Bug#702102: fails to upgrade (cowbuilder) chroot

[Junichi Uekawa]

I think that patch is completely borken too.
/dev/shm seems to be a symlink to /run/shm and /run/shm doesn't seem to exist (in older debootstrap ? how did they get generated?)

anyway, I've committed a change to only look at /run/shm.


At Tue, 26 Mar 2013 18:48:40 +0000 (UTC),
Thorsten Glaser wrote:
> 
> Chris Knadle dixit:
> 
> >+    if [ "$DEB_BUILD_ARCH_OS" = "linux" ] && [ "$USEDEVSHM" = "yes" ]; then
> >+       SHM_PATH="run/shm"                                                                                                               
> >+       [ ! -d "/$SHM_PATH" ] && SHM_PATH="dev/shm"
> 
> This logic is totally and utterly borken.
> 
> Correct logic would be:
> 
> if outside_os = linux and inside_os = linux and USEDEVSHM:
> 	if exists CHROOT/run/shm:
> 		targetdir=run/shm
> 	else if exists CHROOT/dev/shm:
> 		targetdir=dev/shm
> 	else:
> 		abort
> 	mount --bind $(realpath /dev/shm) CHROOT/$targetdir
> 
> This separates the resolution of /dev/shm outside (Linux)
> from the resolution of /dev/shm or /run/shm inside.
> 
> I still believe this is not entirely right. If CHROOT/run
> exists, it should be populated with a tmpfs, including
> creation of subdirectories.
> 
> Does CHROOT/dev/shm/. need to be bind-mounted to whatever
> /dev/shm/. is outside of the chroot, for the eglibc shm
> semantics to work? If not… just mount a tmpfs there, too!
> 
> bye,
> //mirabilos
> -- 
> [DJBDNS Zone] TTL 86400 – <Natureshadow> kann man da auch 1d schreiben?
> <mirabilos> nö, außerdem kann ein Deutscher oder ein Japaner mit 1d
>     ja erstmal nix anfangen, oder könntest du 1日 im zone file lesen?
> <Natureshadow> das heißt für mich: ein Regal, das u.U. schiefstehen könnte
> 
> _______________________________________________
> Pbuilder-maint mailing list
> Pbuilder-maint at lists.alioth.debian.org
> http://lists.alioth.debian.org/cgi-bin/mailman/listinfo/pbuilder-maint