| Message ID | 2a49929d166ac88d4c7900c5961be77cbd601fa4.1642601554.git.luke.nowakowskikrijger@canonical.com |
|---|---|
| State | Superseded |
| Headers | show |
| Series | Expand Cgroup lib and modify controller tests | expand |
Hello Luke, Luke Nowakowski-Krijger <luke.nowakowskikrijger@canonical.com> writes: > When the v1 blkio controller is mounted it unmounts the v2 io controller > effectively bringing the amount of cgroupv2 controllers down and > triggering the tst_brk. Because this is exected behaivor it should be > changed to TWARN in case there is something funny still going on and > should be logged. Ah, we are in new territory here, the library assumes this won't happen. What happens if a test author requires the V2 io controller then requires the blkio controller? At best I think we will get TBROK saying some file doesn't exist. Not too bad, but nothing helpful either. Also what happens if the io controller was in use, disappears then returns? (that is, if it returns, because unmounting V1 controllers doesn't guarantee the V1 root is destroyed and it can block V2 from being used afterwards). The basic assumption we originally made was that we can't mess with the current CGroup setup beyond adding to it. LTP doesn't guarantee it won't mess up your system, but it at least tries not to. Pulling controllers out from underneath the system manager seems likely to cause wierd errors for people. Why not treat the blkio controller and io controller as the same thing? Or if they are not remotely compatible, then do tst_brk(TCONF... when we need blkio, but io is on V2. BTW, this is a moot point, but TWARN is really only used when we think an error was just caused by previous errors. Otherwise, if something is expected, then it should be TINFO. > > Signed-off-by: Luke Nowakowski-Krijger <luke.nowakowskikrijger@canonical.com> > --- > lib/tst_cgroup.c | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > > diff --git a/lib/tst_cgroup.c b/lib/tst_cgroup.c > index 3d56a3364..c53b88ed2 100644 > --- a/lib/tst_cgroup.c > +++ b/lib/tst_cgroup.c > @@ -677,7 +677,7 @@ static void cgroup_root_scan(const char *const mnt_type, > goto discard; > > if (root->ctrl_field) > - tst_brk(TBROK, "Available V2 controllers are changing between scans?"); > + tst_res(TWARN, "Available V2 controllers are changing between scans?"); > > root->ver = TST_CGROUP_V2;
diff --git a/lib/tst_cgroup.c b/lib/tst_cgroup.c index 3d56a3364..c53b88ed2 100644 --- a/lib/tst_cgroup.c +++ b/lib/tst_cgroup.c @@ -677,7 +677,7 @@ static void cgroup_root_scan(const char *const mnt_type, goto discard; if (root->ctrl_field) - tst_brk(TBROK, "Available V2 controllers are changing between scans?"); + tst_res(TWARN, "Available V2 controllers are changing between scans?"); root->ver = TST_CGROUP_V2;
When the v1 blkio controller is mounted it unmounts the v2 io controller effectively bringing the amount of cgroupv2 controllers down and triggering the tst_brk. Because this is exected behaivor it should be changed to TWARN in case there is something funny still going on and should be logged. Signed-off-by: Luke Nowakowski-Krijger <luke.nowakowskikrijger@canonical.com> --- lib/tst_cgroup.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)