From 240fd2be6f46ad7d152f11a41f768e89430b5192 Mon Sep 17 00:00:00 2001
From: dqhl76 <dqhl76@gmail.com>
Date: Mon, 1 Dec 2025 14:52:53 +0800
Subject: [PATCH 1/2] chore: post process frames after explain perf

---
 Cargo.lock                                    |  2 +-
 Cargo.toml                                    |  2 +-
 .../base/src/runtime/perf/query_perf.rs       | 45 +++++++++++++++++++
 3 files changed, 47 insertions(+), 2 deletions(-)

diff --git a/Cargo.lock b/Cargo.lock
index b97349d475426..ba2e8dec0c8aa 100644
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -12251,7 +12251,7 @@ checksum = "439ee305def115ba05938db6eb1644ff94165c5ab5e9420d1c1bcedbba909391"
 [[package]]
 name = "pprof"
 version = "0.15.0"
-source = "git+https://github.com/datafuse-extras/pprof-rs?rev=fe22b23#fe22b23cfa294122547a1add1377918aca0493b6"
+source = "git+https://github.com/datafuse-extras/pprof-rs?rev=edecd74#edecd74268ea5dcd31836bc4cb8acdcb21c51b24"
 dependencies = [
  "aligned-vec",
  "backtrace",
diff --git a/Cargo.toml b/Cargo.toml
index f9030d33c3a56..dfcde08e41730 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -425,7 +425,7 @@ petgraph = { version = "0.6.2", features = ["serde-1"] }
 pin-project = "1"
 pin-project-lite = "0.2.9"
 poem = { version = "3.0", features = ["openssl-tls", "multipart", "compression", "cookie"] }
-pprof = { git = "https://github.com/datafuse-extras/pprof-rs", rev = "fe22b23", features = [
+pprof = { git = "https://github.com/datafuse-extras/pprof-rs", rev = "edecd74", features = [
     "flamegraph",
     "protobuf-codec",
     "protobuf",
diff --git a/src/common/base/src/runtime/perf/query_perf.rs b/src/common/base/src/runtime/perf/query_perf.rs
index 1d20c387fab9d..991a601c21dbc 100644
--- a/src/common/base/src/runtime/perf/query_perf.rs
+++ b/src/common/base/src/runtime/perf/query_perf.rs
@@ -17,6 +17,7 @@ use databend_common_exception::Result;
 use log::debug;
 use pprof::ProfilerGuard;
 use pprof::ProfilerGuardBuilder;
+use regex::Regex;
 
 use crate::runtime::ThreadTracker;
 use crate::runtime::TrackingGuard;
@@ -65,6 +66,7 @@ impl QueryPerf {
     pub fn dump(profiler_guard: &ProfilerGuard<'static>) -> Result<String> {
         let reporter = profiler_guard
             .report()
+            .frames_post_processor(frames_post_processor())
             .build()
             .map_err(|_e| ErrorCode::Internal("Failed to report profiler data"))?;
         debug!("perf stop, begin to dump flamegraph");
@@ -126,6 +128,49 @@ impl QueryPerf {
     }
 }
 
+const PPROF_TRACE_SYMBOL: &str =
+    "<pprof::backtrace::backtrace_rs::Trace as pprof::backtrace::Trace>::trace";
+
+fn frames_post_processor() -> impl Fn(&mut pprof::Frames) {
+    // If pprof cannot get thread name, it will use thread id as fallback
+    // this will make the flamegraph hard to read, so we rename such threads to "threads"
+    let thread_rename = [(Regex::new(r"^\d+$").unwrap(), "threads")];
+
+    move |frames| {
+        for (regex, name) in thread_rename.iter() {
+            if regex.is_match(&frames.thread_name) {
+                frames.thread_name = name.to_string();
+            }
+        }
+
+        // Remove frames introduced by pprof's own stack collection to keep user stacks clean.
+        if let Some(pos) = frames.frames.iter().position(|frame| {
+            frame
+                .iter()
+                .any(|symbol| symbol.name() == PPROF_TRACE_SYMBOL)
+        }) {
+            frames.frames.drain(..=pos);
+        }
+
+        // Mark inlined functions with "(inlined)" suffix
+        for inline_frames in frames.frames.iter_mut() {
+            if inline_frames.len() <= 1 {
+                continue;
+            }
+
+            // Mark every symbol except the outermost one as inlined.
+            let last_symbol_index = inline_frames.len() - 1;
+            for symbol in inline_frames.iter_mut().take(last_symbol_index) {
+                let symbol_name = symbol.name();
+                if symbol_name.ends_with(" (inlined)") {
+                    continue;
+                }
+                symbol.name = Some(format!("{symbol_name} (inlined)").into_bytes());
+            }
+        }
+    }
+}
+
 #[cfg(test)]
 mod tests {
     use crate::runtime::QueryPerf;

From 4ad73791158f76dbf1d5a298255b53af072c450e Mon Sep 17 00:00:00 2001
From: Liuqing Yue <dqhl76@gmail.com>
Date: Mon, 1 Dec 2025 20:30:06 +0800
Subject: [PATCH 2/2] ci: add a flag to allow skip trim debug info

---
 .github/actions/build_linux/action.yml | 5 +++++
 .github/workflows/cloud.yml            | 1 +
 2 files changed, 6 insertions(+)

diff --git a/.github/actions/build_linux/action.yml b/.github/actions/build_linux/action.yml
index e1cbac6700c14..3fddf31e56e49 100644
--- a/.github/actions/build_linux/action.yml
+++ b/.github/actions/build_linux/action.yml
@@ -23,6 +23,10 @@ inputs:
     description: "Category to upload"
     required: false
     default: "default"
+  trim_debug:
+    description: "Trim debug info"
+    required: false
+    default: "true"
 runs:
   using: "composite"
   steps:
@@ -95,6 +99,7 @@ runs:
         ldd ./target/${{ inputs.target }}/${{ env.BUILD_PROFILE }}/databend-query || true
 
     - name: Spilt Binary Symbols
+      if: inputs.trim_debug == 'true'
       shell: bash
       run: |
         objcopy --only-keep-debug ./target/${{ inputs.target }}/${{ env.BUILD_PROFILE }}/databend-query ./target/${{ inputs.target }}/${{ env.BUILD_PROFILE }}/databend-query.debug
diff --git a/.github/workflows/cloud.yml b/.github/workflows/cloud.yml
index 6bc75b519f5c0..fe470c1cec314 100644
--- a/.github/workflows/cloud.yml
+++ b/.github/workflows/cloud.yml
@@ -77,6 +77,7 @@ jobs:
           artifacts: meta,query
           category: docker
           features: python-udf
+          trim_debug: false
 
   docker:
     needs: [ info, build ]