Make Dell keyboard repeat quirk apply to a wider range of hardware

Dell laptops fail to send key up events for several of their special keys. There's an existing quirk in the kernel to handle this, but it's limited to the Latitude range. This patch extends it to cover all portable Dells.

Signed-off-by Matthew Garrett <[email protected]>

---

drivers/input/keyboard/atkbd.c +3 -3

index b1ce10f..e32c24d 100644
@@ -834,10 +834,10 @@ static void atkbd_disconnect(struct serio *serio)
 }
 
 /*
- * Most special keys (Fn+F?) on Dell Latitudes do not generate release
+ * Most special keys (Fn+F?) on Dell laptops do not generate release
  * events so we have to do it ourselves.
  */
-static void atkbd_latitude_keymap_fixup(struct atkbd *atkbd)
+static void atkbd_dell_laptop_keymap_fixup(struct atkbd *atkbd)
 {
 	const unsigned int forced_release_keys[] = {
 		0x85, 0x86, 0x87, 0x88, 0x89, 0x8a, 0x8b, 0x8f, 0x93,

@@ -1461,13 +1461,13 @@ static int __init atkbd_setup_fixup(const struct dmi_system_id *id)
 
 static struct dmi_system_id atkbd_dmi_quirk_table[] __initdata = {
 	{
-		.ident = "Dell Latitude series",
+		.ident = "Dell Laptop",
 		.matches = {

Quoted 39 lines

Dell laptops fail to send key up events for several of their special keys. There's an existing quirk in the kernel to handle this, but it's limited to the Latitude range. This patch extends it to cover all portable Dells.

Signed-off-by Matthew Garrett <[email protected]>

---

(quoted diff) +0 -0

b/drivers/input/keyboard/atkbd.c index b1ce10f..e32c24d 100644

drivers/input/keyboard/atkbd.c +0 -1

@@ -834,10 +834,10 @@ static void atkbd_disconnect(struct serio
 *serio) }
 
 /*
- * Most special keys (Fn+F?) on Dell Latitudes do not generate

release
+ * Most special keys (Fn+F?) on Dell laptops do not generate release
* events so we have to do it ourselves.
*/

drivers/input/keyboard/atkbd.c +1 -1

-static void atkbd_latitude_keymap_fixup(struct atkbd *atkbd)
+static void atkbd_dell_laptop_keymap_fixup(struct atkbd *atkbd)
 {
 	const unsigned int forced_release_keys[] = {
 		0x85, 0x86, 0x87, 0x88, 0x89, 0x8a, 0x8b, 0x8f,

0x93, @@ -1461,13 +1461,13 @@ static int __init atkbd_setup_fixup(const struct dmi_system_id *id)
static struct dmi_system_id atkbd_dmi_quirk_table[] __initdata = {

drivers/input/keyboard/atkbd.c +2 -2

 	{
-		.ident = "Dell Latitude series",
+		.ident = "Dell Laptop",
 		.matches = {
 			DMI_MATCH(DMI_SYS_VENDOR, "Dell Inc."),
-			DMI_MATCH(DMI_PRODUCT_NAME, "Latitude"),
+			DMI_MATCH(DMI_CHASSIS_TYPE, "8"), /*

Portable */ },

drivers/input/keyboard/atkbd.c

Quoted 6 lines

drivers/input/keyboard/atkbd.c +1 -1

 		.callback = atkbd_setup_fixup,
-		.driver_data = atkbd_latitude_keymap_fixup,
+		.driver_data = atkbd_dell_laptop_keymap_fixup,
 	},
 	{
 		.ident = "HP 2133",

This issue has hit a lot of Ubuntu Dell uses (like myself), and the patch indeed fixes it up - but it appears not all Dell machines are alike. One user has reported it doesn't work, and his investigation reveals dmicode output shows:

Vendor: Dell Computer Corporation

So I guess this isn't totally fixed yet. As to what the vendor output is al all these machines, I don't know, but I guess now it could be any permatation of Dell Inc., Dell Computer Corp. etc.

Ref. Ubuntu bug report thread:

https://bugs.launchpad.net/ubuntu/+source/acpid/+bug/285323/comments/88

Nick

Does this work?

(quoted diff) +0 -0

b/drivers/input/keyboard/atkbd.c index 22016ca..15479ef 100644

drivers/input/keyboard/atkbd.c +0 -0

@@ -1460,6 +1460,15 @@ static struct dmi_system_id

atkbd_dmi_quirk_table[] __initdata = { .driver_data = atkbd_dell_laptop_keymap_fixup, },

drivers/input/keyboard/atkbd.c +3 -0

 	{
+		.ident = "Dell Laptop",
+		.matches = {
+			DMI_MATCH(DMI_SYS_VENDOR, "Dell Computer

Corporation"),

drivers/input/keyboard/atkbd.c +1 -0

+			DMI_MATCH(DMI_CHASSIS_TYPE, "8"), /*

Portable */

But how you determine if these certain laptops produce a keypress release I don't know with having all of them at hand...

So it needs BOTH 'Dell Inc.' and 'Dell Computer Corporation' to cover these laptops (I guess), as per the info I got from people running dmidecode:

Quoted 2 lines · Nick Warne

But how you determine if these certain laptops produce a keypress release I don't know with having all of them at hand...

The quirk is only applied to a small number of keys which wouldn't want autorepeat anyway, so it's pretty clearly harmless even if the hardware works sanely.

Quoted 3 lines

So it needs BOTH 'Dell Inc.' and 'Dell Computer Corporation' to cover these laptops (I guess), as per the info I got from people running dmidecode:

That's what the patch adds. I don't have appropriate hardware to test.

BTW, what was it that produced this issue in the first place? All previous kernel versions worked OK?

Does this work?

(quoted diff) +0 -0

b/drivers/input/keyboard/atkbd.c index 22016ca..15479ef 100644

drivers/input/keyboard/atkbd.c +0 -0

@@ -1460,6 +1460,15 @@ static struct dmi_system_id

atkbd_dmi_quirk_table[] __initdata = { .driver_data = atkbd_dell_laptop_keymap_fixup, },

drivers/input/keyboard/atkbd.c +3 -0

 	{
+		.ident = "Dell Laptop",
+		.matches = {
+			DMI_MATCH(DMI_SYS_VENDOR, "Dell Computer

Corporation"),

drivers/input/keyboard/atkbd.c +1 -0

+			DMI_MATCH(DMI_CHASSIS_TYPE, "8"), /*

Portable */

Does this work?

(quoted diff) +0 -0

b/drivers/input/keyboard/atkbd.c index 22016ca..15479ef 100644

drivers/input/keyboard/atkbd.c +0 -0

@@ -1460,6 +1460,15 @@ static struct dmi_system_id

atkbd_dmi_quirk_table[] __initdata = { .driver_data = atkbd_dell_laptop_keymap_fixup, },

drivers/input/keyboard/atkbd.c +3 -0

 	{
+		.ident = "Dell Laptop",
+		.matches = {
+			DMI_MATCH(DMI_SYS_VENDOR, "Dell Computer

Corporation"),

drivers/input/keyboard/atkbd.c +1 -0

+			DMI_MATCH(DMI_CHASSIS_TYPE, "8"), /*

Portable */

Some Dells need the dell input quirk applied but have a different vendor string in their DMI tables. Add an extra entry to cover these machines as well.

Signed-off-by Matthew Garrett <[email protected]>

---

Dmitry, any chance of getting this queued for .29? Probably suitable for -stable as well.

Quoted 4 lines · On Monday 29 December 2008 06:29:02 Matthew Garrett wrote:

Some Dells need the dell input quirk applied but have a different vendor string in their DMI tables. Add an extra entry to cover these machines as well.

Quoted 6 lines

Signed-off-by Matthew Garrett <[email protected]>

---

Dmitry, any chance of getting this queued for .29? Probably suitable for -stable as well.

Yep, will send with the rest of stuff for .29. Thank you Matthew.