Age | Commit message (Collapse) | Author |
|
The chromeos6-row3-rack13-host15.cros DUT has been removed and
is pending replacement. This commit removes it from the
default_remotes. This commit may be reverted when the DUT
is replaced.
BUG=b:210575915
TEST=None
Change-Id: Ifc71364b40c92f6a3fd136d2191a79308aa98099
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/third_party/toolchain-utils/+/3359639
Reviewed-by: Denis Nikitin <denik@chromium.org>
Commit-Queue: Jordan Abrahams <ajordanr@google.com>
Tested-by: Jordan Abrahams <ajordanr@google.com>
|
|
Missed a single instance of this, as it's not a violation in
go/pi-terms, but is a violation in go/coil.
BUG=b:204057159
TEST=Presubmit tests
Change-Id: I4bc4997cb42c82c66e4d56281bb16e4452072e01
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/third_party/toolchain-utils/+/3260693
Tested-by: Jordan R Abrahams <ajordanr@google.com>
Auto-Submit: Jordan R Abrahams <ajordanr@google.com>
Commit-Queue: George Burgess <gbiv@chromium.org>
Reviewed-by: George Burgess <gbiv@chromium.org>
|
|
DUTs have been moved to different rows and racks. This updates the
default_remotes accordingly.
These devices are chosen by checking the swarming dimension setting
"label-pool=toolchain".
Additionally, this removes Lulu machines as they are no longer in
the toolchain pool.
BUG=b:178744715
TEST=Presubmit hooks
Change-Id: Ifefb6cb6d2dae58d3c27b55f3a39bccf2fd27eb4
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/third_party/toolchain-utils/+/3252241
Commit-Queue: Jordan R Abrahams <ajordanr@google.com>
Tested-by: Jordan R Abrahams <ajordanr@google.com>
Reviewed-by: Manoj Gupta <manojgupta@chromium.org>
|
|
Wherever possible, this commit removes non-inclusive languages
as per COIL policy. Some of this language exists in dead code,
so this commit also removes confirmed dead code.
Also some minor lint fixes to get the code past the lint checks.
BUG=b:204057159
TEST=Presubmit tests
Change-Id: I299bfbe5dfcbe761efc5dcb29caea2337abe3f30
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/third_party/toolchain-utils/+/3243940
Tested-by: Jordan R Abrahams <ajordanr@google.com>
Reviewed-by: Manoj Gupta <manojgupta@chromium.org>
Reviewed-by: George Burgess <gbiv@chromium.org>
|
|
This crosperf wrapper allows the user to lock devices by board
type. It will query the swarm using the crosfleet API to find
and lock an appropriate board.
The remaining arguments are passed to crosperf, so the command
will act like crosperf otherwise, once the remote has been set.
One limitation is that this wrapper does not conduct any SSH
port forwarding (which would be required for flashing
devices from a cloudtop, for example).
BUG=b:187535367
TEST=crosperf_autolock.py --board chell chell_exp.exp
Change-Id: If8dc4630c42be63d6427716654e7dd7fa4132df5
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/third_party/toolchain-utils/+/3221950
Reviewed-by: Manoj Gupta <manojgupta@chromium.org>
Reviewed-by: George Burgess <gbiv@chromium.org>
Commit-Queue: Jordan R Abrahams <ajordanr@google.com>
Tested-by: Jordan R Abrahams <ajordanr@google.com>
|
|
Currently, these crosperf files do not abide by the pep8 standard
that CrOS requires. This applies the necessary formatting to
get past the commit hook.
BUG=None
TEST=CQ (formatting)
Change-Id: I501fd37f8a1d20fd0b987587eb52eb582f18c1a1
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/third_party/toolchain-utils/+/3215085
Tested-by: Jordan R Abrahams <ajordanr@google.com>
Reviewed-by: Manoj Gupta <manojgupta@chromium.org>
Commit-Queue: Jordan R Abrahams <ajordanr@google.com>
|
|
This global experiment setting prevents crosperf from
attempting to lock the associated devices. This is important
if the devices were locked previously before by crosfleet.
BUG=b:187535367
TEST=CQ
TEST=Lease board from crosfleet. Forward SSH port. Run crosperf on
forwarded port.
Change-Id: I0f5341e07e325a613c6316778c0f096f6d8088a2
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/third_party/toolchain-utils/+/3210525
Reviewed-by: Manoj Gupta <manojgupta@chromium.org>
Tested-by: Jordan R Abrahams <ajordanr@google.com>
Commit-Queue: Jordan R Abrahams <ajordanr@google.com>
|
|
This reuses existing perf reports to identify idle samples instead of
regenerating them.
BUG=b:194737325
TEST=Ran crosperf with cwp benchmarks.
Change-Id: Ib024addbcfb278815fa55dcab20c6d7db90b287a
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/third_party/toolchain-utils/+/3062781
Reviewed-by: George Burgess <gbiv@chromium.org>
Reviewed-by: Manoj Gupta <manojgupta@chromium.org>
Tested-by: Jian Cai <jiancai@google.com>
|
|
Right now Crosperf reports include samples from idle functions.
Filtering these samples from the reports allow users to have more
accurate evaluation of the performance impact of code changes as they
focuson the meaningful samples.
BUG=b:194737325
TEST=unit tests passed.
Change-Id: I22bc3c78999eec9a141440e92ce52457a4bfa0a3
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/third_party/toolchain-utils/+/3054437
Reviewed-by: Manoj Gupta <manojgupta@chromium.org>
Tested-by: Jian Cai <jiancai@google.com>
|
|
This fix the missing symbols in perf report due to the misplaced debug
information.
BUG=b:194540821
TEST=Ran crosperf and verified all the symbols are present in the perf
report.
Change-Id: Ifed66ac6073d175d99a71e64f5e89925fabdaa92
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/third_party/toolchain-utils/+/3050514
Reviewed-by: Manoj Gupta <manojgupta@chromium.org>
Tested-by: Jian Cai <jiancai@google.com>
|
|
Replace chromeos2-row10-rack9-host1 bob machine.
The new name is chromeos6-row3-rack13-host15.
Replace chromeos2-row10-rack9-host21 elm with
chromeos6-row14-rack15-host21
BUG=b:186870515, b:185164408
TEST=cros shell chromeos6-row3-rack13-host15.cros
TEST=cros shell chromeos6-row14-rack15-host21.cros
Change-Id: I6030b6ec2e5dff723a72561daa1b3c3b13bd4ba2
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/third_party/toolchain-utils/+/2876366
Reviewed-by: Caroline Tice <cmtice@chromium.org>
Reviewed-by: Manoj Gupta <manojgupta@chromium.org>
Commit-Queue: Denis Nikitin <denik@chromium.org>
Tested-by: Denis Nikitin <denik@chromium.org>
|
|
Synchronize the list of the lab machines allocated by the hardware lab
to the toolchain pool. Also add a description of this file to README.
BUG=chromium:1201755
TEST=None
Change-Id: Ie10d0a881b34dfd0d1b49e6892a4fdf807830c95
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/third_party/toolchain-utils/+/2847103
Reviewed-by: Caroline Tice <cmtice@chromium.org>
Reviewed-by: Manoj Gupta <manojgupta@chromium.org>
Tested-by: Jian Cai <jiancai@google.com>
|
|
BUG=chromium:1187326
TEST=Tested by hand on chrotomation.mtv
Change-Id: I7e0bd2a99e85f288fb3b1b9ff52c8e38d25df245
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/third_party/toolchain-utils/+/2773632
Reviewed-by: Caroline Tice <cmtice@chromium.org>
Commit-Queue: Caroline Tice <cmtice@chromium.org>
Tested-by: Caroline Tice <cmtice@chromium.org>
|
|
Our old kefka machines got moved out of the lab; they have given
us two new ones. This updates the default_remotes file to
contain the names of the new kefka machines.
BUG=b/180983943
TEST=None
Change-Id: I3972528f5454fe6993aa96a4c91d0a6ade4780ee
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/third_party/toolchain-utils/+/2719301
Reviewed-by: Manoj Gupta <manojgupta@chromium.org>
Commit-Queue: Caroline Tice <cmtice@chromium.org>
Tested-by: Caroline Tice <cmtice@chromium.org>
|
|
Increment a file index when copying multiple files.
This fixes the issue when multiple perf.data files are copied from
the system_health.memory_desktop benchmark with profiling.
BUG=None
TEST=crosperf with system_health.memory_desktop and profiling enabled.
Change-Id: I7d00ec23a4fa595d6cd0df8fcc8b64cf7181f5dd
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/third_party/toolchain-utils/+/2622963
Tested-by: Denis Nikitin <denik@chromium.org>
Reviewed-by: Manoj Gupta <manojgupta@chromium.org>
|
|
Change "tigger" to "tiger". Also update a unit test.
BUG=chromium:1156409
TEST=None.
Change-Id: I2dd394ccdfa79632e637cbefddf559e95296c8c1
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/third_party/toolchain-utils/+/2580398
Reviewed-by: Tiancong Wang <tcwang@google.com>
Tested-by: Jian Cai <jiancai@google.com>
|
|
Replace veyron_minnie with a newer model as it reaches end of life.
BUG=b:170402899
TEST=None
Change-Id: I0caf06b84b465dcc88a3fdd9dc7d19f13fcd11b9
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/third_party/toolchain-utils/+/2574734
Reviewed-by: Tiancong Wang <tcwang@google.com>
Tested-by: Jian Cai <jiancai@google.com>
|
|
Ignore "core id", "apicid", "initial apicid" fields from cpuinfo in
machine checksum calculation. The values may differ on the same type of
machines.
Add more descriptive output for the BadChecksum error.
Unittest is updated accordingly.
BUG=chromium:1145386
TEST=./run_tests.sh in crosperf.
Change-Id: Ifcc91fb70f02c41d77787fbb665741bc130152c0
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/third_party/toolchain-utils/+/2523398
Tested-by: Denis Nikitin <denik@chromium.org>
Reviewed-by: Bob Haarman <inglorion@chromium.org>
Commit-Queue: Denis Nikitin <denik@chromium.org>
|
|
Replace multiple levels of if/else of the chrome src search with a loop.
Add a missing search path .cache/distfiles/chrome-src.
Add checks for the Chrome source location and raise RuntimeError if it
does not exist.
BUG=None
TEST=Tested with crosperf and distfiles/target/chrome-src-internal.
Change-Id: I66dd7c802a4e80238e4d625586198bdf573621dc
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/third_party/toolchain-utils/+/2453595
Reviewed-by: Manoj Gupta <manojgupta@chromium.org>
Reviewed-by: Caroline Tice <cmtice@chromium.org>
Tested-by: Denis Nikitin <denik@chromium.org>
Commit-Queue: George Burgess <gbiv@chromium.org>
|
|
Chrome source location has recently changed and path search in crosperf
is updated accordingly.
BUG=None
TEST=run crosperf after repo sync and ebuild prepare of chrome.
Change-Id: I715b3a9bf4892a993181b0be5864de94f651afe2
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/third_party/toolchain-utils/+/2446895
Commit-Queue: Denis Nikitin <denik@chromium.org>
Commit-Queue: Manoj Gupta <manojgupta@chromium.org>
Tested-by: Denis Nikitin <denik@chromium.org>
Auto-Submit: Denis Nikitin <denik@chromium.org>
Reviewed-by: Manoj Gupta <manojgupta@chromium.org>
|
|
Previous changes resulted in some complaints about formatting
and Python 3 compatibility from the repo hooks. This change fixes
those.
BUG=None
TEST=repo upload --cbr . # check that it no longer complains
Change-Id: I99cc51dcb8d499d59b7b47817f4cef8fa6ba5059
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/third_party/toolchain-utils/+/2419831
Tested-by: Bob Haarman <inglorion@chromium.org>
Reviewed-by: Manoj Gupta (OoO) <manojgupta@chromium.org>
|
|
This rephrases some code and comments to avoid unclear and
non-inclusive wording. For the remaining terms in unblocked_terms.txt,
comments are added to explain why these terms are present in the code.
BUG=chromium:1099035
TEST=repo hooks
Change-Id: I985613d811836b6fb6351eadcf3e08a3290db97f
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/third_party/toolchain-utils/+/2419656
Tested-by: Bob Haarman <inglorion@chromium.org>
Reviewed-by: Manoj Gupta (OoO) <manojgupta@chromium.org>
Reviewed-by: George Burgess <gbiv@chromium.org>
|
|
Verification is very helpful when we record per-process perf profile
and collect top statistics.
Using verification we can make sure that perf profiles are collected
from the hottest functions in top statistics.
BUG=chromium:1098098
TEST=unittests and crosperf with per-process profiling are passing.
Change-Id: Ib79b070babcdea38932ba709ac3e14fa3311f842
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/third_party/toolchain-utils/+/2407223
Reviewed-by: Tiancong Wang <tcwang@google.com>
Reviewed-by: Caroline Tice <cmtice@chromium.org>
Commit-Queue: Denis Nikitin <denik@chromium.org>
Tested-by: Denis Nikitin <denik@chromium.org>
|
|
BUG=b:153872557
TEST=None
Change-Id: Ibc9f16d253ee9bd113fc654846462e647dc0594c
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/third_party/toolchain-utils/+/2327953
Tested-by: Tiancong Wang <tcwang@google.com>
Reviewed-by: George Burgess <gbiv@chromium.org>
Commit-Queue: George Burgess <gbiv@chromium.org>
Auto-Submit: Tiancong Wang <tcwang@google.com>
|
|
Copy cpuinfo and top logs into results directory.
BUG=None
TEST=tested on local DUT.
Change-Id: Ic868894d681db4a27f78e378945e1e565d4a5cc7
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/third_party/toolchain-utils/+/2255007
Reviewed-by: Zhizhou Yang <zhizhouy@google.com>
Commit-Queue: Denis Nikitin <denik@chromium.org>
Tested-by: Denis Nikitin <denik@chromium.org>
|
|
Top statistics was showing commands which could combined multiple
processes. To include PID in topstats we need to split the commands into
separate processes.
The process PID is appended to the command name in topstats. The top
chrome process is the renderer process running the benchmark.
The list size depends on the number of non-chrome processes which we
limit to 5. For example with 10 chrome processes in the top with 10 following
non-chrome processes the list will show 15 entries.
BUG=None
TEST=Tested on eve, bob and cheza.
Change-Id: Ibf1e61c8cb522aba13cd51a590bb7e24597f66a6
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/third_party/toolchain-utils/+/2227626
Reviewed-by: George Burgess <gbiv@chromium.org>
Tested-by: Denis Nikitin <denik@chromium.org>
|
|
Hasty CLs are hasty. What can I say? ¯\_(ツ)_/¯ Zhizhou says we want to
'succeed' in this case with a log message, though.
BUG=chromium:1079048
TEST=Unittests
Change-Id: I70ea71dae42d7bee069d7311e14059061c5fa0a9
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/third_party/toolchain-utils/+/2220215
Reviewed-by: George Burgess <gbiv@chromium.org>
Tested-by: George Burgess <gbiv@chromium.org>
|
|
This consumes potentially tons of disk. Also we'll raise in this
situation from now on, since it never seems to be what we want.
BUG=chromium:1079048
TEST=Unittests
Change-Id: Ie1933b5ff0b939777d7e1d6d35dd5a29039ca07a
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/third_party/toolchain-utils/+/2220291
Reviewed-by: Zhizhou Yang <zhizhouy@google.com>
Tested-by: George Burgess <gbiv@chromium.org>
|
|
Tast results directory has different name from test_that results,
causing a bug when compress_results option is on. This CL fixes it.
BUG=chromium:1079048
TEST=Tested with tast and test_that benchmarks in crosperf
Change-Id: I06e6615c748aeeb2fc4576f4e1bcfdad93526962
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/third_party/toolchain-utils/+/2218541
Reviewed-by: George Burgess <gbiv@chromium.org>
Commit-Queue: Zhizhou Yang <zhizhouy@google.com>
Tested-by: Zhizhou Yang <zhizhouy@google.com>
|
|
Perf data and report file can occupy large disk space when we copy and
store them in the crosperf results directory. This patch introduces a
new field for crosperf so that we create a tarball for it to save space.
By default compress_result is set to True.
BUG=chromium:1079048
TEST=Passed unittest, tested with simple examples.
Change-Id: I70db6f3f70dc33bacfccd8833a4327c494b4d6b5
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/third_party/toolchain-utils/+/2212432
Reviewed-by: George Burgess <gbiv@chromium.org>
Commit-Queue: Zhizhou Yang <zhizhouy@google.com>
Tested-by: Zhizhou Yang <zhizhouy@google.com>
|
|
Remove experiment arguments reducing performance noise from the
buildbot_test_toolchains script and make the values default in
experiment settings.
Add descriptions about default values.
BUG=None
TEST=Tested on local DUT.
Change-Id: I5fa3d5bf36a2ca37da713b42b49c1f2ad4f08c98
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/third_party/toolchain-utils/+/2197549
Tested-by: Denis Nikitin <denik@chromium.org>
Reviewed-by: Zhizhou Yang <zhizhouy@google.com>
|
|
BUG=chromium:1077841
TEST=tested manually with Crosperf & veyron_minnie & it worked.
Change-Id: Ied842f9109ea81d751c2fc5d676d547ef5122271
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/third_party/toolchain-utils/+/2180299
Reviewed-by: Bob Haarman <inglorion@chromium.org>
Commit-Queue: Caroline Tice <cmtice@chromium.org>
Tested-by: Caroline Tice <cmtice@chromium.org>
|
|
This is a format string; without this, --help raises
ValueError: unsupported format character ',' (0x2c) at index 311
BUG=None
TEST=crosperf --help
Change-Id: I5fbdb996888156f8a1531c2b140c3f734be7edff
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/third_party/toolchain-utils/+/2173547
Reviewed-by: Manoj Gupta <manojgupta@chromium.org>
Tested-by: George Burgess <gbiv@chromium.org>
|
|
Adding the tast test platform.ReportDiskUsage when using
'all_toolchain_perf' benchmark set. This will make toolchain nightly
test launch this tast test together with other telemetry benchmarks and
show image size changes.
BUG=chromium:1071937
TEST=Run experiment with all_toolchain_perf
Change-Id: If05893e4ad129f29276a14893024641d03ec1b95
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/third_party/toolchain-utils/+/2161773
Reviewed-by: George Burgess <gbiv@chromium.org>
Reviewed-by: Manoj Gupta <manojgupta@chromium.org>
Commit-Queue: Zhizhou Yang <zhizhouy@google.com>
Tested-by: Zhizhou Yang <zhizhouy@google.com>
|
|
Crosperf used to only support autotest server/client tests, and was
limiting us from running and collecting tast benchmarks.
This patch introduces a simple support for crosperf to run tast. Note
that currently we do not support passing arguments or running tast with
local changes. It basically launches `tast run <ip> <tast_test>` and
collects chart json result and parses it.
BUG=chromium:1071937
TEST=tested with tast.platform.ReportDiskUsage
Change-Id: I1980a48c4ca8e02548e5313d32b8c5ddcf5e3e60
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/third_party/toolchain-utils/+/2160128
Commit-Queue: Zhizhou Yang <zhizhouy@google.com>
Tested-by: Zhizhou Yang <zhizhouy@google.com>
Auto-Submit: Zhizhou Yang <zhizhouy@google.com>
Reviewed-by: George Burgess <gbiv@chromium.org>
|
|
Crosperf always returns 0 no matter benchmarks fail or experiment
interrupted in the middle. Thus we cannot tell if a run succeeded or not
with the exit status of it and it makes our nightly test failures hard
to find.
In this patch, I changed the behavior of crosperf return value:
1) Crosperf will not generate any report or send email if terminated or
all benchmarks fail. It raises RuntimeError stating all benchmarks
failing in the end.
2) Crosperf will generate report if benchmarks (but not all) fail, and
will raise RuntimeError stating benchmarks partially failing.
3) Crosperf will also copy results json files to local results directory
for further info.
BUG=chromium:1063703
TEST=Passed all unittests, tested with different failure situations.
Change-Id: I998bad51cd7301b9451645d22e8734963bc01aed
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/third_party/toolchain-utils/+/2119231
Reviewed-by: Caroline Tice <cmtice@chromium.org>
Commit-Queue: Zhizhou Yang <zhizhouy@google.com>
Tested-by: Zhizhou Yang <zhizhouy@google.com>
Auto-Submit: Zhizhou Yang <zhizhouy@google.com>
|
|
Currently due to crbug.com/1057755, graphics_WebGLAquarium will fail and
take 1 hour to timeout. We do not want to waste 2 * 5 iterations which
means 10 hours in our nightly test everyday waiting for it to fail.
Will re-enable it when the bug is fixed.
BUG=chromium:1057755
TEST=None
Change-Id: I8da4d60b19e95d00be202f206d0d581bc426adf4
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/third_party/toolchain-utils/+/2088536
Reviewed-by: Manoj Gupta <manojgupta@chromium.org>
Commit-Queue: Zhizhou Yang <zhizhouy@google.com>
Tested-by: Zhizhou Yang <zhizhouy@google.com>
Auto-Submit: Zhizhou Yang <zhizhouy@google.com>
|
|
We have enabled --py3 for cros lint and thus reveals some new linting
errors in migrated scripts. This patch fixes them and some other random
errors.
BUG=chromium:1011676
TEST=Passed unittests.
Change-Id: If129e01c21845e1b944a2b64e50e9fed7138c845
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/third_party/toolchain-utils/+/2055972
Tested-by: Zhizhou Yang <zhizhouy@google.com>
Reviewed-by: George Burgess <gbiv@chromium.org>
Commit-Queue: Zhizhou Yang <zhizhouy@google.com>
Auto-Submit: Zhizhou Yang <zhizhouy@google.com>
|
|
With client tests wrapped up by crosperf_Wrapper server tests, we need a
categorizing for all tests lunched by suite runner.
BUG=chromium:984790
TEST=Passed unittest and simple experiment.
Change-Id: Ib7dc848cf5c18c3d79db059163ed1217fd7d318f
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/third_party/toolchain-utils/+/1977040
Reviewed-by: Caroline Tice <cmtice@chromium.org>
Commit-Queue: Zhizhou Yang <zhizhouy@google.com>
Tested-by: Zhizhou Yang <zhizhouy@google.com>
Auto-Submit: Zhizhou Yang <zhizhouy@google.com>
|
|
This patch migrates crosperf and its utils to python 3.
TEST=Passed presubmit check; tested with simple experiment locally.
BUG=chromium:1011676
Change-Id: Ib2a9f9c7cf6a1bb1d0b42a1dd3d9e3cbb4d70a36
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/third_party/toolchain-utils/+/2003796
Tested-by: Zhizhou Yang <zhizhouy@google.com>
Reviewed-by: Mike Frysinger <vapier@chromium.org>
Reviewed-by: Caroline Tice <cmtice@chromium.org>
Commit-Queue: Zhizhou Yang <zhizhouy@google.com>
Auto-Submit: Zhizhou Yang <zhizhouy@google.com>
|
|
This patch removes afe locking mode for crosperf, since:
1) All lab machines has been migrated to skylab.
2) Interaction with afe depends on python 2 code in autotest.
BUG=chromium:1044764
TEST=Passed unittest and simple experiment run.
Change-Id: Ie0a250548d739f32bbc08619f637380104df6c28
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/third_party/toolchain-utils/+/2016008
Commit-Queue: Zhizhou Yang <zhizhouy@google.com>
Tested-by: Zhizhou Yang <zhizhouy@google.com>
Auto-Submit: Zhizhou Yang <zhizhouy@google.com>
Reviewed-by: George Burgess <gbiv@chromium.org>
|
|
This is the first step of python 3 migration for crosperf
TEST=Passed presubmit for all .py files in crosperf
BUG=chromium:1011676
Change-Id: I615d0323c81ba688c8ce2aa0cac307035d0b882b
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/third_party/toolchain-utils/+/2001696
Commit-Queue: Zhizhou Yang <zhizhouy@google.com>
Tested-by: Zhizhou Yang <zhizhouy@google.com>
Auto-Submit: Zhizhou Yang <zhizhouy@google.com>
Reviewed-by: George Burgess <gbiv@chromium.org>
|
|
Current command generated from suite_runner for telemetry_Crosperf run
is not correct, and this patch will fix it. Also added an unittest to
verify it in the future.
It is unfortunate that we currently still cannot run test with args
successfully in skylab, but according to the args passed to skylab, this
change should work.
TEST=Checked command line to launch skylab create-test.
BUG=None
Change-Id: I05f0f4c60e04d19e2a2637192023829d4dd9f48b
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/third_party/toolchain-utils/+/1963355
Reviewed-by: George Burgess <gbiv@chromium.org>
Commit-Queue: Zhizhou Yang <zhizhouy@google.com>
Tested-by: Zhizhou Yang <zhizhouy@google.com>
Auto-Submit: Zhizhou Yang <zhizhouy@google.com>
|
|
We should not image DUT with certain image when the test is going to be
launched via 'skylab create-test'. This is because the DUT may be
occupied by other tasks before Crosperf acutally get it assigned, and
the first step after the test starts is to flash image.
This patch is a temproray solution to ensure no bad behavior happens on
the DUT before testing.
TEST=Tested wth skylab run, w/o caching.
BUG=chromium:984790
Change-Id: Ie2f439d5777813905bdca5f4a2d1e401df2dc83e
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/third_party/toolchain-utils/+/1952172
Tested-by: Zhizhou Yang <zhizhouy@google.com>
Reviewed-by: Caroline Tice <cmtice@chromium.org>
Commit-Queue: Zhizhou Yang <zhizhouy@google.com>
Auto-Submit: Zhizhou Yang <zhizhouy@google.com>
|
|
This is a patch working together with crrev.com/c/1949606 to migrate
device setup functions from suite_runner to autotest telmetry_Crosperf.
In this patch, we modified the logic of different suite types along with
skylab mode for suite_runner.
In result_cache, we handle the wait time log file to accumulate cool
down time per machine.
We also move intel_pstate kernel updating code into SetupDevice() since
it doesn't increase overhead by checking every iteration.
TEST=Passed all unittests; tested with different benchmark modes (Note
that currently skylab server tests cannot be tested because of
regression: crbug.com/984103).
BUG=chromium:1020655
Cq-Depend: chromium:1949606
Change-Id: If8e5099d096d2568d4f54584a8fcfd3c0f99c8f8
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/third_party/toolchain-utils/+/1949626
Reviewed-by: George Burgess <gbiv@chromium.org>
Reviewed-by: Denis Nikitin <denik@chromium.org>
Tested-by: Zhizhou Yang <zhizhouy@google.com>
Auto-Submit: Zhizhou Yang <zhizhouy@google.com>
Commit-Queue: Denis Nikitin <denik@chromium.org>
|
|
Rendering.desktop benchmark which was added into crosperf does not
return results in avg_surface_fps field. Until recently it worked fine.
There is an alternative field in histograms.json "exp_avg_surface_fps"
holding the data.
Add exp_avg_surface_fps in default-telemetry-results.json to filter-in
the benchmark field in crosperf results.
BUG=chromium:1030475
TEST=crosperf with rendering.desktop returns data on chell and lulu
Change-Id: Ia1d25f7f1894ca559f1e14e032fcf274dc4df49d
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/third_party/toolchain-utils/+/1951336
Reviewed-by: George Burgess <gbiv@chromium.org>
Tested-by: Denis Nikitin <denik@chromium.org>
|
|
This patch moves enable_aslr and turbostat into dut_config dictionary,
since they are all device setup steps, and it make code cleaner and
easier to migrate to telemetry_Crosperf.
TEST=Passed unittests.
BUG=chromium:1020655
Change-Id: I29a649e76591e206efc19bb2b8a8df9aead16575
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/third_party/toolchain-utils/+/1947903
Reviewed-by: Denis Nikitin <denik@chromium.org>
Commit-Queue: Zhizhou Yang <zhizhouy@google.com>
Tested-by: Zhizhou Yang <zhizhouy@google.com>
Auto-Submit: Zhizhou Yang <zhizhouy@google.com>
|
|
There was a bug when crosperf failed due to StopUI failure.
This happened when in a preceding run crosperf exits with an exception
in DeviceSetup (for example Keyboard interrupt) and fails to call
StartUI.
Current change fixes this problem and includes a unittest case testing
the exception case.
BUG=None
TEST=unittest and HW test on DUT with stopped ui pass
Change-Id: Id6c69aebefe21b12ec7ee3a7c7f9dff92d143908
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/third_party/toolchain-utils/+/1941036
Commit-Queue: Denis Nikitin <denik@chromium.org>
Tested-by: Denis Nikitin <denik@chromium.org>
Reviewed-by: Zhizhou Yang <zhizhouy@google.com>
|
|
Crosperf provides a mode to use `skylab create-test` to launch tests
when user specify 'skylab: true' in experiment file.
Since Skylab has been updated in these months, the command line itself
changes. This patch fixes it:
1) bb mode becomes deprecated, do not use it.
2) Task id becomes a recipe id, we need to parse it instead of swarming.
3) Real result location can be found in the json output of `wait-task`
from key "child-result".
TEST=Passed unittest, tested with simple experiment file.
BUG=chromium:984790
Change-Id: Iee114b1e6c623bb64a0ca784df9c2ba6d95e3d07
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/third_party/toolchain-utils/+/1916542
Reviewed-by: Zhizhou Yang <zhizhouy@google.com>
Reviewed-by: Caroline Tice <cmtice@chromium.org>
Tested-by: Zhizhou Yang <zhizhouy@google.com>
Auto-Submit: Zhizhou Yang <zhizhouy@google.com>
|
|
After locking mechanism changed, crosperf checks if locks_dir is set or
not and by default uses X20 location. This requires prodaccess thus
unittests will fail without prodaccess if locks_dir is not explicitly
set. This patch set all experiment files in unittests to use /tmp for
locks_dir.
Also fixed some lint error for deprecated functions.
TEST=Passed all unittest.
BUG=None
Change-Id: Iff4623aad9632842010e5efb85a1e75a38bcf73e
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/third_party/toolchain-utils/+/1912767
Reviewed-by: Manoj Gupta <manojgupta@chromium.org>
Commit-Queue: Zhizhou Yang <zhizhouy@google.com>
Tested-by: Zhizhou Yang <zhizhouy@google.com>
|