iommu/exynos: Abstract getting the fault info

iommu/exynos: Abstract getting the fault info

Fault info obtaining is implemented for SysMMU v1..v5 in a very hardware
specific way, as it relies on:
  - interrupt bits being tied to read or write access
  - having separate registers for the fault address w.r.t. AR/AW ops

Newer SysMMU versions (like SysMMU v7) have different way of providing
the fault info via registers:
  - the transaction type (read or write) should be read from the
    register (instead of hard-coding it w.r.t. corresponding interrupt
    status bit)
  - there is only one single register for storing the fault address

Because of that, it is not possible to add newer SysMMU support into
existing paradigm. Also it's not very effective performance-wise:
  - checking SysMMU version in ISR each time is not necessary
  - performing linear search to find the fault info by interrupt bit can
    be replaced with a single lookup operation

Pave the way for adding support for new SysMMU versions by abstracting
the getting of fault info in ISR. While at it, do some related style
cleanups as well.

This is mostly a refactoring patch, but there are some minor functional
changes:
  - fault message format is a bit different; now instead of AR/AW
    prefixes for the fault's name, the request direction is printed as
    [READ]/[WRITE]. It has to be done to prepare an abstraction for
    SysMMU v7 support
  - don't panic on unknown interrupts; print corresponding message and
    continue
  - if fault wasn't recovered, panic with some sane message instead of
    just doing BUG_ON()

The whole fault message looks like this now:

    [READ] PAGE FAULT occurred at 0x12341000

Signed-off-by: Sam Protsenko <semen.protsenko@linaro.org>
Acked-by: Marek Szyprowski <m.szyprowski@samsung.com>
Link: https://lore.kernel.org/r/20220726200739.30017-2-semen.protsenko@linaro.org
Signed-off-by: Joerg Roedel <jroedel@suse.de>