This sets continue-on-error to true on any scheduled test that uses a
matrix so we can have all sets of tests run regardless if one of them
fails or not.
- Updates the list of current gem5 Maintainers.
- Updates the MAINATAINERS.yaml comments to reflect gem5's contribution
policy and maintainer responsibilities since moving to GitHub.
- Updates the subsystem maintainers, based on maintainer's stated
preferences.
- Adds the optional 'expert' field to subsystems.
This sets continue-on-error to true on any scheduled test that
uses a matrix so we can have all sets of tests run regardless
if one of them fails or not.
Change-Id: I8f6137ebdf62a5cecd582387316c330c8a1401ca
In an SMT CPU, upon a squash, the mis-predicted(squashing) instructions
can still be executing at IEW and own phys registers. If these registers
are added back to the rename freelist on this Tick, the registers may be
renamed to be used by other SMT thread(s). This causes register
ownership hazards, which may eventually freeze the CPU. This problem
seems to date back to 2014
(https://www.mail-archive.com/gem5-users@gem5.org/msg10180.html).
This patch delays the freelist update to avoid the hazard.
I tested that this patch does not cause any performance impact for my
set of benchmarks on default non-SMT O3CPU.
In an SMT CPU, upon a squash, the phys regs used by
mispredicted instructions can still be owned by executing
instructions in IEW. If the regs are added back to freelist
on this tick, the reg may be renamed to be used by another
SMT thread. This causes reg ownership hazard, which may
eventually freeze the CPU.
This patch delays the freelist update to avoid the hazard.
Change-Id: I993b3c7d357269f01146db61fc8a7b83a989ea45
If status.vs == OFF, the RVV instruction should raise Illegal
Instruction according to RISC-V V spec. If RVV is not implemented, all
of the RVV instruction need to raise exception.
This moves the clean runner step in our yaml files to be at the
beginning of a job, so that if a runner goes down and is unable to clean
at the end, we can ensure that
subsequent jobs still run as expected.
Change-Id: Iba52694aefe03c550ad0bfdb5b5f938305273988
Due to a 60GB limit on the VMs the gem5 project's GitHub Actions
self-hosted Runners execute within, we cannot run tests which need to
download the gem5 Resource's PARSEC disk image (v1.0.0,
http://resources.gem5.org/resources/x86-parsec?version=1.0.0). This
image, 33GB, is too big and causes our runners to run out of disk space
and fail.
These changes can be reverted when we are able to increase the size of
our VMs.
The x86-parsec gem5 Resource (v1.0.0,
http://resources.gem5.org/resources/x86-parsec?version=1.0.0) is 33GB.
The gem5 GitHub Actions self-hosted runners do not have enough Disk
Space in the VMs they are run to download this. Ergo we skip it.
Change-Id: I290fe265f03ceca65b2bed87e9f4a4ad601e0fc1
This argument allows the passing of IDs of resources which should be
skipped for this check.
Note: A current limitation here is you cannot specify the version of a
resource. Passing the ID of a resource to this will skip the downloading
for all versions of that resource.
Change-Id: Ifdb7c2b71553126fd52a3d286897ed5dd8e98f7c
These tests are disabled due our GitHub Actions self-hosted Runners
having a 60GB of disk space. The PARSEC Disk Image Resource (v1.0.0,
http://resources.gem5.org/resources/x86-parsec?version=1.0.0) is 33GB
and is simply too big to download and unzip for these tests.
These tests can be reenabled when this issue is resolved.
Change-Id: I9a63aa1903cea3ce7942bdc85bcd0b24761d2f29
This moves the clean runner step in our yaml files to be at the
beginning of a job, so that if a runner goes down and is
unable to clean at the end, we can ensure that
subsequent jobs still run as expected.
Change-Id: Iba52694aefe03c550ad0bfdb5b5f938305273988
There was another missing file in the configs for the fs tests,
so this should allow the switcheroo tests to pass.
Change-Id: Ic4e26cceeb9209f176158b80eaaba88b47968c39
The dailies timed out as they were running the entire directory of tests
due to a wrong variable name being used. In addition, the names of tests
were adjusted to include the matrix type so the artifacts won't
overwrite each other
The configs directory for the fs tests was missing the
checkpoint.py file, causing some of the CI tests to fail.
Change-Id: Ifbd775ad658f96d06bea7bee554fe3bedcf5a5b5
* Solves issue #106 by updating the cpts with the necessary vector
registers.
Change-Id: Ifeda90e96097f0b0a65338c6b22a8258c932c585
util: clear vector_element field
Change-Id: I6c9ec4e71f66722b26de030fa139cd626bdb24dc
"tests/gem5/configs/download_check.py" is used by the
"test-resource-downloading" test (defined in
"tests/gem5/gem5-resources/test_download_resources.py" and ran as part
of the "very-long" suite).
Prior to this change "download_check.py" would download each resource,
check it's md5, then at the end of the script remove all the downloaded
resources. This is inefficient on disk space and was causing our
"very-long" suite of tests to require a machines with a lot of disk
space to run.
This change alters 'download_check.py" to remove each resource after the
md5 check. Thus, only one resource is ever downloaded and present at any
given time during the running of this script.
The dailies timed out as they were running the entire directory
of tests due to a wrong variable named being used. In addition,
the names of tests were adjusted to include the matrix type so
the artifacts won't overwrite each other
Change-Id: Iaa1be8e0cfcbf9d64f1a674590bfe2bf1f0dae90
Updates the directories in which tests are run in accordance
with the refactoring of the testing directory
Change-Id: I93f5c5b0236c5180da04deb425ec2ed6804fa003
This field was added to give gem5 community members a change to register
that they have expertise in a particular subsystem but do not much to
assign themselves the responsibilities of a subsystem maintainer.
Those who have registered interest on being an subsystem expert have
been added.
Change-Id: I8f532e381e8e42257b2a68ac48204131479d8cd0
This change incorporates changes to the set of maintainers and the
maintainers assigned to each subsystem based on individual maintainers'
preferences.
Change-Id: Ic2c39907763282e89936fa0d90e3c1a105a0d917
This comment is updated to reflect new gem5 policy and its move to
GitHub and a Pull Request contribution model.
Change-Id: Iec909ffa0cca254fdbe56ce3165cb948cdd0cbce
"tests/gem5/configs/download_check.py" is used by the
"test-resource-downloading" test (defined in
"tests/gem5/gem5-resources/test_download_resources.py" and ran as part
of the "very-long" suite).
Prior to this change "download_check.py" would download each resource,
check it's md5, then at the end of the script remove all the downloaded
resources. This is inefficient on disk space and was causing our
"very-long" suite of tests to require a machines with a lot of disk
space to run.
This change alters 'download_check.py" to remove each resource after the
md5 check. Thus, only one resource is ever downloaded and present at any
given time during the running of this script.
Change-Id: I38fce100ab09f66c256ccddbcb6f29763839ac40
This merges initial support for RVV. Currently, only the simple CPUs are supported.
The decoder stalls for every vsetvl instruction.
In the future, we will implement vsetvl as a control instruction as described in #144
