On-going discussions if monitoring available physical memory is a good metric to decide TCG available memory, due to possible capping from containers/cgroups.
I'm retrying systemd tests (autopkgtests failed for previous qemu version) to help in finishing the migrations.. and will wait a bit more time to see if they come up into a final conclusion.
I could tackle this with a temporary patch using patch from Alex, or even the similar approach - sysconf(_SC_PHYS_PAGES) - proposed by Christian... but I would be only allowing v5 not to block other packages in migration without a final solution...
If needed I can do it, but let's wait a bit longer for upstream for now...
I'm currently following this due to on-going nettle transitions being blocked by current v5 (removed from archive so migrations could happen).
Latest upstream status:
https:/ /lists. nongnu. org/archive/ html/qemu- devel/2020- 07/msg05547. html
On-going discussions if monitoring available physical memory is a good metric to decide TCG available memory, due to possible capping from containers/cgroups.
I'm retrying systemd tests (autopkgtests failed for previous qemu version) to help in finishing the migrations.. and will wait a bit more time to see if they come up into a final conclusion.
I could tackle this with a temporary patch using patch from Alex, or even the similar approach - sysconf( _SC_PHYS_ PAGES) - proposed by Christian... but I would be only allowing v5 not to block other packages in migration without a final solution...
If needed I can do it, but let's wait a bit longer for upstream for now...