This reverts commit 218ad8f.

(corresponds to commit 810a019)

This reverts commit 36f3225.

(corresponds to commit a13b184)

This reverts commit 4bb6b4b.

(corresponds to commit 71867ba)

This reverts commit b026a27.

(corresponds to commit 09372ba)

(cherry-picked from commit 434376f)
[changed commit title from "Revert downstream_hpd_needs_d0 changes
in i915_legacy" because 5.3 kernel used i915_legacy but 4.19 kernel
uses i915.]

This commit will revert all the changes made by ipts patch on 4.19.

(corresponds to commit 464d29d)
(am from patch https://github.com/kitakar5525/linux-surface-patches/blob/v3.7/patch-4.19/0000-not-merged-into-qzed_linux-surface-yet/0006-ipts-hid-multitouch-revert-all-the-changes-for-4.19.patch)

The previous commit reverted the changes to hid-multitouch that were done by
the IPTS patch. The original implementation modified hid-multitouch to make the
single-touch mode work. However, window managers such as GNOME's mutter cannot
differentiate between two touchscreens on the same panel. This results in broken
touch input rotation (the single-touch panel gets rotated, even if it is not
active).

jakeday/linux-surface#549 had some discussion on this
and multiple fixes. But I think the less changes to unrelated kernel code, the
better. I don't think that anyone actually uses single-touch mode.

Since single-touch mode was disabled by the previous commit, remove all user-
space APIs to enable it, since that just sends all touch input to /dev/null

Because IPTS does switch it's mode during initialization, it is not possible to
remove all references to the sensor mode, at least not without a more in-depth
review.

Signed-off-by: Dorian Stoll <dorian.stoll@tmsp.io>

(cherry-picked from commit 86f95d2)

This makes it easier to extend the companion driver, since everything is
stored in a single struct.

The ipts_companion_t struct is stored in two places, the IPTS main driver,
and the driver registering it. It is still passed through to the functions
that the companion driver implements to allow for device specific behaviour.

Signed-off-by: Dorian Stoll <dorian.stoll@tmsp.io>

(cherry-picked from commit a1531a7)

Since ipts_fw_config.bin is mapped directly onto C structs by the IPTS
driver, we can declare those values ourselves, instead of putting them
into an obscure binary file that was provided by Intel ages ago.

This could also open up the door to modifying the firmware config in
case anyone ever understands IPTS that much.

Signed-off-by: Dorian Stoll <dorian.stoll@tmsp.io>

(cherry-picked from commit e1cfd84)

* Removed some of the gotos in the simpler companion function
* Made the companion driver the first parameter in the firmware handler
* Changed ipts-surface to warn when adding the companion driver failed
* Made ipts_alloc_fw_list static
* Streamlined ipts_request_firmware_config
* Added explicit values for ipts_bin_data_file_flags_t elements

Signed-off-by: Dorian Stoll <dorian.stoll@tmsp.io>

(cherry-picked from commit 636e23d)

Signed-off-by: Dorian Stoll <dorian.stoll@tmsp.io>

(cherry-picked from commit a16bb16)

Signed-off-by: Dorian Stoll <dorian.stoll@tmsp.io>

(cherry-picked from commit bda1092)

Signed-off-by: Dorian Stoll <dorian.stoll@tmsp.io>

(cherry-picked from commit 30a9d30)

Signed-off-by: Dorian Stoll <dorian.stoll@tmsp.io>

(cherry-picked from commit 52e26dd)

Signed-off-by: Dorian Stoll <dorian.stoll@tmsp.io>

(cherry-picked commit from 9c9dfad)

Signed-off-by: Dorian Stoll <dorian.stoll@tmsp.io>

(cherry-picked from commit 05f51be)

Signed-off-by: Dorian Stoll <dorian.stoll@tmsp.io>

(cherry-picked from commit 4aaed39)

Signed-off-by: Dorian Stoll <dorian.stoll@tmsp.io>

(cherry-picked from commit 24f97ef)

Signed-off-by: Dorian Stoll <dorian.stoll@tmsp.io>

(cherry-picked from commit f63f758)

Signed-off-by: Dorian Stoll <dorian.stoll@tmsp.io>

(cherry-picked from commit 0774af9)

Signed-off-by: Dorian Stoll <dorian.stoll@tmsp.io>

(cherry-picked from commit c0556e1)

Signed-off-by: Dorian Stoll <dorian.stoll@tmsp.io>

(cherry-picked from commit 1fb6e17)

Signed-off-by: Dorian Stoll <dorian.stoll@tmsp.io>

(cherry-picked from commit 1050bfc)

Signed-off-by: Dorian Stoll <dorian.stoll@tmsp.io>

(cherry-picked from commit 863de54)

Signed-off-by: Dorian Stoll <dorian.stoll@tmsp.io>

(cherry-picked from commit 9f369e2)

Signed-off-by: Dorian Stoll <dorian.stoll@tmsp.io>

(cherry-picked from commit 1623943)

Signed-off-by: Dorian Stoll <dorian.stoll@tmsp.io>

(cherry-picked from commit b7c60a6)
[resolved conflicts caused by changes between 4.19 and 5.3]

When cherry-picking the commit b7c60a6
(ipts: Cleanup ipts-i915 to pass checkpatch), the following line added
accidentally.

        drivers/gpu/drm/i915/i915_irq.c:40:10: fatal error: intel_psr.h: No such file or directory
           40 | #include "intel_psr.h"
              |          ^~~~~~~~~~~~~

This commit fixes up the accidentally added line by removing it.

Fixup the following commits:
    BACKPORT: ipts: Cleanup ipts-sensor-regs.h to pass checkpatch
    BACKPORT: ipts: Clean up ipts-mei-msg.h to pass checkpatch

The static_assert macro is not available on 4.19. So, there are two ways
to fixup those commits:
    1. Backport upstream commit[1] that adds the static_assert
    2. Add the definition to ipts header files

I chose the latter way because backporting the upstream commit causes
conflicts and I have to a) backport other commit(s) or b) resolve the
conflict manually.

Also, the original IPTS patch[2] is doing the same way; they defined
C_ASSERT in those files [3][4].

References:
[1] torvalds/linux@6bab69c
[2] https://github.com/ipts-linux-org/ipts-linux-new
[3] https://github.com/ipts-linux-org/ipts-linux-new/blob/master/drivers/misc/ipts/ipts-sensor-regs.h
[4] https://github.com/ipts-linux-org/ipts-linux-new/blob/master/drivers/misc/ipts/ipts-mei-msgs.h

Signed-off-by: Dorian Stoll <dorian.stoll@tmsp.io>

(cherry-picked from commit ec34a39)

Signed-off-by: Dorian Stoll <dorian.stoll@tmsp.io>

(cherry-picked from commit 15585f5)

Signed-off-by: Dorian Stoll <dorian.stoll@tmsp.io>

(cherry-picked from commit 6fad9aa)

Having a DMI match for some surface devices doesn't really match with
the concept of having a device-specific companion driver, so this commit
moves the matching code to the companion.

ipts-surface is now setup in a way that allows us to add device-specific
data pretty easily. All you need is the interface to IPTS, which can
just be copied from no_feedback.

Signed-off-by: Dorian Stoll <dorian.stoll@tmsp.io>

(cherry-picked from commit 40f343b)

Signed-off-by: Dorian Stoll <dorian.stoll@tmsp.io>

(cherry-picked from commit 55e49e1)

This reverts commit 7f59405.

(corresponds to commit d95adde)

(cherry-picked from commit 8dbbfa6)

We don't want IPTS to set itself to singletouch mode in case it thinks
that might fix anything, because it doesn't. Even a malfunctioning
multitouch IPTS has a higher chance of working than singletouch.

Signed-off-by: Dorian Stoll <dorian.stoll@tmsp.io>

(cherry-picked from commit 894fbba)