July 2024 - ruby-core - ml.ruby-lang.org

[ruby-core:117037] [Ruby master Bug#20322] rb_enc_interned_str_cstr doesn't accept null pointer for encoding
by thomasmarshall (Thomas Marshall) 15 Jul '24

15 Jul '24

Issue #20322 has been reported by thomasmarshall (Thomas Marshall). ---------------------------------------- Bug #20322: rb_enc_interned_str_cstr doesn't accept null pointer for encoding https://bugs.ruby-lang.org/issues/20322 * Author: thomasmarshall (Thomas Marshall) * Status: Open * Backport: 3.0: UNKNOWN, 3.1: UNKNOWN, 3.2: UNKNOWN, 3.3: UNKNOWN ---------------------------------------- The [header documentation](https://github.com/ruby/ruby/blob/93556d46203545bc2364b1c0dd… for `rb_enc_interned_str_cstr` notes: > `enc` can be a null pointer. When [adding support for this function to TruffleRuby](https://github.com/oracle/truffleruby/pull/3427) we noticed that the behaviour did not match that documentation. I think this did work previously, but it currently causes a segmentation fault when trying to [autoload the encoding](https://github.com/ruby/ruby/blob/93556d46203545bc2364b1c0dd1281b… because it calls `rb_enc_mbmaxlen` which [expects a non-null encoding](https://github.com/ruby/ruby/blob/93556d46203545bc2364b1c0dd1281b…. I'm not sure how important this behaviour is, whether extensions ever actually call this with a null pointer, or if the documentation is instead just incorrect. I think it's a straightforward fix to make the behaviour match the documentation, so I would like to open a PR, but also happy to just remove the note if that is preferable. -- https://bugs.ruby-lang.org/

4 4

[ruby-core:117531] [Ruby master Bug#20431] Ruby 3.3.0 build fail with make: *** [io_buffer.o] Error 1
by shubham_yadav (Shubham Yadav) 15 Jul '24

15 Jul '24

Issue #20431 has been reported by shubham_yadav (Shubham Yadav). ---------------------------------------- Bug #20431: Ruby 3.3.0 build fail with make: *** [io_buffer.o] Error 1 https://bugs.ruby-lang.org/issues/20431 * Author: shubham_yadav (Shubham Yadav) * Status: Open * Backport: 3.0: UNKNOWN, 3.1: UNKNOWN, 3.2: UNKNOWN, 3.3: UNKNOWN ---------------------------------------- I'm trying to build ruby from scratch with openss1.1.0. But it's failing with below error. In file included from ./include/ruby/internal/arithmetic/char.h:23, from ./include/ruby/internal/arithmetic.h:24, from ./include/ruby/ruby.h:28, from ./include/ruby/encoding.h:19, from ./include/ruby/io.h:16, from io_buffer.c:9: ./include/ruby/internal/arithmetic/int.h: In function ‘rb_int2num_inline’: ./include/ruby/internal/arithmetic/int.h:241: warning: comparison is always true due to limited range of data type ./include/ruby/internal/arithmetic/int.h:241: warning: comparison is always true due to limited range of data type ./include/ruby/internal/arithmetic/int.h: In function ‘rb_uint2num_inline’: ./include/ruby/internal/arithmetic/int.h:256: warning: comparison is always true due to limited range of data type io_buffer.c: In function ‘io_buffer_read_U8’: io_buffer.c:1872: error: ‘RB_IO_BUFFER_HOST_ENDIAN’ undeclared (first use in this function) io_buffer.c:1872: error: (Each undeclared identifier is reported only once io_buffer.c:1872: error: for each function it appears in.) io_buffer.c: In function ‘io_buffer_write_U8’: io_buffer.c:1872: error: ‘RB_IO_BUFFER_HOST_ENDIAN’ undeclared (first use in this function) io_buffer.c: In function ‘io_buffer_read_S8’: io_buffer.c:1873: error: ‘RB_IO_BUFFER_HOST_ENDIAN’ undeclared (first use in this function) io_buffer.c: In function ‘io_buffer_write_S8’: io_buffer.c:1873: error: ‘RB_IO_BUFFER_HOST_ENDIAN’ undeclared (first use in this function) io_buffer.c: In function ‘io_buffer_read_u16’: io_buffer.c:1875: error: ‘RB_IO_BUFFER_HOST_ENDIAN’ undeclared (first use in this function) io_buffer.c: In function ‘io_buffer_write_u16’: io_buffer.c:1875: error: ‘RB_IO_BUFFER_HOST_ENDIAN’ undeclared (first use in this function) io_buffer.c: In function ‘io_buffer_read_U16’: io_buffer.c:1876: error: ‘RB_IO_BUFFER_HOST_ENDIAN’ undeclared (first use in this function) io_buffer.c: In function ‘io_buffer_write_U16’: io_buffer.c:1876: error: ‘RB_IO_BUFFER_HOST_ENDIAN’ undeclared (first use in this function) io_buffer.c: In function ‘io_buffer_read_s16’: io_buffer.c:1877: error: ‘RB_IO_BUFFER_HOST_ENDIAN’ undeclared (first use in this function) io_buffer.c: In function ‘io_buffer_write_s16’: io_buffer.c:1877: error: ‘RB_IO_BUFFER_HOST_ENDIAN’ undeclared (first use in this function) io_buffer.c: In function ‘io_buffer_read_S16’: io_buffer.c:1878: error: ‘RB_IO_BUFFER_HOST_ENDIAN’ undeclared (first use in this function) io_buffer.c: In function ‘io_buffer_write_S16’: io_buffer.c:1878: error: ‘RB_IO_BUFFER_HOST_ENDIAN’ undeclared (first use in this function) io_buffer.c: In function ‘io_buffer_read_u32’: io_buffer.c:1880: error: ‘RB_IO_BUFFER_HOST_ENDIAN’ undeclared (first use in this function) io_buffer.c: In function ‘io_buffer_write_u32’: io_buffer.c:1880: error: ‘RB_IO_BUFFER_HOST_ENDIAN’ undeclared (first use in this function) io_buffer.c: In function ‘io_buffer_read_U32’: io_buffer.c:1881: error: ‘RB_IO_BUFFER_HOST_ENDIAN’ undeclared (first use in this function) io_buffer.c: In function ‘io_buffer_write_U32’: io_buffer.c:1881: error: ‘RB_IO_BUFFER_HOST_ENDIAN’ undeclared (first use in this function) io_buffer.c: In function ‘io_buffer_read_s32’: io_buffer.c:1882: error: ‘RB_IO_BUFFER_HOST_ENDIAN’ undeclared (first use in this function) io_buffer.c: In function ‘io_buffer_write_s32’: io_buffer.c:1882: error: ‘RB_IO_BUFFER_HOST_ENDIAN’ undeclared (first use in this function) io_buffer.c: In function ‘io_buffer_read_S32’: io_buffer.c:1883: error: ‘RB_IO_BUFFER_HOST_ENDIAN’ undeclared (first use in this function) io_buffer.c: In function ‘io_buffer_write_S32’: io_buffer.c:1883: error: ‘RB_IO_BUFFER_HOST_ENDIAN’ undeclared (first use in this function) io_buffer.c: In function ‘io_buffer_read_u64’: io_buffer.c:1885: error: ‘RB_IO_BUFFER_HOST_ENDIAN’ undeclared (first use in this function) io_buffer.c: In function ‘io_buffer_write_u64’: io_buffer.c:1885: error: ‘RB_IO_BUFFER_HOST_ENDIAN’ undeclared (first use in this function) io_buffer.c: In function ‘io_buffer_read_U64’: io_buffer.c:1886: error: ‘RB_IO_BUFFER_HOST_ENDIAN’ undeclared (first use in this function) io_buffer.c: In function ‘io_buffer_write_U64’: io_buffer.c:1886: error: ‘RB_IO_BUFFER_HOST_ENDIAN’ undeclared (first use in this function) io_buffer.c: In function ‘io_buffer_read_s64’: io_buffer.c:1887: error: ‘RB_IO_BUFFER_HOST_ENDIAN’ undeclared (first use in this function) io_buffer.c: In function ‘io_buffer_write_s64’: io_buffer.c:1887: error: ‘RB_IO_BUFFER_HOST_ENDIAN’ undeclared (first use in this function) io_buffer.c: In function ‘io_buffer_read_S64’: io_buffer.c:1888: error: ‘RB_IO_BUFFER_HOST_ENDIAN’ undeclared (first use in this function) io_buffer.c: In function ‘io_buffer_write_S64’: io_buffer.c:1888: error: ‘RB_IO_BUFFER_HOST_ENDIAN’ undeclared (first use in this function) io_buffer.c: In function ‘io_buffer_read_f32’: io_buffer.c:1890: error: ‘RB_IO_BUFFER_HOST_ENDIAN’ undeclared (first use in this function) io_buffer.c: In function ‘io_buffer_write_f32’: io_buffer.c:1890: error: ‘RB_IO_BUFFER_HOST_ENDIAN’ undeclared (first use in this function) io_buffer.c: In function ‘io_buffer_read_F32’: io_buffer.c:1891: error: ‘RB_IO_BUFFER_HOST_ENDIAN’ undeclared (first use in this function) io_buffer.c: In function ‘io_buffer_write_F32’: io_buffer.c:1891: error: ‘RB_IO_BUFFER_HOST_ENDIAN’ undeclared (first use in this function) io_buffer.c: In function ‘io_buffer_read_f64’: io_buffer.c:1892: error: ‘RB_IO_BUFFER_HOST_ENDIAN’ undeclared (first use in this function) io_buffer.c: In function ‘io_buffer_write_f64’: io_buffer.c:1892: error: ‘RB_IO_BUFFER_HOST_ENDIAN’ undeclared (first use in this function) io_buffer.c: In function ‘io_buffer_read_F64’: io_buffer.c:1893: error: ‘RB_IO_BUFFER_HOST_ENDIAN’ undeclared (first use in this function) io_buffer.c: In function ‘io_buffer_write_F64’: io_buffer.c:1893: error: ‘RB_IO_BUFFER_HOST_ENDIAN’ undeclared (first use in this function) io_buffer.c: In function ‘Init_IO_Buffer’: io_buffer.c:3702: error: ‘RB_IO_BUFFER_HOST_ENDIAN’ undeclared (first use in this function) At top level: cc1: warning: unrecognized command line option "-Wno-maybe-uninitialized" cc1: warning: unrecognized command line option "-Wno-tautological-compare" cc1: warning: unrecognized command line option "-Wno-self-assign" cc1: warning: unrecognized command line option "-Wno-parentheses-equality" cc1: warning: unrecognized command line option "-Wno-constant-logical-operand" cc1: warning: unrecognized command line option "-Wno-cast-function-type" make: *** [io_buffer.o] Error 1 -- https://bugs.ruby-lang.org/

6 10

[ruby-core:116114] [Ruby master Bug#20169] `GC.compact` can raises `EFAULT` on IO
by ko1 (Koichi Sasada) 15 Jul '24

15 Jul '24

Issue #20169 has been reported by ko1 (Koichi Sasada). ---------------------------------------- Bug #20169: `GC.compact` can raises `EFAULT` on IO https://bugs.ruby-lang.org/issues/20169 * Author: ko1 (Koichi Sasada) * Status: Open * Priority: Normal * Backport: 3.0: UNKNOWN, 3.1: UNKNOWN, 3.2: UNKNOWN, 3.3: UNKNOWN ---------------------------------------- 1. `GC.compact` introduces read barriers to detect read accesses to the pages. 2. I/O operations release GVL to pass the control while their execution, and another thread can call `GC.compact` (or auto compact feature I guess, but not checked yet). 3. Call `write(ptr)` can return `EFAULT` when `GC.compact` is running because `ptr` points read-barrier protected pages. Reproducible steps: Apply the following patch to increase possibility: ```patch diff --git a/io.c b/io.c index f6cd2c1a56..83d67ba2dc 100644 --- a/io.c +++ b/io.c @@ -1212,8 +1212,12 @@ internal_write_func(void *ptr) } } + int cnt = 0; retry: - do_write_retry(write(iis->fd, iis->buf, iis->capa)); + for (; cnt < 1000; cnt++) { + do_write_retry(write(iis->fd, iis->buf, iis->capa)); + if (result <= 0) break; + } if (result < 0 && !iis->nonblock) { int e = errno; ``` Run the following code: ```ruby t1 = Thread.new{ 10_000.times.map{"#{_1}"}; GC.compact while true } t2 = Thread.new{ i=0 $stdout.write "<#{i+=1}>" while true } t2.join ``` and ``` $ make run (snip) 4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4><4>#<Thread:0x00007fa61b4dd758 ../../src/trunk/test.rb:3 run> terminated with exception (report_on_exception is true): ../../src/trunk/test.rb:5:in `write': Bad address @ io_write - <STDOUT> (Errno::EFAULT) from ../../src/trunk/test.rb:5:in `block in <main>' ../../src/trunk/test.rb:5:in `write': Bad address @ io_write - <STDOUT> (Errno::EFAULT) from ../../src/trunk/test.rb:5:in `block in <main>' make: *** [uncommon.mk:1383: run] Error 1 ``` I think this is why we get `EFAULT` on CI. To increase possibilities running many busy processes (`ruby -e 'loop{}'` for example) will help (and on CI environment there are such busy processes accidentally). -- https://bugs.ruby-lang.org/

6 13

[ruby-core:116908] [Ruby master Bug#20292] Abort ruby by `String#initialize`
by ksss (Yuki Kurihara) 15 Jul '24

15 Jul '24

Issue #20292 has been reported by ksss (Yuki Kurihara). ---------------------------------------- Bug #20292: Abort ruby by `String#initialize` https://bugs.ruby-lang.org/issues/20292 * Author: ksss (Yuki Kurihara) * Status: Open * ruby -v: ruby 3.4.0dev (2024-02-22T06:43:46Z master e1c684e471) [arm64-darwin22] * Backport: 3.0: UNKNOWN, 3.1: UNKNOWN, 3.2: UNKNOWN, 3.3: UNKNOWN ---------------------------------------- I found a code that causes an anomaly. ``` $ uname -a Darwin 20208671n 22.6.0 Darwin Kernel Version 22.6.0: Wed Oct 4 21:26:55 PDT 2023; root:xnu-8796.141.3.701.17~4/RELEASE_ARM64_T6020 arm64 $ ruby -v ruby 3.4.0dev (2024-02-22T06:43:46Z master e1c684e471) [arm64-darwin22] $ ruby -e '100.times { "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa".__send__(:initialize, capacity: -1) }' ruby(13817,0x1f22b2080) malloc: Heap corruption detected, free list is damaged at 0x600000e27bc0 *** Incorrect guard value: 29830901955328 ruby(13817,0x1f22b2080) malloc: *** set a breakpoint in malloc_error_break to debug [1] 13817 abort ruby -e $ ruby -e '100.times { "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa".__send__(:initialize, capacity: 0) }' # never return .... ``` My research indicates the following conditions. - String must be at least 64 characters long - Call `#initialize` with capacity: 0 or less - Repeat this several times. -- https://bugs.ruby-lang.org/

4 4

[ruby-core:116930] [Ruby master Bug#20296] Complex(:sym, exception: false) generate exception with weird timing
by MaxLap (Maxime Lapointe) 15 Jul '24

15 Jul '24

Issue #20296 has been reported by MaxLap (Maxime Lapointe). ---------------------------------------- Bug #20296: Complex(:sym, exception: false) generate exception with weird timing https://bugs.ruby-lang.org/issues/20296 * Author: MaxLap (Maxime Lapointe) * Status: Open * Backport: 3.0: UNKNOWN, 3.1: UNKNOWN, 3.2: UNKNOWN, 3.3: UNKNOWN ---------------------------------------- Happens in Ruby 3.2.2, 3.2.3 and 3.3.0 (didn't test others). Add this to a file and execute it as-is: ``` 3.times { Complex(:sym, exception: false) } puts "hi" ``` The output will be: ``` $ ruby local2.rb hi local2.rb:1:in `Complex': can't convert Array into Complex (TypeError) 3.times { Complex([], exception: false) } ^^^^^^^^^^^^^^^^^^^^ from local2.rb:1:in `block in <main>' from local2.rb:1:in `times' from local2.rb:1:in `<main>' local2.rb:1:in `Complex': can't convert Array into Complex (TypeError) 3.times { Complex([], exception: false) } ^^^^^^^^^^^^^^^^^^^^ from local2.rb:1:in `block in <main>' from local2.rb:1:in `times' from local2.rb:1:in `<main>' local2.rb:1:in `Complex': can't convert Array into Complex (TypeError) 3.times { Complex([], exception: false) } ^^^^^^^^^^^^^^^^^^^^ from local2.rb:1:in `block in <main>' from local2.rb:1:in `times' from local2.rb:1:in `<main>' ``` So the exception is still reported (instead of being swallow), but the puts, which is after, had the time to execute first. I can even put a `sleep()` in between and the exceptions are still reported after the puts. In a IRB, there is no exception at all and everything seems to work as expected. Happens if I pass in an Array too. Doesn't seem to happen with bad strings. There is also a quadratic behavior somewhere in there. Running the code in an IRB (which doesn't show exceptions), if you double the number of `times` in the loop, it takes 4x as long. This is probably related... -- https://bugs.ruby-lang.org/

3 2

[ruby-core:118590] [Ruby master Feature#17279] Allow a negative step in Range#step with a block
by zverok (Victor Shepelev) 14 Jul '24

14 Jul '24

Issue #17279 has been updated by zverok (Victor Shepelev). FWIW, my PR with adjusting `Range#step` behavior fixes this, too: https://github.com/ruby/ruby/pull/7444 (as part of the unificatioin/simplification process): From the PR description: > Consistent support for negative step: ```ruby p (1..-10).step(-3).to_a #=> [1, -2, -5, -8] -- ArithmeticSequence backward iteration, on Ruby 3.3 and my code (1..-10).step(-3) { p _1 } # Ruby 3.3: step can't be negative (ArgumentError) -- inconsistent with ArithmeticSequence behavior # my code: prints 1, -2, -5, -8, consistent with ArithmeticSequence ``` ---------------------------------------- Feature #17279: Allow a negative step in Range#step with a block https://bugs.ruby-lang.org/issues/17279#change-109117 * Author: mrkn (Kenta Murata) * Status: Assigned * Assignee: matz (Yukihiro Matsumoto) ---------------------------------------- `Range#step` prohibits a negative step when a block is given. ``` >> (6..3).step(-1) {|i| p i } Traceback (most recent call last): 5: from /home/mrkn/.rbenv/versions/2.7/bin/irb:23:in `<main>' 4: from /home/mrkn/.rbenv/versions/2.7/bin/irb:23:in `load' 3: from /home/mrkn/.rbenv/versions/2.7.1/lib/ruby/gems/2.7.0/gems/irb-1.2.4/exe/irb:11:in `<top (required)>' 2: from (irb):1 1: from (irb):1:in `step' ArgumentError (step can't be negative) ``` But `Range#step` allows a negative step when it is called without a block. In this case, `Range#step` creates an ArithmeticSequence, and `ArithmeticSequence#each` can iterate with a negative step. ``` >> (6..3).step(-1).each {|i| p i } 6 5 4 3 => ((6..3).step(-1)) ``` I think the prohibition of a negative step in `Range#step` has already been meaningless, so it may be better to permit it for consistency. -- https://bugs.ruby-lang.org/

1 0

[ruby-core:116965] [Ruby master Bug#20307] `Hash#update` from compare_by_identity hash can have unfrozen string keys
by nobu (Nobuyoshi Nakada) 13 Jul '24

13 Jul '24

Issue #20307 has been reported by nobu (Nobuyoshi Nakada). ---------------------------------------- Bug #20307: `Hash#update` from compare_by_identity hash can have unfrozen string keys https://bugs.ruby-lang.org/issues/20307 * Author: nobu (Nobuyoshi Nakada) * Status: Open * Backport: 3.0: REQUIRED, 3.1: REQUIRED, 3.2: REQUIRED, 3.3: REQUIRED ---------------------------------------- I don't think this behavior is expected. ```ruby i = Hash.new.compare_by_identity k = "a" i[k] = 0 h = {}.update(i) p h["a"] # => 0 k.upcase! p h.fetch(k) # in 'Hash#fetch': key not found: "A" (KeyError) ``` -- https://bugs.ruby-lang.org/

5 6

[ruby-core:116643] [Ruby master Bug#20250] Crash with "Object ID seen, but not in mapping table: proc" error
by zetter (Chris Zetter) 13 Jul '24

13 Jul '24

Issue #20250 has been reported by zetter (Chris Zetter). ---------------------------------------- Bug #20250: Crash with "Object ID seen, but not in mapping table: proc" error https://bugs.ruby-lang.org/issues/20250 * Author: zetter (Chris Zetter) * Status: Open * Priority: Normal * ruby -v: ruby 3.3.0 (2023-12-25 revision 5124f9ac75) [arm64-darwin23] * Backport: 3.0: UNKNOWN, 3.1: UNKNOWN, 3.2: UNKNOWN, 3.3: UNKNOWN ---------------------------------------- Hello, I experienced a crash which I was able to reliably reproduce with the following: ``` ruby require 'bundler/inline' gemfile(true) do source 'https://rubygems.org' gem 'activesupport', '7.1.3' end require 'active_support' logger = ActiveSupport::Logger.new('log/log.log', 1, 100 * 1024 * 1024) logger.formatter = proc {|_, _, _, message| "#{message}\n" } logger = ActiveSupport::TaggedLogging.new(logger) logger.tagged("TAG").info "hello" logger.tagged("TAG").info "hello" # usually crashes here GC.start # sometimes need to trigger crash with GC ``` It looks like this is caused by the interaction with the formatter proc and the tagged logging features of activesupport. Let me know if a more minimal example would be useful. I can reproduce this with: * ruby 3.3.0 on arm64-darwin23 * ruby 3.3.0 on x86_64-linux * latest ruby HEAD (5e12b75716) on arm64-darwin23 I cannot reproduce this on ruby 3.2.2. Thanks for any help ---Files-------------------------------- ruby-2024-02-09-092829.ips (13.2 KB) output.log (142 KB) crash_test_2.rb (465 Bytes) -- https://bugs.ruby-lang.org/

4 5

[ruby-core:118513] [Ruby master Bug#20619] Backport bd583ca645ea348a2894f4227fcb1af650ee8dec to ruby 3.3
by ko1 (Koichi Sasada) 13 Jul '24

13 Jul '24

Issue #20619 has been reported by ko1 (Koichi Sasada). ---------------------------------------- Bug #20619: Backport bd583ca645ea348a2894f4227fcb1af650ee8dec to ruby 3.3 https://bugs.ruby-lang.org/issues/20619 * Author: ko1 (Koichi Sasada) * Status: Closed * Target version: 3.3 * Backport: 3.1: DONTNEED, 3.2: DONTNEED, 3.3: REQUIRED ---------------------------------------- commit:bd583ca645ea348a2894f4227fcb1af650ee8dec (https://github.com/ruby/ruby/pull/11037) should be merged. -- https://bugs.ruby-lang.org/

2 1

[ruby-core:118561] [Ruby master Bug#20629] Ruby infinite loop when raising+rescuing inside a RUBY_INTERNAL_EVENT_NEWOBJ
by ivoanjo (Ivo Anjo) 12 Jul '24

12 Jul '24

Issue #20629 has been reported by ivoanjo (Ivo Anjo). ---------------------------------------- Bug #20629: Ruby infinite loop when raising+rescuing inside a RUBY_INTERNAL_EVENT_NEWOBJ https://bugs.ruby-lang.org/issues/20629 * Author: ivoanjo (Ivo Anjo) * Status: Open * ruby -v: ruby 3.3.4 (2024-07-09 revision be1089c8ec) [x86_64-linux] * Backport: 3.1: UNKNOWN, 3.2: UNKNOWN, 3.3: UNKNOWN ---------------------------------------- ### Background Hey! While working on the [Datadog Ruby Profiler](https://github.com/datadog/dd-trace-rb) I ran into a weird situation where the Ruby VM locks up as it gets stuck in an infinite loop in vm_sync.c: ```c while (recorded_lock_rec < current_lock_rec) { RB_VM_LOCK_LEAVE_LEV(&current_lock_rec); } ``` ...because leave is this: ```c static inline void rb_vm_lock_leave(unsigned int *lev, const char *file, int line) { if (rb_multi_ractor_p()) { rb_vm_lock_leave_body(lev APPEND_LOCATION_PARAMS); } } ``` ...and `rb_multi_ractor_p()` is always `false` as I'm not using Ractors. ### How to reproduce (I include a zip with all the files needed to reproduce this, no need to copy-paste from here!) Inside a native extension, I'm creating a `RUBY_INTERNAL_EVENT_NEWOBJ`. Inside it, I'm raising and catching an exception, to simulate a (handled) bug in our code: ```c #include <ruby.h> #include <ruby/thread.h> #include <ruby/thread_native.h> #include <ruby/debug.h> static VALUE raise_exception(VALUE unused) { fprintf(stderr, "Raising!\n"); rb_raise(rb_eRuntimeError, "This is a test!"); } static VALUE ignore_failure(VALUE self_instance, VALUE exception) { fprintf(stderr, "Caught!\n"); return Qnil; } static void on_newobj_event(VALUE tracepoint_data, void *unused) { fprintf(stderr, "Inside on_newobj_event\n"); // VALUE klass = rb_class_of(rb_tracearg_object(rb_tracearg_from_tracepoint(tracepoint_data))); // rb_p(klass); rb_rescue2( raise_exception, Qnil, ignore_failure, Qnil, rb_eException, // rb_eException is the base class of all Ruby exceptions 0 // Required by API to be the last argument ); fprintf(stderr, "Finished on_newobj_event\n"); } static VALUE install_tracepoint(VALUE unused) { VALUE tp = rb_tracepoint_new(Qnil, RUBY_INTERNAL_EVENT_NEWOBJ, on_newobj_event, NULL /* unused */); rb_tracepoint_enable(tp); fprintf(stderr, "Tracepoint installed!\n"); return Qnil; } void Init_exception_bug_repro(void) { rb_define_singleton_method(rb_define_module("Repro"), "install_tracepoint", install_tracepoint, 0); } ``` And my test Ruby script is: ```ruby require_relative './exception_bug_repro' Repro.install_tracepoint while true begin raise('hello') rescue print '.' end end ``` Here's how running it looks: ``` $ bundle exec rake clean compile $ bundle exec ruby testcase.rb Tracepoint installed! Inside on_newobj_event Raising! Caught! Finished on_newobj_event Inside on_newobj_event Raising! Caught! Finished on_newobj_event Inside on_newobj_event Raising! Caught! Finished on_newobj_event Inside on_newobj_event Raising! Caught! Finished on_newobj_event Inside on_newobj_event Raising! # <--- Infinite loop ``` That is, we should always see the raise -> caught, but sometimes the raise doesn't go very well, and from then Ruby eventually locks up. Here's the full backtrace from gdb: ``` #0 rb_multi_ractor_p () at ruby-3_3/vm_sync.h:39 #1 0x00005fdde2534665 in rb_vm_lock_leave (lev=0x7ffc2f1b5f70, file=0x5fdde2783bb5 "vm_sync.c", line=231) at ruby-3_3/vm_sync.h:92 #2 0x00005fdde2534b28 in rb_ec_vm_lock_rec_release (ec=0x5fdde3e9b570, recorded_lock_rec=0, current_lock_rec=1) at vm_sync.c:231 #3 0x00005fdde23088f2 in rb_ec_vm_lock_rec_check (ec=0x5fdde3e9b570, recorded_lock_rec=0) at ruby-3_3/eval_intern.h:144 #4 0x00005fdde2308937 in rb_ec_tag_state (ec=0x5fdde3e9b570) at ruby-3_3/eval_intern.h:155 #5 0x00005fdde230c71e in setup_exception (ec=0x5fdde3e9b570, tag=6, mesg=126196681963080, cause=4) at eval.c:564 #6 0x00005fdde230cdc5 in rb_longjmp (ec=0x5fdde3e9b570, tag=6, mesg=126196681963080, cause=4) at eval.c:660 #7 0x00005fdde230d398 in rb_raise_jump (mesg=126196681963080, cause=36) at eval.c:853 #8 0x00005fdde230d0cc in rb_f_raise (argc=1, argv=0x72c690294048) at eval.c:752 #9 0x00005fdde230d0f4 in f_raise (c=1, v=0x72c690294048, _=126197147684320) at eval.c:789 #10 0x00005fdde2502d24 in ractor_safe_call_cfunc_m1 (recv=126197147684320, argc=1, argv=0x72c690294048, func=0x5fdde230d0cc <f_raise>) at ruby-3_3/vm_insnhelper.c:3311 #11 0x00005fdde250395b in vm_call_cfunc_with_frame_ (ec=0x5fdde3e9b570, reg_cfp=0x72c690393fa0, calling=0x7ffc2f1b66a0, argc=1, argv=0x72c690294048, stack_bottom=0x72c690294040) at ruby-3_3/vm_insnhelper.c:3502 #12 0x00005fdde2503b30 in vm_call_cfunc_with_frame (ec=0x5fdde3e9b570, reg_cfp=0x72c690393fa0, calling=0x7ffc2f1b66a0) at ruby-3_3/vm_insnhelper.c:3530 #13 0x00005fdde2503c48 in vm_call_cfunc_other (ec=0x5fdde3e9b570, reg_cfp=0x72c690393fa0, calling=0x7ffc2f1b66a0) at ruby-3_3/vm_insnhelper.c:3556 #14 0x00005fdde2504070 in vm_call_cfunc (ec=0x5fdde3e9b570, reg_cfp=0x72c690393fa0, calling=0x7ffc2f1b66a0) at ruby-3_3/vm_insnhelper.c:3638 #15 0x00005fdde2506b9f in vm_call_method_each_type (ec=0x5fdde3e9b570, cfp=0x72c690393fa0, calling=0x7ffc2f1b66a0) at ruby-3_3/vm_insnhelper.c:4429 #16 0x00005fdde25076e0 in vm_call_method (ec=0x5fdde3e9b570, cfp=0x72c690393fa0, calling=0x7ffc2f1b66a0) at ruby-3_3/vm_insnhelper.c:4566 #17 0x00005fdde250784d in vm_call_general (ec=0x5fdde3e9b570, reg_cfp=0x72c690393fa0, calling=0x7ffc2f1b66a0) at ruby-3_3/vm_insnhelper.c:4599 #18 0x00005fdde2509ed2 in vm_sendish (ec=0x5fdde3e9b570, reg_cfp=0x72c690393fa0, cd=0x5fdde3f4f7e0, block_handler=0, method_explorer=mexp_search_method) at ruby-3_3/vm_insnhelper.c:5593 #19 0x00005fdde25114b9 in vm_exec_core (ec=0x5fdde3e9b570) at ruby-3_3/insns.def:834 #20 0x00005fdde252829b in rb_vm_exec (ec=0x5fdde3e9b570) at vm.c:2486 #21 0x00005fdde25290fa in rb_iseq_eval_main (iseq=0x72c674667ad0) at vm.c:2752 #22 0x00005fdde230be59 in rb_ec_exec_node (ec=0x5fdde3e9b570, n=0x72c674667ad0) at eval.c:287 #23 0x00005fdde230bfd6 in ruby_run_node (n=0x72c674667ad0) at eval.c:328 #24 0x00005fdde2307245 in rb_main (argc=2, argv=0x7ffc2f1b7618) at ./main.c:39 #25 0x00005fdde23072bd in main (argc=2, argv=0x7ffc2f1b7618) at ./main.c:58 ``` In this minimal reproducer, usually it's when allocating a `Thread::Backtrace` object that triggers the issue, but on the original codebase where I was experimenting I saw this issue even with other objects (I think??). To be honest, maybe the problem is on the extension side -- this code assumes it's ok to raise/rescue in the `RUBY_INTERNAL_EVENT_NEWOBJ` tracepoint as it usually works fine, but maybe it's not? ---Files-------------------------------- exception-bug-repo.zip (3.79 KB) -- https://bugs.ruby-lang.org/

2 2