gcc7/gcc7-pr88345-min-func-alignment.diff

From 874b0e7d0c16653beeed2ce8d92363cc0c1580cd Mon Sep 17 00:00:00 2001
From: Martin Jambor <mjambor@suse.cz>
Date: Thu, 25 Jan 2024 14:15:10 +0100
Subject: [PATCH] Add -fmin-function-alignmnet
To: gcc-patches@gcc.gnu.org

This is a manual backport of gcc master ommit r14-8395-g0f5a9a00e3ab1f
without alphabetical reordering of parameters in common.opt.  Commit
message of that commit (without Changelog):

  commit 0f5a9a00e3ab1fe96142f304cfbcf3f63b15f326
  Author: Jan Hubicka <jh@suse.cz>
  Date:   Wed Jan 24 18:13:17 2024 +0100

      Add -fmin-function-alignmnet

      -falign-functions is ignored in cold code, since it is an
      optimization intended to improve instruction prefetch.  In some
      case it is necessary to force alignment for all functions, so
      this patch adds -fmin-function-alignment for this purpose.

gcc/ChangeLog:

2024-01-25  Jan Hubicka <jh@suse.cz>
	    Martin Jambor  <mjambor@suse.cz>

	* common.opt (fmin-function-alignment): New parameter.
        * doc/invoke.texi: (-fmin-function-alignment): Document.
        (-falign-functions,-falign-loops,-falign-labels): Mention that
	* varasm.c (assemble_start_function): Handle min-function-alignment.
	* lto-streamer.h (LTO_minor_version): Bump.
---
 gcc/common.opt      |  4 ++++
 gcc/doc/invoke.texi | 17 +++++++++++++++++
 gcc/lto-streamer.h  |  2 +-
 gcc/varasm.c        |  5 +++++
 4 files changed, 27 insertions(+), 1 deletion(-)

diff --git a/gcc/common.opt b/gcc/common.opt
index 437db8e8615..8eadae558f3 100644
--- a/gcc/common.opt
+++ b/gcc/common.opt
@@ -1824,6 +1824,10 @@ fmessage-length=
 Common RejectNegative Joined UInteger
 -fmessage-length=<number>	Limit diagnostics to <number> characters per line.  0 suppresses line-wrapping.
 
+fmin-function-alignment=
+Common Joined RejectNegative UInteger Var(flag_min_function_alignment) Optimization
+Align the start of every function.
+
 fmodulo-sched
 Common Report Var(flag_modulo_sched) Optimization
 Perform SMS based modulo scheduling before the first scheduling pass.
diff --git a/gcc/doc/invoke.texi b/gcc/doc/invoke.texi
index 8f279e454b0..bed117fa542 100644
--- a/gcc/doc/invoke.texi
+++ b/gcc/doc/invoke.texi
@@ -355,6 +355,7 @@ Objective-C and Objective-C++ Dialects}.
 @gccoptlist{-faggressive-loop-optimizations  -falign-functions[=@var{n}] @gol
 -falign-jumps[=@var{n}] @gol
 -falign-labels[=@var{n}]  -falign-loops[=@var{n}] @gol
+-fmin-function-alignment=[@var{n}] @gol
 -fassociative-math  -fauto-profile  -fauto-profile[=@var{path}] @gol
 -fauto-inc-dec  -fbranch-probabilities @gol
 -fbranch-target-load-optimize  -fbranch-target-load-optimize2 @gol
@@ -8688,6 +8689,9 @@ Align the start of functions to the next power-of-two greater than
 @option{-falign-functions=32} aligns functions to the next 32-byte
 boundary, but @option{-falign-functions=24} aligns to the next
 32-byte boundary only if this can be done by skipping 23 bytes or less.
+This is an optimization of code performance and alignment is ignored for
+functions considered cold.  If alignment is required for all functions,
+use @option{-fmin-function-alignment}.
 
 @option{-fno-align-functions} and @option{-falign-functions=1} are
 equivalent and mean that functions are not aligned.
@@ -8733,6 +8737,8 @@ Align loops to a power-of-two boundary, skipping up to @var{n} bytes
 like @option{-falign-functions}.  If the loops are
 executed many times, this makes up for any execution of the dummy
 operations.
+This is an optimization of code performance and alignment is ignored for
+loops considered cold.
 
 @option{-fno-align-loops} and @option{-falign-loops=1} are
 equivalent and mean that loops are not aligned.
@@ -8749,6 +8755,8 @@ Align branch targets to a power-of-two boundary, for branch targets
 where the targets can only be reached by jumping, skipping up to @var{n}
 bytes like @option{-falign-functions}.  In this case, no dummy operations
 need be executed.
+This is an optimization of code performance and alignment is ignored for
+jumps considered cold.
 
 @option{-fno-align-jumps} and @option{-falign-jumps=1} are
 equivalent and mean that loops are not aligned.
@@ -8758,6 +8766,15 @@ The maximum allowed @var{n} option value is 65536.
 
 Enabled at levels @option{-O2}, @option{-O3}.
 
+@item -fmin-function-alignment
+@itemx -fmin-function-alignment=@var{n}
+@opindex fmin-function-alignment
+Specify minimal alignment of functions to the next power-of-two greater than or
+equal to @var{n}. Unlike @option{-falign-functions} this alignment is applied
+also to all functions (even those considered cold).  The alignment is also not
+affected by @option{-flimit-function-alignment}
+
+
 @item -funit-at-a-time
 @opindex funit-at-a-time
 This option is left for compatibility reasons. @option{-funit-at-a-time}
diff --git a/gcc/lto-streamer.h b/gcc/lto-streamer.h
index 689fbcf64ee..ef0aec79863 100644
--- a/gcc/lto-streamer.h
+++ b/gcc/lto-streamer.h
@@ -129,7 +129,7 @@ along with GCC; see the file COPYING3.  If not see
      form followed by the data for the string.  */
 
 #define LTO_major_version 6
-#define LTO_minor_version 2
+#define LTO_minor_version 3
 
 typedef unsigned char	lto_decl_flags_t;
 
diff --git a/gcc/varasm.c b/gcc/varasm.c
index 1b9030029ac..311b899ced7 100644
--- a/gcc/varasm.c
+++ b/gcc/varasm.c
@@ -1778,6 +1778,11 @@ assemble_start_function (tree decl, const char *fnname)
 
   /* Tell assembler to move to target machine's alignment for functions.  */
   align = floor_log2 (align / BITS_PER_UNIT);
+  /* Handle forced alignment.  This really ought to apply to all functions,
+     since it is used by patchable entries.  */
+  if (flag_min_function_alignment)
+    align = MAX (align, floor_log2 (flag_min_function_alignment));
+
   if (align > 0)
     {
       ASM_OUTPUT_ALIGN (asm_out_file, align);
-- 
2.35.3