This is the mail archive of the
gcc-patches@gcc.gnu.org
mailing list for the GCC project.
[PATCH] Handle many consecutive location notes more efficiently in dwarf2.
- From: David Miller <davem at davemloft dot net>
- To: gcc-patches at gcc dot gnu dot org
- Date: Sun, 30 Oct 2011 21:55:42 -0400 (EDT)
- Subject: [PATCH] Handle many consecutive location notes more efficiently in dwarf2.
Tests such as limits-fndefn.c spend %97 of their time in
next_real_insn() when -g is on the command line.
It's the invocation from dwarf2out_var_location(). These test cases
are basically 100,000 consecutive var location notes, followed by a
real instruction.
This is similar to the reorg issue I dealt with yesterday, we
basically scan the entire function over and over again.
How I handle this here is if we see that the next insn is yet another
var location note we cache the next_real_insn() result.
Now limits-fndefn.c with "-O3 -g" takes about 20 seconds instead of
several minutes on my Niagara-T3 box.
Committed to trunk.
gcc/
* dwarf2out.c (dwarf2out_var_location): When processing several
consecutive location notes, cache the result of next_real_insn().
git-svn-id: svn+ssh://gcc.gnu.org/svn/gcc/trunk@180695 138bc75d-0d04-0410-961f-82ee72b054a4
---
gcc/ChangeLog | 5 +++++
gcc/dwarf2out.c | 34 ++++++++++++++++++++++++++++++++--
2 files changed, 37 insertions(+), 2 deletions(-)
diff --git a/gcc/ChangeLog b/gcc/ChangeLog
index 26bb1a9..037138a 100644
--- a/gcc/ChangeLog
+++ b/gcc/ChangeLog
@@ -1,3 +1,8 @@
+2011-10-30 David S. Miller <davem@davemloft.net>
+
+ * dwarf2out.c (dwarf2out_var_location): When processing several
+ consecutive location notes, cache the result of next_real_insn().
+
2011-10-30 Uros Bizjak <ubizjak@gmail.com>
* config/i386/i386.md (avx2_vec_dup<mode>): Macroize insn from
diff --git a/gcc/dwarf2out.c b/gcc/dwarf2out.c
index 8d5a9f0..478952f 100644
--- a/gcc/dwarf2out.c
+++ b/gcc/dwarf2out.c
@@ -20127,10 +20127,12 @@ dwarf2out_var_location (rtx loc_note)
{
char loclabel[MAX_ARTIFICIAL_LABEL_BYTES + 2];
struct var_loc_node *newloc;
- rtx next_real;
+ rtx next_real, next_note;
static const char *last_label;
static const char *last_postcall_label;
static bool last_in_cold_section_p;
+ static rtx expected_next_loc_note;
+ static rtx cached_next_real_insn;
tree decl;
bool var_loc_p;
@@ -20149,7 +20151,35 @@ dwarf2out_var_location (rtx loc_note)
if (var_loc_p && !DECL_P (NOTE_VAR_LOCATION_DECL (loc_note)))
return;
- next_real = next_real_insn (loc_note);
+ /* Optimize processing a large consecutive sequence of location
+ notes so we don't spend too much time in next_real_insn. If the
+ next insn is another location note, remember the next_real_insn
+ calculation for next time. */
+ next_real = cached_next_real_insn;
+ if (next_real)
+ {
+ if (expected_next_loc_note != loc_note)
+ next_real = NULL_RTX;
+ }
+
+ next_note = NEXT_INSN (loc_note);
+ if (! next_note
+ || INSN_DELETED_P (next_note)
+ || GET_CODE (next_note) != NOTE
+ || (NOTE_KIND (next_note) != NOTE_INSN_VAR_LOCATION
+ && NOTE_KIND (next_note) != NOTE_INSN_CALL_ARG_LOCATION))
+ next_note = NULL_RTX;
+
+ if (! next_real)
+ next_real = next_real_insn (loc_note);
+
+ if (next_note)
+ {
+ expected_next_loc_note = next_note;
+ cached_next_real_insn = next_real;
+ }
+ else
+ cached_next_real_insn = NULL_RTX;
/* If there are no instructions which would be affected by this note,
don't do anything. */
--
1.7.6.401.g6a319