-
Stripping perf out from kernel source
You may have heard about perf, a tool made by Linux kernel developers to help debugging, benchmarking and profiling.
The problem is you have to download the whole kernel source in order to compile perf. If you are not used to follow kernel developping, most likely you will not have its source. But it’s still possible to compile perf.
As I’m helping to maintain perf package in Archlinux, I made the following script to strip perf from a given kernel version:
#!/bin/bash function usage() { cat - <<EOF USAGE: make-perf-util.sh git-tag-version EOF } function die() { echo -e "ERROR\n\tSee usage\n" usage exit 1 } VERSION=$1 BASENAME=perf-util NAME=$BASENAME-$VERSION PKGNAME=$NAME.tar.bz2 if [ $# -ne 1 ]; then die fi git checkout $VERSION || die [ -a $NAME ] && rm -rf $NAME cd tools/perf && make clean cd - rsync -Ra tools lib include/linux arch/x86/include/asm include/asm-generic \ --link-dest=. $NAME tar -cjf $PKGNAME $NAME rm -rf $NAME echo "created $PKGNAME"
This script works fine with latest 2.6.33 kernel. Regularly I’m uploading to this site the stripped version of perf. The latest one, you find here.
2 responses to “Stripping perf out from kernel source”
-
Is this something like OpenSolaris/FreeBSD DTrace? Does it need a lot of dependencies, doesn’t it? Is this suitable for embedded apps?
How different is perf from Valgrind?
-
Lucas De Marchi March 21st, 2010 at 21:47
Yes, it’s related to opensolaris/freebsd DTrace. There are a few dependencies, namely: elfutils, zlib, perl (perl dependency was introduced with latest version). To compile it you also need asciidoc and xmlto for documentation and libdwarf.
Valgrind is used mostly to find bugs in memory accesses like using an uninitialized var, freed memory, wrong pointer dereference etc. It does can be used to profile cache references, but it uses a simulation model to give you the miss/hit-rate. Perf, instead, relies on hardware counters and thus you can profile your application with the architecture it is running on.
It was first born as an access to hardware counters, but now it’s becoming the de facto performance debugging utility for linux kernel.
-
Leave a reply
-
Ricardo Guiraldelli March 16th, 2010 at 03:17