| Age | Commit message (Collapse) | Author |
|---|
|
|
|
f23d5028059078a346efc977287b669d494a5a3f, a0f7de814ae5c299d6ce99bed5fb308a05d50ba0: [Backport #20296]
[Bug #20296] Clear errinfo when `exception: false`
[Bug #20296] Refine the test
[Bug #20296] Fix the default assertion message
|
|
d5080f6e8b77364483ff6727b1065e45e180f05d: [Backport #20292]
[Bug #20292] Truncate embedded string to new capacity
Fix -Wsign-compare on String#initialize
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
../string.c:1886:57: warning: comparison of integer expressions of different signedness: ‘size_t’ {aka ‘long unsigned int’} and ‘long int’ [-Wsign-compare]
1886 | if (STR_EMBED_P(str)) RUBY_ASSERT(osize <= str_embed_capa(str));
| ^~
|
|
Make io_fwrite safe for compaction
[Bug #20169]
Embedded strings are not safe for system calls without the GVL because
compaction can cause pages to be locked causing the operation to fail
with EFAULT. This commit changes io_fwrite to use rb_str_tmp_frozen_no_embed_acquire,
which guarantees that the return string is not embedded.
|
|
|
|
[Bug #20307] Fix `Hash#update` to make frozen copy of string keys
|
|
|
|
|
|
rb_obj_setup: do not copy RUBY_FL_SEEN_OBJ_ID
[Bug #20250]
We're seting up a new instance, so it never had an associated
object_id.
|
|
Fix memory leak when duplicating too complex object
[Bug #20162]
Creating a ST table then calling st_replace leaks memory because the
st_replace overwrites the ST table without freeing any of the existing
memory. This commit changes it to use st_copy instead.
For example:
RubyVM::Shape.exhaust_shapes
o = Object.new
o.instance_variable_set(:@a, 0)
10.times do
100_000.times { o.dup }
puts `ps -o rss= -p #{$$}`
end
Before:
23264
33600
42672
52160
61600
71728
81056
90528
100560
109840
After:
14752
14816
15584
15584
15664
15664
15664
15664
15664
15664
|
|
f3af5ae7e6c1c096bbfe46d69de825a02b1696cf: [Backport #20311]
Don't pin named structs defined in Ruby
[Bug #20311]
`rb_define_class_under` assumes it's called from C and that the
reference might be held in a C global variable, so it adds the
class to the VM root.
In the case of `Struct.new('Name')` it's wasteful and make
the struct immortal.
Make Struct memory leak test faster
[Bug #20311]
It times out on some platform, so we can reduce iterations.
On my machine it completes in 250ms and RSS grows 8X.
|
|
ensure safe profiling
**What does this PR do?**
This PR tweaks the `vm_push_frame` function to add an explicit compiler
fence (`atomic_signal_fence`) to ensure profilers that use signals
to interrupt applications (stackprof, vernier, pf2, Datadog profiler)
can safely sample from the signal handler.
This is a backport of #11036 to Ruby 3.2 .
**Motivation:**
The `vm_push_frame` was specifically tweaked in
https://github.com/ruby/ruby/pull/3296 to initialize the a frame
before updating the `cfp` pointer.
But since there's nothing stopping the compiler from reordering
the initialization of a frame (`*cfp =`) with the update of the cfp
pointer (`ec->cfp = cfp`) we've been hesitant to rely on this on
the Datadog profiler.
In practice, after some experimentation + talking to folks, this
reordering does not seem to happen.
But since modern compilers have a way for us to exactly tell them
not to do the reordering (`atomic_signal_fence`), this seems even
better.
I've actually extracted `vm_push_frame` into the "Compiler Explorer"
website, which you can use to see the assembly output of this function
across many compilers and architectures: https://godbolt.org/z/3oxd1446K
On that link you can observe two things across many compilers:
1. The compilers are not reordering the writes
2. The barrier does not change the generated assembly output
(== has no cost in practice)
**Additional Notes:**
The checks added in `configure.ac` define two new macros:
* `HAVE_STDATOMIC_H`
* `HAVE_DECL_ATOMIC_SIGNAL_FENCE`
Since Ruby generates an arch-specific `config.h` header with
these macros upon installation, this can be used by profilers
and other libraries to test if Ruby was compiled with the fence enabled.
**How to test the change?**
As I mentioned above, you can check https://godbolt.org/z/3oxd1446K
to confirm the compiled output of `vm_push_frame` does not change
in most compilers (at least all that I've checked on that site).
|
|
[Bug #20239] Fix overflow at down-casting
|
|
[Bug #20305] Fix matching against an incomplete character
When matching against an incomplete character, some `enclen` calls are
expected not to exceed the limit, and some are expected to return the
required length and then the results are checked if it exceeds.
|
|
04729fe68dceddab045be7324e26c2bb15aa62c7: [Backport #20286] [Backport #20286]
Ensure that exiting thread invokes end-of-life behaviour. (#10039)
Fix exception handling in `rb_fiber_scheduler_set`. (#10042)
|
|
Don't clear pending interrupts in the parent process. (#10365)
|
|
Enumerator should use a non-blocking fiber, change `rb_fiber_new` to be non-blocking by default. (#10481)
|
|
[Bug #20499] Use Xcode owned tools for Xcode clang
Xcode has its own version tools that may be incompatible with genuine
LLVM tools, use the tools in the same directory.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Separate the directly from the customized test-unit, since it may not
work with bundled gems.
|
|
9f708d48f6df37ee9600db9d51b57a156609a13b,0301473fb523c71d8cdc4966971f31f502001185,7f0e26b7f99bf76408569892ce20318501f74729: [Backport #20516]
Clear runtime dependencies if default gems is specified.
The current build system uses runtime dependencies from only
`.bundle` directory. We shouldn't install runtime dependencies
from rubygems.org when `make test-bundled-gems` is invoked.
Fixed dependencies list format
Re-use strscan with ruby repo
|
|
[Backport #20516]
Update bundled_gems
|
|
[Backport #20494]"
This reverts commit ec30266267b2a063089b8ce7967203bcbadbdf4e.
|
|
1ab7c412d2e3880a7ad233c32e93961888f8145c, fd549b229b0822198ddc847703194263a2186ed1: [Backport #20515]"
This reverts commit dc4ca25e0a01e072ba3cf1fc47612aff72c980af.
|
|
1ab7c412d2e3880a7ad233c32e93961888f8145c, fd549b229b0822198ddc847703194263a2186ed1: [Backport #20515]
ci: Test whether GMP is working in compilers.yml (#10875)
Avoid reoccurence of [Bug #20515]
Requires https://github.com/ruby/ruby/pull/10876 since 18eaf0be905e3e251423b42d6f4e56b7cae1bc3b
bug: https://bugs.ruby-lang.org/issues/20515
RUBY_CHECK_HEADER didn't define HAVE_{header-file} (#10876)
--with-gmp is not working at all because HAVE_GMP_H
was missing since 18eaf0be90. [Bug #20515]
bug: https://bugs.ruby-lang.org/issues/20515
follow-up: https://bugs.ruby-lang.org/issues/20494
follow-up: 18eaf0be905e3e251423b42d6f4e56b7cae1bc3b
follow-up: https://github.com/ruby/ruby/pull/10805
test_bignum: defined? returns String (#10880)
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit
didn't verify the test is working properly due to mistaken auto-merge… [Bug #20515]
bug: https://bugs.ruby-lang.org/issues/20515
follow-up: 22e4eeda6561693367fc7a00b92b90f46b09cabd
follow-up: https://github.com/ruby/ruby/pull/10875
|
|
[Bug #20494] Search non-default directories for GMP
Co-Authored-by: lish82 (Hiroki Katagiri)
|
|
There's an exhaustive explanation of this in the linked redmine bug, but
the short version is as follows:
blocking_region_begin can spill callee-saved registers to the stack for
its own use. That means they're not saved to ec->machine by the call to
setjmp, since by that point they're already on the stack and new,
different values are in the real registers. ec->machine's end-of-stack
pointer is also bumped to accomodate this, BUT, after
blocking_region_begin returns, that points past the end of the stack!
As far as C is concerned, that's fine; the callee-saved registers are
restored when blocking_region_begin returns. But, if another thread
triggers GC, it is relying on finding references to Ruby objects by
walking the stack region pointed to by ec->machine.
If the C code in exec; subsequently does things that use that stack
memory, then the value will be overwritten and the GC might prematurely
collect something it shouldn't.
[Bug #20493]
|
|
[Bug #20180]
Backports #9624.
|
|
[Bug #20569]
`putobject RubyVM::FrozenCore`, is not serializable, we
have to use `putspecialobject VM_SPECIAL_OBJECT_VMCORE`.
|
|
README.ja.md: Remove Cirrus CI badge image. folloup for 15ee9c7c1b693f29b6b2dbe7b47488bf154e481c.
|
|
Cirrus CI has been removed [ci skip]
|
|
[DOC] README.md: Remove Cirrus CI badge image. (#8785)
We removed the `.cirrus.yml` at the commit
<01b5d1d2ff6ca91b2909dfa67295f59b53e6f065>.
Let's remove the badge image too.
|
|
YJIT: Add a cargo job for Arm64 and remove .cirrus.yml (#8679)
* YJIT: Add a cargo job for Arm64
* YJIT: Use command names as job names
They look more consistent with jobs that use the command name as a job
name, such as `make (check)`, `make (test-bundled-gems)`, etc.
* Remove .cirrus.yml
|
|
70ad58cb62b195ba86a5ef07a565b22b02a040ea: [Backport #20516]"
This reverts commit 519ac9c5fd7bfa75655231188ac1da9589563618.
|
|
9f708d48f6df37ee9600db9d51b57a156609a13b,0301473fb523c71d8cdc4966971f31f502001185,7f0e26b7f99bf76408569892ce20318501f74729: [Backport #20516]"
This reverts commit 0de1ff12973ffc04bc9a8b0799e963922986d1eb.
|
|
9f708d48f6df37ee9600db9d51b57a156609a13b,0301473fb523c71d8cdc4966971f31f502001185,7f0e26b7f99bf76408569892ce20318501f74729: [Backport #20516]
Clear runtime dependencies if default gems is specified.
The current build system uses runtime dependencies from only
`.bundle` directory. We shouldn't install runtime dependencies
from rubygems.org when `make test-bundled-gems` is invoked.
Fixed dependencies list format
Re-use strscan with ruby repo
|
|
[Backport #20516]
Update bundled_gems
|
|
This is a combination of main Ruby commit
https://github.com/ruby/ruby/commit/aefc98891c42024039f19ef45bdfe93fbc590b7c
and my PR correcting the regex https://github.com/ruby/ruby/pull/10819.
Upstream Ruby requests that changes to this test go to readline-ext repo
before being backported to 3.2 branch.
https://github.com/ruby/readline-ext/commit/868f873a78
|
|
* Fix `io_buffer_get_string` default length computation.
When an offset bigger than the size is given, the resulting length will be
computed incorrectly. Raise an argument error in this case.
* Validate all arguments.
|
|
- Fix IO::Buffer `read`/`write` to use a minimum length.
|
|
|
|
|
|
Alias init functions
The extension library has each initialization function named "Init_" +
basename. If multiple extensions have the same base name (such as
cgi/escape and erb/escape), the same function will be registered for
both names.
To fix this conflict, rename the initialization functions under sub
directories using using parent names, when statically linking.
|
|
[Bug #20421]
The bug was fixed in Ruby 3.3 via 9dcdffb8bf8a3654fd78bf1a58b30c8e13888a7a
|
|
Because `optflags` is pasted into the invocation line after `CC`, we were
building with -O1 unintentionally. You can see this in the configuration
summary: https://github.com/ruby/ruby/actions/runs/3933391169/jobs/6727044423#step:9:753
The check actually fails with -O2. To make it pass, upstream suggest
that we use the annocheck GCC plugin. Since it requires building from
source as the debian package for it isn't ready yet, punt on it for now
and use `--skip-gaps`.
Co-authored-by: Jun Aruga <jaruga@redhat.com>
|
