It appears that the driver has been developed without proper thinking
of what the difference between HW IRQ and Linux IRQ (also known as vIRQ).
This misunderstanding led to the 0 being used as vIRQ which is no-no and
platform APIs unveil this after the commit a85a6c86c25b ("driver core:
platform: Clarify that IRQ 0 is invalid"). With this the Intel Broxton
Whiskey Cove PMIC driver has to be fixed all over the places.
In this series two immediate fixes as reported recently and we have
drivers for (patches 1 & 2) followed by the patch that prepares the
same fix for the devices that have no in-tree drivers yet. At the end
it is a fix for IRQ domain debugging (when enabled), but it requires
quite recent patches in IRQ domain and regmap IRQ subsystems, which is
reflected in the Depends-on tag (I can't use Fixes, as this doesn't
fix anything related to that).
The series touches a couple of subsystems as the respective drivers
has to be fixed simultaneously. The idea is to proceed the series via
MFD tree to where the main driver belongs.
I have tested this on Intel Joule SoM (Apollo Lake SoC).
In v2:
- added tags (Hans, Zhang)
- fixed typo in the commit messages
Andy Shevchenko (4):
mfd: intel_soc_pmic_bxtwc: Use IRQ domain for USB Type-C device
mfd: intel_soc_pmic_bxtwc: Use IRQ domain for TMU device
mfd: intel_soc_pmic_bxtwc: Use IRQ domain for PMIC devices
mfd: intel_soc_pmic_bxtwc: Fix IRQ domain names duplication
drivers/mfd/intel_soc_pmic_bxtwc.c | 144 ++++++++++++++++---------
drivers/platform/x86/intel/bxtwc_tmu.c | 22 ++--
drivers/usb/typec/tcpm/wcove.c | 4 -
3 files changed, 97 insertions(+), 73 deletions(-)
--
2.43.0.rc1.1336.g36b5255a03ac
