trans_hugepage_relocated: fix some issues to make it run correctly

[yanan-fu]

depends on: https://github.com/avocado-framework/avocado-vt/pull/3335

id: 2032517 Signed-off-by: Yanan Fu [email protected]

[fbq815]

On s390x: (5/5) Host_RHEL.m9.u0.nographic.qcow2.virtio_scsi.up.virtio_net.Guest.RHEL.9.0.0.s390x.io-github-autotest-qemu.trans_hugepage.relocated.s390-virtio: ERROR: Hugepages allocated can not reach a half: 0/1024

@yanan-fu I've add the log in our bz, please check

[zhenyzha]

Very strange, the program seems to sleep after entering the guest and does not continue to run. No error is reported after waiting for more than 4 hours, and the test can only be interrupted manually.

Add detailed log to bug.

[zhenyzha]

BTW https://github.com/autotest/tp-qemu/blob/master/generic/tests/cfg/trans_hugepage.cfg#L21 The thp_test_config path needs to be modified on ARM, I have corrected it locally, but the auto case is still not completed. on ARM thp_test_config = "/sys/kernel/mm/transparent_hugepage/khugepaged/pages_to_scan:4096;/sys/kernel/mm/transparent_hugepage/khugepaged/scan_sleep_millisecs:10000;/sys/kernel/mm/transparent_hugepage/khugepaged/alloc_sleep_millisecs:60000"

[fbq815]

On s390x: (5/5) Host_RHEL.m9.u0.nographic.qcow2.virtio_scsi.up.virtio_net.Guest.RHEL.9.0.0.s390x.io-github-autotest-qemu.trans_hugepage.relocated.s390-virtio: ERROR: Hugepages allocated can not reach a half: 0/1024

@yanan-fu I've add the log in our bz, please check

@yanan-fu On s390x, VM memory backing will not cost THP in the host, please add no s390x for this case as well

[mcasquer]

@yanan-fu any update about this? In my last comment I suggested to change the vmsm variable value, is there any other suggestion?