This patch contains several ideas:
(1) Disposable inline method cache (IMC) for race-free inline method cache
* Making call-cache (CC) as a RVALUE (GC target object) and allocate new
CC on cache miss.
* This technique allows race-free access from parallel processing
elements like RCU.
(2) Introduce per-Class method cache (pCMC)
* Instead of fixed-size global method cache (GMC), pCMC allows flexible
cache size.
* Caching CCs reduces CC allocation and allow sharing CC's fast-path
between same call-info (CI) call-sites.
(3) Invalidate an inline method cache by invalidating corresponding method
entries (MEs)
* Instead of using class serials, we set "invalidated" flag for method
entry itself to represent cache invalidation.
* Compare with using class serials, the impact of method modification
(add/overwrite/delete) is small.
* Updating class serials invalidate all method caches of the class and
sub-classes.
* Proposed approach only invalidate the method cache of only one ME.
See [Feature #16614] for more details.
Now, rb_call_info contains how to call the method with tuple of
(mid, orig_argc, flags, kwarg). Most of cases, kwarg == NULL and
mid+argc+flags only requires 64bits. So this patch packed
rb_call_info to VALUE (1 word) on such cases. If we can not
represent it in VALUE, then use imemo_callinfo which contains
conventional callinfo (rb_callinfo, renamed from rb_call_info).
iseq->body->ci_kw_size is removed because all of callinfo is VALUE
size (packed ci or a pointer to imemo_callinfo).
To access ci information, we need to use these functions:
vm_ci_mid(ci), _flag(ci), _argc(ci), _kwarg(ci).
struct rb_call_info_kw_arg is renamed to rb_callinfo_kwarg.
rb_funcallv_with_cc() and rb_method_basic_definition_p_with_cc()
is temporary removed because cd->ci should be marked.
ISO/IEC 9899:1999 section 6.5.7 states that "If the value of the right
operand is negative or is greater than or equal to the width of the
promoted left operand, the behavior is undefined". So we have to take
care of such situations.
This has not been a problem because contemporary C compilers are
extraordinary smart to compile the series of shifts into a single
ROTLQ/ROTRQ machine instruction. In contrast to what C says those
instructions have fully defined behaviour for all possible inputs.
Hence it has been quite difficult to observe the undefined-ness of such
situations. But undefined is undefined. We should not rely on such
target-specific assumptions.
We are fixing the situation by carefully avoiding shifts with out-of-
range values. At least GCC since 4.6.3 and Clang since 8.0 can issue
the exact same instructions like before the changeset.
Also in case of Intel processors, there supposedly be intrinsics named
_rotr/_rotl that do exactly what we need. They, in practice, are absent
on Clang before 9.x so we cannot blindly use. But we can at least save
MSVC.
See also:
https://gcc.gnu.org/bugzilla/show_bug.cgi?id=57157https://bugs.llvm.org/show_bug.cgi?id=17332
Replaces `#ifdef _MSC_VER` with more accurate version checks. Also,
`defined(_WIN64) && defined(__AVX2__)` is redundant because there is no
such tihng like a 32bit AVX2 machine.
This removes the warnings added in 2.7, and changes the behavior
so that a final positional hash is not treated as keywords or
vice-versa.
To handle the arg_setup_block splat case correctly with keyword
arguments, we need to check if we are taking a keyword hash.
That case didn't have a test, but it affects real-world code,
so add a test for it.
This removes rb_empty_keyword_given_p() and related code, as
that is not needed in Ruby 3. The empty keyword case is the
same as the no keyword case in Ruby 3.
This changes rb_scan_args to implement keyword argument
separation for C functions when the : character is used.
For backwards compatibility, it returns a duped hash.
This is a bad idea for performance, but not duping the hash
breaks at least Enumerator::ArithmeticSequence#inspect.
Instead of having RB_PASS_CALLED_KEYWORDS be a number,
simplify the code by just making it be rb_keyword_given_p().
Noticed that internal/stdbool.h and addr2line.c are the only two place
where missing/stdbool.h is included. Why not delete the file so that
we can merge internal/stdbool.h and missing/stdbool.h into one.
Saves comitters' daily life by avoid #include-ing everything from
internal.h to make each file do so instead. This would significantly
speed up incremental builds.
We take the following inclusion order in this changeset:
1. "ruby/config.h", where _GNU_SOURCE is defined (must be the very
first thing among everything).
2. RUBY_EXTCONF_H if any.
3. Standard C headers, sorted alphabetically.
4. Other system headers, maybe guarded by #ifdef
5. Everything else, sorted alphabetically.
Exceptions are those win32-related headers, which tend not be self-
containing (headers have inclusion order dependencies).
These headers need no rewrite. Just add some minor tweaks, like
addition of #include lines. Mainly cosmetic.
TIMET_MAX_PLUS_ONE was deleted because the macro was used from only
one place (directly write expression there).
Annotated MJIT_FUNC_EXPORTED functions as such. Declaration of
rb_sym_to_proc is moved into this file because the function is defined
in proc.c rather than string.c.
Improved readability by reducing the use of macros. Also moved some
part of internal/compilers.h into this file, because it seems to be the
right place for them.
This file containes other materials than in compile.c. I could perhaps
split them into files, but felt overkill. Just add comments that
describe the situations.
This file has almost nothing to do. Added some #ifdef lines and
rearranged file contents.
Those macros are unable to translate into inline functions, because they
are used as lvalues of assignments.
Reduce macros to make them inline functions, as well as mark
MJIT_FUNC_EXPORTED functions explicitly as such.
Definition of ar_hint_t is simplified. This has been the only possible
definition so far.
Improving readability by converting some macros into inline functions.
Also improved support for recent x86_64 processors, which have better
instructions for the purposes.
With these macros implemented we can write codes just like we can assume
the compiler being clang. MSC_VERSION_SINCE is defined to implement
those macros, but turned out to be handy for other places. The -fdeclspec
compiler flag is necessary for clang to properly handle __has_declspec().
Now that we no longer support old compilers, we can safely delete
several obsolete #ifdef gurads. Also because (as of writing) it is
impossible to compile the program using C++ compilers, lets just
entirely prohibit __cplusplus to reduce # of LOCs.
Note however that we still cannot eliminate __STDC_VERSION__ checks,
because MSVC does not define it, saying its C99 support is partial.
See also https://social.msdn.microsoft.com/Forums/vstudio/en-US/53a4fd75-9f97-48b2-aa63-2e2e5a15efa3
One day, I could not resist the way it was written. I finally started
to make the code clean. This changeset is the beginning of a series of
housekeeping commits. It is a simple refactoring; split internal.h into
files, so that we can divide and concur in the upcoming commits. No
lines of codes are either added or removed, except the obvious file
headers/footers. The generated binary is identical to the one before.
Koichi Sasada
Nobuyoshi Nakada
Tanaka Akira
Kenta Murata
卜部昌平
Jeremy Evans
Kenta Murata
Kazuhiro NISHIYAMA