On Tue, Jul 19, 2022 at 6:57 AM Rob Landley <[email protected]> wrote:
> On 7/18/22 19:02, enh wrote:> ack. i tried to take an update but hit > another > -Werror=format-security issue > > with one of your diff.c changes. i've sent a patch (and a separate > patch to > > add that -Werror= to the default toybox configure, since that's one > we > > always have to fix in the end anyway; may as well catch them fresh?). > > > > i'll try again tomorrow... (i want to try to use `timeout -i` too!) > > > > heh, this doesn't seem intentional (especially because it happens > without `-i`), > > but it wasn't obvious to me what the fix is? > > > > ~/toybox$ time ./toybox timeout -s SEGV 10 ./foo > > timeout: exec ./foo: No such file or directory > > > > real 0m10.035s > > user 0m0.009s > > sys 0m0.013s > > It isn't? If the child process exits, the pipe should close? Even if > SIGCHLD is > delivered before we can set the signal handler, that shouldn't... zombie > process... > sorry, i was in a rush to shut my laptop for the day... the point was real 0m10.035s i don't think a failure should wait for the full timeout to exit. but it wasn't obvious to me where to put the missing exit. check whether the global exit status has been set after xpopen_both()? or should xpopen_both() be handling this? i don't think this bug is a blocker, and i'll try to take an update today, but i noticed the first time i tried to use timeout (albeit via user error), so i likely won't be the last :-) > Lemme get back to you on this. There's is an elderly cat. > > Rob >
_______________________________________________ Toybox mailing list [email protected] http://lists.landley.net/listinfo.cgi/toybox-landley.net
