OpenE2K
/
gcc
2
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

re PR tree-optimization/93199 (Compile time hog in sink_clobbers) 

2020-01-10  Richard Biener  <rguenther@suse.de>

	PR middle-end/93199
	* tree-eh.c (redirect_eh_edge_1): Avoid some work if possible.
	(cleanup_all_empty_eh): Walk landing pads in reverse order to
	avoid quadraticness.

From-SVN: r280101
This commit is contained in:
Richard Biener 2020-01-10 10:49:57 +00:00 committed by Richard Biener
parent 1a315435db
commit 5eaf0c498f
2 changed files with 21 additions and 7 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

7
gcc/ChangeLog
View File

@ -1,3 +1,10 @@
2020-01-10 Richard Biener <rguenther@suse.de>
PR middle-end/93199
* tree-eh.c (redirect_eh_edge_1): Avoid some work if possible.
(cleanup_all_empty_eh): Walk landing pads in reverse order to
avoid quadraticness.
2020-01-10 Martin Jambor <mjambor@suse.cz>
* params.opt (param_ipa_sra_max_replacements): Mark as Optimization.

21
gcc/tree-eh.c
View File

@ -2310,7 +2310,7 @@ redirect_eh_edge_1 (edge edge_in, basic_block new_bb, bool change_region)
old_lp = get_eh_landing_pad_from_number (old_lp_nr);
throw_stmt = last_stmt (edge_in->src);
gcc_assert (lookup_stmt_eh_lp (throw_stmt) == old_lp_nr);
gcc_checking_assert (lookup_stmt_eh_lp (throw_stmt) == old_lp_nr);
new_label = gimple_block_label (new_bb);
@ -4307,9 +4307,10 @@ cleanup_empty_eh_merge_phis (basic_block new_bb, basic_block old_bb,
| | EH
<..>
which CFG verification would choke on. See PR45172 and PR51089. */
FOR_EACH_EDGE (e, ei, old_bb->preds)
if (find_edge (e->src, new_bb))
return false;
if (!single_pred_p (new_bb))
FOR_EACH_EDGE (e, ei, old_bb->preds)
if (find_edge (e->src, new_bb))
return false;
FOR_EACH_EDGE (e, ei, old_bb->preds)
redirect_edge_var_map_clear (e);
@ -4698,9 +4699,15 @@ cleanup_all_empty_eh (void)
eh_landing_pad lp;
int i;
for (i = 1; vec_safe_iterate (cfun->eh->lp_array, i, &lp); ++i)
if (lp)
changed |= cleanup_empty_eh (lp);
/* Ideally we'd walk the region tree and process LPs inner to outer
to avoid quadraticness in EH redirection. Walking the LP array
in reverse seems to be an approximation of that. */
for (i = vec_safe_length (cfun->eh->lp_array) - 1; i >= 1; --i)
{
lp = (*cfun->eh->lp_array)[i];
if (lp)
changed |= cleanup_empty_eh (lp);
}
return changed;
}