| Message ID | 20230630193742.310416-1-pvorel@suse.cz |
|---|---|
| State | Accepted |
| Headers | show |
| Series | [COMMITTED,1/1] runtest: Move fanotify23 from staging to syscalls | expand |
On Fri, Jun 30, 2023 at 10:38 PM Petr Vorel <pvorel@suse.cz> wrote: > > fanotify23 is a test for FAN_MARK_EVICTABLE, prepared for v5.19-rc1. > We forget to move it after v5.19 being released. > > Signed-off-by: Petr Vorel <pvorel@suse.cz> > --- > Obviously we forget to check the file after several kernel releases :(. > I'm sorry. Oh, missed your note on commit: [ pvorel: move test into staging ] If I'd known, I might have remembered to remind you ;-) fanotify23 and fanotify10 both added tests for FAN_MARK_EVICTABLE at the same time. But since then there were two attempts to improve the reliability of fanotify10: 48cfd7a99 syscalls/fanotify10: Make evictable marks test more reliable 4fefdf340 fanotify10: Make evictable marks tests more reliable I wonder if fanotify23's reliability did not come up so far because 1. It was in staging list 2. It has this hack: /* Shrinkers on other fs do not work reliably enough to guarantee mark eviction on drop_caches */ .dev_fs_type = "ext2", I guess we will know soon ... Thanks, Amir.
Hi Amir, > On Fri, Jun 30, 2023 at 10:38 PM Petr Vorel <pvorel@suse.cz> wrote: > > fanotify23 is a test for FAN_MARK_EVICTABLE, prepared for v5.19-rc1. > > We forget to move it after v5.19 being released. > > Signed-off-by: Petr Vorel <pvorel@suse.cz> > > --- > > Obviously we forget to check the file after several kernel releases :(. > > I'm sorry. > Oh, missed your note on commit: > [ pvorel: move test into staging ] > If I'd known, I might have remembered to remind you ;-) > fanotify23 and fanotify10 both added tests for > FAN_MARK_EVICTABLE at the same time. > But since then there were two attempts to improve the reliability > of fanotify10: > 48cfd7a99 syscalls/fanotify10: Make evictable marks test more reliable > 4fefdf340 fanotify10: Make evictable marks tests more reliable > I wonder if fanotify23's reliability did not come up so far because > 1. It was in staging list > 2. It has this hack: > /* Shrinkers on other fs do not work reliably enough to > guarantee mark eviction on drop_caches */ > .dev_fs_type = "ext2", > I guess we will know soon ... I wonder myself :). But I actually filled openSUSE bug [1] about fanotify23 randomly failing, when running more times (e.g. ./fanotify23 -i5). I was not sure if it's openSUSE specific or not (I suspect the problem is generic, but 2 tested Debian systems don't suffer this), thus I first put it to the Jan's queue. Kind regards, Petr [1] https://bugzilla.suse.com/1212906 > Thanks, > Amir.
diff --git a/runtest/staging b/runtest/staging index 88dcea79b..ef1cdea15 100644 --- a/runtest/staging +++ b/runtest/staging @@ -1,2 +1 @@ # Tests for features that are not yet in the stable kernel ABI -fanotify23 fanotify23 diff --git a/runtest/syscalls b/runtest/syscalls index d902af500..b29151186 100644 --- a/runtest/syscalls +++ b/runtest/syscalls @@ -612,6 +612,7 @@ fanotify19 fanotify19 fanotify20 fanotify20 fanotify21 fanotify21 fanotify22 fanotify22 +fanotify23 fanotify23 ioperm01 ioperm01 ioperm02 ioperm02
fanotify23 is a test for FAN_MARK_EVICTABLE, prepared for v5.19-rc1. We forget to move it after v5.19 being released. Signed-off-by: Petr Vorel <pvorel@suse.cz> --- Obviously we forget to check the file after several kernel releases :(. I'm sorry. Kind regards, Petr runtest/staging | 1 - runtest/syscalls | 1 + 2 files changed, 1 insertion(+), 1 deletion(-)