diff --git a/README.md b/README.md
index baad208d..71aad9cc 100644
--- a/README.md
+++ b/README.md
@@ -32,6 +32,8 @@ Please see https://pocketpy.dev for details and try the following resources.
 + [Live Python Demo](https://pocketpy.dev/static/web/): Run Python code in your browser
 + [Live C Examples](https://pocketpy.github.io/examples/): Explore C-APIs in your browser
 + [Godot Extension](https://github.com/pocketpy/godot-pocketpy): Use pocketpy in Godot Engine
++ [VSCode Extension](https://marketplace.visualstudio.com/items?itemName=pocketpy.pocketpy): Debug and profile pocketpy scripts in VSCode
++ [Flutter Plugin](https://pub.dev/packages/pocketpy): Use pocketpy in Flutter apps
 
 ## Supported Platforms
 
diff --git a/docs/features/debugging.md b/docs/features/debugging.md
index f7383a7a..2bac8f62 100644
--- a/docs/features/debugging.md
+++ b/docs/features/debugging.md
@@ -1,24 +1,88 @@
 ---
 icon: dot
 title: Debugging
+order: 80
 ---
 
+## Install VSCode Extension
+
+To debug a pocketpy program, you need to install our VSCode extension first:
+
+https://marketplace.visualstudio.com/items?itemName=pocketpy.pocketpy
+
 !!!
-This feature is not available in `v2.0` yet.
+The VSCode extension requires pocketpy version >= `2.1.1`
 !!!
 
-You can invoke `breakpoint()` in your python code to start a PDB-like session.
+## Create a `launch.json` file
 
-The following commands are supported:
+Navigate to the Debug view in VSCode, and click on "create a launch.json file" link.
+In the dropdown menu, select "pocketpy".
 
-+ `h, help`: show this help message
-+ `q, quit`: exit the debugger
-+ `n, next`: execute next line
-+ `s, step`: step into
-+ `w, where`: show current stack frame
-+ `c, continue`: continue execution
-+ `a, args`: show local variables
-+ `l, list`: show lines around current line
-+ `ll, longlist`: show all lines
-+ `p, print <expr>`: evaluate expression
-+ `!, execute statement`: execute statement
+![launch_json](../static/debugger/launch_json.png)
+
+Then a default `launch.json` file will be created in the `.vscode` folder
+with a sample pocketpy debug configuration.
+
+## How does it work?
+
+pocketpy provides a C-API `py_debugger_waitforattach`,
+which starts a debug server and waits for the VSCode extension to attach.
+When the debugger is attached, the program will continue to run.
+
++ If you are using pocketpy's standalone executable `main.exe`, you can pass `--debug` flag to it. This will automatically call `py_debugger_waitforattach("localhost", 6110)` before running your program.
++ If you are embedding pocketpy as a library, you need to call `py_debugger_waitforattach` manually in your C/C++ code.
+
+## Configuration
+
++ `type`: must be `pocketpy`
++ `request`: can be `attach` or `launch`
++ `name`: the name of this configuration
++ `port`: the port number of the debug server, must match the one in `py_debugger_waitforattach`
++ `host`: the host of the debug server, must match the one in `py_debugger_waitforattach`
++ `sourceFolder`: the root folder of your python source code, default to `${workspaceFolder}`. However,
+sometimes you may run your program from a subfolder, in this case you need to set `sourceFolder` to the correct path. If this is not set correctly, breakpoints will not be hit.
++ `program`: (for launch mode only) the path to the executable file which calls `py_debugger_waitforattach`, e.g. the pocketpy standalone executable `main.exe`.
++ `args`: (for launch mode only) the arguments to pass to the executable file, e.g. `--debug` and the script path if you are using `main.exe`.
++ `cwd`: (for launch mode only) the working directory to launch the executable file, default to `${workspaceFolder}`.
+
+### For attach mode
+
+In this mode, you need to start your pocketpy program manually which must call `py_debugger_waitforattach` first.
+After the program starts, you can let VSCode attach to the debug server.
+
+```json
+{
+    "type": "pocketpy",
+    "request": "attach",
+    "name": "Attach to pocketpy program",
+    "port": 6110,
+    "host": "localhost",
+    "sourceFolder": "${workspaceFolder}"
+}
+```
+
+### For launch mode
+
+In this mode, VSCode will start your program with the specified `program`, `args` and `cwd`.
+After the program starts, VSCode attempts to attach to the debug server automatically.
+
+```json
+{
+    "type": "pocketpy",
+    "request": "launch",
+    "name": "Launch pocketpy program",
+    "port": 6110,
+    "host": "localhost",
+    "sourceFolder": "${workspaceFolder}",
+    "program": "${workspaceFolder}/pocketpy/main.exe",
+    "args": [
+        "--debug"
+    ],
+    "cwd": "${workspaceFolder}"
+}
+```
+
+## Showcase
+
+![debugger_demo](../static/debugger/debugger_demo.png)
diff --git a/docs/features/profiling.md b/docs/features/profiling.md
new file mode 100644
index 00000000..4799129d
--- /dev/null
+++ b/docs/features/profiling.md
@@ -0,0 +1,29 @@
+---
+icon: dot
+title: Profiling
+order: 79
+---
+
+To profile your pocketpy program, you can run `main.exe` with `--profile` flag.
+
+For example, to profile `test/test_math.py`, run
+
+```
+main.exe --profile test/test_math.py
+```
+
+This will output a JSON report file named `profile_report.json` in the current directory,
+which records the time spent for each line. To visualize the report, please install our VSCode extension.
+
+https://marketplace.visualstudio.com/items?itemName=pocketpy.pocketpy
+
+!!!
+The VSCode extension requires pocketpy version >= `2.1.1`
+!!!
+
+With pocketpy VSCode extension, press `F1` and type `pocketpy: Load Line Profiler Report`,
+select **1. the `profile_report.json` file; 2. the source root of the program**. Then you will see a nice visualization of the profiling result.
+
+![profiler_report](../static/profiler_demo.png)
+
+Press `ESC` to exit the report view.
diff --git a/docs/index.md b/docs/index.md
index e112d063..5a342559 100644
--- a/docs/index.md
+++ b/docs/index.md
@@ -13,6 +13,7 @@ Developers are able to write Python bindings via C-API or pybind11 compatible in
 + [Live Python Demo](https://pocketpy.dev/static/web/): Run Python code in your browser
 + [Live C Examples](https://pocketpy.github.io/examples/): Explore C-APIs in your browser
 + [Godot Extension](https://github.com/pocketpy/godot-pocketpy): Use pocketpy in Godot Engine
++ [VSCode Extension](https://marketplace.visualstudio.com/items?itemName=pocketpy.pocketpy): Debug and profile pocketpy scripts in VSCode
 + [Flutter Plugin](https://pub.dev/packages/pocketpy): Use pocketpy in Flutter apps
 
 ## What it looks like
diff --git a/docs/retype.yml b/docs/retype.yml
index 89232f96..7d05eb7e 100644
--- a/docs/retype.yml
+++ b/docs/retype.yml
@@ -3,7 +3,7 @@ output: .retype
 url: https://pocketpy.dev
 branding:
   title: pocketpy
-  label: v2.1.0
+  label: v2.1.1
   logo: "./static/logo.png"
 favicon: "./static/logo.png"
 meta:
diff --git a/docs/static/debugger/debugger_demo.png b/docs/static/debugger/debugger_demo.png
new file mode 100644
index 00000000..e3d72827
Binary files /dev/null and b/docs/static/debugger/debugger_demo.png differ
diff --git a/docs/static/debugger/launch_json.png b/docs/static/debugger/launch_json.png
new file mode 100644
index 00000000..540aebbe
Binary files /dev/null and b/docs/static/debugger/launch_json.png differ
diff --git a/docs/static/profiler_demo.png b/docs/static/profiler_demo.png
new file mode 100644
index 00000000..3bb742ad
Binary files /dev/null and b/docs/static/profiler_demo.png differ
diff --git a/include/pocketpy/common/smallmap.h b/include/pocketpy/common/smallmap.h
index 74556202..3a007268 100644
--- a/include/pocketpy/common/smallmap.h
+++ b/include/pocketpy/common/smallmap.h
@@ -22,12 +22,11 @@
 #define K c11_sv
 #define V int
 #define NAME c11_smallmap_v2d
-#define less(a, b)      (c11_sv__cmp((a), (b)) <  0)
-#define equal(a, b)     (c11_sv__cmp((a), (b)) == 0)
+#define less(a, b) (c11_sv__cmp((a), (b)) < 0)
+#define equal(a, b) c11__sveq((a), (b))
 #include "pocketpy/xmacros/smallmap.h"
 #undef SMALLMAP_T__HEADER
 
-
 #define SMALLMAP_T__HEADER
 #define K void*
 #define V py_i64
diff --git a/src/objects/namedict.c b/src/objects/namedict.c
index 7e89c6df..ea37932d 100644
--- a/src/objects/namedict.c
+++ b/src/objects/namedict.c
@@ -5,7 +5,20 @@
 #include <string.h>
 #include <assert.h>
 
-#define HASH_KEY(__k) ((uintptr_t)(__k) >> 3U)
+// https://jfdube.wordpress.com/2011/10/12/hashing-strings-and-pointers-avoiding-common-pitfalls/
+uintptr_t ThomasWangInt32Hash(void* Ptr) {
+    // Here we think only the lower 32 bits are useful
+    uint32_t Value = (uint32_t)(uintptr_t)Ptr;
+    Value = ~Value + (Value << 15);
+    Value = Value ^ (Value >> 12);
+    Value = Value + (Value << 2);
+    Value = Value ^ (Value >> 4);
+    Value = Value * 2057;
+    Value = Value ^ (Value >> 16);
+    return Value;
+}
+
+#define HASH_KEY(__k) ThomasWangInt32Hash(__k)
 
 #define HASH_PROBE_0(__k, ok, i)                                                                   \
     ok = false;                                                                                    \
diff --git a/src/public/py_dict.c b/src/public/py_dict.c
index a3870cca..fee464db 100644
--- a/src/public/py_dict.c
+++ b/src/public/py_dict.c
@@ -136,16 +136,17 @@ static void Dict__set_index(Dict* self, uint32_t index, uint32_t value) {
     }
 }
 
+// Dict__probe won't raise exception for string keys
 static bool Dict__probe(Dict* self,
                         py_TValue* key,
                         uint64_t* p_hash,
                         uint32_t* p_idx,
                         DictEntry** p_entry) {
-    py_i64 h_user;
-    if(!py_hash(key, &h_user)) return false;
     if(py_isstr(key)) {
-        *p_hash = (uint64_t)h_user;
+        *p_hash = c11_sv__hash(py_tosv(key));
     } else {
+        py_i64 h_user;
+        if(!py_hash(key, &h_user)) return false;
         *p_hash = Dict__hash_2nd((uint64_t)h_user);
     }
     uint32_t mask = self->capacity - 1;
@@ -155,13 +156,23 @@ static bool Dict__probe(Dict* self,
         if(idx2 == self->null_index_value) break;
         DictEntry* entry = c11__at(DictEntry, &self->entries, idx2);
         if(entry->hash == (*p_hash)) {
-            int res = py_equal(&entry->key, key);
-            if(res == 1) {
-                *p_idx = idx;
-                *p_entry = entry;
-                return true;
+            if(py_isstr(&entry->key) && py_isstr(key)) {
+                c11_sv lhs = py_tosv(&entry->key);
+                c11_sv rhs = py_tosv(key);
+                if(c11__sveq(lhs, rhs)) {
+                    *p_idx = idx;
+                    *p_entry = entry;
+                    return true;
+                }
+            } else {
+                int res = py_equal(&entry->key, key);
+                if(res == 1) {
+                    *p_idx = idx;
+                    *p_entry = entry;
+                    return true;
+                }
+                if(res == -1) return false;  // error
             }
-            if(res == -1) return false;  // error
         }
         // try next index
         idx = Dict__step(idx);