

Some keyboards have a fn-key (and/or fn-mode lock) that can generate
0xe0-prefixed scan codes for certain keys.  Previously the 0xe0 prefix
state wasn't tracked (it was ignored), so these scan codes could be
misinterpreted as non-0xe0 presses (e.g., 'mute' generates 0xe0 0x20
but this would get interpreted as 0x20 and be treated as a 'D' press).
Most 0xe0-prefixed codes (besides those for the 10-key) are irrelevant
for firmware so we'll ignore them.

BUG=none
BRANCH=none
TEST=local build and flash.  Verified that special keys are no longer
interpreted as other keystrokes.

Change-Id: Ie6a828b6a0a3ab47456d6db2db504aa0d7aefc28
Signed-off-by: Matt Delco 
---
M payloads/libpayload/drivers/i8042/keyboard.c
1 file changed, 23 insertions(+), 5 deletions(-)

git pull ssh://review.coreboot.org:29418/coreboot refs/changes/03/33303/1

diff --git a/payloads/libpayload/drivers/i8042/keyboard.c b/payloads/libpayload/drivers/i8042/keyboard.c
index 3e5f988..bca916f 100644
--- a/payloads/libpayload/drivers/i8042/keyboard.c
+++ b/payloads/libpayload/drivers/i8042/keyboard.c
@@ -235,12 +235,30 @@
    unsigned char ch;
    int shift;
    int ret = 0;
-
-	while (!keyboard_havechar()) ;
+	static int prior_special = 0;

ch = keyboard_get_scancode();

-	if (!(ch & 0x80) && ch < 0x59) {
+	if (prior_special) {
+		if (ch == 0x5e)
+			ret = POWER_BUTTON;
+		/*
+		 * For now 0xe0 with 0x1c, 0x35, or 0x47 - 0x53 is the same as
+		 * non-0xe0. For 10-key scan codes 0x47-0x53 it's probably more
+		 * accurate for the non-0xe0 presses to use the numeric
+		 * interpretation of the key (i.e., when num lock is on) and
+		 * the 0xe0 prefixed versions should use the non-numeric
+		 * interpretation (i.e., num lock off). However, the existing
+		 * tables are using the non-numeric interpretation so for now
+		 * the 0xe0 presses can use what's in the table for non-0xe0
+		 * presses.  The main exception is that shift+'/' on the keypad
+		 * (0xe0 0x35) should still produce a '/' (vs. a '?' like on
+		 * the key that's to the left of the right-shift on English
+		 * keyboards) so we'll ignore the shift status.
+		 */
+		else if (ch == 0x1c || ch == 0x35 || (ch >= 0x47 && ch <= 0x53))
+			ret = map->map[0][ch];
+	} else if (ch < 0x59) { /* implies !(ch & 0x80), i.e., key press */
    	shift =
    	    (modifier & KB_MOD_SHIFT) ^ (modifier & KB_MOD_CAPSLOCK) ? 1 : 0;

@@ -265,8 +283,8 @@
    	}
    }

-	if (ch == 0x5e)
-		ret = POWER_BUTTON;
+	/* sequence 0xe0 0xe0 should not leave prior_special set. */
+	prior_special = !prior_special && ch == 0xe0;

return ret;
 }

[S] Change in coreboot[master]: libpayload/drivers/i8042: ignore special keys on keyboard

[Stefan Reinauer (Code Review)]

Stefan Reinauer has abandoned this change. ( https://review.coreboot.org/c/coreboot/+/33303?usp=email )

Change subject: libpayload/drivers/i8042: ignore special keys on keyboard
......................................................................

Abandoned