access to nbd-mounted volume causes kernel panic, oops, shock and death
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
linux (Ubuntu) |
Expired
|
High
|
Unassigned |
Bug Description
Clients:
various, with Ubuntu 10.04.1 LTS i386 (desktop), or 10.04.2 LTS i386 (installed as terminal-client)
Last tested Ethernet Controller: nVidia Corporation MCP61 Ethernet (forcedeth driver)
Server:
DistroRelease: Ubuntu 10.04
InstallationMedia: Ubuntu-Server 10.04 LTS "Lucid Lynx" - Release amd64 (20100427)
Package: nbd-server 1:2.9.14-2ubuntu1
PackageArchitec
ProblemType: Bug
ProcEnviron:
LANG=en_US.UTF-8
SHELL=/bin/bash
ProcVersionSign
SourcePackage: nbd
Tags: lucid
Uname: Linux 2.6.32-30-server x86_64
Ethernet controller: Realtek Semiconductor Co., Ltd. RTL8111/8168B PCI Express Gigabit Ethernet controller
(driver r8169 driver)
Steps to reproduce:
on server:
mkdir /var/tmp/cargo
dd if=/dev/urandom of=/var/tmp/cargo/1 bs=1048576 count=128
dd if=/dev/urandom of=/var/tmp/cargo/2 bs=1048576 count=128
dd if=/dev/urandom of=/var/tmp/cargo/3 bs=1048576 count=128
dd if=/dev/urandom of=/var/tmp/cargo/4 bs=1048576 count=128
mksquashfs /var/tmp/cargo /var/tmp/cargo.img -noI -noD -noF -b 4096 # image size is 513Mb, comparable with LTSP image
nbd-server 2000 /var/tmp/cargo.img -C /dev/null -r
on client:
nbd-client ${SERVER_IP} 2000 /dev/nbd0 # tried with options bs=4096 and bs=1024
mount -o ro -t squashfs /dev/nbd0 /var/tmp/nbdtmp
while true; do cat /var/tmp/nbdtmp/* > /dev/null; done # any access will trigger
expected:
succesfull mounted volume on client;
in reproducing scenario: high load of cpu making stupid things in endless, but interruptable process
in ltsp booting: ltsp_nbd boot scenario to continue after mounting root device
observed:
mounted with "nbd0: unknown partition table in kernel.log"
random kernel segfaults, Oops, panic, and some_random_
stack trace is different each time, at time of reporting bug it failed to get logged.
attaching kernel log as an example
ProblemType: Bug
DistroRelease: Ubuntu 10.04
Package: nbd-client 1:2.9.14-2ubuntu1
ProcVersionSign
Uname: Linux 2.6.32-27-generic i686
NonfreeKernelMo
Architecture: i386
Date: Tue Apr 12 18:13:48 2011
ProcEnviron:
LANGUAGE=
LANG=en_US.UTF-8
SHELL=/bin/bash
SourcePackage: nbd
Kernel oopsed, so I reckon this is best in the linux package (even if it did happen in response to being tickled by nbd).
You'll probably get some mails asking for apport information etc and possibly to try other kernels.