Since the fix for Bug #21441, ec->cfp is set to NULL before clearing
the VM stack so that thread_profile_frames() called from a signal
handler returns early instead of walking a stack that is no longer set
on the ec. But these are plain stores to distinct fields of the same
struct, so the compiler is free to reorder them.
With Apple clang 21 and GCC 16 on macOS arm64, and with clang 18 and
GCC 13 on Linux aarch64, both stores of rb_ec_set_vm_stack() were
combined into a single stp emitted before the store to ec->cfp
(vm_stack is at offset 0, vm_stack_size at 0x8 and cfp at 0x10):
stp xzr, xzr, [x0] ; ec->vm_stack = NULL, vm_stack_size = 0
str xzr, [x0, #0x10] ; ec->cfp = NULL
ret
So a signal in between would see a non-NULL ec->cfp with a NULL
ec->vm_stack. Stopping there in lldb when a thread terminates and
calling rb_profile_frames() segfaults in thread_profile_frames(), as
end_cfp is computed from the NULL ec->vm_stack and the loop walks past
the control frames. With this change, ec->cfp is already NULL at that
point and rb_profile_frames() returns 0.
Like in vm_push_frame(), use atomic_signal_fence() to keep the store to
ec->cfp first. With it, all these compilers emit the store to ec->cfp no
later than the stores clearing the stack. The rest of vm.o is unchanged
with Apple clang 21, GCC 16 and GCC 13 on aarch64. Clang 21 and 18:
stp xzr, xzr, [x0, #0x8] ; ec->vm_stack_size = 0, cfp = NULL
str xzr, [x0] ; ec->vm_stack = NULL
ret
GCC 16 and 13:
str xzr, [x0, #0x10] ; ec->cfp = NULL
stp xzr, xzr, [x0] ; ec->vm_stack = NULL, vm_stack_size = 0
ret
With GCC 13 on x86_64, the store to ec->cfp is already emitted first
without the fence, and vm.o is identical with and without it:
endbr64
movq $0x0, 0x10(%rdi) ; ec->cfp = NULL
movq $0x0, (%rdi) ; ec->vm_stack = NULL
movq $0x0, 0x8(%rdi) ; ec->vm_stack_size = 0
ret
[Bug #22477]
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
test_aspawn_too_long_path just wants to verify that passing a very long
string to spawn does not cause a crash. Use an input string that is long
enough for this purpose and does not ends up with spawning too many
processes.
This change makes this test run much faster on systems where /bin/sh is
bash, which retries fork() several times with long delays before giving
up.
(cherry picked from commit fa8df43bd1)
On my computer, setting RLIMIT_NPROC to the low value of 2048 prevents
the forked process from creating any new native threads and it causes
make test-all to hang forever.
Remove it, as it does not seem to serve any purpose here.
(cherry picked from commit 7d1be0888a)
With a subexp call, the result of unset_addr_list_fix overwrote the error, so a pattern over the limit compiled without raising RegexpError.
Follow-up to cf35145fc4.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
ruby/prism's ruby-4.0 branch triggers sync_default_gems.yml on ruby_4_0
since ruby/prism#4005, but the sync only fetched prism's main branch.
The ruby-4.0 commits were never fetched, so every sync found no commits
to pick and finished successfully without syncing anything.
I've been seeing this for a while now:
```
../../../../ext/prism/../../src/memchr.c: In function ‘pm_memchr’:
../../../../ext/prism/../../src/memchr.c:35:16: warning: return discards ‘const’ qualifier from pointer target type [-Wdiscarded-qualifiers]
35 | return memchr(memory, character, number);
| ^~~~~~
```
https://github.com/ruby/prism/commit/a3b1f10dbc
(cherry picked from commit ce398d7545)
A range of single character Strings told that it includes its end
even when it is empty, since the end was not compared with the begin.
("z".."a").to_a #=> []
("z".."a").include?("a") #=> true
(cherry picked from commit 1bf9130943)
prefork() flushed rb_stdout and rb_stderr, which are the main Ractor's IOs.
From another Ractor that takes the write_lock of an IO the Ractor does not
own. When a main Ractor thread holds it (any thread logging to $stdout will
do), the spawning Ractor's only thread sleeps on the mutex and
rb_check_deadlock, which only looks at the Ractor's own threads, raises
"No live threads left. Deadlock?" (fatal) from Kernel#spawn/system/popen.
Use rb_ractor_stdout()/rb_ractor_stderr() so each Ractor flushes its own.
[Bug #22434]
(cherry picked from commit 28f2eae4cf)
Interrupts are checked at arbitrary points, including while a jump
(break/next/redo/return/raise/throw) is being propagated, and the payload
of that jump lives in ec->errinfo.
Handling an interrupt can run Ruby code: rb_signal_exec() runs a trap
handler and threadptr_interrupt_exec_exec() runs interrupt_exec tasks.
Neither preserves ec->errinfo, and any rescue or ensure clause that code
enters sets it to nil.
Thread.handle_interrupt checks for interrupts after its block returns but
before re-raising the jump that left the block. If a signal handler runs
in that window the throw data is replaced with nil and the VM dereferences
nil as a vm_throw_data:
THROW_DATA_CATCH_FRAME (obj=0x4) at vm_insnhelper.h:217
vm_exec_handle_exception (ec=..., errinfo=4) at vm.c:2880
vm_exec_loop / rb_vm_exec at vm.c:2829 / vm.c:2812
invoke_iseq_block_from_c / rb_yield / rb_ary_each
rb_postponed_job_flush() and rb_exec_event_hooks() already save and restore
ec->errinfo around the Ruby code they run; do the same for the rest of
rb_threadptr_execute_interrupts(). The paths that set errinfo on purpose
(rb_exc_raise, rb_threadptr_to_kill) jump out of the function and so do not
reach the restore.
C23 has qualifier-preserving standard library functions, so calling
strchr with a `const char *` will return a `const char *`. We can change
the type of the local variables because we don't mutate the strings.
(cherry picked from commit 42b3cdc51a)
A thread that is waiting for the fork lock when another thread forks
exists only in the parent, but on macOS the child's copy of the lock
still counts it. After the child unlocks its copy,
pthread_rwlock_destroy() and pthread_rwlock_init() fail with EBUSY, so
the child aborts with
"[BUG] pthread_rwlock_destroy: Resource busy (EBUSY)". Lookup threads
wait for the lock whenever a fork holds it, so a process that resolves
names in background threads while it forks loses a child now and then.
Reset the child's copy by assigning PTHREAD_RWLOCK_INITIALIZER, as
rb_internal_thread_event_hooks_rw_lock_atfork() does for its lock, and
do not unlock it first. With a waiter counted, macOS's unlock enters the
kernel to wake it, and the state that leaves behind for the lock's
address survives the assignment, so a later waiter in the child can
sleep forever although no thread holds the lock. FreeBSD already skipped
the unlock in the child (dc8c382266), and on glibc the child's unlock
does not release the write lock anyway.
[Bug #22445]
A getaddrinfo thread calls getaddrinfo(3) inside rb_thread_prevent_fork,
but on glibc its thread exit takes the resolver configuration lock again
after the thread has released the fork lock. A fork that lands then
leaves the child with that lock held, and every later name lookup in the
child blocks forever. getnameinfo(3) attaches the same resolver state,
even for a name in /etc/hosts, and its thread ran outside
rb_thread_prevent_fork altogether.
Make the lookup threads joinable and have fork join any that may still
be exiting. As a lookup thread finishes, still inside
rb_thread_prevent_fork, it swaps itself into a single slot with an
atomic exchange and joins the thread it displaced, which has usually
exited by then, and rb_thread_acquire_fork_lock joins whatever is left
in the slot. Once lookups settle, at most one finished thread is left
unjoined. The slot holds pthread_t as a pointer, with NULL for empty,
which assumes pthread_t is scalar and never 0.
getnameinfo threads now run within rb_thread_prevent_fork and register
as they exit, as getaddrinfo threads do, and where getnameinfo runs
without a helper thread it is called within rb_thread_prevent_fork too.
So fork also waits for a reverse lookup in progress, as it already did
for a forward one. raddrinfo_pthread_create now always creates joinable
threads, which drops its pthread_detach fallback for platforms without
pthread_attr_setdetachstate.
The slot is kept across fork, so every fork path that keeps running
Ruby code has to take the fork lock. On the BSDs, Process.daemon forks
with daemon(3) and did not take it, so the daemon would inherit the
registered thread, and its next fork would abort in pthread_join with
ESRCH. Take and release the fork lock around daemon(3) as rb_fork_ruby
does. daemon(3) can fail in the child as well as in the parent, so
compare pids to tell which process returned.
[Bug #22406]
(cherry picked from commit 44f7c6b00d)
box_entry_free could run before other T_DATA objects are finalized in
rb_objspace_call_finalizer, and FreeLibrary there unmapped the
rb_data_type_t and dfree functions that later finalization still
dereferences, crashing any boxed process that leaves extension objects
alive at exit. Keep the copied DLLs loaded until ruby_vm_destruct, then
unload and delete them.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Timezone objects can return abbreviations in encodings beyond the
locale. Respect their encoding when formatting `%Z` to avoid
misinterpreting their bytes.
Array#values_at can execute arbitrary code during execution which may
modify the source array. This can cause a buffer overflow if the new length
of the array has decreased. The following script crashes:
a = (0..100_000).to_a
obj = Object.new
obj.define_singleton_method(:begin) do
a.clear
0
end
obj.define_singleton_method(:end) { 10_000 }
obj.define_singleton_method(:exclude_end?) { false }
puts a.values_at(obj).length
FIX2INT does not call to_int where int and long have the same size, so `method(:puts).curry(1r)` waited for a garbage number of arguments on Windows. Convert the arity once and keep the result as a Fixnum, which also limits the to_int call to one.
IO#gets first searches for separator candidates by looking for the last
byte of the record separator, and then checks whether it is an actual
match.
If a candidate is found at limit-th byte of the stream but is rejected
because it does not start at a character boundary, the limit check is
incorrectly skipped.
[Bug #22342]
The Enumerable version did not crash an unfixed build on Linux. Kernel
does on both Linux and macOS, but only while nothing in the main box has
written to it yet, and loading test/unit already has. So run the code
through assert_in_out_err instead of assert_separately.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
rb_class_duplicate_classext builds the whole copy first and the caller only
then puts it in the class's table, so a GC in the middle collects the tables
it has made and marking the class afterwards walks into a freed object. Put
the classext in the table first instead. Marking it while it is still half
built is fine, since ZALLOC leaves every field zero.
$ RUBY_BOX=1 ruby --disable-gems -e 'GC.stress = true; module Enumerable; def m; end; end'
-e:1: [BUG] try to mark T_NONE object (obj: out-of-heap:0x..., parent: ... T_MODULE/Module Enumerable)
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
`RUBY_FREE_AT_EXIT=1` frees every object in an arbitrary order, so a
class a box refers to can go before the box itself, and freeing a class
already frees its classext for that box. The box then reached rb_bug
with a T_NONE object, and the report crashed and looped in the signal
handler instead of aborting, so the process spun at exit.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
ensure_origin read the m_tbl before the first write created the box-local classext, so the origin ICLASS captured the prime m_tbl while the subclass ICLASSes referred the box-local copy made by the CoW. rb_prepend_module then missed the shared m_tbl check and inserted the prepended module after the target module instead of before it.
Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
With more than 31 keyword arguments, tracking unevaluated defaults
uses an integer-keyed Hash whose collisions can invoke methods when
`Integer#==` is refined or redefined, corrupting argument locals above
`cfp->sp`. Use identity hashes to prevent these calls during argument
setup and `checkkeyword` execution.
Making `rb_any_cmp` compare special constants directly would also
bypass redefined `eql?` methods in ordinary hashes. Limit identity
comparison to the internal keyword tracking hashes to preserve that
behavior for now.
[Bug #22330]
The following script demonstrates a use-afer-free where we see corruption:
s = "あ" * 10_000
s.encode("US-ASCII", fallback: proc { |c| s.clear; "?" })
Raises may different errors such as:
"\xCB" followed by "\xF5" on UTF-8 (Encoding::InvalidByteSequenceError)
"\xF4" followed by "r" on UTF-8 (Encoding::InvalidByteSequenceError)
"\x8B" on UTF-8 (Encoding::InvalidByteSequenceError)
The flag also covers drive-relative and root-relative targets, which are
not absolute paths. "independent" describes that the target is not
interpreted relative to the link's directory.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
A target starting with a single backslash is relative to the root of the
current drive, not to the link's directory. Treat any backslash-leading
target (root-relative or UNC) as non-relative when deciding the directory
flag; otherwise w32_symlink builds a malformed path and fails to detect
the directory attribute.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
File.symlink decided whether to create a directory symlink by calling
GetFileAttributesW on the target as given, which resolves a relative
target against the current directory instead of the link's directory.
When the two differ, a relative target pointing at a directory became a
file symlink, so Dir operations on it failed. Resolve relative targets
against the link's directory before testing for a directory.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
[Bug #22334]
If we return a with a keyword splat, then each element does not necessarily
push two values onto the stack, so we cannot compile this as a simple sequence
of key-value pairs combined with newhash. The following script reproduces
the issue and crashes:
def r
return **{foo: 1}
end
puts r()
[Bug #22319]
Array#zip triggers an use-after-free it does not account for when the call
to rb_check_array_type modifies the source array. The following script crashes:
a = (1..100_000).to_a
evil = Object.new; $a = a
def evil.to_ary; $a.clear; [1,2,3]; end
a.zip(evil)
[Bug #22318]
Array#flatten triggers an use-after-free it does not account for when the
call to rb_check_array_type modifies the source array. The following script
crashes:
a = (1..100_000).to_a
$a = a
bad = Object.new
def bad.to_ary; $a.clear; [1,2,3]; end
a << bad
a.flatten
