table-generator
computes scores according to SV-COMP'24 scoring scheme.
This changes only the scoring for witness-validation results.- Support for property files at HTTP(S) URLs in
table-generator
.
Tables can already be produced not only from local result files, but also from files that are downloaded on the fly. Now this also works for results with property files. - Fix for tool-info module
witnesslint
.
- Two tool-info modules improved.
- Tool-info modules can now provide URLs that will be used for links in HTML tables.
There are two new methods,project_url()
andurl_for_version()
that can be implemented, andtable-generator
will put links to these URLs in the summary table of the HTML tables (for the tool name and the version). Most existing tool-info modules were extended withproject_url()
. - Many new and improved tool-info modules.
The big change in this release is the long-awaited support for cgroups v2! Please refer to the installation instructions for how to use it (on Ubuntu/Debian, installing our package is enough). Note that this of course has not been tested yet on as many different systems as our support for cgroups v1, so there might still be some rough edges. Please provide feedback if you encounter any problems or have questions. On systems with cgroups v1, everything should work the same way as before.
There are also some other minor improvements:
- If the system administrator has configured a lower frequency limit than what the CPU supports, BenchExec now reports this limit as the CPU speed.
- Fixes in
table-generator
for result files that conform to the specification but were not produced bybenchexec
. - A few minor improvements for better integration with Podman containers.
- Even more robust handling of child cgroups created within a run
Despite the improvements from BenchExec 3.13 there could be crashes because sometimes cgroups vanish while we iterate over them. Avoiding this did not fully work, maybe due to some delays in the kernel, so now we handle this. - Actually fix handling of non-printable characters in environment variables.
The fix that was part of BenchExec 3.12 only worked if a non-printable character was the first character of the value of the environment variable.
This release works only on Python 3.7 and newer!
- Improve performance of the "Merging results" step of
table-generator
.
In the common case of all run sets having the same set of tasks, this step is now much faster. - Avoid deadlock in
table-generator
for cases with many child processes and heavy system load. - On systems with many cores and if there are many input files or columns,
table-generator
will now use more than 32 child processes. - Support the scoring schema used in SV-COMP'23 for witness validators,
and show the witness category as part of the task identifier in tables.
Note that just like for the category
correct-unconfirmed
external postprocessing of the results is necessary for this. - If the pattern inside a
<propertyfile>
tag in the benchmark definition does not match a file, BenchExec now terminates with an error instead of just logging a warning. - Improved logging in
table-generator
.
The log messages about missing run results and missing properties, which could occur many times in certain use cases, are now omitted. Instead, some numbers about the size of the resulting tables and the number of missing results will be logged.
- Updated installation instructions for Debian.
Manual installation of the package is needed becausedpkg
on Debian is incompatible with our PPA on Launchpad. - Some improvements to tool-info modules.
This release does not change the minimum supported Python version, but we would like to remind you that BenchExec will soon stop supporting Python 3.6.
- Added a workaround for the known glibc deadlock described in #656.
In most cases BenchExec should detect the deadlock and continue benchmarking after a timeout of 60s. - We noticed a performance problem in Python
that affects
benchexec
in container mode on machines with many cores and added an optimization that reduces its impact. - Improved handling of non-existent mount points.
This makes BenchExec easier to use within Podman containers, and we now recommend Podman over Docker in our documentation. table-generator
no longer attempts to spawn a large number of threads (failing due to the limit on open files) on machines with many cores.- Many new and improved tool-info modules.
This release does not change the minimum supported Python version, but we would like to remind you that BenchExec will soon stop supporting Python 3.6.
- More robust handling of child cgroups created within a run
Note that it is not safe to execute an untrusted tool inside a BenchExec run and give it access to cgroups (e.g., with--no-container
or--full-access /sys/fs/cgroups
). However, some users want to execute trusted tools with cgroup access (e.g., nestingrunexec
) and we aim to support this. It was found that BenchExec could hang or crash in certain cases where child cgroups were created by the process within a run and used to freeze processes (done by nestedrunexec
, for example), or were made inaccessible using file-system permissions. This lead to incomplete run cleanup and processes left running or hanging in an unkillable state. This release fixes such situations and users who allow cgroup access within runs are strongly recommended to upgrade. - Fix incomplete rewriting of numbers into roman numerals in the new LaTeX output for statistics.
This release does not change the minimum supported Python version, but we would like to remind you that BenchExec will soon stop supporting Python 3.6.
- Compatibility with Python 3.10
Note that it is expected that this is one of the last releases that supports Python 3.6. - Export of statistics as LaTex commands in table-generator:
All those values that exist in the statistics table on the Summary tab of our HTML tables can be exported as a series of LaTeX commands with--format=statistics-tex
. These can then be included in a paper and the commands can be used to retrieve the statistics values (documentation). The scriptcontrib/statistics-tex.py
that provided a subset of this functionality so far is removed. Thanks to @Sowasvonbot for implementing this! - Slight improvements for mounting overlayfs:
- Avoid redundant mount points that could prevent certain nested mounts.
- Avoid a warning in the kernel log.
- Fix handling of non-printable characters in environment variables.
- Slight improvements for the HTML tables:
- Use of color for categories in filters
- More informative tooltip in quantile plots (thanks @leventeBajczi)
- Added and improved tool-info modules
This release brings one major feature for the HTML tables: The statistics on the summary tab are now updated on-the-fly if filters are applied and always take those filters into account (i.e., only show statistics about the selected rows). We update the row name in the statistics table to indicate this, but still be aware of it if you are accustomed to the previous behavior! This was also the last place in the HTML tables where filters were not respected, so now filters are taking into account consistently wherever data are shown or used. Thanks to @DennisSimon for implementing this!
There are also a few other improvements and fixes:
- Fix loading logs from local HTML tables in Chrome/Chromium.
- Update hint on which Firefox setting to change for access to logs from local HTML tables.
- When loading logs from local HTML tables fails,
show a hint on how to start a small HTTP server with Python
and let it serve the table and logs.
Using this server to access the table from the browser
instead of
file://
URLs will make the log access work regardless of browser versions and configuration options. The shown hint even includes a command line that just needs to be copy-and-pasted. - Fix sliders jumping around in the filter sidebar when defining filters for numeric columns.
- Some tool-info modules were improved.
- Fix bug in HTML tables where content of a cell could be visible through another cell when scrolling horizontally.
- Slightly improved error handling and error messages regarding cgroup permissions, cgroupsv2, and invalid directory modes.
- Improved handling of debug log (
--debug
). Previously, every transferred output file of the tool was logged, which would lead to large debug logs in case a tool produced many files. Now we log only the names of at most 1000 such output files. - Many new and improved tool-info modules.
- The default branch of the git repository has been renamed to
main
. Please adjust forks and checkouts if required.
- Improved container mode to make it work more easily inside LXC containers.
- The scatter plot in HTML tables produced by table-generator can now show a linear-regression graph using ordinary least squares. A tooltip shows more information such as the regression coefficient.
- The library that is used for rendering the actual tables in the HTML tables got a major upgrade, which required some work. If you notice any regression in table behavior, please file an issue.
- One new tool-info module.
For people using the git repository of BenchExec:
Immediately after the release of BenchExec 3.9,
the default branch of the repository will be renamed to main
.
Please adjust forks and checkouts if required.
This release works only on Python 3.6 and newer!
- Add possibility to have a
close()
method to tool-info modules. - The
test_tool_info
utility now has a--debug
argument that will show the debug log, e.g., from the tool-info module. - Fix bug in detection of CPU throttling during benchmarking, for which we warn if we detect it. This did not work for cases with core numbers longer than one digit.
- Properly escape the suggested command line for running table-generator that is shown by benchexec before it terminates.
- Allow specifying path to libseccomp via environment variable
LIBSECCOMP
. This is useful for environments like NixOS. - Fix handling of empty result files in table-generator, the generated HTML tables will work again.
- Fix handling of empty run results when filtering rows in HTML tables.
- Make filters for status and textual columns in HTML tables work if filter string contains a colon.
- When entering a filter in the HTML tables' filter row, do not loose focus on the input field when applying the filter, such that users can continue typing.
- Fix invalid values shown in score-based quantile plot if some runs of a table do not have a score value. Now a run set is only shown in a score-based quantile plot if all values have scores (otherwise the plot would be misleading).
- Text selection now works as expected while an overlay window is open in the HTML tables (only text in the overlay will be selected).
This is expected to be the last BenchExec release that supports Python 3.5, newer releases will require Python 3.6 or newer. Please cf. issue #717 for our plan on dropping support for further Python versions.
We would like to note that Linux kernel 5.11
brings a major improvement for BenchExec users not on Ubuntu:
Now it should be possible to use the overlayfs feature as a regular user,
no need to pass --read-only-dir /
and similar parameters.
We updated our installation instructions
accordingly and also clarified that BenchExec requires x86 or ARM machines
and recommend Linux kernel 4.14 or newer due to reduced cgroups overhead.
Changes in this release:
- In HTML tables, the following settings are now stored in the hash part of the URL:
- Column sorting
- Page size of the table, i.e., how many rows are shown
- Filters for task names that are defined by entering text into the left-most input field of the filter row of the table. Previously this would only work for task-name filters defined in the filter sidebar. This means that using the back/forward navigation of the browser will change these settings and that they can be present in shared links.
- Fix a few cases of printing of statistics information in HTML tables.
This affects corner cases like the number of visible decimal digits for
0
and trailing zeroes for the standard deviation in the tooltip. - When a user requests rounding to a certain number of decimal digits, the filtering functionality of the HTML tables will now use the raw values, not the rounded values. This is consistent with the behavior when rounding is not explicitly requested and BenchExec applies the default rounding rules.
- Fix harmless stack trace printed at end of
benchexec
execution in cases like of early termination, e.g., if the tool could not be found. - Some improvements to tool-info modules.
- Several updates of JS libraries, but this should not bring user-visible changes.
- One tool-info module improved.
- One tool-info module improved.
-
BenchExec is now available in a PPA for easy installation on Ubuntu. Just run the following commands
sudo add-apt-repository ppa:sosy-lab/benchmarking sudo apt install benchexec
-
Column filters are now reflected in the URL of HTML tables. This makes it possible to open a table, configure some filters, and share a link with others that will apply these filters on load. Furthermore, using the back and forward buttons of the browser will now also update the applied filters.
-
Add parameter
--initial-table-state
totable-generator
, which allows to define the default state of HTML tables (e.g., filters, opened tab, etc.). -
Category-specific statistics are shown more often again on first table tab. Since BenchExec 3.0 these were removed in some cases where we cannot compute them, but this accidentally removed them from more than the desired cases.
-
Improved rounding in table-generator.
-
SV-COMP scoring schema updated according to rules of SV-COMP'21.
-
Many tool-info modules updated to use the new API from BenchExec 3.3 and improvements for SV-COMP'21 and Test-Comp'21.
-
Improved warnings in certain cases where a benchmark definition does not make sense (e.g.,
<exclude>
tags that do not match anything). -
HTML tables now show a proper error message if the browser is not supported and also a loading message.
-
Several smaller bug fixes like avoiding crashes in corner cases.
-
New API for tool-info modules (needed by
benchexec
for getting information about the benchmarked tool). The new API is defined by classbenchexec.tools.template.BaseTool2
and is similar to the old API, but more convenient to use and provides more useful information to the tool-info module. The old API is still supported and will be removed no sooner than in BenchExec 4.0. We also provide a migration guide. -
A new parameter
--tool-directory
forbenchexec
allows to specify the installation directory of the benchmarked tool easily without having to modifyPATH
or change into the tool's directory. Note that this only works if the respective tool-info module makes use of the newBaseTool2
API. -
New version 2.0 of the task-definition format for
benchexec
. This format allows to specify arbitrary additional information in a key namedoptions
andbenchexec
will pass everything in this key to the tool-info module, but note that this only works if the respective tool-info module makes use of the newBaseTool2
API. This is useful to add domain-specific information about tasks, for example in the SV-Benchmarks repository it is used to declare the program language. BenchExec also still supports version 1.0 of the format. -
table-generator
is now defined to work on Windows and we test this in continuous integration. Previously, it probably was working on Windows most of the time but we did not systematically test this. -
Fix a crash in
benchexec
for task with property but without task-definition file.
- The HTML tables produced by
table-generator
now provide a score-based quantile plot in addition to the regular quantile plot if scores are used. If available, it is shown by default on the tab for quantile plots. Score-based quantile plots are for example used by SV-COMP to visualize results. - Better axis labels in scatter plot of HTML tables.
- More auxiliary lines available in scatter plot of HTML tables.
- New tool-info module added.
Bug fixes:
- Fix crash in
benchexec
if a non-SV-COMP property was used. - Fix for empty property files being treated as SV-COMP properties.
- Fix unnecessarily large I/O for text file with results of
benchexec
during benchmarking. The.results.txt
file is now written incrementally. - Fix incorrect handling of
<withoutfile>
tasks if the tool-info module declared a non-standard working directory. - Small fix for the new filter overlay in the HTML tables when the first run set has no filter.
- Fix our
benchexec.check_cgroups
installation check, which showed invalid warnings since BenchExec 2.7. - Improve handling of inaccessible mountpoints in containers.
This should make it possible to use nested containers on most systems
using the default arguments (e.g., no need for
--hidden-dir /sys
). - Improved row filters of HTML tables (thanks to @DennisSimon).
In addition to filtering via drop-down fields in the table header,
it is now also possible to define filters on a separate overlay,
which can be opened from all tabs via a button in the top-right corner
(e.g., also while looking at plots).
The filters for status and category in the filter overlay are more flexible
because several values can be selected for status and category.
This allows to define filters like
category = "correct" AND (status = "false" OR status = "false(unreach-call)")
. Furthermore, the filter overlay allows to filter the parts of the task id (left-most column) individually and makes it easier to define filters with numeric ranges. - Redesigned UI for changing the plot settings of quantile and scatter plots in the HTML tables (thanks to @lachnerm).
- Hiding columns in HTML tables is now reflected in the URL. This makes it possible to create links to tables that hide columns.
This release contains only one new feature compared to BenchExec 2.7:
- Tables produced by
table-generator
now show the expected verdict of each task, if it is known and it is not the same for all rows.
However, there are several deprecated features removed and other backwards-incompatible changes to make BenchExec more consistent and user-friendly:
- Support for Python 2.7 and 3.4 is removed, the minimal Python version is now 3.5 for all components of BenchExec. We plan to remove support for Python 3.5 after Ubuntu 16.04 goes out of support in 2021.
- If a tool-info module returns
UNKNOWN
for a run result, BenchExec will no longer overwrite that if it thinks the tool terminated abnormally. It will continue to do so ifERROR
is returned. - Result values named
cpuenergy-pkg[0-9]+
are renamed tocpuenergy-pkg[0-9]+-package
because these are not a sum of all the other CPU-energy measurements. - Names of result files produced by
benchexec
now contain timestamps with seconds in order to avoid problems when startingbenchexec
in quick succession. - Support for generating the old-style static HTML tables
(with
table-generator --static-table
) is removed. Only the modern tables that are available since BenchExec 2.3 and CSV tables can be generated. - More metadata are stored in result files of
benchexec
, sotable-generator
no longer needs access to the task-definition files, and changes to the expected verdict that are made after benchmarking will not be reflected in tables. - The Python library Tempita is no longer a dependency of BenchExec.
- We do not create and distribute
.egg
packages for BenchExec releases anymore, only the more modern.whl
packages, as well as Debian/Ubuntu packages and Tar archives.
Furthermore, BenchExec no longer contains hard-coded knowledge about any specific property, all properties are treated in the same way. (The only exception is that score computation is enabled for SV-COMP properties.) This simplification implies several more changes:
- For checking expected verdicts and computing scores it is now required that task-definition files are used. Expected verdicts encoded in the task name are no longer supported.
- Tool-info modules need to return results
true
orfalse
, the resultssat
andunsat
are no longer supported (these were allowed only for the propertySATISFIABLE
). - There is no special handling for composite properties like SV-COMP's property
for memory safety anymore. Previously this property would be represented as
a collection of its subproperties, now it is treated as one property.
Task-definition files can still contain a violated subproperty,
and
benchexec
will continue to use this information for checking the tool result, but this does not depend on which property is used. - Score computation is fixed for tables where property files have uncommon names.
The name of property files is now no longer relevant (as it should have been).
Because of this,
table-generator
needs to have access to the property files that were used during benchmarking.
- The supplied file
benchexec-cgroup.service
for cgroup configuration on systems with systemd now works with systemd 240 or newer (e.g., on Ubuntu 20.04). This also affects the Debian package of BenchExec. - Error messages about failed cgroup access were improved.
- Buttons below plots in the HTML table do not need to be clicked twice.
- Directly opening the quantile tab of HTML tables via the URL works now.
- First line of logs shown in overlay of HTML tables is selectable again.
This release brings several improvements for the new kind of HTML tables
produced by table-generator
, in particular:
- Add hash routing, i.e., the possibility to navigate to certain parts
of the application directly by adding a suffix to the URL.
For example, opening
...table.html#/table
will directly open the table. While navigating through the application, the URL automatically adjusts. This also means that it is possible to use the "Back" button of the browser for going back to previously opened tabs or for closing an overlay window. Thanks @DennisSimon for this! - Make references to files in task-definition files clickable. When clicking on a cell in the first column of table, it shows the task-definition file in an overlay. Now the file's YAML content is parsed and links to input files are added. Thanks @lachnerm for this!
- Fix filtering of negative values in half-open intervals.
- More tooltips and hover effects on table headers to improve usability.
- The table tab now appropriately adjusts if the browser window is resized.
- Fix legend of quantile plot if some columns are empty/missing, and show disabled columns in gray.
- Fix scatter plot if not all data points have valid values.
- Fix layout of column-selection dialog in case not all columns are present for all run sets.
- Fix scrolling behavior of close button of overlay windows.
- In case the property is the same for all tasks of a table, it was not shown so far in the table. Now we show it on the summary tab.
- Improve position of scroll bars across all tabs.
There are also a few changes in other parts of BenchExec:
- Fix mount problems in container mode if mount points with unusual characters
(like
:
) or bind mounts over files exist. The latter is for example relevant when nesting containers (inside another BenchExec or Docker container). - Several new tool-info modules and small improvements to existing ones.
runexec
now creates parent directories of output files if necessary.table-generator
now works if environment variableLANG
is missing.table-generator
should now work on Windows.- It is possible to turn off colored output on stdout by setting the
environment variable
NO_COLOR
(cf. https://no-color.org/). - In the
contrib
folder, we now provide a script for generating task-definition files in YAML format for old-style tasks.
This release does not contain any changes to BenchExec itself,
just for a script in the contrib
directory.
This release contains only a small improvement of one tool-info module.
(not released)
- A complete rewrite of the HTML tables produced by
table-generator
. The tables are now based on React, load much faster, and provide features like pagination, sorting, and more intuitive filters. More information can be found in PR #477. Thanks @bschor for this! Note that the tables are not usable without JavaScript anymore. The old kind of HTML tables can still be produced with the command-line flag--static-table
, but this is deprecated and will be removed in BenchExec 3.0 in January 2020 (cf. #479). - Recursively clean up cgroups after a run.
This enables nesting
runexec
in itself, but only if--full-access-dir /sys/fs/cgroup
is passed to the outerrunexec
, which means that the processes in the outer container have full access to the cgroup hierarchy and could use this to circumvent resource limits. benchexec
filters the tasks to execute depending on the expected verdict, if<propertyfile expectedverdict="...">
in used the benchmark definition.- BenchExec now stores a timestamp for the start time of each run, and timestamps for start and end of reach run set.
benchexec
will store arbitrary user-defined text as benchmark description together with the results if specified withbenchexec --description-file ...
.- Support for execution on Python 3.8.
- Fix crash in
runexec
if the tool's stdout/stderr contain invalid UTF-8. - Fix hanging
benchexec
in container mode if tool cannot be executed (e.g., if executable is missing). - New tool-info modules and updates for SV-COMP'20 and Test-Comp'20.
This release fixes two security issues, all users are encouraged to update:
- Since BenchExec 2.1, the setup of the container for the tool-info module (which was added in BenchExec 1.20) could silently fail, for example if user namespaces are disabled on the system. In this case the tool-info module would be executed outside of the container. Run execution was not affected.
- The kernel offers a keyring feature for storage of keys related to features like Kerberos and ecryptfs. Before Linux 5.2, there existed one keyring per user, and BenchExec did not prevent access from the tool inside the container to the kernel keyring of the user who started BenchExec. Now such accesses are forbidden (on all kernel versions) using seccomp if libseccomp2 is installed, which should the case on any standard distribution. Note that seccomp filters do have a slight performance impact and could prevent some binaries on exotic architectures from working. In such a case please file a bug report.
benchexec
can now partition the Level 3 cache of the CPU for parallel runs
and measure cache usage and memory bandwidth,
at least on some Intel CPUs and if the pqos
and pqos_wrapper are installed.
More information is in the documentation.
Furthermore, some error messages for systems without container support were improved.
This release does not add new features compared to BenchExec 1.22, but removes several deprecated features and brings several other backwards-incompatible changes to make BenchExec more consistent and user-friendly:
- Support for Python 3.2 and 3.3 is removed, the minimal Python version is now 3.4.
Additionally,
runexec
/RunExecutor
continue to support Python 2.7 until end of 2019. - Support for running benchmarks as a different user with
sudo
is removed (parameters--user
/--users
). Use container mode as better method for isolating runs. - Container mode is enabled by default.
It can be disabled with
--no-container
, but this decreases reliability of benchmarking. - If the
cpuacct
cgroup is not available, CPU-time measurements and limits are not supported. - Either container mode or the
freezer
cgroup are required to ensure protection against fork bombs. - Niceness of benchmarked process is not changed, previously it was increased by 5.
- Changes to input of
benchexec
:- The memory limit given to
benchexec
requires an explicitly specified unit. - Support for
<test>
tags,<sourcefiles>
tags, and variables named${sourcefile_*}
removed from benchmark definitions. Use<rundefinition>
,<tasks>
, and${inputfile_*}
instead. - Variables named
${taskdef_*}
are defined only if task-definition files are used, and variables named${inputfile_*}
only otherwise.
- The memory limit given to
- Changes to
table-generator
:- A column named
memUsage
is automatically renamed tomemory
. - A column named
memory
is automatically converted to Megabytes. Both conversions are only applied if no<column>
tags are used.
- A column named
- Changes to run-result data:
- In case of aborted or failed runs, no dummy results (e.g.,
cputime
of 0s) are present. - The memory results of
benchexec
are namedmemory
, notmemUsage
. - Memory results have the unit
B
explicitly specified. Furthermore, units are present in all attributes of the result XML files where they were still missing. - Result item
exitcode
is removed, onlyRunExecutor.execute_run()
still returns it, but as an object instance instead of anint
. Usereturnvalue
andexitsignal
instead.
- In case of aborted or failed runs, no dummy results (e.g.,
- Module
benchexec.test_tool_wrapper
is removed, usebenchexec.test_tool_info
instead. - BenchExec (both
benchexec
andrunexec
) terminates itself cleanly after aborting all runs if it receives one of the signalsSIGTERM
,SIGINT
(Ctrl+C), orSIGQUIT
.
Additionally, this release adds a fix for the container
that is used since BenchExec 1.20 for the tool-info module.
In this container, the environment variable HOME
did not point to /home/benchexec
as expected but to the user's real home directory.
This broke tools like Ultimate if the /home
was configured to be hidden or read-only.
Furthermore, we declare the following features deprecated and plan on removing them for BenchExec 3.0, which is expected to be released in January 2020:
- Support for Python 2.7 and 3.4 (cf. #438)
- Support for checking correctness of run results and computing scores if task-definition files are not used (cf. #439)
Please respond in the respective issue if one of these deprecations is a problem for you.
- More robust handling of Ctrl+C in
benchexec
. For example, output files are now always fully written, whereas previously pressing Ctrl+C at the wrong time could result in truncated files. A side effect of this is that if you callbenchexec.benchexec.BenchExec().start()
in own Python code, you must now add a signal handler forSIGINT
. The same was already true for users ofRunExecutor
, this is now documented. - Fix Ctrl+C for
benchexec
in container mode. In BenchExec 1.21, one would need to press Ctrl+C twice to stopbenchexec
. - Fix unreliable container mode on Python 3.7.
- Some robustness improvements and fixes of rare deadlocks.
- Decreased overhead of
benchexec
while runs are executing.
This release contains only a few bug fixes:
- Forwarding signals to the benchmarked process (and thus, stopping runs via Ctrl+C), was broken on Python 2.
- If the freezer cgroup was available but mounted in a separate hierarchy, it was not used reliably as protection against fork bombs when killing processes.
- Since BenchExec 1.19, an exception would occur if a non-existing command was started in container mode.
- Since BenchExec 1.19, copying output files from a container would occur while subprocesses are still running and would be counted towards the walltime limit. This is fixed, although subprocesses will still be running if the freezer cgroup is not available (cf. #433).
- If
benchexec --container
is used, all code that is part of the tool-info module (as well as all processes started by it) are now run in a separate container with the same layout and restrictions as the run container. Note, however, that it is not the same container, so any modifications made by the tool-info module to files on disk are not visible in the runs! Thetest_tool_info
utility also has gained a parameter--container
for testing how a tool-info module behaves in a container. - Nested containers are now supported.
Due to a change to the internal implementation of the container mode,
commands like the following succeed now:
containerexec -- containerexec --hidden-dir /sys -- /bin/bash
. (Some parts of/sys
need to be excluded because of kernel limitations.) Note that nestingrunexec
orbenchexec
is still not supported, because nested cgroups are not implemented, so any cgroup-related features (resource limitations and measurements) are missing. But nestingcontainerexec
andrunexec --container
(or vice-versa) now works. /etc/hostname
in container now also shows the container's host name that exists since BenchExec 1.19.- Change how CPUs with several NUMA nodes per CPU are handled: BenchExec will now treat each NUMA node like a separate CPU package and avoid creating runs that span several NUMA nodes. Thanks @alohamora!
- In container mode, all temp directories are now on a
tmpfs
"RAM disk". This affects everything written to directories in the hidden or overlay modes. Files written there are now included in the memory measurements and the memory limit! The advantage is that performance should be more deterministic, especially if several runs use much I/O in parallel. This feature can be disabled with--no-tmpfs
. /dev/shm
and/run/shm
are now available inside the container and provide atmpfs
instance (even with--no-tmpfs
) as required by some tools for shared memory.- Container mode now recommends LXCFS
and automatically uses it if available for a better container isolation
(e.g., uptime measures container uptime, not host uptime).
On Debian/Ubuntu, just use
sudo apt install lxcfs
. - Several small bug fixes and other improvements of isolation for container mode (e.g., host name in container is no longer the real host name).
- Add
benchexec --no-hyperthreading
, which restricts core assignments to a single virtual core per physical CPU core (all other sibling cores will stay unused). Thanks @alohamora!
- Add result
done
that tools can output if the standard resultstrue
/false
/unknown
are not applicable (for example because no property was checked), and the run completed successfully. - In container mode,
--keep-system-config
is no longer necessary if overlayfs is not used for/etc
, and thus it is is no longer automatically implied in such cases. - Benchmark definitions support a new attribute
displayName
with a human-readable name that will be shown in tables. - A new variable
${taskdef_name}
can now be used in places where variable substitution is supported. - Table-generator supports
%
as unit for numerical values. - Some improvements for score handling outside of SV-COMP (i.e., if scores are not calculated by BenchExec).
- New tool-info modules for Test-Comp'19
- Several small bug fixes and improvements
- Tasks can now be defined in a YAML-based format, cf. the documentation This supports tasks with several input files, and allows providing metadata such as expected verdicts in a structured format instead of encoded in the file name. The format will be extended to handle more information in the future.
- The wall-time limit can now be specified separately from the CPU-time limit
for
benchexec
as command-line parameter or in the benchmark definition. - Support for SV-COMP'19 property
memcleanup
. - In containers, properly handle
/run/systemd/resolve
, which is necessary for DNS resolution on systems withsystemd-resolved
. - Avoid warnings for mountpoints below inaccessible directories in containers.
- Improvements for handling
NaN
andInf
values intable-generator
. - Log output of BenchExec will now have colors if
coloredlogs
is installed. - New tool-info modules and updates for SV-COMP'19.
- Support for energy measurements if cpu-energy-meter is installed.
- Several small bug fixes and improvements
- Updated tool-info modules for all participants of SV-COMP'18.
- Extended support for variable replacements in table-definitions of table-generator.
- For Debian/Ubuntu, the
.deb
package is now the recommended way of installation, because it automatically configures cgroups as necessary. - BenchExec now automatically attempts to use the sub-cgroup
system.slice/benchexec-cgroup.service
if it does not have access to the current cgroup. This means that if you followed our installation instructions for systems with systemd, there is no need anymore to manually put your shell into the correct cgroup. - Several smaller bug fixes for table-generator: #249, #250, #259, #260, #271, #272
- For users of the Python API of RunExecutor, different file names can now be specified for stdout and stderr of the tool.
- Some new tool-info modules and updates for SV-COMP'18.
- Fix execution of runs specified with
<withoutfile>
tags in the benchmark definition: the name of the run was missing from the command-line in BenchExec 1.11.
table-generator
can now be given result XML files as arguments in addition to a table-definition XML file (with parameter-x
). In this case, it will use the column definitions from the latter for tables with the separately given results.- The directory
contrib
of the repository now contains a scriptstatistics-tex.py
, which can export summary data for benchmark results (e.g., number of solved tasks, average CPU time, etc.) to LaTeX. - The dummy tools
true
andfalse
, which could be used for testing a BenchExec installation, are replaced with a more generic dummy tool calleddummy
. - A few minor bug fixes and performance optimizations.
A new paper about BenchExec called Reliable Benchmarking: Requirements and Solutions is now available.
Please note that support for Python 3.2 and 3.3 is deprecated.
Furthermore, the support for "sudo mode" (parameter --user
/--users
)
is also deprecated.
All deprecated features will be removed in BenchExec 2.0.
This release brings several smaller and medium-sized features:
- Tool-info modules for all participants of SV-COMP'17,
and support for results of the category
correct-unconfirmed
, which is used by SV-COMP if witness validation was not successful. To conform with SV-COMP's definitions, violations of the SV-COMP reachability propertyunreach-call
will now be reported asfalse(unreach-call)
instead offalse(reach)
. - Measurement of block I/O if the
blkio
cgroup controller is available (experimental, please read the documentation!). - Measurement of the energy used by the CPU for a run, if the tool cpu-energy-meter is installed on the system (experimental, please read the documentation!).
- Ability to limit the disk space a tool can occupy in container mode.
- Various minor improvements to make container mode more robust.
- The feature for executing benchmarks under different user accounts with sudo is now marked as deprecated and may be removed in the future, consider using the container mode instead for isolating runs (cf. issue #215).
table-generator
is now more flexible:- Builtin support for certain unit conversions, such that the scale factor does not always need to be explicitly specified. Furthermore, unit conversions now work even if the values already have a unit.
- Column titles can be manually specified with the
displayTitle
attribute. - What columns are relevant for the "diff" table can be configured.
Please also note that we are considering dropping the support for Python 3.2 and maybe 3.3 in BenchExec 2.0 (to be released in a few weeks). If this is a problem for you, please tell us in issue #207.
The main feature of this release is the addition of a container mode that allows to isolate runs from each other and from the host, for example preventing filesystem and network accesses. It also allows to collect and store all files created by the tool in a run. The container mode is still in beta and disabled by default for now, it will be enabled by default in BenchExec 2.0. Please try it out and tell us your experiences!
Further changes:
table-generator
now supports HTTP(S) URLs to be given for result XML files to allow generating tables for results without needing to download them first. The HTML tables will contain correct links to the log files.- New SV-COMP property deadlock supported by
benchexec
. - The parameters
--rundefinition
and--tasks
ofbenchexec
now support wildcards. - Rounding of very small and very large values in
table-generator
has been fixed. - The default font for HTML tables has changed, it is now a font that supports correctly aligned digits.
benchexec
now compresses results by default: XML result files are compressed with BZip2, and log files are stored within a ZIP archive. This can reduce the necessary disk space significantly (typically these logs compress very well), and for large benchmark sets it reduces the number of necessary files, which can make dealing with the results much faster. The previous behavior can be restored with the parameter--no-compress-results
.table-generator
now supports benchmark results where the log files are stored in a ZIP file instead of a regular directory. All features continue to work with compressed results, including extraction of values from log files and viewing log files from HTML tables (cf. table-generator documentation for more details). Compressed and uncompressed results are handled transparently and can be mixed, and using results that were manually compressed or decompressed is also supported.
- Fix
table-generator
behavior for columns where different cells have different units: The release notes for 1.6 claimed that these columns are treated as text column, when instead they were rejected. Now they are treated as text. Note that BenchExec does not create such columns itself, so this should not affect most users. - Fix computation of scores according to the SV-COMP scoring scheme:
if the expected result is for example
false(valid-deref)
and the tool returnsfalse(valid-free)
, the resulting score is the one for a wrong false answer (-16 points), not the one for a wrong true answer (-32 points). The latter score is only given if the tool actually answerstrue
incorrectly. - Change result classification, if the returned answer does not belong to the property of the task,
for example, if the tool returns
true
instead ofsat
for a task with categorysatisfiability
, or if the tool returnsfalse(no-overflow)
when it should not even check for overflows. Now these results are classified as unknown (with score 0), previously these were treated as wrong answers. - Fix escaping of links in HTML tables, e.g., to log files with special characters in their name. This was broken in 1.6.
This release brings several improvements to table-generator
:
table-generator
now rounds measurement values in a scientifically correct way, i.e., with a fixed number of significant digits, not with a fixed number of decimal places. The attributenumberOfDigits
of<column>
tags in table-definition files now also specifies significant digits, not decimal places. By default, in HTML tables all fractional values are now rounded (e.g., time measurements) and all integer values continue without rounding (e.g., memory measurements), previously only "time" columns were rounded. The remaining rounding-related behavior stays unchanged: In CSV tables, values are not rounded by default, and ifnumberOfDigits
is explicitly given for a column, it's value will always be rounded in both HTML and CSV tables.table-generator
now automatically extracts units from the cells in a column and puts them into the table header.- In HTML tables, numeric values are now aligned at the decimal point, and text values are left aligned (previously both were right aligned).
table-generator
now allows to convert values from one unit into another. So far this is only implemented for values that do not have a unit attached to them, and both the target unit and the scale factor need to be specified explicitly in the<column>
tag. This can be used for example to show memory measurements in MB instead of Bytes in tables.table-generator
now allows columns with links to arbitrary files to be added to tables.table-generator
does not handle columns where cells have differing units wrongly anymore. Previously, the unit was simply dropped, leading to wrong values for statistics. Now such columns are treated as text and no statistics are generated. (Note that BenchExec never creates such columns by itself, only if values are extracted from the tool output this could happen).
Other changes:
- The behavior of
benchexec --timelimit
was changed slightly, if a value forhardtimelimit
was given in the benchmark-definition file. If a time limit is specified on the command line, this now overrides both soft and hard time limit. - Implementation of tool-info modules got easier because the
test_tool_info
helper got improved (it now allows to test the function for extracting results from tool outputs). - Several tool-info modules of tools participating in SV-COMP got improved.
- Simplified cgroups setup for systemd systems.
- Improved documentation.
- Improved definition of time and memory limits: Both can now be specified including units such as "s", "min" / "MB", "GB". to make them easier to read and less ambiguous. The old input format without units is still valid.
- runexec now allows enabling other cgroup subsystems and setting arbitrary cgroup options.
- HTML tables gained the possibility for inverting row filters.
- Improve detection of out-of-memory situations (were not always reported as OOM).
- External resources in HTML tables are loaded from HTTPS URLs such that browsers do not complain because of mixed content when viewing tables via HTTPS.
- Improved warnings for swapping and CPU throttling for benchexec.
- Various improvements to internal handling of memory values, they are not consistently stored as bytes (this only affects extensions of BenchExec, not regular input and output for users).
- BenchExec moved to https://github.com/sosy-lab/benchexec
- Fix several bugs in table-generator introduced in version 1.3.
- BenchExec now creates fresh empty directories for $HOME and $TMPDIR of all runs, and removes them afterwards.
- table-generator now transparently supports result XML files as input that are compressed with GZip or BZip2.
- benchexec now reports some more information as status when a tool crashes, e.g. whether it segfaulted or aborted, and what the exit code was (previously this was only done for some tools).
- If a tool produces a result but still violates a resource limit, this is now shown in the status (but still counted as timeout / out of memory).
- Added dummy tool "calculatepi" that needs no input files and no installation, but can be used to create some CPU load and test benchmarking (it calculates Pi up some arbitrary number of digits using the tool "bc").
- Renaming "tool wrapper" to "tool info".
This is mostly an internal and documentation change, but the utility
benchexec.test_tool_wrapper
is now namedbenchexec.test_tool_info
.
- Fix core assignment on AMD Bulldozer/Piledriver Opterons.
- Measure and report CPU time usage per core
(hidden by default in tables, use
table-generator --all-columns
to show). - Parameter
--user
allows executing benchmarks under a different user (cf. https://github.com/sosy-lab/benchexec/blob/master/doc/separate-user.md). - Performance improvements for table-generator, including parallel processing of input and output files and statistics.
- HTML Tables support filtering rows by task name.
- Improved statistics in HTML tables: median is now the arithmetic median, unnecessary rounding removed, standard deviation added, and missing results are not counted as "0" but ignored in calculation.
- New utility for testing tool wrappers, making it easier to add support for new tools.
- Several new modules for integration of various software verifiers.
- BenchExec now records whether TurboBoost was enabled during benchmarking.
- Updated SV-COMP scoring scheme to SV-COMP 2016.
- Support new property 'no-overflow' for SV-COMP 2016.
- Several new modules for integration of various software verifiers.
- Some improvements to CPU-core assignment.
- HTML tables produced by table-generator now have a header that stays always visible, even when scrolling through the table.
- A Debian package is now created for releases and made available on GitHub.
- Small bug fixes.
- Multiple runs for the same file can now be shown in the table in different rows if they have different properties or ids.
- Helper files for generating scatter and quantile plots with Gnuplot added.
- Doctype declarations are now used in all XML files.
- Statistics output at end of benchexec run was wrong.
- Allow to redirect stdin of the benchmarked tool in runexec / RunExecutor
- Fix bug in measurement of CPU time (only occurred in special cases and produced a wrong value below 0.5s)
- Improve utility command for checking cgroups to work around a problem with cgrulesngd not handlings threads correctly.
- Support for integrating SMTLib 2 compliant SMT solvers and checking the expected output.
- runexec now supports Python 2 again.
- table-generator allows to selected desired output formats and supports output to stdout.
- Added utility command for checking if cgroups have been set up correctly.
- Avoid "false posititive/negative" and use "incorrect false/true" instead.
- Command-line arguments to all tools can be read from a file given with prefix "@".
- Bug fixes and performance improvements.
- HTML tables now have header with direct access to plots.
- Maximum score of table is generated again.
- table-generator can now extract statistic values for other tools, too (not only CPAchecker).
- More flexible time limit specifications.
- Warnings shown if system swaps or throttles during benchmarking.
- Improved reliability of benchmarking: forbid swapping, use freezer to kill processes atomically.
- Renamed
<sourcefiles>
tag to<tasks>
in benchexec input. - Bug fixes.
- Added documentation.
- Added more tests.
- bug fixes
- switch to Python 3 completely
Initial version of BenchExec as taken from the repository of CPAchecker.