Speed up --yjit-trace-exits code (#6106)

In a small script the speed of this feature isn't really noticeable but
on Rails it's very noticeable how slow this can be. This PR aims to
speed up two parts of the functionality.

1) The Rust exit recording code

Instead of adding all samples as we see them to the yjit_raw_samples and
yjit_line_samples, we can increment the counter on the ones we've seen
before. This will be faster on traces where we are hitting the same
stack often. In a crude measurement of booting just the active record
base test (`test/cases/base_test.rb`) we found that this improved the
speed by 1 second.

This also results in a smaller marshal dump file which sped up the test
boot time by 4 seconds with trace exits on.

2) The Ruby parsing code

Previously we were allocating new arrays using `shift` and
`each_with_index`. This change avoids allocating new arrays by using an
index. This change saves us the most amount of time, gaining 11 seconds.

Before this change the test boot time took 62 seconds, after it took 47
seconds. This is still too long but it's a step closer to faster
functionality. Next we're going to tackle allowing you to collect trace
exits for a specific instruction. There is also some potential slowness
in the GC code that I'd like to take a second look at.

Co-authored-by: Aaron Patterson <tenderlove@ruby-lang.org>

Co-authored-by: Aaron Patterson <tenderlove@ruby-lang.org>

1 files changed, 54 insertions, 8 deletions

diff --git a/yjit/src/stats.rs b/yjit/src/stats.rs
index 4118f894b4..4843cecf92 100644
--- a/yjit/src/stats.rs
+++ b/yjit/src/stats.rs
@@ -432,23 +432,67 @@ pub extern "C" fn rb_yjit_record_exit_stack(exit_pc: *const VALUE)
         // Call frame info is stored in the frames_buffer, line number information
         // in the lines_buffer. The first argument is the start point and the second
         // argument is the buffer limit, set at 2048.
-        let num = unsafe { rb_profile_frames(0, BUFF_LEN as i32, frames_buffer.as_mut_ptr(), lines_buffer.as_mut_ptr()) };
+        let stack_length = unsafe { rb_profile_frames(0, BUFF_LEN as i32, frames_buffer.as_mut_ptr(), lines_buffer.as_mut_ptr()) };
+        let samples_length = (stack_length as usize) + 3;
 
-        let mut i = num - 1;
         let yjit_raw_samples = YjitExitLocations::get_raw_samples();
         let yjit_line_samples = YjitExitLocations::get_line_samples();
 
-        yjit_raw_samples.push(VALUE(num as usize));
-        yjit_line_samples.push(num);
+        // If yjit_raw_samples is less than or equal to the current length of the samples
+        // we might have seen this stack trace previously.
+        if yjit_raw_samples.len() >= samples_length {
+            let prev_stack_len_index = yjit_raw_samples.len() - samples_length;
+            let prev_stack_len = i64::from(yjit_raw_samples[prev_stack_len_index]);
+            let mut idx = stack_length - 1;
+            let mut prev_frame_idx = 0;
+            let mut seen_already = true;
+
+            // If the previous stack lenght and current stack length are equal,
+            // loop and compare the current frame to the previous frame. If they are
+            // not equal, set seen_already to false and break out of the loop.
+            if prev_stack_len == stack_length as i64 {
+                while idx >= 0 {
+                    let current_frame = frames_buffer[idx as usize];
+                    let prev_frame = yjit_raw_samples[prev_stack_len_index + prev_frame_idx + 1];
+
+                    // If the current frame and previous frame are not equal, set
+                    // seen_already to false and break out of the loop.
+                    if current_frame != prev_frame {
+                        seen_already = false;
+                        break;
+                    }
+
+                    idx -= 1;
+                    prev_frame_idx += 1;
+                }
+
+                // If we know we've seen this stack before, increment the counter by 1.
+                if seen_already {
+                    let prev_idx = yjit_raw_samples.len() - 1;
+                    let prev_count = i64::from(yjit_raw_samples[prev_idx]);
+                    let new_count = prev_count + 1;
+
+                    yjit_raw_samples[prev_idx] = VALUE(new_count as usize);
+                    yjit_line_samples[prev_idx] = new_count as i32;
+
+                    return;
+                }
+            }
+        }
+
+        yjit_raw_samples.push(VALUE(stack_length as usize));
+        yjit_line_samples.push(stack_length);
+
+        let mut idx = stack_length - 1;
 
-        while i >= 0 {
-            let frame = frames_buffer[i as usize];
-            let line = lines_buffer[i as usize];
+        while idx >= 0 {
+            let frame = frames_buffer[idx as usize];
+            let line = lines_buffer[idx as usize];
 
             yjit_raw_samples.push(frame);
             yjit_line_samples.push(line);
 
-            i -= 1;
+            idx -= 1;
         }
 
         // Push the insn value into the yjit_raw_samples Vec.
@@ -459,6 +503,8 @@ pub extern "C" fn rb_yjit_record_exit_stack(exit_pc: *const VALUE)
         let line = yjit_line_samples.len() - 1;
         yjit_line_samples.push(line as i32);
 
+        // Push number of times seen onto the stack, which is 1
+        // because it's the first time we've seen it.
         yjit_raw_samples.push(VALUE(1 as usize));
         yjit_line_samples.push(1);
     }