Merge branch 'pgo-clang-11'

Closes #418

CHANGELOG.md

@@ -27,6 +27,11 @@
 ## Unreleased
 ### Added
 ### Changed
+
+* Update PGO build instructions in `INSTALL.md`
+  (https://codeberg.org/dnkl/foot/issues/418).
+
+
 ### Deprecated
 ### Removed
 ### Fixed

INSTALL.md

@@ -65,6 +65,7 @@ In addition to the dev variant of the packages above, you need:
 * wayland protocols
 * ncurses (needed to generate terminfo)
 * scdoc (for man page generation)
+* llvm (for PGO builds with Clang)
 * [tllist](https://codeberg.org/dnkl/tllist) [^1]
 
 A note on compilers; in general, foot runs **much** faster when
@@ -184,21 +185,15 @@ slower!) binary.
 First, configure the build directory:
 
 ```sh
-export CFLAGS="$CFLAGS -O3 -Wno-missing-profile"
+export CFLAGS="$CFLAGS -O3"
 meson --buildtype=release --prefix=/usr -Db_lto=true ../..
 ```
 
 It is **very** important `-O3` is being used here, as GCC-10.1.x and
 later have a regression where PGO with `-O2` is **much** slower.
 
-If you are using Clang instead of GCC, use the following `CFLAGS` instead:
-
-```sh
-export CFLAGS="$CFLAGS -O3 \
-    -Wno-ignored-optimization-argument \
-    -Wno-profile-instr-out-of-date \
-    -Wno-profile-instr-unprofiled"
-```
+Clang users **must** add `-Wno-ignored-optimization-argument` to
+`CFLAGS`.
 
 Then, tell meson we want to _generate_ profiling data, and build:
 
@@ -234,6 +229,8 @@ We will use the `pgo` binary along with input corpus generated by
 `scripts/generate-alt-random-writes.py`:
 
 ```sh
+./footclient --version
+./foot --version
 tmp_file=$(mktemp)
 ../../scripts/generate-alt-random-writes \
     --rows=67 \
@@ -253,7 +250,12 @@ tmp_file=$(mktemp)
 rm ${tmp_file}
 ```
 
-The snippet above first creates an (empty) temporary file. Then, it
+The first step, running `./foot --version` and `./footclient
+--version` might seem unnecessary, but is needed to ensure we have
+_some_ profiling data for functions not covered by the PGO helper
+binary. Without this, the final link phase will fail.
+
+The snippet above then creates an (empty) temporary file. Then, it
 runs a script that generates random escape sequences (if you cat
 `${tmp_file}` in a terminal, you’ll see random colored characters all
 over the screen). Finally, we feed the randomly generated escape
@@ -271,14 +273,19 @@ This method requires a running Wayland session.
 We will use the script `scripts/generate-alt-random-writes.py`:
 
 ```sh
+./footclient --version
 foot_tmp_file=$(mktemp)
-./foot --config=/dev/null --term=xterm sh -c "<path-to-generate-alt-random-writes.py> --scroll --scroll-region --colors-regular --colors-bright --colors-256 --colors-rgb --attr-bold --attr-italic --attr-underline ${foot_tmp_file} && cat ${foot_tmp_file}"
+./foot --config=/dev/null --term=xterm sh -c "<path-to-generate-alt-random-writes.py> --scroll --scroll-region --colors-regular --colors-bright --colors-256 --colors-rgb --attr-bold --attr-italic --attr-underline --sixel ${foot_tmp_file} && cat ${foot_tmp_file}"
 rm ${foot_tmp_file}
 ```
 
 You should see a foot window open up, with random colored text. The
 window should close after ~1-2s.
 
+The first step, `./footclient --version` might seem unnecessary, but
+is needed to ensure we have _some_ profiling data for
+`footclient`. Without this, the final link phase will fail.
+
 
 ##### Use the generated PGO data
 

PKGBUILD

@@ -14,37 +14,70 @@ pkgver() {
 }
 
 build() {
+  local compiler=other
+  local do_pgo=no
+
   # makepkg uses -O2 by default, but we *really* want -O3
-  # -Wno-missing-profile since we're not exercising everything when doing PGO builds
-  export CFLAGS+=" -O3 -Wno-missing-profile"
+  CFLAGS+=" -O3"
+
+  # Figure out which compiler we're using, and whether or not to do PGO
+  case $(${CC-cc} --version) in
+    *GCC*)
+      compiler=gcc
+      do_pgo=yes
+      ;;
+
+    *clang*)
+      compiler=clang
+
+      # We need llvm to be able to manage the profiling data
+      if command -v llvm-profdata > /dev/null; then
+        do_pgo=yes
+
+        # Meson adds -fprofile-correction, which Clang doesn't
+        # understand
+        CFLAGS+=" -Wno-ignored-optimization-argument"
+      fi
+      ;;
+  esac
 
   meson --prefix=/usr --buildtype=release --wrap-mode=nofallback -Db_lto=true ..
 
-  find -name "*.gcda" -delete
-  meson configure -Db_pgo=generate
-  ninja
+  if [[ ${do_pgo} == yes ]]; then
+    find -name "*.gcda" -delete
+    meson configure -Db_pgo=generate
+    ninja
 
-  script_options="--scroll --scroll-region --colors-regular --colors-bright --colors-256 --colors-rgb --attr-bold --attr-italic --attr-underline --sixel"
+    local script_options="--scroll --scroll-region --colors-regular --colors-bright --colors-256 --colors-rgb --attr-bold --attr-italic --attr-underline --sixel"
 
-  tmp_file=$(mktemp)
+    tmp_file=$(mktemp)
 
-  if [[ -v WAYLAND_DISPLAY ]]; then
-    ./foot \
-      --config /dev/null \
-      --term=xterm \
-      sh -c "../scripts/generate-alt-random-writes.py ${script_options} ${tmp_file} && cat ${tmp_file}"
-  else
-    ../scripts/generate-alt-random-writes.py \
-      --rows=67 \
-      --cols=135 \
-      ${script_options} \
-      ${tmp_file}
-    ./pgo ${tmp_file} ${tmp_file} ${tmp_file}
+    if [[ -v WAYLAND_DISPLAY ]]; then
+      ./footclient --version
+      ./foot \
+        --config /dev/null \
+        --term=xterm \
+        sh -c "../scripts/generate-alt-random-writes.py ${script_options} ${tmp_file} && cat ${tmp_file}"
+    else
+      ./footclient --version
+      ./foot --version
+      ../scripts/generate-alt-random-writes.py \
+        --rows=67 \
+        --cols=135 \
+        ${script_options} \
+        ${tmp_file}
+      ./pgo ${tmp_file} ${tmp_file} ${tmp_file}
+    fi
+
+    rm "${tmp_file}"
+
+    if [[ ${compiler} == clang ]]; then
+      llvm-profdata merge default_*profraw --output=default.profdata
+    fi
+
+    meson configure -Db_pgo=use
   fi
 
-  rm "${tmp_file}"
-
-  meson configure -Db_pgo=use
   ninja
 }
 

client.c

@@ -35,8 +35,10 @@ static const char *
 version_and_features(void)
 {
     static char buf[256];
-    snprintf(buf, sizeof(buf), "version: %s %cime",
-             FOOT_VERSION, feature_ime() ? '+' : '-');
+    snprintf(buf, sizeof(buf), "version: %s %cime %cpgo",
+             FOOT_VERSION,
+             feature_ime() ? '+' : '-',
+             feature_pgo() ? '+' : '-');
     return buf;
 }
 

foot-features.h

@@ -10,3 +10,12 @@ static inline bool feature_ime(void)
     return false;
 #endif
 }
+
+static inline bool feature_pgo(void)
+{
+#if defined(FOOT_PGO_ENABLED) && FOOT_PGO_ENABLED
+    return true;
+#else
+    return false;
+#endif
+}

main.c

@@ -45,8 +45,10 @@ static const char *
 version_and_features(void)
 {
     static char buf[256];
-    snprintf(buf, sizeof(buf), "version: %s %cime",
-             FOOT_VERSION, feature_ime() ? '+' : '-');
+    snprintf(buf, sizeof(buf), "version: %s %cime %cpgo",
+             FOOT_VERSION,
+             feature_ime() ? '+' : '-',
+             feature_pgo() ? '+' : '-');
     return buf;
 }
 

meson.build

@@ -24,6 +24,9 @@ add_project_arguments(
   (get_option('ime')
     ? ['-DFOOT_IME_ENABLED=1']
     : []) +
+  (get_option('b_pgo') == 'use'
+    ? ['-DFOOT_PGO_ENABLED=1']
+    : []) +
   cc.get_supported_arguments(
     ['-pedantic',
      '-fstrict-aliasing',
@@ -114,16 +117,20 @@ version = custom_target(
   output: 'version.h',
   command: [generate_version_sh, meson.project_version(), '@SOURCE_ROOT@', '@OUTPUT@'])
 
+common = static_library(
+  'common',
+  'log.c', 'log.h',
+  'debug.c', 'debug.h',
+  'xmalloc.c', 'xmalloc.h',
+  'xsnprintf.c', 'xsnprintf.h'
+)
+
 misc = static_library(
   'misc',
-  'debug.c', 'debug.h',
   'hsl.c', 'hsl.h',
-  'log.c', 'log.h',
   'macros.h',
   'misc.c', 'misc.h',
-  'uri.c', 'uri.h',
-  'xmalloc.c', 'xmalloc.h',
-  'xsnprintf.c', 'xsnprintf.h',
+  'uri.c', 'uri.h'
 )
 
 vtlib = static_library(
@@ -137,7 +144,7 @@ vtlib = static_library(
   wl_proto_src + wl_proto_headers,
   version,
   dependencies: [libepoll, pixman, fcft, tllist, wayland_client],
-  link_with: misc,
+  link_with: [common, misc],
 )
 
 pgolib = static_library(
@@ -150,13 +157,15 @@ pgolib = static_library(
   link_with: vtlib,
 )
 
-executable(
-  'pgo',
-  'pgo/pgo.c',
-  wl_proto_src + wl_proto_headers,
-  dependencies: [math, threads, libepoll, pixman, wayland_client, fcft, tllist],
-  link_with: pgolib,
-)
+if get_option('b_pgo') == 'generate'
+  executable(
+    'pgo',
+    'pgo/pgo.c',
+    wl_proto_src + wl_proto_headers,
+    dependencies: [math, threads, libepoll, pixman, wayland_client, fcft, tllist],
+    link_with: pgolib,
+  )
+endif
 
 executable(
   'foot',
@@ -192,14 +201,11 @@ executable(
 executable(
   'footclient',
   'client.c', 'client-protocol.h',
-  'debug.c', 'debug.h',
   'foot-features.h',
-  'log.c', 'log.h',
   'macros.h',
   'util.h',
-  'xmalloc.c', 'xmalloc.h',
-  'xsnprintf.c', 'xsnprintf.h',
   version,
+  link_with: common,
   install: true)
 
 if tic.found()