On Tue, Apr 21, 2026 at 01:59:21PM +0100, Mark Brown wrote: > On Tue, Apr 21, 2026 at 11:19:34AM +0300, Mike Rapoport wrote: > > On Mon, Apr 20, 2026 at 06:54:31PM +0100, Mark Brown wrote: > > > > If that fflush() is needed it should go into the helper, particularly > > > since ksft_finshed() calls exit() so the fflush() will never be reachd. > > > Looks like I'm missing prints because of something in my environment, I'll > > just drop the fflush(). > > Interesting. Usually the culprit would be calls to _exit() that bypass > the libc cleanup but I'm not seeing any right now...
Interesting indeed :) My observations up till now that if I run the test on the host, it prints everything (without fflush()) changes at all). If I run the same test in a VM either using virtio-ng or ssh'ing into a debian VM, only the first part of prints is displayed. However if I run protection_keys_64 | tee in the same VM everything is printed. I'll just drop the fflush() from this patch and will investigate later. -- Sincerely yours, Mike.
