lists.openwall.net   lists  /  announce  owl-users  owl-dev  john-users  john-dev  passwdqc-users  yescrypt  popa3d-users  /  oss-security  kernel-hardening  musl  sabotage  tlsify  passwords  /  crypt-dev  xvendor  /  Bugtraq  Full-Disclosure  linux-kernel  linux-netdev  linux-ext4  linux-hardening  linux-cve-announce  PHC 
Open Source and information security mailing list archives
 
Hash Suite: Windows password security audit tool. GUI, reports in PDF.
[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <20150501141322.GA9936@thunk.org>
Date:	Fri, 1 May 2015 10:13:22 -0400
From:	Theodore Ts'o <tytso@....edu>
To:	"U.Mutlu" <for-gmane@...luit.com>
Cc:	linux-ext4@...r.kernel.org, fstests@...r.kernel.org
Subject: Re: ext4 non-kernel standalone dev version for testing

On Fri, May 01, 2015 at 03:29:11AM +0200, U.Mutlu wrote:
> can the developer-version (ie. that with the crypto extension) be tested
> outside the kernel?
> Ie. testing with file type filesystem, ie. file.img, on a system with old kernel.

No, but you can test under KVM.  This is how I normally do most of my
ext4 development.  Things are a little complicated at the moment
because some e2fsprogs changes are needed that aren't in the binaries
in the kvm-xfstests image, but I'll try to get a new version sent out
soon.

						- Ted


Quick start instructions for kvm-xfstests
==========================================

0.  Make sure qemu with kvm support is installed on your system.

1.  Run the following commands:

    git clone git://git.kernel.org/pub/scm/fs/ext2/xfstests-bld.git fstests
    cd fstests/kvm-xfstests
    wget -O test_appliance/root_fs.img https://www.kernel.org/pub/linux/kernel/people/tytso/kvm-xfstests/root_fs.img.i386

Substitute "i386" for "x86_64" in the above URL if you want to test
using a native 64-bit userspace.  It should be possible to use the
32-bit userspace with a 64-bit kernel, although you may run into some
rough spots via-a-vis compat ioctls working correctly.  (If you do,
let me know; those are bugs that should be fixed.  :-)

2.  In the fstests/kvm-xfstests directory, take a look at the "config"
    file and either edit that file in place, or put override values in
    ~/.config/kvm-xfstests or config.custom.  The latter is preferred.

3.  Build a kernel with all of the necessary drives for kvm built into
    the kernel (i.e., no modules!).  There are some sample kernel
    configs in the kernel-configs directory.

4.  Run "./kvm-xfstests smoke" to do a quick test.  Or "./kvm-xfstests
    -g auto" to do a full test.  You can also run specific tests on
    specific configurations, i.e., "./kvm-xfstests -c bigalloc
    generic/013 generic/127".   To run a shell, use "./kvm-xfstests shell"

For more information, please see:

https://git.kernel.org/cgit/fs/ext2/xfstests-bld.git/tree/kvm-xfstests/README


Building a 32-bit kernel
========================


To build a 32-bit kernel, create a /build/ext4 directory (or change
directory name as you wish) and place a kernel configuration file in
/build/ext4/.config.  (There are some example config files in the
kernel-configs directory of the xfstests-bld git repo).  Then run:

make O=/build/ext4 ARCH=i386 -j16 $*

The reason for using a separate build directory is you can maintain a
separate build directory for 64-bit builds, which you can now build
using:

make O=/build/ext4-64 ARCH=x86_64 -j16 $*

I generally create shell scripts which look like this:

#!/bin/dash
cd /usr/projects/linux/ext4
make O=/build/ext4-64 ARCH=x86_64 -j16 $*

And store them as "make-ext4", "make-ext4-64", "make-ext4-crypto",
etc., in /usr/projects/linux.  The reason for having the "cd" command
is so that if you accidentally run "../make-ext4" while you are in the
/usr/projects/linux/ext4-crypto directory, instead of the
/usr/projects/linux/ext4 directory (for example), you won't end up
accidentally contaminating the build directory.

The remote ports feature
========================

While kvm-xfstests is running, you can telnet to a number of ports
(which are bound to localhost).  Ports 7500, 7501, and 7502 will
connect you to a shell prompts while the tests are running (if you
want to check on /proc/slabinfo, enable tracing, etc.)  You can also
use these ports in conjuction with "kvm-xfstests shell" if you want
additional windows to capture traces using ftrace.

You can also access the qemu monitor on port 7498, and you debug the
kernel using remote gdb on localhost port 7499.  Just run "gdb
/path/to/vmlinux", and then use the command "target remote
localhost:7499".  (Pro tip: it's helpful to temporarily add
"EXTRA_CFLAGS += -O0" to fs/{ext4,jbd2}/Makefile, and to make sure you
have CONFIG_DEBUG_INFO, CONFIG_DEBUG_INFO_DWARF4, and
CONFIG_FRAME_POINTER enabled.)


Miscellaneous kernel hacking hints
==================================

The following hints may be useful to folks who are new to kernel
development.

Using ccache
------------

Enabling ccache can really help speed up your kernel builds.  I
strongly recommend it.

Examples of using ftrace
------------------------

This is an example of how to debug the lazytime feature (which is
when you mount a file system using -m lazytime using a 4.0-rcX and
later kernels).

cd /sys/kernel/debug/tracing
echo 1 > events/writeback/writeback_lazytime/enable
echo 1 > events/writeback/writeback_lazytime_iput/enable
echo "state & 2048" > events/writeback/writeback_dirty_inode_enqueue/filter
echo 1 > events/writeback/writeback_dirty_inode_enqueue/enable
echo 1 > events/ext4/ext4_other_inode_update_time/enable
cat trace_pipe

The definition of the tracepoints can be found in
include/linux/trace/events.  The tracepoints used by ext4 and be found
in ext4.h, and by jbd2 in the jbd2.h files in that directory, and so
on.

For more information, please see:

http://lwn.net/Articles/365835/
http://lwn.net/Articles/366796/
http://lwn.net/Articles/370423/

--
To unsubscribe from this list: send the line "unsubscribe linux-ext4" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html

Powered by blists - more mailing lists

Powered by Openwall GNU/*/Linux Powered by OpenVZ