Rebuilding the loaded feature index slowed down with the bug fix
for #17885 in 79a4484a072e9769b603e7b4fbdb15b1d7eccb15. The
slowdown was extreme if realpath emulation was used, but even when
not emulated, it could be about 10x slower.
This adds loaded_features_realpath_map to rb_vm_struct. This is a
hidden hash mapping loaded feature paths to realpaths. When
rebuilding the loaded feature index, look at this hash to get
cached realpath values, and skip calling rb_check_realpath if a
cached value is found.
Fixes [Bug #19246]
The stackprof-format raw samples are suffixed with a count (ie. how many
times did the previously recorded side-exit repeat). Previously we were
correctly using this value to increment the :samples count, but not the
:total_samples count or edges.
This made the stackprof aggregate results incorrect (though any
flamegraphs generated should have been correct, since those only rely on
raw samples).
[Feature #19538]
This new `peformance` warning category is disabled by default.
It needs to be specifically enabled via `-W:performance` or `Warning[:performance] = true`
Add a sampling option to trace exits
Running YJIT with trace exits enabled can make very large metrics files.
This allows us to configure a sample rate to make tracing exits possible
on larger tests. This also updates the documented YJIT options.
Co-authored-by: Alan Wu <XrXr@users.noreply.github.com>
Co-authored-by: John Hawthorn <john@hawthorn.email>
Co-authored-by: Maxime Chevalier-Boisvert <maximechevalierb@gmail.com>
[Bug #19593]
rb_ec_setup_exception did not check if errinfo is a throw_data. This can
cause crashes in code since it is assumed that id_cause is an object.
We saw a crash in show_cause due to id_cause of errinfo being a
throw_data. It crashes on rb_obj_is_kind_of since it cannot be called on
T_IMEMO objects.
Unfortunately, we couldn't find a reproduction script, however we
debugged the core dump and rb_ec_setup_exception is the only place where
id_cause is assigned from errinfo without checking if it is a
throw_data.
```
0x0000556c5708e6dd in sigsegv (sig=11, info=0x7f301befa3f0, ctx=0x7f301befa2c0) at signal.c:964
0x00007f301d046420 in <signal handler called> () at /lib/x86_64-linux-gnu/libpthread.so.0
class_search_class_ancestor (c=139844586301760, cl=<optimized out>) at object.c:810
rb_obj_is_kind_of (obj=obj@entry=139839221734880, c=139844586301760) at object.c:861
0x0000556c56f2f00f in show_cause
(errinfo=errinfo@entry=139838840645160, str=str@entry=139839221730520, opt=139839221730480, highlight=0, reverse=reverse@entry=0, backtrace_limit=backtrace_limit@entry=-1, shown_causes=0x7ffe9d1a2d68) at ./include/ruby/internal/special_consts.h:175
```
Co-Authored-By: Jean Boussier <byroot@ruby-lang.org>
OpenSSL's man page previously stated that "the application is
responsible for seeding the PRNG by calling RAND_add" (see [1]).
So we had this code. However things changed. They no longer
say so, instead "manual (re-)seeding of the default OpenSSL
random generator is not necessary" now (see [2]). It seems all
OpenSSL versions that we support now already behaves like this.
Let's follow that.
[1]: https://www.openssl.org/docs/man1.0.2/man3/RAND_add.html
[2]: https://www.openssl.org/docs/manmaster/man3/RAND_add.html
Previously we were missing a compile-time check that the known cfuncs
receive the correct number of arguments.
We noticied this because in particular when using ARGS_SPLAT, which also
wasn't checked, YJIT would crash on code which was otherwise correct
(didn't raise exceptions in the VM).
This still supports vararg (argc == -1) cfuncs. I added an additional
assertion that when we use the specialized codegen for one of these
known functions that the argc are popped off the stack correctly, which
should help ensure they're implemented correctly (previously the crash
was usually observed on a future `leave` insn).
[Bug #19595]
* YJIT: Let Assembler own Context
* Update a comment
Co-authored-by: Maxime Chevalier-Boisvert <maximechevalierb@gmail.com>
---------
Co-authored-by: Maxime Chevalier-Boisvert <maximechevalierb@gmail.com>
