Pre-processing /proc data during the build considerably reduces the
amount of data written to disk: 176KB instead of 4.7MB for a 20
minuted build. Parsing also becomes faster.
buildstats.bbclass only writes the reduced logs now, but support for
the full /proc files is kept around as reference.
(From OE-Core rev: b5e47df9af1ebbb477074587fdeae17eb2f55582)
Signed-off-by: Patrick Ohly <patrick.ohly@intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The internal representation after parsing now matches exactly
what the drawing code needs, thus speeding up drawing a bit.
However, the main motivation is to store exactly that required
information in a more compact file.
(From OE-Core rev: ca06e67a0bb5820b38fda4c8dfee20764c1e59ae)
Signed-off-by: Patrick Ohly <patrick.ohly@intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This adds a new, separate chart showing the amount of disk space used
over time for each volume monitored during the build. The hight of the
graph entries represents the delta between current usage and minimal
usage during the build.
That's more useful than showing just the current usage, because then a
graph showing changes in the order of MBs in a volume that is several
GB large would be just flat.
The legend shows the maximum of those deltas, i.e. maximum amount of
space needed for the build. Minor caveat: sampling of disk space usage
starts a bit later than the initial task, so the displayed value may
be slightly lower than the actual amount of space needed because
sampling does not record the actual initial state.
(From OE-Core rev: 263d189d066b578debf08b2bd07494a69b70f70d)
Signed-off-by: Patrick Ohly <patrick.ohly@intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
When matching fails, m.group(0) is invalid and can't be used in the
error message.
(From OE-Core rev: ddfea21e06a2e6b1a1e766969f8c134a6de7388a)
Signed-off-by: Patrick Ohly <patrick.ohly@intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This enables rendering of the original bootchart charts for CPU, disk
and memory usage. It depends on the /proc samples recorded by the
updated buildstats.bbclass. Currently, empty charts CPU and disk usage
charts are drawn if that data is not present; the memory chart already
gets skipped when there's no data, which will also have to be added
for the other two.
(From OE-Core rev: 233d3e50b361feea07803a9c0f2a691e687c6cd5)
Signed-off-by: Patrick Ohly <patrick.ohly@intel.com>
Signed-off-by: Ross Burton <ross.burton@intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Removed print_function and with_statement imports from __future__
as they're supported by python 3 by default.
(From OE-Core rev: 77ae2daad5d775d710b953cf0c623ce74cb2c274)
Signed-off-by: Ed Bartosh <ed.bartosh@linux.intel.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
When --full-time (or -T) is used, the graph allways shows the full
time regardless of which processes are currently shown. This is
especially useful in combinationm with the -s flag when outputting to
multiple files.
(From OE-Core rev: c6e88199ddf2c4ae243d42afc403d28ab56f00f0)
Signed-off-by: Peter Kjellerstedt <peter.kjellerstedt@axis.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
While "Show more" is enabled, all processes are shown, regardless of
--mintime.
This also has the added benefit of making the first shown bar start at
its correct offset from the start time, rather than always starting at
0.
(From OE-Core rev: 5f1b8730f90099c0f73a6b08599990ee71e831b5)
Signed-off-by: Peter Kjellerstedt <peter.kjellerstedt@axis.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This update the pybootchartgui code to the latest release from its new
location at "https://github.com/mmeeks/bootchart". This only imports
the relevant parts, and not all of bootchart2.
(From OE-Core rev: 6f1568e54a7808b2ab568618fc5bb244249579f1)
Signed-off-by: Peter Kjellerstedt <peter.kjellerstedt@axis.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Rather than hardcode the value of "8", allow the minimum task length to be
configured from the commandline using the -m option. "-m 0" means all
tasks will be graphed.
(From OE-Core rev: 30001153d3ce7dadf8f1ec79e634a638a9994518)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
If two entries have the same start time, the data store used will cause
all but one of the entries to be lost. This patch enhances the data
storage structure to avoid this problem and allow more than one
event to start at the same time.
(From OE-Core rev: 220b071fd8d1cc6bdbca58f75489e3c9b34921ca)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Split the output chart into multiple ones to make it more readable, it
only works with "-o path", which means that it doesn't work if the user
doesn't want to save the chart to the disk. For example:
$ ./pybootchartgui.py /path/to/tmp/buildstats/core-image-sato-qemux86/201205301810/ -f svg -s 5 -o /tmp/
bootchart written to /tmp/bootchart_1.svg
bootchart written to /tmp/bootchart_2.svg
bootchart written to /tmp/bootchart_3.svg
bootchart written to /tmp/bootchart_4.svg
bootchart written to /tmp/bootchart_5.svg
[YOCTO #2403]
(From OE-Core rev: 04a34899e1c15a70babd97a3a59ccb9f8af05bad)
Signed-off-by: Robert Yang <liezhi.yang@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
The original patch is from Richard, I rebased it to the up-to-date
upstream code, here are the original messages from him:
We have just merged Beth's initial buildstats logging work. I was
sitting wondering how to actually evaluate the numbers as I wanted to
know "where are we spending the time?".
It occurred to me that I wanted a graph very similar to that generated
by bootchart. I looked around and found pyboootchartgui and then hacked
it around a bit and coerced it to start producing charts like:
http://tim.rpsys.net/bootchart.png
which is the initial "pseudo-native" part of the build. This was simple
enough to test with.
I then tried graphing a poky-image-sato. To get a graph I could actually
read, I stripped out any task taking less than 8 seconds and scaled the
x axis from 25 units per second to one unit per second. The result was:
http://tim.rpsys.net/bootchart2.png
(warning this is a 2.7MB png)
I also added in a little bit of colour coding for the second chart.
Interestingly it looks like there is more yellow than green meaning
configure is a bigger drain on the build time not that its
unexpected :/.
I quite enjoyed playing with this and on a serious note, the gradient of
the task graph makes me a little suspicious of whether the overhead of
launching tasks in bitbake itself is having some effect on build time.
Certainly on the first graph there are some interesting latencies
showing up.
Anyhow, I think this is the first time bitbake's task execution has been
visualised and there are some interesting things we can learn from it.
I'm hoping this is a start of a much more detailed understanding of the
build process with respect to performance.
[YOCTO #2403]
(From OE-Core rev: 6ea0c02d0db08f6b4570769c6811ecdb051646ad)
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Signed-off-by: Robert Yang <liezhi.yang@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
This is from:
http://pybootchartgui.googlecode.com/files/pybootchartgui-r124.tar.gz
Will modify it to make the build profiling in pictures.
Remove the examples since they would not work any more, and they cost
much disk space.
[YOCTO #2403]
(From OE-Core rev: 1f0791109e1aed715f02945834d6d7fdb9a411b4)
Signed-off-by: Robert Yang <liezhi.yang@windriver.com>
Signed-off-by: Richard Purdie <richard.purdie@linuxfoundation.org>
Patrick Ohly
Ed Bartosh
Peter Kjellerstedt
Peter Kjellerstedt
Richard Purdie
Robert Yang
