Previously this was listed as a suppression, but we actually want this
permanently unsanitized. This should be faster and more reliable since
TASN won't have to match against symbolicated backtraces.
This introduces a new method to encapsulate checking whether the current
Ractor owns the vm->ractor.sync lock. This allows us to disable TSan on
it since that operation should be safe, and still get validation of
other uses.
* Add support for `cause:` argument to `Fiber#raise` and `Thread#raise`.
The implementation behaviour is consistent with `Kernel#raise` and
`Exception#initialize` methods, allowing the `cause:` argument to be
passed to `Fiber#raise` and `Thread#raise`. This change ensures that
the `cause:` argument is handled correctly, providing a more consistent
and expected behavior when raising exceptions in fibers and threads.
[Feature #21360]
* Shared specs for Fiber/Thread/Kernel raise.
---------
Co-authored-by: Samuel Williams <samuel.williams@shopify.com>
Previously we had an assertion that the method table was only set on
young objects, and a comment stating that was how it needed to be used.
I think that confused the complexity of the write barriers that may be
needed here.
* Setting an empty M_TBL never needs a write barrier
* T_CLASS and T_MODULE should always fire a write barrier to newly added
methods
* T_ICLASS only needs a write barrier to methods when
RCLASSEXT_ICLASS_IS_ORIGIN(x) && !RCLASSEXT_ICLASS_ORIGIN_SHARED_MTBL(x)
We shouldn't assume that the object being young is sufficient, because
we also need write barriers for incremental marking and it's unreliable.
When creating a new origin in ensure_origin, we need to fire a write
barrier after RCLASS_WRITE_ORIGIN. rb_class_set_super allocates, so GC
could happen there, either incrementally marking or promoting the newly
allocated class, and only after RCLASS_WRITE_ORIGIN will origin mark
object in the M_TBL.
I've enabled the [Predictive Test Selection](https://www.launchableinc.com/docs/features/predictive-test-selection/) feature, which let machine-learning model selects the appropriate tests to reduce CI execution time in https://github.com/ruby/ruby/pull/12617.
However, I noticed that there are some problems for enabling PTS in Ruby CI after several experiments. Until fixing the problem, I'll disable this feature by removing `launchable subset` command.
objspace->flags.immediate_sweep shares the same word as
objspace->flags.during_incremental_marking. So in gc_start we need to
assign it after gc_enter() so that we hold the VM lock and have issued a
barrier, as rb_gc_impl_writebarrier is reading
objspace->flags.during_incremental_marking.
Strings concatenated with backslash may end up being frozen when they
shouldn't be. This commit fixes the issue. It required a change
upstream in Prism, but also a change to the Prism compiler in CRuby.
https://github.com/ruby/prism/pull/3606
[Bug #21187]
Old versions of BUndler need a workaround to support nested `bundle
exec` invocations by overriding `Gem.activate_bin_path`. However,
RubyGems now uses this new `Gem.activate_and_load_bin_path` helper in
binstubs, which is of course not overridden in those Bundler versions
since it didn't exist at the time.
So, include the override here to workaround that.
e5ed95e242
