public inbox for gcc-patches@gcc.gnu.org
 help / color / mirror / Atom feed
From: "Fei Gao" <gaofei@eswincomputing.com>
To: gcc-patches <gcc-patches@gcc.gnu.org>
Cc: jeffreyalaw <jeffreyalaw@gmail.com>,
	 "Kito Cheng" <kito.cheng@gmail.com>,
	 "Palmer Dabbelt" <palmer@dabbelt.com>
Subject: [PING]  [PATCH 2/3] RISC-V: optimize stack manipulation in save-restore
Date: Fri, 3 Feb 2023 16:52:44 +0800	[thread overview]
Message-ID: <20230203165243892433118@eswincomputing.com> (raw)
In-Reply-To: <20221201100332.22226-3-gaofei@eswincomputing.com>


Gentle ping.

The patch I previously submitted:
| Date: Wed, 30 Nov 2022 00:38:08 -0800
| Subject: [PATCH] RISC-V: optimize stack manipulation in save-restore
| Message-ID: <gaofei@eswincomputing.com>

I split the patches as per Palmer's review comment.

BR
Fei

>The stack that save-restore reserves is not well accumulated in stack allocation and deallocation.
>This patch allows less instructions to be used in stack allocation and deallocation if save-restore enabled.
>
>before patch:
>  bar:
>    call	t0,__riscv_save_4
>    addi	sp,sp,-64
>    ...
>    li	t0,-12288
>    addi	t0,t0,-1968 # optimized out after patch
>    add	sp,sp,t0 # prologue
>    ...
>    li	t0,12288 # epilogue
>    addi	t0,t0,2000 # optimized out after patch
>    add	sp,sp,t0
>    ...
>    addi	sp,sp,32
>    tail	__riscv_restore_4
>
>after patch:
>  bar:
>    call	t0,__riscv_save_4
>    addi	sp,sp,-2032
>    ...
>    li	t0,-12288
>    add	sp,sp,t0 # prologue
>    ...
>    li	t0,12288 # epilogue
>    add	sp,sp,t0
>    ...
>    addi	sp,sp,2032
>    tail	__riscv_restore_4
>
>gcc/ChangeLog:
>
>        * config/riscv/riscv.cc (riscv_expand_prologue): consider save-restore in stack allocation.
>        (riscv_expand_epilogue): consider save-restore in stack deallocation.
>
>gcc/testsuite/ChangeLog:
>
>        * gcc.target/riscv/stack_save_restore.c: New test.
>---
> gcc/config/riscv/riscv.cc                     | 50 ++++++++++---------
> .../gcc.target/riscv/stack_save_restore.c     | 40 +++++++++++++++
> 2 files changed, 66 insertions(+), 24 deletions(-)
> create mode 100644 gcc/testsuite/gcc.target/riscv/stack_save_restore.c
>
>diff --git a/gcc/config/riscv/riscv.cc b/gcc/config/riscv/riscv.cc
>index f0bbcd6d6be..a50f2303032 100644
>--- a/gcc/config/riscv/riscv.cc
>+++ b/gcc/config/riscv/riscv.cc
>@@ -5010,12 +5010,12 @@ void
> riscv_expand_prologue (void)
> {
>   struct riscv_frame_info *frame = &cfun->machine->frame;
>-  poly_int64 size = frame->total_size;
>+  poly_int64 remaining_size = frame->total_size;
>   unsigned mask = frame->mask;
>   rtx insn;
>
>   if (flag_stack_usage_info)
>-    current_function_static_stack_size = constant_lower_bound (size);
>+    current_function_static_stack_size = constant_lower_bound (remaining_size);
>
>   if (cfun->machine->naked_p)
>     return;
>@@ -5026,7 +5026,7 @@ riscv_expand_prologue (void)
>       rtx dwarf = NULL_RTX;
>       dwarf = riscv_adjust_libcall_cfi_prologue ();
>
>-      size -= frame->save_libcall_adjustment;
>+      remaining_size -= frame->save_libcall_adjustment;
>       insn = emit_insn (riscv_gen_gpr_save_insn (frame));
>       frame->mask = 0; /* Temporarily fib that we need not save GPRs.  */
>
>@@ -5037,16 +5037,14 @@ riscv_expand_prologue (void)
>   /* Save the registers.  */
>   if ((frame->mask | frame->fmask) != 0)
>     {
>-      HOST_WIDE_INT step1 = riscv_first_stack_step (frame, frame->total_size);
>-      if (size.is_constant ())
>-	step1 = MIN (size.to_constant(), step1);
>+      HOST_WIDE_INT step1 = riscv_first_stack_step (frame, remaining_size);
>
>       insn = gen_add3_insn (stack_pointer_rtx,
>     stack_pointer_rtx,
>     GEN_INT (-step1));
>       RTX_FRAME_RELATED_P (emit_insn (insn)) = 1;
>-      size -= step1;
>-      riscv_for_each_saved_reg (size, riscv_save_reg, false, false);
>+      remaining_size -= step1;
>+      riscv_for_each_saved_reg (remaining_size, riscv_save_reg, false, false);
>     }
>
>   frame->mask = mask; /* Undo the above fib.  */
>@@ -5055,29 +5053,29 @@ riscv_expand_prologue (void)
>   if (frame_pointer_needed)
>     {
>       insn = gen_add3_insn (hard_frame_pointer_rtx, stack_pointer_rtx,
>-	    GEN_INT ((frame->hard_frame_pointer_offset - size).to_constant ()));
>+	    GEN_INT ((frame->hard_frame_pointer_offset - remaining_size).to_constant ()));
>       RTX_FRAME_RELATED_P (emit_insn (insn)) = 1;
>
>       riscv_emit_stack_tie ();
>     }
>
>   /* Allocate the rest of the frame.  */
>-  if (known_gt (size, 0))
>+  if (known_gt (remaining_size, 0))
>     {
>       /* Two step adjustment:
> 1.scalable frame. 2.constant frame.  */
>       poly_int64 scalable_frame (0, 0);
>-      if (!size.is_constant ())
>+      if (!remaining_size.is_constant ())
> {
>   /* First for scalable frame.  */
>-	  poly_int64 scalable_frame = size;
>-	  scalable_frame.coeffs[0] = size.coeffs[1];
>+	  poly_int64 scalable_frame = remaining_size;
>+	  scalable_frame.coeffs[0] = remaining_size.coeffs[1];
>   riscv_v_adjust_scalable_frame (stack_pointer_rtx, scalable_frame, false);
>-	  size -= scalable_frame;
>+	  remaining_size -= scalable_frame;
> }
>
>       /* Second step for constant frame.  */
>-      HOST_WIDE_INT constant_frame = size.to_constant ();
>+      HOST_WIDE_INT constant_frame = remaining_size.to_constant ();
>       if (constant_frame == 0)
> return;
>
>@@ -5142,6 +5140,8 @@ riscv_expand_epilogue (int style)
>   HOST_WIDE_INT step2 = 0;
>   bool use_restore_libcall = ((style == NORMAL_RETURN)
>       && riscv_use_save_libcall (frame));
>+  unsigned libcall_size = use_restore_libcall ?
>+                            frame->save_libcall_adjustment : 0;
>   rtx ra = gen_rtx_REG (Pmode, RETURN_ADDR_REGNUM);
>   rtx insn;
>
>@@ -5212,13 +5212,18 @@ riscv_expand_epilogue (int style)
>       REG_NOTES (insn) = dwarf;
>     }
>
>+  if (use_restore_libcall)
>+    frame->mask = 0; /* Temporarily fib for GPRs.  */
>+
>   /* If we need to restore registers, deallocate as much stack as
>      possible in the second step without going out of range.  */
>   if ((frame->mask | frame->fmask) != 0)
>-    {
>-      step2 = riscv_first_stack_step (frame, frame->total_size);
>-      step1 -= step2;
>-    }
>+    step2 = riscv_first_stack_step (frame, frame->total_size - libcall_size);
>+
>+  if (use_restore_libcall)
>+    frame->mask = mask; /* Undo the above fib.  */
>+
>+  step1 -= step2 + libcall_size;
>
>   /* Set TARGET to BASE + STEP1.  */
>   if (known_gt (step1, 0))
>@@ -5272,15 +5277,12 @@ riscv_expand_epilogue (int style)
>     frame->mask = 0; /* Temporarily fib that we need not save GPRs.  */
>
>   /* Restore the registers.  */
>-  riscv_for_each_saved_reg (frame->total_size - step2, riscv_restore_reg,
>+  riscv_for_each_saved_reg (frame->total_size - step2 - libcall_size,
>+                            riscv_restore_reg,
>     true, style == EXCEPTION_RETURN);
>
>   if (use_restore_libcall)
>-    {
>       frame->mask = mask; /* Undo the above fib.  */
>-      gcc_assert (step2 >= frame->save_libcall_adjustment);
>-      step2 -= frame->save_libcall_adjustment;
>-    }
>
>   if (need_barrier_p)
>     riscv_emit_stack_tie ();
>diff --git a/gcc/testsuite/gcc.target/riscv/stack_save_restore.c b/gcc/testsuite/gcc.target/riscv/stack_save_restore.c
>new file mode 100644
>index 00000000000..522e706cfbf
>--- /dev/null
>+++ b/gcc/testsuite/gcc.target/riscv/stack_save_restore.c
>@@ -0,0 +1,40 @@
>+/* { dg-do compile } */
>+/* { dg-options "-march=rv32imafc -mabi=ilp32f -msave-restore -O2 -fno-schedule-insns -fno-schedule-insns2 -fno-unroll-loops -fno-peel-loops -fno-lto" } */
>+/* { dg-final { check-function-bodies "**" "" } } */
>+
>+char my_getchar();
>+float getf();
>+
>+/*
>+**bar:
>+**	call	t0,__riscv_save_4
>+**	addi	sp,sp,-2032
>+**	...
>+**	li	t0,-12288
>+**	add	sp,sp,t0
>+**	...
>+**	li	t0,12288
>+**	add	sp,sp,t0
>+**	...
>+**	addi	sp,sp,2032
>+**	tail	__riscv_restore_4
>+*/
>+int bar()
>+{
>+  float volatile farray[3568];
>+
>+  float sum = 0;
>+  float f1 = getf();
>+  float f2 = getf();
>+  float f3 = getf();
>+  float f4 = getf();
>+
>+  for (int i = 0; i < 3568; i++)
>+  {
>+    farray[i] = my_getchar() * 1.2;
>+    sum += farray[i];
>+  }
>+
>+  return sum + f1 + f2 + f3 + f4;
>+}
>+
>--
>2.17.1

  reply	other threads:[~2023-02-03  8:52 UTC|newest]

Thread overview: 16+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2022-12-01 10:03 [PATCH 0/3] " Fei Gao
2022-12-01 10:03 ` [PATCH 1/3] RISC-V: add a new parameter in riscv_first_stack_step Fei Gao
2023-02-03  8:52   ` [PING][PATCH " Fei Gao
2023-04-16 16:40   ` [PATCH " Jeff Law
2023-04-17 18:09   ` Jeff Law
2022-12-01 10:03 ` [PATCH 2/3] RISC-V: optimize stack manipulation in save-restore Fei Gao
2023-02-03  8:52   ` Fei Gao [this message]
2023-04-16 16:45   ` Jeff Law
2023-04-17 22:51   ` Jeff Law
2022-12-01 10:03 ` [PATCH 3/3] RISC-V: make the stack manipulation codes more readable Fei Gao
2023-02-03  8:52   ` [PING] " Fei Gao
2023-04-18  0:14   ` Jeff Law
2023-02-03  8:52 ` [PING] [PATCH 0/3] RISC-V: optimize stack manipulation in save-restore Fei Gao
2023-02-09  2:21   ` [PING 2] " Fei Gao
2023-02-16  7:17   ` Fei Gao
2023-02-16 14:39     ` Jeff Law

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=20230203165243892433118@eswincomputing.com \
    --to=gaofei@eswincomputing.com \
    --cc=gcc-patches@gcc.gnu.org \
    --cc=jeffreyalaw@gmail.com \
    --cc=kito.cheng@gmail.com \
    --cc=palmer@dabbelt.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for read-only IMAP folder(s) and NNTP newsgroup(s).