[PATCH 0/3] [linaro/gcc-4_9-branch] backport fixes to jump-threading

[PATCH 0/3] [linaro/gcc-4_9-branch] backport fixes to jump-threading

[Sebastian Pop]

Hi,

As I was backporting the jump-thread patches to the AOSP gcc 4.9, I found that
the linaro branch does not have the following three fixes.  Ok to commit to the
linaro/gcc-4_9-branch?

Thanks,
Sebastian

Sebastian Pop (3):
  backport patch to fix PR65048
  backport patch to fix PR65177
  backport fix for PR65735

 .../gcc/testsuite/gcc.c-torture/compile/pr65735.c  |  21 ++++
 .../testsuite/gcc.dg/tree-ssa/ssa-dom-thread-10.c  |  24 ++++
 .../testsuite/gcc.dg/tree-ssa/ssa-dom-thread-9.c   |  50 ++++++++
 gcc-4.9/gcc/tree-ssa-threadedge.c                  |  12 +-
 gcc-4.9/gcc/tree-ssa-threadupdate.c                | 133 ++++++++++++++++-----
 5 files changed, 202 insertions(+), 38 deletions(-)
 create mode 100644 gcc-4.9/gcc/testsuite/gcc.c-torture/compile/pr65735.c
 create mode 100644 gcc-4.9/gcc/testsuite/gcc.dg/tree-ssa/ssa-dom-thread-10.c
 create mode 100644 gcc-4.9/gcc/testsuite/gcc.dg/tree-ssa/ssa-dom-thread-9.c

-- 
2.1.0.243.g30d45f7

[PATCH 3/3] backport fix for PR65735

[Sebastian Pop]

	PR tree-optimization/65735
	* tree-ssa-threadedge.c (fsm_find_control_statement_thread_paths):
	Remove visited_phis argument, add visited_bbs, avoid recursing into the
	same bb rather than just into the same phi node.
	(thread_through_normal_block): Adjust caller.

	* gcc.c-torture/compile/pr65735.c: New test.
---
 .../gcc/testsuite/gcc.c-torture/compile/pr65735.c   | 21 +++++++++++++++++++++
 gcc-4.9/gcc/tree-ssa-threadedge.c                   | 12 ++++++------
 2 files changed, 27 insertions(+), 6 deletions(-)
 create mode 100644 gcc-4.9/gcc/testsuite/gcc.c-torture/compile/pr65735.c

diff --git a/gcc-4.9/gcc/testsuite/gcc.c-torture/compile/pr65735.c b/gcc-4.9/gcc/testsuite/gcc.c-torture/compile/pr65735.c
new file mode 100644
index 0000000..c30de8e
--- /dev/null
+++ b/gcc-4.9/gcc/testsuite/gcc.c-torture/compile/pr65735.c
@@ -0,0 +1,21 @@
+/* PR tree-optimization/65735 */
+
+int foo (void);
+
+void
+bar (int a, int b, int c)
+{
+  while (!a)
+    {
+      c = foo ();
+      if (c == 7)
+	c = b;
+      switch (c)
+	{
+	case 1:
+	  a = b++;
+	  if (b)
+	    b = 1;
+	}
+    }
+}
diff --git a/gcc-4.9/gcc/tree-ssa-threadedge.c b/gcc-4.9/gcc/tree-ssa-threadedge.c
index cad91ef..604123e 100644
--- a/gcc-4.9/gcc/tree-ssa-threadedge.c
+++ b/gcc-4.9/gcc/tree-ssa-threadedge.c
@@ -947,7 +947,7 @@ static int max_threaded_paths;
 
 static void
 fsm_find_control_statement_thread_paths (tree expr,
-					 pointer_set_t *visited_phis,
+					 pointer_set_t *visited_bbs,
 					 vec<basic_block, va_gc> *&path,
 					 bool seen_loop_phi)
 {
@@ -966,7 +966,7 @@ fsm_find_control_statement_thread_paths (tree expr,
     return;
 
   /* Avoid infinite recursion.  */
-  if (pointer_set_insert (visited_phis, def_stmt))
+  if (pointer_set_insert (visited_bbs, var_bb))
     return;
 
   int next_path_length = 0;
@@ -1040,7 +1040,7 @@ fsm_find_control_statement_thread_paths (tree expr,
 	{
 	  vec_safe_push (path, bbi);
 	  /* Recursively follow SSA_NAMEs looking for a constant definition.  */
-	  fsm_find_control_statement_thread_paths (arg, visited_phis, path,
+	  fsm_find_control_statement_thread_paths (arg, visited_bbs, path,
 						   seen_loop_phi);
 
 	  path->pop ();
@@ -1306,13 +1306,13 @@ thread_through_normal_block (edge e,
       vec<basic_block, va_gc> *bb_path;
       vec_alloc (bb_path, n_basic_blocks_for_fn (cfun));
       vec_safe_push (bb_path, e->dest);
-      pointer_set_t *visited_phis = pointer_set_create ();
+      pointer_set_t *visited_bbs = pointer_set_create ();
 
       max_threaded_paths = PARAM_VALUE (PARAM_MAX_FSM_THREAD_PATHS);
-      fsm_find_control_statement_thread_paths (cond, visited_phis, bb_path,
+      fsm_find_control_statement_thread_paths (cond, visited_bbs, bb_path,
 					       false);
 
-      pointer_set_destroy (visited_phis);
+      pointer_set_destroy (visited_bbs);
       vec_free (bb_path);
     }
   return 0;
-- 
2.1.0.243.g30d45f7

[PATCH 1/3] backport patch to fix PR65048

[Sebastian Pop]

PR tree-optimization/65048
	* tree-ssa-threadupdate.c (valid_jump_thread_path): New.
	(thread_through_all_blocks): Call valid_jump_thread_path.      Remove invalid
	FSM jump-thread paths.

 PR tree-optimization/65048
 * gcc.dg/tree-ssa/ssa-dom-thread-9.c: New.
---
 .../testsuite/gcc.dg/tree-ssa/ssa-dom-thread-9.c   | 50 ++++++++++++++++++++++
 gcc-4.9/gcc/tree-ssa-threadupdate.c                | 40 ++++++++++++++---
 2 files changed, 84 insertions(+), 6 deletions(-)
 create mode 100644 gcc-4.9/gcc/testsuite/gcc.dg/tree-ssa/ssa-dom-thread-9.c

diff --git a/gcc-4.9/gcc/testsuite/gcc.dg/tree-ssa/ssa-dom-thread-9.c b/gcc-4.9/gcc/testsuite/gcc.dg/tree-ssa/ssa-dom-thread-9.c
new file mode 100644
index 0000000..6be4203
--- /dev/null
+++ b/gcc-4.9/gcc/testsuite/gcc.dg/tree-ssa/ssa-dom-thread-9.c
@@ -0,0 +1,50 @@
+/* PR 65048 */
+/* { dg-do compile } */
+/* { dg-options "-O3" } */
+
+int a, b, c, d;
+void fn (void);
+
+int
+foo (x)
+{
+  switch (x)
+    {
+    case 'A':
+      return 'T';
+    case 'U':
+      return 'A';
+    }
+}
+
+void
+bar (int x, int y)
+{
+  switch (c)
+    {
+    case 'U':
+      switch (x)
+	{
+	default:
+	  fn ();
+	case 'G':
+	  switch (y)
+	    {
+	    case 'A':
+	      d = 7;
+	    }
+	}
+    }
+}
+
+void
+baz (void)
+{
+  while (1)
+    {
+      a = foo ();
+      b = foo ();
+      bar (a, b);
+    }
+}
+
diff --git a/gcc-4.9/gcc/tree-ssa-threadupdate.c b/gcc-4.9/gcc/tree-ssa-threadupdate.c
index d1b289f..3c7df9a 100644
--- a/gcc-4.9/gcc/tree-ssa-threadupdate.c
+++ b/gcc-4.9/gcc/tree-ssa-threadupdate.c
@@ -1772,6 +1772,21 @@ duplicate_seme_region (edge entry, edge exit,
   return true;
 }
 
+/* Return true when PATH is a valid jump-thread path.  */
+
+static bool
+valid_jump_thread_path (vec<jump_thread_edge *> *path)
+{
+  unsigned len = path->length ();
+
+  /* Check that the path is connected.  */
+  for (unsigned int j = 0; j < len - 1; j++)
+    if ((*path)[j]->e->dest != (*path)[j+1]->e->src)
+      return false;
+
+  return true;
+}
+
 /* Walk through all blocks and thread incoming edges to the appropriate
    outgoing edge for each edge pair recorded in THREADED_EDGES.
 
@@ -1807,12 +1822,25 @@ thread_through_all_blocks (bool may_peel_loop_headers)
       vec<jump_thread_edge *> *path = paths[i];
       edge entry = (*path)[0]->e;
 
-      if ((*path)[0]->type != EDGE_FSM_THREAD
-	  /* Do not jump-thread twice from the same block.  */
-	  || bitmap_bit_p (threaded_blocks, entry->src->index)) {
-	i++;
-	continue;
-      }
+      /* Only code-generate FSM jump-threads in this loop.  */
+      if ((*path)[0]->type != EDGE_FSM_THREAD)
+	{
+	  i++;
+	  continue;
+	}
+
+      /* Do not jump-thread twice from the same block.  */
+      if (bitmap_bit_p (threaded_blocks, entry->src->index)
+	  /* Verify that the jump thread path is still valid: a
+	     previous jump-thread may have changed the CFG, and
+	     invalidated the current path.  */
+	  || !valid_jump_thread_path (path))
+	{
+	  /* Remove invalid FSM jump-thread paths.  */
+	  delete_jump_thread_path (path);
+	  paths.unordered_remove (i);
+	  continue;
+	}
 
       unsigned len = path->length ();
       edge exit = (*path)[len - 1]->e;
-- 
2.1.0.243.g30d45f7

[PATCH 0/3] [linaro/gcc-4_9-branch] backport fixes to jump-threading

[Sebastian Pop]

Hi,

As I was backporting the jump-thread patches to the AOSP gcc 4.9, I found that
the linaro branch does not have the following three fixes.  Ok to commit to the
linaro/gcc-4_9-branch?

Thanks,
Sebastian

Sebastian Pop (3):
  backport patch to fix PR65048
  backport patch to fix PR65177
  backport fix for PR65735

 .../gcc/testsuite/gcc.c-torture/compile/pr65735.c  |  21 ++++
 .../testsuite/gcc.dg/tree-ssa/ssa-dom-thread-10.c  |  24 ++++
 .../testsuite/gcc.dg/tree-ssa/ssa-dom-thread-9.c   |  50 ++++++++
 gcc-4.9/gcc/tree-ssa-threadedge.c                  |  12 +-
 gcc-4.9/gcc/tree-ssa-threadupdate.c                | 133 ++++++++++++++++-----
 5 files changed, 202 insertions(+), 38 deletions(-)
 create mode 100644 gcc-4.9/gcc/testsuite/gcc.c-torture/compile/pr65735.c
 create mode 100644 gcc-4.9/gcc/testsuite/gcc.dg/tree-ssa/ssa-dom-thread-10.c
 create mode 100644 gcc-4.9/gcc/testsuite/gcc.dg/tree-ssa/ssa-dom-thread-9.c

-- 
2.1.0.243.g30d45f7

[PATCH 2/3] backport patch to fix PR65177

[Sebastian Pop]

        PR tree-optimization/65177
        * tree-ssa-threadupdate.c (verify_seme): Renamed verify_jump_thread.
        (bb_in_bbs): New.
        (duplicate_seme_region): Renamed duplicate_thread_path.  Redirect all
        edges not adjacent on the path to the original code.

        * gcc.dg/tree-ssa/ssa-dom-thread-10.c: New.
---
 .../testsuite/gcc.dg/tree-ssa/ssa-dom-thread-10.c  | 24 ++++++
 gcc-4.9/gcc/tree-ssa-threadupdate.c                | 93 ++++++++++++++++------
 2 files changed, 91 insertions(+), 26 deletions(-)
 create mode 100644 gcc-4.9/gcc/testsuite/gcc.dg/tree-ssa/ssa-dom-thread-10.c

diff --git a/gcc-4.9/gcc/testsuite/gcc.dg/tree-ssa/ssa-dom-thread-10.c b/gcc-4.9/gcc/testsuite/gcc.dg/tree-ssa/ssa-dom-thread-10.c
new file mode 100644
index 0000000..4acf580
--- /dev/null
+++ b/gcc-4.9/gcc/testsuite/gcc.dg/tree-ssa/ssa-dom-thread-10.c
@@ -0,0 +1,24 @@
+/* PR 65177 */
+/* { dg-do compile } */
+/* { dg-options "-O3" } */
+
+typedef struct p7_profile_s {} P7_PROFILE;
+enum p7t_statetype_e {
+  p7T_S = 4,   p7T_N = 5,   p7T_E = 7,   p7T_C = 8,   p7T_J = 10, };
+typedef struct p7_trace_s {} P7_TRACE;
+typedef struct p7_gmx_s {
+  int L;
+} P7_GMX;
+static inline int select_c(const P7_PROFILE *gm, const P7_GMX *pp, const P7_GMX *gx, int i) {
+  float path[2];
+  return ((path[0] > path[1]) ? p7T_C : p7T_E);
+}
+void p7_GOATrace(const P7_PROFILE *gm, const P7_GMX *pp, const P7_GMX *gx, P7_TRACE *tr) {
+  int i = gx->L;
+  int sprv, scur;
+  while (sprv != p7T_S)     {
+    switch (sprv) {       case p7T_C: scur = select_c(gm, pp, gx, i); break;       }
+    if ( (scur == p7T_N || scur == p7T_J || scur == p7T_C) && scur == sprv) i--;
+    sprv = scur;
+  }
+}
diff --git a/gcc-4.9/gcc/tree-ssa-threadupdate.c b/gcc-4.9/gcc/tree-ssa-threadupdate.c
index 3c7df9a..c2f02a6 100644
--- a/gcc-4.9/gcc/tree-ssa-threadupdate.c
+++ b/gcc-4.9/gcc/tree-ssa-threadupdate.c
@@ -1623,36 +1623,32 @@ bb_ends_with_multiway_branch (basic_block bb ATTRIBUTE_UNUSED)
   return false;
 }
 
-/* Verify that the REGION is a Single Entry Multiple Exits region: make sure no
-   edge other than ENTRY is entering the REGION.  */
+/* Verify that the REGION is a valid jump thread.  A jump thread is a special
+   case of SEME Single Entry Multiple Exits region in which all nodes in the
+   REGION have exactly one incoming edge.  The only exception is the first block
+   that may not have been connected to the rest of the cfg yet.  */
 
 DEBUG_FUNCTION void
-verify_seme (edge entry, basic_block *region, unsigned n_region)
+verify_jump_thread (basic_block *region, unsigned n_region)
 {
-  bitmap bbs = BITMAP_ALLOC (NULL);
-
   for (unsigned i = 0; i < n_region; i++)
-    bitmap_set_bit (bbs, region[i]->index);
+    gcc_assert (EDGE_COUNT (region[i]->preds) <= 1);
+}
 
-  for (unsigned i = 0; i < n_region; i++)
-    {
-      edge e;
-      edge_iterator ei;
-      basic_block bb = region[i];
+/* Return true when BB is one of the first N items in BBS.  */
 
-      /* All predecessors other than ENTRY->src should be in the region.  */
-      for (ei = ei_start (bb->preds); (e = ei_safe_edge (ei)); ei_next (&ei))
-	if (e != entry)
-	  gcc_assert (bitmap_bit_p (bbs, e->src->index));
-    }
+static inline bool
+bb_in_bbs (basic_block bb, basic_block *bbs, int n)
+{
+  for (int i = 0; i < n; i++)
+    if (bb == bbs[i])
+      return true;
 
-  BITMAP_FREE (bbs);
+  return false;
 }
 
-/* Duplicates a Single Entry Multiple Exit REGION (set of N_REGION basic
-   blocks).  The ENTRY edge is redirected to the duplicate of the region.  If
-   REGION is not a Single Entry region, ignore any incoming edges other than
-   ENTRY: this makes the copied region a Single Entry region.
+/* Duplicates a jump-thread path of N_REGION basic blocks.
+   The ENTRY edge is redirected to the duplicate of the region.
 
    Remove the last conditional statement in the last basic block in the REGION,
    and create a single fallthru edge pointing to the same destination as the
@@ -1664,7 +1660,7 @@ verify_seme (edge entry, basic_block *region, unsigned n_region)
    Returns false if it is unable to copy the region, true otherwise.  */
 
 static bool
-duplicate_seme_region (edge entry, edge exit,
+duplicate_thread_path (edge entry, edge exit,
 		       basic_block *region, unsigned n_region,
 		       basic_block *region_copy)
 {
@@ -1726,7 +1722,53 @@ duplicate_seme_region (edge entry, edge exit,
     }
 
   copy_bbs (region, n_region, region_copy, &exit, 1, &exit_copy, loop,
-	    split_edge_bb_loc (entry), 0);
+	    split_edge_bb_loc (entry), false);
+
+  /* Fix up: copy_bbs redirects all edges pointing to copied blocks.  The
+     following code ensures that all the edges exiting the jump-thread path are
+     redirected back to the original code: these edges are exceptions
+     invalidating the property that is propagated by executing all the blocks of
+     the jump-thread path in order.  */
+
+  for (i = 0; i < n_region; i++)
+    {
+      edge e;
+      edge_iterator ei;
+      basic_block bb = region_copy[i];
+
+      if (single_succ_p (bb))
+	{
+	  /* Make sure the successor is the next node in the path.  */
+	  gcc_assert (i + 1 == n_region
+		      || region_copy[i + 1] == single_succ_edge (bb)->dest);
+	  continue;
+	}
+
+      /* Special case the last block on the path: make sure that it does not
+	 jump back on the copied path.  */
+      if (i + 1 == n_region)
+	{
+	  FOR_EACH_EDGE (e, ei, bb->succs)
+	    if (bb_in_bbs (e->dest, region_copy, n_region - 1))
+	      {
+		basic_block orig = get_bb_original (e->dest);
+		if (orig)
+		  redirect_edge_and_branch_force (e, orig);
+	      }
+	  continue;
+	}
+
+      /* Redirect all other edges jumping to non-adjacent blocks back to the
+	 original code.  */
+      FOR_EACH_EDGE (e, ei, bb->succs)
+	if (region_copy[i + 1] != e->dest)
+	  {
+	    basic_block orig = get_bb_original (e->dest);
+	    if (orig)
+	      redirect_edge_and_branch_force (e, orig);
+	  }
+    }
+
   if (total_count)
     {
       scale_bbs_frequencies_gcov_type (region, n_region,
@@ -1743,8 +1785,7 @@ duplicate_seme_region (edge entry, edge exit,
     }
 
 #ifdef ENABLE_CHECKING
-  /* Make sure no edge other than ENTRY is entering the copied region.  */
-  verify_seme (entry, region_copy, n_region);
+  verify_jump_thread (region_copy, n_region);
 #endif
 
   /* Remove the last branch in the jump thread path.  */
@@ -1849,7 +1890,7 @@ thread_through_all_blocks (bool may_peel_loop_headers)
       for (unsigned int j = 0; j < len - 1; j++)
 	region[j] = (*path)[j]->e->dest;
 
-      if (duplicate_seme_region (entry, exit, region, len - 1, NULL))
+      if (duplicate_thread_path (entry, exit, region, len - 1, NULL))
 	{
 	  /* We do not update dominance info.  */
 	  free_dominance_info (CDI_DOMINATORS);
-- 
2.1.0.243.g30d45f7

Re: [PATCH 0/3] [linaro/gcc-4_9-branch] backport fixes to jump-threading

[Yvan Roux]

Hi Sebastian,

On 31 August 2015 at 17:33, Sebastian Pop <s.pop@samsung.com> wrote:
> Hi,
>
> As I was backporting the jump-thread patches to the AOSP gcc 4.9, I found that
> the linaro branch does not have the following three fixes.  Ok to commit to the
> linaro/gcc-4_9-branch?

Our 4.9 branch is now in maintenance mode, which means that we're only
backporting bugfixes on it, and synchronizing it with upstream 4.9
branch.  So, if you'll only backport them to AOSP and not FSF gcc 4.9,
yes it can be beneficial for our branch, but we first have to pass
them into our validation process.

Cheers,
Yvan

> Thanks,
> Sebastian
>
> Sebastian Pop (3):
>   backport patch to fix PR65048
>   backport patch to fix PR65177
>   backport fix for PR65735
>
>  .../gcc/testsuite/gcc.c-torture/compile/pr65735.c  |  21 ++++
>  .../testsuite/gcc.dg/tree-ssa/ssa-dom-thread-10.c  |  24 ++++
>  .../testsuite/gcc.dg/tree-ssa/ssa-dom-thread-9.c   |  50 ++++++++
>  gcc-4.9/gcc/tree-ssa-threadedge.c                  |  12 +-
>  gcc-4.9/gcc/tree-ssa-threadupdate.c                | 133 ++++++++++++++++-----
>  5 files changed, 202 insertions(+), 38 deletions(-)
>  create mode 100644 gcc-4.9/gcc/testsuite/gcc.c-torture/compile/pr65735.c
>  create mode 100644 gcc-4.9/gcc/testsuite/gcc.dg/tree-ssa/ssa-dom-thread-10.c
>  create mode 100644 gcc-4.9/gcc/testsuite/gcc.dg/tree-ssa/ssa-dom-thread-9.c
>
> --
> 2.1.0.243.g30d45f7
>

RE: [PATCH 0/3] [linaro/gcc-4_9-branch] backport fixes to jump-threading

[Sebastian Paul Pop]

I do not intend to back-port the jump-thread patches to the fsf gcc 4.9 branch: the patches I sent are only for the linaro 4.9 branch.
Yes, please test the patches with your validation suite, and then commit to the linaro branch.

Thanks,
Sebastian

-----Original Message-----
From: Yvan Roux [mailto:yvan.roux@linaro.org] 
Sent: Monday, August 31, 2015 10:43 AM
To: Sebastian Pop
Cc: gcc-patches@gcc.gnu.org; Sebastian Pop; Christophe Lyon
Subject: Re: [PATCH 0/3] [linaro/gcc-4_9-branch] backport fixes to jump-threading

Hi Sebastian,

On 31 August 2015 at 17:33, Sebastian Pop <s.pop@samsung.com> wrote:
> Hi,
>
> As I was backporting the jump-thread patches to the AOSP gcc 4.9, I found that
> the linaro branch does not have the following three fixes.  Ok to commit to the
> linaro/gcc-4_9-branch?

Our 4.9 branch is now in maintenance mode, which means that we're only
backporting bugfixes on it, and synchronizing it with upstream 4.9
branch.  So, if you'll only backport them to AOSP and not FSF gcc 4.9,
yes it can be beneficial for our branch, but we first have to pass
them into our validation process.

Cheers,
Yvan

> Thanks,
> Sebastian
>
> Sebastian Pop (3):
>   backport patch to fix PR65048
>   backport patch to fix PR65177
>   backport fix for PR65735
>
>  .../gcc/testsuite/gcc.c-torture/compile/pr65735.c  |  21 ++++
>  .../testsuite/gcc.dg/tree-ssa/ssa-dom-thread-10.c  |  24 ++++
>  .../testsuite/gcc.dg/tree-ssa/ssa-dom-thread-9.c   |  50 ++++++++
>  gcc-4.9/gcc/tree-ssa-threadedge.c                  |  12 +-
>  gcc-4.9/gcc/tree-ssa-threadupdate.c                | 133 ++++++++++++++++-----
>  5 files changed, 202 insertions(+), 38 deletions(-)
>  create mode 100644 gcc-4.9/gcc/testsuite/gcc.c-torture/compile/pr65735.c
>  create mode 100644 gcc-4.9/gcc/testsuite/gcc.dg/tree-ssa/ssa-dom-thread-10.c
>  create mode 100644 gcc-4.9/gcc/testsuite/gcc.dg/tree-ssa/ssa-dom-thread-9.c
>
> --
> 2.1.0.243.g30d45f7
>

Re: [PATCH 0/3] [linaro/gcc-4_9-branch] backport fixes to jump-threading

[Yvan Roux]

On 31 August 2015 at 18:08, Sebastian Paul Pop <s.pop@samsung.com> wrote:
> I do not intend to back-port the jump-thread patches to the fsf gcc 4.9 branch: the patches I sent are only for the linaro 4.9 branch.
> Yes, please test the patches with your validation suite, and then commit to the linaro branch.

Ok great. Thanks for these backports Sebastian.

Cheers,
Yvan

> Thanks,
> Sebastian
>
> -----Original Message-----
> From: Yvan Roux [mailto:yvan.roux@linaro.org]
> Sent: Monday, August 31, 2015 10:43 AM
> To: Sebastian Pop
> Cc: gcc-patches@gcc.gnu.org; Sebastian Pop; Christophe Lyon
> Subject: Re: [PATCH 0/3] [linaro/gcc-4_9-branch] backport fixes to jump-threading
>
> Hi Sebastian,
>
> On 31 August 2015 at 17:33, Sebastian Pop <s.pop@samsung.com> wrote:
>> Hi,
>>
>> As I was backporting the jump-thread patches to the AOSP gcc 4.9, I found that
>> the linaro branch does not have the following three fixes.  Ok to commit to the
>> linaro/gcc-4_9-branch?
>
> Our 4.9 branch is now in maintenance mode, which means that we're only
> backporting bugfixes on it, and synchronizing it with upstream 4.9
> branch.  So, if you'll only backport them to AOSP and not FSF gcc 4.9,
> yes it can be beneficial for our branch, but we first have to pass
> them into our validation process.
>
> Cheers,
> Yvan
>
>> Thanks,
>> Sebastian
>>
>> Sebastian Pop (3):
>>   backport patch to fix PR65048
>>   backport patch to fix PR65177
>>   backport fix for PR65735
>>
>>  .../gcc/testsuite/gcc.c-torture/compile/pr65735.c  |  21 ++++
>>  .../testsuite/gcc.dg/tree-ssa/ssa-dom-thread-10.c  |  24 ++++
>>  .../testsuite/gcc.dg/tree-ssa/ssa-dom-thread-9.c   |  50 ++++++++
>>  gcc-4.9/gcc/tree-ssa-threadedge.c                  |  12 +-
>>  gcc-4.9/gcc/tree-ssa-threadupdate.c                | 133 ++++++++++++++++-----
>>  5 files changed, 202 insertions(+), 38 deletions(-)
>>  create mode 100644 gcc-4.9/gcc/testsuite/gcc.c-torture/compile/pr65735.c
>>  create mode 100644 gcc-4.9/gcc/testsuite/gcc.dg/tree-ssa/ssa-dom-thread-10.c
>>  create mode 100644 gcc-4.9/gcc/testsuite/gcc.dg/tree-ssa/ssa-dom-thread-9.c
>>
>> --
>> 2.1.0.243.g30d45f7
>>
>