Bug#853918: syslinux-utils: Unable to build writable installation thumb drive
Package: syslinux-utils
Version: 5.00+dfsg-1
Severity: grave
Justification: renders package unusable
Dear Maintainer,
I've followed the instructions at
https://www.debian.org/releases/stable/amd64/ch04s03.html.en section
4.3.3 (essentially
http://hyper.to/blog/link/debian-installer-on-a-usb-key/) for several
years to make Debian installer thumb drives that are writable. I
discovered that the ones I made on a Jessie machine will not boot into
the installer whereas ones made on a Wheezy machine will.
I traced the break to when syslinux was upgraded from 4.06 to 5.00. If
I boot a thumb drive created using 4.06 in QEMU
(sudo qemu-system-x86_64 -hdb /dev/sdb), it begins like this:
Booting from Hard Disk...
MBR
Loading vmlinuz.....
Loading initrd.gz............ready.
Probing EDD (edd=off to disable)...ok
After a few seconds, the Debian installer appears. Good. If I use
5.00, this is how the boot begins:
Booting from Hard Disk...
MBR
Loading vmlinuz... ok
Probing EDD (edd=off to disable)...ok
Note how initrd.gz is not mentioned. Eventually the boot process will
end with a kernel panic that looks like one of these two snippets:
example 1
[ 0.832110] DR3: 0000000000000000 DR6: 0000000000000000 DR7: 0000000000000000
[ 0.832110] Stack:
[ 0.832110] ffff880000000010 ffff8800070b7eb0 ffff8800070b7e50 ffff8800070b7ea0
[ 0.832110] ffff8800070b7eb8 0000000000000012 0000000000000001 000000000000000a
[ 0.832110] 000000000000fffe ffff880000088000 0000000000008001 ffffffff81704fb5
[ 0.832110] Call Trace:
[ 0.832110] [<ffffffff819035a7>] ? mount_block_root+0x2a9/0x2b8
[ 0.832110] [<ffffffff811bae95>] ? SyS_mknod+0x185/0x210
[ 0.832110] [<ffffffff81903739>] ? prepare_namespace+0x133/0x169
[ 0.832110] [<ffffffff81903258>] ? kernel_init_freeable+0x1d7/0x1e1
[ 0.832110] [<ffffffff8190295e>] ? initcall_blacklist+0xb2/0xb2
[ 0.832110] [<ffffffff81507da0>] ? rest_init+0x80/0x80
[ 0.832110] [<ffffffff81507daa>] ? kernel_init+0xa/0xf0
[ 0.832110] [<ffffffff8151ad18>] ? ret_from_fork+0x58/0x90
[ 0.832110] [<ffffffff81507da0>] ? rest_init+0x80/0x80
[ 0.832110] Code: c3 64 eb b1 83 3d 48 4d 55 00 00 74 05 e8 81 d0 b7 ff 48 c7 c6 c0 67 a6 81 48 c7 c7 f8 68 71 81 31 c0 e8 66 06 00 00 fb 66 66 90 <66> 66 90 45 31 e4 e8 9d ce be ff 4d 39 ec 7c 18 41 83 f6 01 44
[ 0.832110] RIP [<ffffffff81511a58>] panic+0x1c2/0x206
[ 0.832110] RSP <ffff8800070b7e38>
[ 0.832110] ---[ end trace c6e5ec37ea66b262 ]---
example 2
[ 0.801766] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS1.7.5-20140531_083030-gandalf 04/01/2014
[ 0.801930] 0000000000000000 ffffffff81514c11 ffffffff817054c8 ffff8800070b7ea0
[ 0.802072] ffffffff8151195e ffff880000000010 ffff8800070b7eb0 ffff8800070b7e50
[ 0.802166] ffff8800070b7ea0 ffff8800070b7eb8 0000000000000012 0000000000000001
[ 0.802282] Call Trace:
[ 0.802572] [<ffffffff81514c11>] ? dump_stack+0x5d/0x78
[ 0.802654] [<ffffffff8151195e>] ? panic+0xc8/0x206
[ 0.802734] [<ffffffff819035a7>] ? mount_block_root+0x2a9/0x2b8
[ 0.802788] [<ffffffff811bae95>] ? SyS_mknod+0x185/0x210
[ 0.802841] [<ffffffff81903739>] ? prepare_namespace+0x133/0x169
[ 0.802893] [<ffffffff81903258>] ? kernel_init_freeable+0x1d7/0x1e1
[ 0.802945] [<ffffffff8190295e>] ? initcall_blacklist+0xb2/0xb2
[ 0.802996] [<ffffffff81507da0>] ? rest_init+0x80/0x80
[ 0.803046] [<ffffffff81507daa>] ? kernel_init+0xa/0xf0
[ 0.803096] [<ffffffff8151ad18>] ? ret_from_fork+0x58/0x90
[ 0.803146] [<ffffffff81507da0>] ? rest_init+0x80/0x80
[ 0.803506] Kernel Offset: 0x0 from 0xffffffff81000000 (relocation range:0xffffffff80000000-0xffffffff9fffffff)
[ 0.803738] ---[ end Kernel panic - not syncing: VFS: Unable to mountroot fs on unknown-block(0,0)
The final line from the second panic spew dovetails with fact that the
beginning of the bad boot is missing a reference to initrd.gz. I'm now
going through the syslinux git repo to see exactly what caused this
change or if the problem is due to something bad in the debianification
process.
I feel strongly that this needs to be fixed before Stretch is finalized
and hopefully backported to Jessie.
Reply to: