While testing region autodetect with physical hardware a few fixes fell out. The most interesting being evidence that a device is sensitive to 8-byte reads of 2 consecutive 4-byte registers. The other is a long reported issue by Jonathan on how "passthrough" decoders are detected, and having an example with physical hardware to reinforce the observation from QEMU.
The rest are ancillary fixes and new debug messages.
---
Dan Williams (5): cxl/hdm: Fail upon detecting 0-sized decoders cxl/hdm: Use 4-byte reads to retrieve HDM decoder base+limit cxl/core: Drop unused io-64-nonatomic-lo-hi.h cxl/port: Scan single-target ports for decoders cxl/hdm: Add more HDM decoder debug messages at startup
drivers/cxl/core/hdm.c | 52 ++++++++++++++++++++++++++++++++++++----------- drivers/cxl/core/mbox.c | 1 - drivers/cxl/core/port.c | 1 - drivers/cxl/port.c | 18 ++++++++++++---- 4 files changed, 53 insertions(+), 19 deletions(-)
base-commit: 24b18197184ac39bb8566fb82c0bf788bcd0d45b