public inbox for gentoo-commits@lists.gentoo.org
 help / color / mirror / Atom feed
From: "Sam James" <sam@gentoo.org>
To: gentoo-commits@lists.gentoo.org
Subject: [gentoo-commits] proj/gcc-patches:master commit in: 15.0.0/gentoo/
Date: Sun, 29 Sep 2024 22:56:13 +0000 (UTC)	[thread overview]
Message-ID: <1727650528.891af478cc93ae3e2e176ed9739166799b814a05.sam@gentoo> (raw)

commit:     891af478cc93ae3e2e176ed9739166799b814a05
Author:     Sam James <sam <AT> gentoo <DOT> org>
AuthorDate: Sun Sep 29 22:55:02 2024 +0000
Commit:     Sam James <sam <AT> gentoo <DOT> org>
CommitDate: Sun Sep 29 22:55:28 2024 +0000
URL:        https://gitweb.gentoo.org/proj/gcc-patches.git/commit/?id=891af478

15.0.0: add 71_all_PR116878-libcall.patch

It's been approved but not yet merged.

Bug: https://gcc.gnu.org/PR116616
Bug: https://gcc.gnu.org/PR116878
Signed-off-by: Sam James <sam <AT> gentoo.org>

 15.0.0/gentoo/71_all_PR116878-libcall.patch | 215 ++++++++++++++++++++++++++++
 15.0.0/gentoo/README.history                |   4 +
 2 files changed, 219 insertions(+)

diff --git a/15.0.0/gentoo/71_all_PR116878-libcall.patch b/15.0.0/gentoo/71_all_PR116878-libcall.patch
new file mode 100644
index 0000000..781196d
--- /dev/null
+++ b/15.0.0/gentoo/71_all_PR116878-libcall.patch
@@ -0,0 +1,215 @@
+https://gcc.gnu.org/PR116878
+https://gcc.gnu.org/PR116616
+
+From b9883d33e6338b1a44247b3f8772a5e5d2757593 Mon Sep 17 00:00:00 2001
+Message-ID: <b9883d33e6338b1a44247b3f8772a5e5d2757593.1727650422.git.sam@gentoo.org>
+From: Filip Kastl <fkastl@suse.cz>
+Date: Tue, 10 Sep 2024 19:22:27 +0200
+Subject: [PATCH] gimple ssa: Don't use __builtin_popcount in switch exp
+ transform
+
+Switch exponential transformation in the switch conversion pass
+currently generates
+
+tmp1 = __builtin_popcount (var);
+tmp2 = tmp1 == 1;
+
+when inserting code to determine if var is power of two.  If the target
+doesn't support expanding the builtin as special instructions switch
+conversion relies on this whole pattern being expanded as bitmagic.
+However, it is possible that other GIMPLE optimizations move the two
+statements of the pattern apart.  In that case the builtin becomes a
+libgcc call in the final binary.  The call is slow and in case of
+freestanding programs can result in linking error (this bug was
+originally found while compiling Linux kernel).
+
+This patch modifies switch conversion to insert the bitmagic
+(var ^ (var - 1)) > (var - 1) instead of the builtin.
+
+gcc/ChangeLog:
+
+	PR tree-optimization/116616
+	* tree-switch-conversion.cc (can_pow2p): Remove this function.
+	(gen_pow2p): Generate bitmagic instead of a builtin.  Remove the
+	TYPE parameter.
+	(switch_conversion::is_exp_index_transform_viable): Don't call
+	can_pow2p.
+	(switch_conversion::exp_index_transform): Call gen_pow2p without
+	the TYPE parameter.
+	* tree-switch-conversion.h: Remove
+	m_exp_index_transform_pow2p_type.
+
+gcc/testsuite/ChangeLog:
+
+	PR tree-optimization/116616
+	* gcc.target/i386/switch-exp-transform-1.c: Don't test for
+	presence of the POPCOUNT internal fn call.
+
+Signed-off-by: Filip Kastl <fkastl@suse.cz>
+---
+ .../gcc.target/i386/switch-exp-transform-1.c  |  7 +-
+ gcc/tree-switch-conversion.cc                 | 84 +++++--------------
+ gcc/tree-switch-conversion.h                  |  6 +-
+ 3 files changed, 23 insertions(+), 74 deletions(-)
+
+diff --git a/gcc/testsuite/gcc.target/i386/switch-exp-transform-1.c b/gcc/testsuite/gcc.target/i386/switch-exp-transform-1.c
+index a8c9e03e515f..4832f5b52c33 100644
+--- a/gcc/testsuite/gcc.target/i386/switch-exp-transform-1.c
++++ b/gcc/testsuite/gcc.target/i386/switch-exp-transform-1.c
+@@ -1,10 +1,8 @@
+ /* { dg-do compile } */
+-/* { dg-options "-O2 -fdump-tree-switchconv -fdump-tree-widening_mul -mpopcnt -mbmi" } */
++/* { dg-options "-O2 -fdump-tree-switchconv -mbmi" } */
+ 
+ /* Checks that exponential index transform enables switch conversion to convert
+-   this switch into an array lookup.  Also checks that the "index variable is a
+-   power of two" check has been generated and that it has been later expanded
+-   into an internal function.  */
++   this switch into an array lookup.  */
+ 
+ int foo(unsigned bar)
+ {
+@@ -30,4 +28,3 @@ int foo(unsigned bar)
+ }
+ 
+ /* { dg-final { scan-tree-dump "CSWTCH" "switchconv" } } */
+-/* { dg-final { scan-tree-dump "POPCOUNT" "widening_mul" } } */
+diff --git a/gcc/tree-switch-conversion.cc b/gcc/tree-switch-conversion.cc
+index c1332a260943..00426d400006 100644
+--- a/gcc/tree-switch-conversion.cc
++++ b/gcc/tree-switch-conversion.cc
+@@ -133,75 +133,33 @@ gen_log2 (tree op, location_t loc, tree *result, tree type)
+   return stmts;
+ }
+ 
+-/* Is it possible to efficiently check that a value of TYPE is a power of 2?
+-
+-   If yes, returns TYPE.  If no, returns NULL_TREE.  May also return another
+-   type.  This indicates that logarithm of the variable can be computed but
+-   only after it is converted to this type.
+-
+-   Also see gen_pow2p.  */
+-
+-static tree
+-can_pow2p (tree type)
+-{
+-  /* __builtin_popcount supports the unsigned type or its long and long long
+-     variants.  Choose the smallest out of those that can still fit TYPE.  */
+-  int prec = TYPE_PRECISION (type);
+-  int i_prec = TYPE_PRECISION (unsigned_type_node);
+-  int li_prec = TYPE_PRECISION (long_unsigned_type_node);
+-  int lli_prec = TYPE_PRECISION (long_long_unsigned_type_node);
+-
+-  if (prec <= i_prec)
+-    return unsigned_type_node;
+-  else if (prec <= li_prec)
+-    return long_unsigned_type_node;
+-  else if (prec <= lli_prec)
+-    return long_long_unsigned_type_node;
+-  else
+-    return NULL_TREE;
+-}
+-
+-/* Build a sequence of gimple statements checking that OP is a power of 2.  Use
+-   special optabs if target supports them.  Return the result as a
+-   boolean_type_node ssa name through RESULT.  Assumes that OP's value will
+-   be non-negative.  The generated check may give arbitrary answer for negative
+-   values.
+-
+-   Before computing the check, OP may have to be converted to another type.
+-   This should be specified in TYPE.  Use can_pow2p to decide what this type
+-   should be.
+-
+-   Should only be used if can_pow2p returns true for type of OP.  */
++/* Build a sequence of gimple statements checking that OP is a power of 2.
++   Return the result as a boolean_type_node ssa name through RESULT.  Assumes
++   that OP's value will be non-negative.  The generated check may give
++   arbitrary answer for negative values.  */
+ 
+ static gimple_seq
+-gen_pow2p (tree op, location_t loc, tree *result, tree type)
++gen_pow2p (tree op, location_t loc, tree *result)
+ {
+   gimple_seq stmts = NULL;
+   gimple_stmt_iterator gsi = gsi_last (stmts);
+ 
+-  built_in_function fn;
+-  if (type == unsigned_type_node)
+-    fn = BUILT_IN_POPCOUNT;
+-  else if (type == long_unsigned_type_node)
+-    fn = BUILT_IN_POPCOUNTL;
+-  else
+-    {
+-      fn = BUILT_IN_POPCOUNTLL;
+-      gcc_checking_assert (type == long_long_unsigned_type_node);
+-    }
++  tree type = TREE_TYPE (op);
++  tree utype = unsigned_type_for (type);
+ 
+-  tree orig_type = TREE_TYPE (op);
++  /* Build (op ^ (op - 1)) > (op - 1).  */
+   tree tmp1;
+-  if (type != orig_type)
+-    tmp1 = gimple_convert (&gsi, false, GSI_NEW_STMT, loc, type, op);
+-  else
++  if (types_compatible_p (type, utype))
+     tmp1 = op;
+-  /* Build __builtin_popcount{l,ll} (op) == 1.  */
+-  tree tmp2 = gimple_build (&gsi, false, GSI_NEW_STMT, loc,
+-			    as_combined_fn (fn), integer_type_node, tmp1);
+-  *result = gimple_build (&gsi, false, GSI_NEW_STMT, loc, EQ_EXPR,
+-			  boolean_type_node, tmp2,
+-			  build_one_cst (integer_type_node));
++  else
++    tmp1 = gimple_convert (&gsi, false, GSI_NEW_STMT, loc, utype, op);
++  tree tmp2 = gimple_build (&gsi, false, GSI_NEW_STMT, loc, MINUS_EXPR, utype,
++			    tmp1, build_one_cst (utype));
++  tree tmp3 = gimple_build (&gsi, false, GSI_NEW_STMT, loc, BIT_XOR_EXPR,
++			    utype, tmp1, tmp2);
++  *result = gimple_build (&gsi, false, GSI_NEW_STMT, loc, GT_EXPR,
++			  boolean_type_node, tmp3, tmp2);
++
+   return stmts;
+ }
+ 
+@@ -371,9 +329,6 @@ switch_conversion::is_exp_index_transform_viable (gswitch *swtch)
+   m_exp_index_transform_log2_type = can_log2 (index_type, opt_type);
+   if (!m_exp_index_transform_log2_type)
+     return false;
+-  m_exp_index_transform_pow2p_type = can_pow2p (index_type);
+-  if (!m_exp_index_transform_pow2p_type)
+-    return false;
+ 
+   /* Check that each case label corresponds only to one value
+      (no case 1..3).  */
+@@ -467,8 +422,7 @@ switch_conversion::exp_index_transform (gswitch *swtch)
+   new_edge2->probability = profile_probability::even ();
+ 
+   tree tmp;
+-  gimple_seq stmts = gen_pow2p (index, UNKNOWN_LOCATION, &tmp,
+-				m_exp_index_transform_pow2p_type);
++  gimple_seq stmts = gen_pow2p (index, UNKNOWN_LOCATION, &tmp);
+   gsi = gsi_last_bb (cond_bb);
+   gsi_insert_seq_after (&gsi, stmts, GSI_LAST_NEW_STMT);
+   gcond *stmt_cond = gimple_build_cond (NE_EXPR, tmp, boolean_false_node,
+diff --git a/gcc/tree-switch-conversion.h b/gcc/tree-switch-conversion.h
+index 14610499e5fb..6468995eb316 100644
+--- a/gcc/tree-switch-conversion.h
++++ b/gcc/tree-switch-conversion.h
+@@ -920,11 +920,9 @@ public:
+   bool m_exp_index_transform_applied;
+ 
+   /* If switch conversion decided exponential index transform is viable, here
+-     will be stored the types to which index variable has to be converted
+-     before the logarithm and the "is power of 2" operations which are part of
+-     the transform.  */
++     will be stored the type to which index variable has to be converted
++     before the logarithm operation which is a part of the transform.  */
+   tree m_exp_index_transform_log2_type;
+-  tree m_exp_index_transform_pow2p_type;
+ };
+ 
+ void
+-- 
+2.46.2
+

diff --git a/15.0.0/gentoo/README.history b/15.0.0/gentoo/README.history
index a035c01..6c31450 100644
--- a/15.0.0/gentoo/README.history
+++ b/15.0.0/gentoo/README.history
@@ -1,3 +1,7 @@
+13	29 September 2024
+
+	+ 71_all_PR116878-libcall.patch
+
 12	1 September 2024
 
 	- 32_all_PR116470-debug-bootstrap.patch


             reply	other threads:[~2024-09-29 22:56 UTC|newest]

Thread overview: 224+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2024-09-29 22:56 Sam James [this message]
  -- strict thread matches above, loose matches on Subject: below --
2025-04-18 19:21 [gentoo-commits] proj/gcc-patches:master commit in: 15.0.0/gentoo/ Sam James
2025-04-16 13:30 Sam James
2025-04-16 10:04 Sam James
2025-04-16  9:58 Sam James
2025-04-15 21:09 Sam James
2025-04-15 16:40 Sam James
2025-04-14 22:59 Sam James
2025-04-14 22:47 Sam James
2025-04-14 22:44 Sam James
2025-04-14 21:06 Sam James
2025-04-14 20:51 Sam James
2025-04-14 16:08 Sam James
2025-04-14 16:08 Sam James
2025-04-13 22:54 Sam James
2025-04-10 21:59 Sam James
2025-04-10  9:08 Sam James
2025-04-09 15:27 Sam James
2025-04-09 13:27 Sam James
2025-04-09 12:42 Sam James
2025-04-07 18:02 Sam James
2025-04-07  7:07 Sam James
2025-04-07  6:54 Sam James
2025-04-06 23:08 Sam James
2025-04-05 15:33 Sam James
2025-04-05  8:02 Sam James
2025-04-05  1:43 Sam James
2025-04-04 19:06 Sam James
2025-04-02 18:48 Sam James
2025-04-02 18:03 Sam James
2025-04-02 16:14 Sam James
2025-04-02 13:56 Sam James
2025-04-02  4:59 Sam James
2025-04-01 14:46 Sam James
2025-04-01 14:46 Sam James
2025-03-31 22:16 Sam James
2025-03-31 22:03 Sam James
2025-03-31  4:05 Sam James
2025-03-29 20:31 Sam James
2025-03-29 14:33 Sam James
2025-03-29 13:51 Sam James
2025-03-26  6:25 Sam James
2025-03-25 10:27 Sam James
2025-03-25  8:38 Sam James
2025-03-25  2:32 Sam James
2025-03-25  1:27 Sam James
2025-03-24  0:35 Sam James
2025-03-21 19:31 Sam James
2025-03-21 17:21 Sam James
2025-03-21 16:23 Sam James
2025-03-21 11:20 Sam James
2025-03-21  8:51 Sam James
2025-03-21  6:07 Sam James
2025-03-20 22:08 Sam James
2025-03-20  1:59 Sam James
2025-03-20  1:59 Sam James
2025-03-16 22:37 Sam James
2025-03-14 14:46 Sam James
2025-03-14 13:37 Sam James
2025-03-13 16:48 Sam James
2025-03-13 10:08 Sam James
2025-03-11 10:32 Sam James
2025-03-07 16:54 Sam James
2025-03-03 16:38 Sam James
2025-03-01 10:33 Sam James
2025-03-01  6:50 Sam James
2025-02-17  1:30 Sam James
2025-02-13  9:23 Sam James
2025-02-12 15:12 Sam James
2025-02-10 21:22 Sam James
2025-02-09 23:58 Sam James
2025-02-07 23:37 Sam James
2025-02-07 21:19 Sam James
2025-02-03 22:04 Sam James
2025-02-02 22:41 Sam James
2025-01-29 20:21 Sam James
2025-01-26 22:52 Sam James
2025-01-22 16:27 Sam James
2025-01-19 22:43 Sam James
2025-01-16 23:11 Sam James
2025-01-16 23:11 Sam James
2025-01-15 11:41 Sam James
2025-01-14 16:22 Sam James
2025-01-14 15:06 Sam James
2025-01-14 15:06 Sam James
2025-01-14 12:29 Sam James
2025-01-14  8:43 Sam James
2025-01-14  8:40 Sam James
2025-01-13 13:58 Sam James
2025-01-13  6:00 Sam James
2025-01-13  3:40 Sam James
2025-01-13  3:23 Sam James
2025-01-13  3:20 Sam James
2025-01-13  0:20 Sam James
2025-01-12 18:53 Sam James
2025-01-11 12:53 Sam James
2025-01-08 21:51 Sam James
2025-01-06 10:50 Sam James
2025-01-06 10:03 Sam James
2025-01-06  4:49 Sam James
2025-01-06  4:44 Sam James
2025-01-06  4:13 Sam James
2025-01-06  4:13 Sam James
2025-01-06  4:13 Sam James
2025-01-06  4:03 Sam James
2025-01-05 23:19 Sam James
2025-01-03  3:07 Sam James
2024-12-30  1:05 Sam James
2024-12-29 10:00 Sam James
2024-12-27 15:14 Sam James
2024-12-24 20:48 Sam James
2024-12-22 22:46 Sam James
2024-12-20 11:25 Sam James
2024-12-20  5:57 Sam James
2024-12-20  1:55 Sam James
2024-12-19 18:34 Sam James
2024-12-13 13:23 Sam James
2024-12-13 11:52 Sam James
2024-12-13  5:08 Sam James
2024-12-12 12:28 Sam James
2024-12-11  4:41 Sam James
2024-12-11  0:58 Sam James
2024-12-10 19:19 Sam James
2024-12-10 14:55 Sam James
2024-12-10  5:19 Sam James
2024-12-10  5:13 Sam James
2024-12-10  5:11 Sam James
2024-12-10  5:07 Sam James
2024-12-09  3:05 Sam James
2024-12-08 22:41 Sam James
2024-12-06 17:33 Sam James
2024-12-04 20:40 Sam James
2024-12-01 22:51 Sam James
2024-12-01 22:51 Sam James
2024-11-30 11:30 Sam James
2024-11-27 17:42 Sam James
2024-11-25 15:10 Sam James
2024-11-25  3:01 Sam James
2024-11-25  3:00 Sam James
2024-11-25  3:00 Sam James
2024-11-24 22:42 Sam James
2024-11-18 17:25 Sam James
2024-11-18 10:42 Sam James
2024-11-18 10:42 Sam James
2024-11-18  9:25 Sam James
2024-11-18  9:25 Sam James
2024-11-14 18:38 Sam James
2024-11-13  4:26 Sam James
2024-11-13  0:16 Sam James
2024-11-12  2:33 Sam James
2024-11-11 19:46 Sam James
2024-11-11 19:46 Sam James
2024-11-10 22:41 Sam James
2024-11-09 16:24 Sam James
2024-11-09  7:55 Sam James
2024-11-08  8:22 Sam James
2024-11-07 16:13 Sam James
2024-11-03 23:16 Sam James
2024-11-01  8:24 Sam James
2024-11-01  8:24 Sam James
2024-11-01  8:18 Sam James
2024-11-01  8:17 Sam James
2024-10-30 16:03 Sam James
2024-10-29 19:17 Sam James
2024-10-28 21:32 Sam James
2024-10-28  8:09 Sam James
2024-10-23 15:40 Sam James
2024-10-22 19:09 Sam James
2024-10-22 18:34 Sam James
2024-10-21 12:33 Sam James
2024-10-21 12:27 Sam James
2024-10-21 12:26 Sam James
2024-10-21 11:45 Sam James
2024-10-20 22:42 Sam James
2024-10-18 14:05 Sam James
2024-10-18 10:35 Sam James
2024-10-17 23:33 Sam James
2024-10-17 23:03 Sam James
2024-10-17  5:01 Sam James
2024-10-17  4:15 Sam James
2024-10-13 22:48 Sam James
2024-10-07  2:45 Sam James
2024-10-04 10:37 Sam James
2024-10-04  9:28 Sam James
2024-10-02 19:45 Sam James
2024-09-30 14:05 Sam James
2024-09-24  1:41 Sam James
2024-09-23 15:23 Sam James
2024-09-02  2:28 Sam James
2024-08-26 13:44 Sam James
2024-08-26  6:24 Sam James
2024-08-23 13:51 Sam James
2024-08-20 20:31 Sam James
2024-08-19 18:43 Sam James
2024-08-14  9:48 Sam James
2024-08-14  2:57 Sam James
2024-08-11 22:40 Sam James
2024-08-09 19:54 Sam James
2024-08-09 19:54 Sam James
2024-08-09 19:47 Sam James
2024-08-09 19:25 Sam James
2024-08-08 11:10 Sam James
2024-08-08 11:06 Sam James
2024-08-08 11:03 Sam James
2024-08-05  9:09 Sam James
2024-08-05  1:54 Sam James
2024-08-05  1:51 Sam James
2024-08-02 20:39 Sam James
2024-08-01 14:40 Sam James
2024-07-28 23:34 Sam James
2024-07-22  1:11 Sam James
2024-07-19 11:14 Sam James
2024-07-18  0:45 Sam James
2024-07-14 23:36 Sam James
2024-06-28 12:49 Sam James
2024-06-27  0:02 Sam James
2024-06-26 23:57 Sam James
2024-06-16 22:45 Sam James
2024-06-10 20:18 Sam James
2024-06-10 17:28 Sam James
2024-06-10 17:28 Sam James
2024-06-10  2:08 Sam James
2024-06-08 17:03 Sam James
2024-06-08 17:03 Sam James

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=1727650528.891af478cc93ae3e2e176ed9739166799b814a05.sam@gentoo \
    --to=sam@gentoo.org \
    --cc=gentoo-commits@lists.gentoo.org \
    --cc=gentoo-dev@lists.gentoo.org \
    /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