Fix memory overread in registry.rb
The terminator is not actually getting filled in; we're simply passing
(two) bytes of empty memory as the NUL terminator. This can lead to
garbage characters getting written to registry values.
Fix this by explicitly putting a WCHAR_NUL character into the string to
be sent to the registry API, like we do in the MULTI_SZ case.
[Bug #20096]
---
ext/win32/lib/win32/registry.rb | 7 ++-----
1 file changed, 2 insertions(+), 5 deletions(-)
* Merge RubyGems-3.5.4 and Bundler-2.5.4
* Merge RubyGems-3.5.5 and Bundler-2.5.5
* Make tests play with upstream Ruby tests
CI broke in https://github.com/ruby/ruby/pull/9604 because if any Ruby
tests run `require 'net/http'`, they will pollute the
`$LOADED_FEATURES` for the RubyGems tests. We can fix this by renaming
the test default gem from `net-http` to `my-http`.
See https://github.com/rubygems/rubygems/pull/7379#issuecomment-1901241299
for more details.
---------
Co-authored-by: Stan Hu <stanhu@gmail.com>
[Bug #20085] Use consistent default options for `-mbranch-protection`
We need to use the same options for both C compiler and assembler
when `-mbranch-protection` is guessed by configure. Otherwise,
`coroutine/arm64/Context.{h,S}` will use incompatible PAC strategies.
* YJIT: reduce default exec mem size to 48MiB based
Based on user feedback from @jhawthorn and others.
Better for small and memory-constrained deployments.
NOTE: This commit should be included in the next Ruby 3.3.x point
release. @xrxr should we tag someone specific?
* YJIT: Update yjit.md about mem size (#9687)
---------
Co-authored-by: Takashi Kokubun <takashikkbn@gmail.com>
Avoid reading unused lvars in Primitive.cexpr
Previously on builds with optimizations disabled, this could result in
an out of bounds read. When we had all of:
* built with -O0
* Leaf builtin
* Primitive.mandatory_only
* "no args builtin", called by vm_call_single_noarg_inline_builti
* The stack is escaped to the heap via binding or a proc
This is because mk_builtin_loader generated reads for all locals
regardless of whether they were used and in the case we generated a
mandatory_only iseq that would include more variables than were actually
available.
On optimized builds, the invalid accesses would be optimized away, and
this also was often unnoticed as the invalid access would just hit
another part of the stack unless it had been escaped to the heap.
The fix here is imperfect, as this could have false positives, but since
Primitive.cexpr! is only available within the cruby codebase itself
that's probably fine as a proper fix would be much more challenging (the
only false positives we found were in rjit.rb).
Fixes [Bug #20178]
Co-authored-by: Adam Hess <HParker@github.com>
---
bootstraptest/test_method.rb | 9 +++++++++
tool/mk_builtin_loader.rb | 6 ++++++
2 files changed, 15 insertions(+)
Remove setaffinity of pthread for getaddrinfo
It looks like `sched_getcpu(3)` returns a strange number on some
(virtual?) environments.
I decided to remove the setaffinity mechanism because the performance
does not appear to degrade on a quick benchmark even if removed.
[Bug #20172]
---
ext/socket/extconf.rb | 2 --
ext/socket/raddrinfo.c | 48 ++++--------------------------------------------
2 files changed, 4 insertions(+), 46 deletions(-)
merge revision(s) 82b57d7bfeefd717c10f7a5a3484aca6b3e708a3: [Backport #20162]
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 #{$$}`
end
Before:
23264
33600
42672
52160
61600
71728
81056
90528
100560
109840
After:
14752
14816
15584
15584
15664
15664
15664
15664
15664
15664
---
object.c | 3 +--
test/ruby/test_shapes.rb | 13 +++++++++++++
2 files changed, 14 insertions(+), 2 deletions(-)
* Fix GC.measure_total_time regression
Commit 93ac7405b8 introduced a regression
where measurements would still be taken after setting
GC.measure_total_time = false.
Fixes [Bug #20157]
* Add test case for GC.measure_total_time
---------
Co-authored-by: Rian McGuire <rian@rian.id.au>
Memory leak when duplicating identhash
[Bug #20145]
Before this commit, both copy_compare_by_id and hash_copy will create a
copy of the ST table, so the ST table created in copy_compare_by_id will
be leaked.
h = { 1 => 2 }.compare_by_identity
10.times do
1_000_000.times do
h.select { false }
end
puts `ps -o rss= -p #{$$}`
end
Before:
110736
204352
300272
395520
460704
476736
542000
604704
682624
770528
After:
15504
16048
16144
16256
16320
16320
16752
16752
16752
16752
---
hash.c | 10 +++++++++-
test/ruby/test_hash.rb | 10 ++++++++++
2 files changed, 19 insertions(+), 1 deletion(-)
Use index for referring to symbols in `args` rule instead of named
references
In `args: args ',' arg_splat`, `args` is not unique name.
Currently the associated rule is interpreted as
`$$ = rest_arg_append(p, $$, $3, &@$);`.
The action works as expected because `$$` is initialized with
`$1` before each action is executed.
However it's misleading then change to use index.
---
parse.y | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
* Bump up net-ftp to 0.3.4
* Bump up net-smtp to 0.4.0.1
* Bump up net-imap to 0.4.9.1
* [ruby/net-http] Renew test certificates
The private key is replaced with a public known test key published at
[RFC 9500].
Also lifetime has been extended to 10 years from 4 years.
[RFC 9500]: https://www.rfc-editor.org/rfc/rfc9500.htmlhttps://github.com/ruby/net-http/commit/4ab6c4a500
* Bump up net-http to 0.4.1
---------
Co-authored-by: Sorah Fukumori <her@sorah.jp>
RUBY_FL_PROMOTED is used by the garbage collector to track when an
object becomes promoted to the old generation. rb_obj_setup must not
copy that flag over because then it may become out-of-sync with the age
of the object.
This fixes a bug in Method#clone where the cloned Method object may get
RUBY_FL_PROMOTED incorrectly set.
The test fails when RGENGC_CHECK_MODE is turned on:
1) Failure:
TestSymbol#test_inspect_under_gc_compact_stress [test/ruby/test_symbol.rb:123]:
<":testing"> expected but was
<":\x00\x00\x00\x00\x00\x00\x00">.
The for loops for marking and reference updating declaratively marked
TypedData objects did not mark/reference update the very last element.
When RGENGC_CHECK_MODE is turned on, this caused the test in Enumerator
to fail with:
tool/lib/test/unit/testcase.rb:173:in `rescue in run': failed to allocate memory (NoMemoryError)
rb_reg_desc was not safe for GC compaction because it took in the C
string and length but not the backing String object so it get moved
during compaction. This commit changes rb_reg_desc to use the string
from the Regexp object.
The test fails when RGENGC_CHECK_MODE is turned on:
TestRegexp#test_inspect_under_gc_compact_stress [test/ruby/test_regexp.rb:474]:
<"(?-mix:\\/)|"> expected but was
<"/\\x00\\x00\\x00\\x00\\x00\\x00\\x00\\x00\\x00\\x00/">.
The test fails when RGENGC_CHECK_MODE is turned on:
TestRegexp#test_match_under_gc_compact_stress:
NoMethodError: undefined method `match' for nil
test_regexp.rb:878:in `block in test_match_under_gc_compact_stress'
* Restore experimental warnings.
* Documentation and code structure improvements.
* Improved validation of flags, clarified documentation of argument handling.
* Remove inconsistent use of `Example:` and add example to `null?`.
* Expose `IO::Buffer#private?` and add test.
On platforms where `typeof` is unsupported, `ccan_container_off_var`
calculates the offset of member by pointer subtraction. Although this
will result in the compile-time invariant value regardless the pointer
value, the loop variable will be used before assignment and may cause
an using uninitialized variable warning.
NARUSE, Yui
NARUSE, Yui
Takashi Kokubun
Hiroshi SHIBATA
Yuta Saito
Maxime Chevalier-Boisvert
KJ Tsanaktsidis
Nobuyoshi Nakada
Peter Zhu
Samuel Williams
BurdetteLamar
git
Koichi Sasada