Hi guys, I found a bug in Busybox 1.20.2 (Debian 1:1.20.0-7), but before I report it, I want to make sure it's not a dup of #367605 (https://groups.google.com/forum/?fromgroups=#!searchin/linux.debian.bugs.dist/busybox$20loop/linux.debian.bugs.dist/fT4wUlqfCV4/d920VARKhCwJ).
It seems very similar, but there are a number of significant differences, so I think the cause for the same symptom may be something entirely unrelated. So, here's what I did: Boot debian-7.0.0-amd64-netinst.iso (or any other image, I also tested ...CD-1.iso) via grub2 installed on my USB thumb drive. This of course only loads the kernel and initrd, so the installer fails to "Detect and mount CD-ROM". This is perfectly expected, so I press Alt+F2, mount my USB (mount /dev/sdb1 /mnt) and try to mount the image: mount -o loop /mnt/debian-7.0.0-amd64-netinst.iso /cdrom This is where the trouble begins. The command never finishes, gives no output whatsoever, but does not hang, so you can cancel it anytime pressing Ctrl+C. I discovered that after this, there are tons of loop device nodes in /dev, so I suspect what happens is something like this: Busybox tries to setup the .iso to /dev/loop0 (or any other number), this fails, so it increases the number (creating the device node if it does not exist) and tries again. The real bug seems to be that setting up the loop device always fails, so the above logic enters an infinite loop. You can work around the bug by unpacking the iso into a folder on the USB drive and symlinking (or bind-mounting) that to /cdrom. Additional info: Kernel 3.2.0-4-amd64 SMP #1 Debian 3.2.41-2 x86_64 Busybox 1.20.2 (Debian 1:1.20.0-7) eglibc 2.13-38 I tested this on multiple machines, including a Dell PowerEdge R520 and some Dell laptops. Ubuntu 13.04 (server) ships Busybox 1.20.2 (Ubuntu 1:1.20.0-8ubuntu1) + Linux 3.8.0-19, and is not affected, mounting the image works fine, I can continue normal installation pressing Alt+F1 and selecting "Load installer components from CD". Likewise for Ubuntu 12.04.2 (Busybox 1.18.5, Linux 3.5.0-23). I tried to confirm it was not a kernel bug by building a new initrd with Ubuntu's busybox (and Ubuntu's eglibc 2.17 that it depends on), but ran into various incompatibilities and gave up. Can someone confirm the bug and/or my assumptions? I would also appreciate some insight into whether this could be the same problem as in bug #367605. Thank you in advance. Best regards, Julian PS: please CC me, as I'm not subscribed to the list. -- To UNSUBSCRIBE, email to debian-user-requ...@lists.debian.org with a subject of "unsubscribe". Trouble? Contact listmas...@lists.debian.org Archive: http://lists.debian.org/1367927748.4347.31.camel@xpsubuntu
