Hi, On Sat, Sep 10, 2022 at 11:16:15PM +0200, Paul Gevers wrote: > Hi, > > On 10-09-2022 04:23, Antoni Villalonga wrote: > > Hi Paul, > > > > Thanks for reporting the issue. > > > > I've tested the package on a virtual ppc64el Debian/Sid fresh installation. > > > > Testing the coreutils seq takes around 0.8 seconds: > > $ echo -ne '' | timeout 10 seq 10000000 > /dev/null > > > > Testing the toybox seq takes around 1.4 seconds: > > $ echo -ne '' | timeout 10 toybox seq 10000000 > /dev/null > > > > In that case, Coreutils seems to perform twice faster than Toybox. In any > > case > > both have taken lot less than 10 seconds specified for the test. > > And in an unstable container on one of our ppc64el host: > root@elbrus:/# time echo -ne '' | timeout 10 seq 10000000 > /dev/null > > real 0m0.183s > user 0m0.177s > sys 0m0.007s > root@elbrus:/# time echo -ne '' | timeout 10 toybox seq 10000000 > /dev/null > > real 0m0.426s > user 0m0.411s > sys 0m0.012s
Thanks for testing. > Are you sure it's this commmand that times out in the test? You are right the comand I issued doesn't reproduce the problem correctly. The problem was on toybox timeout implementation itself. It's already fixed upstream and I've pushed a patch to salsa. https://salsa.debian.org/debian/toybox/-/blob/2c4a6613ca8d579841224b6b762cc07680f23183/debian/patches/timeout.patch Thanks for your time! -- Antoni Villalonga https://friki.cat/