On Mon, 20 Feb 2017, Linus Torvalds wrote:
I agree; that's in line with what we usually try to stick to (forceI'll have a more specific commit (or range) soon.Hmm. It's commit 279967a65b32 ("HID: rmi: Handle all Synaptics
touchpads using hid-rmi").
And the reason seems to be stupid: I don't have RMI enabled at all,
because that didn't use to work or make a difference.
Maybe that "let's use RMI" code should depend on RMI actually being
enabled? Because as-is, that code now breaks existing configurations.
specific drivers if the device doesn't work with the generic at all and
switch over to generic in compile-time for devices that have limited
functionality with the generic driver).
Andrew, Benjamin, how about the patch below?
From: Jiri Kosina <jkosina@xxxxxxx>
Subject: [PATCH] HID: rmi: fallback to generic/multitouch if hid-rmi is not built
Commit 279967a65b32 ("HID: rmi: Handle all Synaptics touchpads using hid-rmi")
unconditionally switches over handling of all Synaptics touchpads to hid-rmi
(to make use of extended features of the HW); in case CONFIG_HID_RMI is
disabled though this renders the touchpad unusable, as the
HID_DEVICE(HID_BUS_ANY, HID_GROUP_RMI, HID_ANY_ID, HID_ANY_ID)
match doesn't exist and generic/multitouch doesn't bind to it either (due
to hid group mismatch).
Fix this by switching over to hid-rmi only if it has been actually built.
Fixes: 279967a65b32 ("HID: rmi: Handle all Synaptics touchpads using hid-rmi")
Reported-by: Linus Torvalds <torvalds@xxxxxxxxxxxxxxxxxxxx>
Signed-off-by: Jiri Kosina <jkosina@xxxxxxx>
drivers/hid/hid-core.c | 3 ++-
1 file changed, 2 insertions(+), 1 deletion(-)
diff --git a/drivers/hid/hid-core.c b/drivers/hid/hid-core.c
index 538ff697a4cf..e9e87d337446 100644
@@ -827,7 +827,8 @@ static int hid_scan_report(struct hid_device *hid)
* hid-rmi should take care of them,
* not hid-generic
- hid->group = HID_GROUP_RMI;
+ if (IS_ENABLED(CONFIG_HID_RMI))
+ hid->group = HID_GROUP_RMI;