Re: [PATCH] Replace evrp use in loop versioning with ranger.

[Richard Sandiford <richard.sandiford@arm.com>]

Aldy Hernandez <aldyh@redhat.com> writes:
> On Mon, Jul 26, 2021 at 7:28 PM Richard Sandiford
> <richard.sandiford@arm.com> wrote:
>>
>> Aldy Hernandez <aldyh@redhat.com> writes:
>> > On Mon, Jul 26, 2021 at 4:18 PM Richard Sandiford
>> > <richard.sandiford@arm.com> wrote:
>> >>
>> >> Aldy Hernandez <aldyh@redhat.com> writes:
>> >> > This patch replaces the evrp_range_analyzer in the loop versioning code
>> >> > with an on-demand ranger.
>> >> >
>> >> > Everything was pretty straightforward, except that range_of_expr requires
>> >> > a gimple statement as context to provide context aware ranges.  I didn't see
>> >> > a convient place where the statement was saved, so I made a vector indexed
>> >> > by SSA names.  As an alternative, I tried to use the loop's first statement,
>> >> > but that proved to be insufficient.
>> >>
>> >> The mapping is one-to-many though: there can be multiple statements
>> >> for each SSA name.  Maybe that doesn't matter in this context and
>> >> any of the statements can act as a representative.
>> >>
>> >> I'm surprised that the loop's first statement didn't work though,
>> >> since the SSA name is supposedly known to be loop-invariant.  What went
>> >> wrong when you tried that?
>> >
>> > I was looking at the first statement of loop_info->block_list and one
>> > of the dg.exp=loop-versioning* tests failed.  Perhaps I should have
>> > used the loop itself, as in the attached patch.  With this patch all
>> > of the loop-versioning tests pass.
>> >
>> >>
>> >> > I am not familiar with loop versioning, but if the DOM walk was only
>> >> > necessary for the calls to record_ranges_from_stmt, this too could be
>> >> > removed as the ranger will work without it.
>> >>
>> >> Yeah, that was the only reason.  If the information is available at
>> >> version_for_unity (I guess it is) then we should just avoid recording
>> >> the versioning there if so.
>> >>
>> >> How expensive is the check?  If the result is worth caching, perhaps
>> >> we should have two bitmaps: the existing one, and one that records
>> >> whether we've checked a particular SSA name.
>> >>
>> >> If the check is relatively cheap then that won't be worth it though.
>> >
>> > If you're asking about the range_of_expr check, that's all cached, so
>> > it should be pretty cheap.  Besides, we're no longer calculating
>> > ranges for each statement in the IL, as we were doing in lv_dom_walker
>> > with evrp's record_ranges_from_stmt.  Only statements of interest are
>> > queried.
>>
>> Sounds good.  If the results are already cached then another level
>> of caching (via the second bitmap I mentioned above) would obviously
>> be a waste of time.
>
> My callgrind harness for performance testing wasn't able to pick up
> enough samples to measure the time spent in
> pass_loop_versioning::execute.  I've seen this happen before with
> passes that run too fast.  I'm afraid I don't have enough cycles to
> continue working on this.

Yeah, any testing of this was above and beyond IMO.  Hearing that the
range query does its own caching was enough for me. :-)

>> > How about this patch, pending tests?
>>
>> OK, thanks, as a strict improvement over the status quo.  But it'd be
>> even better without the dom walk :-)
>
> I've removed the DOM walk, and re-tested.
>
> OK to push?

Sorry for asking for another iteration, but…

> Aldy
>
> From 9b1cba95377e7b26b4f0495b1b5998d2f7f33a14 Mon Sep 17 00:00:00 2001
> From: Aldy Hernandez <aldyh@redhat.com>
> Date: Sat, 24 Jul 2021 12:29:28 +0200
> Subject: [PATCH] Replace evrp use in loop versioning with ranger.
>
> This patch replaces the evrp_range_analyzer in the loop versioning code
> with a ranger.
>
> Tested on x86-64 Linux.
>
> gcc/ChangeLog:
>
> 	* gimple-loop-versioning.cc (lv_dom_walker::lv_dom_walker): Remove.
> 	(loop_versioning::lv_dom_walker::before_dom_children): Remove.
> 	(loop_versioning::lv_dom_walker::after_dom_children): Remove.
> 	(loop_versioning::prune_loop_conditions): Replace vr_values use
> 	with range_query interface.
> 	(loop_versioning::prune_conditions): Replace dom walk with
> 	straight iteration.
> 	(pass_loop_versioning::execute): Use ranger.
> ---
>  gcc/gimple-loop-versioning.cc | 78 ++++++++---------------------------
>  1 file changed, 18 insertions(+), 60 deletions(-)
>
> diff --git a/gcc/gimple-loop-versioning.cc b/gcc/gimple-loop-versioning.cc
> index 4b70c5a4aab..52eb6429171 100644
> --- a/gcc/gimple-loop-versioning.cc
> +++ b/gcc/gimple-loop-versioning.cc
> @@ -30,19 +30,17 @@ along with GCC; see the file COPYING3.  If not see
>  #include "tree-ssa-loop.h"
>  #include "ssa.h"
>  #include "tree-scalar-evolution.h"
> -#include "tree-chrec.h"
>  #include "tree-ssa-loop-ivopts.h"
>  #include "fold-const.h"
>  #include "tree-ssa-propagate.h"
>  #include "tree-inline.h"
>  #include "domwalk.h"
> -#include "alloc-pool.h"
> -#include "vr-values.h"
> -#include "gimple-ssa-evrp-analyze.h"
>  #include "tree-vectorizer.h"
>  #include "omp-general.h"
>  #include "predict.h"
>  #include "tree-into-ssa.h"
> +#include "gimple-range.h"
> +#include "tree-cfg.h"
>  
>  namespace {
>  
> @@ -253,24 +251,6 @@ public:
>    unsigned int run ();
>  
>  private:
> -  /* Used to walk the dominator tree to find loop versioning conditions
> -     that are always false.  */
> -  class lv_dom_walker : public dom_walker
> -  {
> -  public:
> -    lv_dom_walker (loop_versioning &);
> -
> -    edge before_dom_children (basic_block) FINAL OVERRIDE;
> -    void after_dom_children (basic_block) FINAL OVERRIDE;
> -
> -  private:
> -    /* The parent pass.  */
> -    loop_versioning &m_lv;
> -
> -    /* Used to build context-dependent range information.  */
> -    evrp_range_analyzer m_range_analyzer;
> -  };
> -
>    /* Used to simplify statements based on conditions that are established
>       by the version checks.  */
>    class name_prop : public substitute_and_fold_engine
> @@ -308,7 +288,7 @@ private:
>    bool analyze_block (basic_block);
>    bool analyze_blocks ();
>  
> -  void prune_loop_conditions (class loop *, vr_values *);
> +  void prune_loop_conditions (class loop *);
>    bool prune_conditions ();
>  
>    void merge_loop_info (class loop *, class loop *);
> @@ -499,36 +479,6 @@ loop_info::worth_versioning_p () const
>  	  && (!bitmap_empty_p (&unity_names) || subloops_benefit_p));
>  }
>  
> -loop_versioning::lv_dom_walker::lv_dom_walker (loop_versioning &lv)
> -  : dom_walker (CDI_DOMINATORS), m_lv (lv), m_range_analyzer (false)
> -{
> -}
> -
> -/* Process BB before processing the blocks it dominates.  */
> -
> -edge
> -loop_versioning::lv_dom_walker::before_dom_children (basic_block bb)
> -{
> -  m_range_analyzer.enter (bb);
> -
> -  if (bb == bb->loop_father->header)
> -    m_lv.prune_loop_conditions (bb->loop_father, &m_range_analyzer);
> -
> -  for (gimple_stmt_iterator si = gsi_start_bb (bb); !gsi_end_p (si);
> -       gsi_next (&si))
> -    m_range_analyzer.record_ranges_from_stmt (gsi_stmt (si), false);
> -
> -  return NULL;
> -}
> -
> -/* Process BB after processing the blocks it dominates.  */
> -
> -void
> -loop_versioning::lv_dom_walker::after_dom_children (basic_block bb)
> -{
> -  m_range_analyzer.leave (bb);
> -}
> -
>  /* Decide whether to replace VAL with a new value in a versioned loop.
>     Return the new value if so, otherwise return null.  */
>  
> @@ -1483,18 +1433,21 @@ loop_versioning::analyze_blocks ()
>     LOOP.  */
>  
>  void
> -loop_versioning::prune_loop_conditions (class loop *loop, vr_values *vrs)
> +loop_versioning::prune_loop_conditions (class loop *loop)
>  {
>    loop_info &li = get_loop_info (loop);
>  
>    int to_remove = -1;
>    bitmap_iterator bi;
>    unsigned int i;
> +  int_range_max r;
>    EXECUTE_IF_SET_IN_BITMAP (&li.unity_names, 0, i, bi)
>      {
>        tree name = ssa_name (i);
> -      const value_range_equiv *vr = vrs->get_value_range (name);
> -      if (vr && !vr->may_contain_p (build_one_cst (TREE_TYPE (name))))
> +      gimple *stmt = first_stmt (loop->header);
> +
> +      if (get_range_query (cfun)->range_of_expr (r, name, stmt)
> +	  && !r.contains_p (build_one_cst (TREE_TYPE (name))))
>  	{
>  	  if (dump_enabled_p ())
>  	    dump_printf_loc (MSG_NOTE, find_loop_location (loop),
> @@ -1519,9 +1472,11 @@ loop_versioning::prune_conditions ()
>    AUTO_DUMP_SCOPE ("prune_loop_conditions",
>  		   dump_user_location_t::from_function_decl (m_fn->decl));
>  
> -  calculate_dominance_info (CDI_DOMINATORS);
> -  lv_dom_walker dom_walker (*this);
> -  dom_walker.walk (ENTRY_BLOCK_PTR_FOR_FN (m_fn));
> +  basic_block bb;
> +  FOR_EACH_BB_FN (bb, m_fn)
> +    if (bb == bb->loop_father->header)
> +      prune_loop_conditions (bb->loop_father);

If we were going to keep pruning as a separate step, I think we should
iterate over loops rather than blocks.

However, what I meant by;

>> >> If the information is available at
>> >> version_for_unity (I guess it is) then we should just avoid recording
>> >> the versioning there if so.

is that we should instead put the get_range_query (cfun)->range_of_expr
and !r.contains_p test…

------------------------------------------------------------------------
void
loop_versioning::version_for_unity (gimple *stmt, tree name)
{
  class loop *loop = loop_containing_stmt (stmt);
  loop_info &li = get_loop_info (loop);

…here
  if (bitmap_set_bit (&li.unity_names, SSA_NAME_VERSION (name)))
------------------------------------------------------------------------

and report that the value can't be 1 at that point.  There would then
be no need for a separate pruning step.  Having this range information
on tap makes the pass much simpler than it used to be. :-)

FAOD, I think it would be good to keep using first_stmt (loop->header)
(as in your patch) rather than use the stmt argument to version_for_unity.

Thanks,
Richard


> +
>    return m_num_conditions != 0;
>  }
>  
> @@ -1810,7 +1765,10 @@ pass_loop_versioning::execute (function *fn)
>    if (number_of_loops (fn) <= 1)
>      return 0;
>  
> -  return loop_versioning (fn).run ();
> +  enable_ranger (fn);
> +  unsigned int ret = loop_versioning (fn).run ();
> +  disable_ranger (fn);
> +  return ret;
>  }
>  
>  } // anon namespace