On Mon, 2014-07-21 at 04:59 -0600, Gary Thomas wrote: > On 2014-07-21 02:27, Richard Purdie wrote: > > On Thu, 2014-07-17 at 15:53 +0800, Chen Qi wrote: > >> The following changes since commit > >> 846bc50fde11bbb36c8eb5b2e3ae6bb644c037f3: > >> > >> ltp: use "foreign" automake strictness (2014-07-16 10:27:16 +0100) > >> > >> are available in the git repository at: > >> > >> git://git.openembedded.org/openembedded-core-contrib ChenQi/shadow-4.2.1 > >> > >> http://cgit.openembedded.org/cgit.cgi/openembedded-core-contrib/log/?h=ChenQi/shadow-4.2.1 > >> > >> Chen Qi (2): > >> shadow: upgrade from 4.1.4.3 to 4.2.1 > >> oeqa: fix return status in pam.py to match shadow-4.2.1 > > > > I think but am not 100% sure this has introduced another regression: > > > > https://autobuilder.yoctoproject.org/main/builders/nightly-oecore/builds/177 > > > > The issue is that X doesn't start in oe-core built images. It does work > > correctly in poky. The issue is that the /etc/init.d/xserver-nodm script > > returns: > > > > root@qemuarm:/etc# /etc/init.d/xserver-nodm start > > Starting Xserver > > su: applet not found > > > > or to be more specific: > > > > root@qemuarm:/etc# su -l -c '/etc/X11/Xserver&' xuser > > su: applet not found > > > > > > The poky images appear to work since we install bash. If you remove bash > > from packagegroup-core-device-devel.bb, the poky images will show the > > same error. > > > > I think this is something to do with the none-suid busybox binary not > > containing su support, but why su is being called there, I don't know. > > su is being provided by shadow in the image. If I make su provided by > > busybox the problem also goes away. > > The problem (reported last week on the Yocto list) is su.shadow > changed how it behaves. The old version would end up passing > the script off by > 6761 execve("/bin/sh", ["/bin/sh", "-c", "/etc/X11/Xserver"], [/* 7 vars > */]) = 0 > whereas the new version does > 961 execve("/bin/sh", ["-su", "-c", "/etc/X11/Xserver"], [/* 7 vars */]) > = 0
Hmm, busybox is interpreting that "-su" as an argument to call su, which busybox.nosuid doesn't have within it as an applet, hence the error message. > I still haven't figured out why su.shadow now behaves this way. That does indeed look like what we need to figure out... Cheers, Richard -- _______________________________________________ Openembedded-core mailing list Openembedded-core@lists.openembedded.org http://lists.openembedded.org/mailman/listinfo/openembedded-core
