In many cases kicking watchdog is required from the very beginning, since otherwise the system gets rebooted before it manages to mount /usr (especially if the latter requires fsck). A notable example would be a system with no initrd, with / partition on UBI and /usr on an embedded/external mmc card. Is this reason good enough? Reproducible: Always
Created attachment 359288 [details] ebuild
Comment on attachment 359288 [details] ebuild --- watchdog-5.13-r1.ebuild 2013-09-23 14:17:15.171886504 +0200 +++ - 2013-09-23 14:17:55.176345321 +0200 @@ -24,7 +24,7 @@ append-cppflags $(${PKG_CONFIG} libtirpc --cflags) export LIBS+=" $(${PKG_CONFIG} libtirpc --libs)" fi - econf $(use_enable nfs) + econf --sbindir=/sbin $(use_enable nfs) } src_install() {
Based on the council's decision in the 24-Sep meeting and the news item that is going out on 27-Sep, I would recommend closing this as wontfix.
(In reply to William Hubbs from comment #3) > Based on the council's decision in the 24-Sep meeting and the news item > that is going out on 27-Sep, I would recommend closing this as wontfix. Could you be more specific, Mr. Hubbs?
(In reply to grey dot from comment #4) > (In reply to William Hubbs from comment #3) > > Based on the council's decision in the 24-Sep meeting and the news item > > that is going out on 27-Sep, I would recommend closing this as wontfix. > > Could you be more specific, Mr. Hubbs? If you need separate /usr, then you should use initramfs that mounts the /usr early so it's available for the rest of boot process That's what the council voted for However it's up to package maintainers if they want to support it by other means, like moving this binary in /sbin is still doable but the maintainer now has legimate reasons for refusing such request if he wants
busybox provides a watchdog and lives in / if you really want one