Add 1 to number of counters due to PMCCNTR being always on#106
Add 1 to number of counters due to PMCCNTR being always on#106lsf37 merged 1 commit intoseL4:masterfrom
Conversation
|
Please add what you have in the PR description to the commit message -- otherwise this is going to be an extremely mysterious change if someone looks in 10 years. The key bit of info is not that |
JDuchniewicz
force-pushed
the
fix_arm_pmcr
branch
from
8c2d5a7 to
7b4c59e
Compare
Indanz
added
hw-build
lsf37
force-pushed
the
fix_arm_pmcr
branch
3 times, most recently
from
e5f9e75 to
3e207f0
Compare
|
Apologies for the force pushes, I was trying to get the PR checks to run. I have tweaked the commit message wording very slightly to get that to happen. |
PMCCNTR is always present, even when PMCR reports 0 as the number of event counters. We therefore always add +1 to the reported number of counters. Signed-off-by: Jakub Duchniewicz <[email protected]> Signed-off-by: Gerwin Klein <[email protected]>
|
Which reveals that the commit title was too long :-) Have fixed that now. |
I tried doing the same by adding and removing the hw label. It's silly that there isn't just a CI rerun button somewhere. You can do it when there are failed tests, but not when it hangs in limbo waiting for tests to finish for months. |
3 participants
Fixes the issue #64 which was uncovered when debugging RPi4B in AArch32 mode where PMCR reports zero meaning there are no dedicated event counters - but PMCCTR is always present. Thus adding 1 to this register's return value.