Closed jsturdy closed 5 years ago
@jsturdy so you want to introduce transactions re-trying? The best place to implement this is in the librwreg, so I need the source code for the librwreg_memhub. BTW, why it is not here? Is it in a different repository? If so, please provide a reference and explanation, why it is not here.
Brief summary of issue
Checking the status of a set of VFAT registers, e.g.,
rwc OHX*VFAT*VFAT*ContReg0
, often(always?) results in at least one or more0xdeaddead
read back valuesvfat_info_uhal.py
only reports these when the chip is actually not present.Types of issue
Expected Behavior
0xdeaddead
should only be reported for truly disconnected hardware, single failed transactions should be eliminated insofar as that is possibleCurrent Behavior
In the
vfat_info_uhal.py
output, the non0x37
values are real aberrations, currently under investigation, while the0xdeaddead
readback fromgem_reg.py
is a spurious failed transaction of some kind.Context (for feature requests)
eagle33
is using thememhub
transaction serializer, so collisions should be minimal, the primary failures likely coming from the timeout limitations of theCTP7
linux image on remote registers (fixed in a future linux image, FW not ported to this image)librwreg.so
:librwreg_memhub.so
ipbus
:ipbus_memhub
This issue is to track any reliability issues that are solely on this side of that divide, assuming that the linux core image issues and overlapping transactions will be addressed in a more appropriate place.