You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Some Julia CI tests are currently incompatible with MMTk. They are related to:
jl_n_gcthreads is set to 0 for MMTk since it uses its own worker threads. The tests in cmdlineargs.jl that are related to --gc-threads=X currently fail.
MMTk does not support dumping a heap snapshot (at least not in the same format as the stock GC), some tests fail for this reason.
When doing fastpath allocation we currently do not have a callback for the allocation profiler; some of the tests related to codegen and the allocation profiler fail when using MMTk.
MMTk uses a different logging style than the stock GC. This test in misc.jl fails.
--heap-size-hint is not implemented by MMTk since it only supports a hard heap limit. The tests related to that also fail.
Some GC tests rely on the FULL_SWEEP_REASONS variable, which is specific to the stock GC. Furthermore, tests that check for page utilization also fail.
This issue is to document these tests until we decide whether they should be specific to the stock GC or until we implement the same semantics for MMTk (or change the semantics such the tests are compatible with both GCs).
The text was updated successfully, but these errors were encountered:
As noted in #57103, there are a
few tests that are specific to the stock GC and will fail for MMTk. For
now, this PR simply skips those tests when not using the stock GC until
we agree on how to approach the incompatibilities described in the
issue.
Updating mmtk-julia version to include
mmtk/mmtk-julia#228 and fix the MMTk CI.
I've also changed the allocation profiler tests to skip all tests
instead of just a few since I've seen some spurious errors - they should
all be related though, we need to make sure the profiler accounts for
fastpath allocation (see
#57103)
This should fix#57306.
Updating mmtk-julia version to include
mmtk/mmtk-julia#228 and fix the MMTk CI.
I've also changed the allocation profiler tests to skip all tests
instead of just a few since I've seen some spurious errors - they should
all be related though, we need to make sure the profiler accounts for
fastpath allocation (see
#57103)
This should fix#57306.
(cherry picked from commit 72f8a10)
Some Julia CI tests are currently incompatible with MMTk. They are related to:
jl_n_gcthreads
is set to 0 for MMTk since it uses its own worker threads. The tests incmdlineargs.jl
that are related to--gc-threads=X
currently fail.misc.jl
fails.--heap-size-hint
is not implemented by MMTk since it only supports a hard heap limit. The tests related to that also fail.FULL_SWEEP_REASONS
variable, which is specific to the stock GC. Furthermore, tests that check for page utilization also fail.This issue is to document these tests until we decide whether they should be specific to the stock GC or until we implement the same semantics for MMTk (or change the semantics such the tests are compatible with both GCs).
The text was updated successfully, but these errors were encountered: