1 |
684 |
jeremybenn |
/* SSA Jump Threading
|
2 |
|
|
Copyright (C) 2005, 2006, 2007, 2008, 2009, 2010, 2011
|
3 |
|
|
Free Software Foundation, Inc.
|
4 |
|
|
Contributed by Jeff Law <law@redhat.com>
|
5 |
|
|
|
6 |
|
|
This file is part of GCC.
|
7 |
|
|
|
8 |
|
|
GCC is free software; you can redistribute it and/or modify
|
9 |
|
|
it under the terms of the GNU General Public License as published by
|
10 |
|
|
the Free Software Foundation; either version 3, or (at your option)
|
11 |
|
|
any later version.
|
12 |
|
|
|
13 |
|
|
GCC is distributed in the hope that it will be useful,
|
14 |
|
|
but WITHOUT ANY WARRANTY; without even the implied warranty of
|
15 |
|
|
MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
|
16 |
|
|
GNU General Public License for more details.
|
17 |
|
|
|
18 |
|
|
You should have received a copy of the GNU General Public License
|
19 |
|
|
along with GCC; see the file COPYING3. If not see
|
20 |
|
|
<http://www.gnu.org/licenses/>. */
|
21 |
|
|
|
22 |
|
|
#include "config.h"
|
23 |
|
|
#include "system.h"
|
24 |
|
|
#include "coretypes.h"
|
25 |
|
|
#include "tm.h"
|
26 |
|
|
#include "tree.h"
|
27 |
|
|
#include "flags.h"
|
28 |
|
|
#include "tm_p.h"
|
29 |
|
|
#include "basic-block.h"
|
30 |
|
|
#include "cfgloop.h"
|
31 |
|
|
#include "output.h"
|
32 |
|
|
#include "function.h"
|
33 |
|
|
#include "timevar.h"
|
34 |
|
|
#include "tree-dump.h"
|
35 |
|
|
#include "tree-flow.h"
|
36 |
|
|
#include "tree-pass.h"
|
37 |
|
|
#include "tree-ssa-propagate.h"
|
38 |
|
|
#include "langhooks.h"
|
39 |
|
|
#include "params.h"
|
40 |
|
|
|
41 |
|
|
/* To avoid code explosion due to jump threading, we limit the
|
42 |
|
|
number of statements we are going to copy. This variable
|
43 |
|
|
holds the number of statements currently seen that we'll have
|
44 |
|
|
to copy as part of the jump threading process. */
|
45 |
|
|
static int stmt_count;
|
46 |
|
|
|
47 |
|
|
/* Array to record value-handles per SSA_NAME. */
|
48 |
|
|
VEC(tree,heap) *ssa_name_values;
|
49 |
|
|
|
50 |
|
|
/* Set the value for the SSA name NAME to VALUE. */
|
51 |
|
|
|
52 |
|
|
void
|
53 |
|
|
set_ssa_name_value (tree name, tree value)
|
54 |
|
|
{
|
55 |
|
|
if (SSA_NAME_VERSION (name) >= VEC_length (tree, ssa_name_values))
|
56 |
|
|
VEC_safe_grow_cleared (tree, heap, ssa_name_values,
|
57 |
|
|
SSA_NAME_VERSION (name) + 1);
|
58 |
|
|
VEC_replace (tree, ssa_name_values, SSA_NAME_VERSION (name), value);
|
59 |
|
|
}
|
60 |
|
|
|
61 |
|
|
/* Initialize the per SSA_NAME value-handles array. Returns it. */
|
62 |
|
|
void
|
63 |
|
|
threadedge_initialize_values (void)
|
64 |
|
|
{
|
65 |
|
|
gcc_assert (ssa_name_values == NULL);
|
66 |
|
|
ssa_name_values = VEC_alloc(tree, heap, num_ssa_names);
|
67 |
|
|
}
|
68 |
|
|
|
69 |
|
|
/* Free the per SSA_NAME value-handle array. */
|
70 |
|
|
void
|
71 |
|
|
threadedge_finalize_values (void)
|
72 |
|
|
{
|
73 |
|
|
VEC_free(tree, heap, ssa_name_values);
|
74 |
|
|
}
|
75 |
|
|
|
76 |
|
|
/* Return TRUE if we may be able to thread an incoming edge into
|
77 |
|
|
BB to an outgoing edge from BB. Return FALSE otherwise. */
|
78 |
|
|
|
79 |
|
|
bool
|
80 |
|
|
potentially_threadable_block (basic_block bb)
|
81 |
|
|
{
|
82 |
|
|
gimple_stmt_iterator gsi;
|
83 |
|
|
|
84 |
|
|
/* If BB has a single successor or a single predecessor, then
|
85 |
|
|
there is no threading opportunity. */
|
86 |
|
|
if (single_succ_p (bb) || single_pred_p (bb))
|
87 |
|
|
return false;
|
88 |
|
|
|
89 |
|
|
/* If BB does not end with a conditional, switch or computed goto,
|
90 |
|
|
then there is no threading opportunity. */
|
91 |
|
|
gsi = gsi_last_bb (bb);
|
92 |
|
|
if (gsi_end_p (gsi)
|
93 |
|
|
|| ! gsi_stmt (gsi)
|
94 |
|
|
|| (gimple_code (gsi_stmt (gsi)) != GIMPLE_COND
|
95 |
|
|
&& gimple_code (gsi_stmt (gsi)) != GIMPLE_GOTO
|
96 |
|
|
&& gimple_code (gsi_stmt (gsi)) != GIMPLE_SWITCH))
|
97 |
|
|
return false;
|
98 |
|
|
|
99 |
|
|
return true;
|
100 |
|
|
}
|
101 |
|
|
|
102 |
|
|
/* Return the LHS of any ASSERT_EXPR where OP appears as the first
|
103 |
|
|
argument to the ASSERT_EXPR and in which the ASSERT_EXPR dominates
|
104 |
|
|
BB. If no such ASSERT_EXPR is found, return OP. */
|
105 |
|
|
|
106 |
|
|
static tree
|
107 |
|
|
lhs_of_dominating_assert (tree op, basic_block bb, gimple stmt)
|
108 |
|
|
{
|
109 |
|
|
imm_use_iterator imm_iter;
|
110 |
|
|
gimple use_stmt;
|
111 |
|
|
use_operand_p use_p;
|
112 |
|
|
|
113 |
|
|
FOR_EACH_IMM_USE_FAST (use_p, imm_iter, op)
|
114 |
|
|
{
|
115 |
|
|
use_stmt = USE_STMT (use_p);
|
116 |
|
|
if (use_stmt != stmt
|
117 |
|
|
&& gimple_assign_single_p (use_stmt)
|
118 |
|
|
&& TREE_CODE (gimple_assign_rhs1 (use_stmt)) == ASSERT_EXPR
|
119 |
|
|
&& TREE_OPERAND (gimple_assign_rhs1 (use_stmt), 0) == op
|
120 |
|
|
&& dominated_by_p (CDI_DOMINATORS, bb, gimple_bb (use_stmt)))
|
121 |
|
|
{
|
122 |
|
|
return gimple_assign_lhs (use_stmt);
|
123 |
|
|
}
|
124 |
|
|
}
|
125 |
|
|
return op;
|
126 |
|
|
}
|
127 |
|
|
|
128 |
|
|
/* We record temporary equivalences created by PHI nodes or
|
129 |
|
|
statements within the target block. Doing so allows us to
|
130 |
|
|
identify more jump threading opportunities, even in blocks
|
131 |
|
|
with side effects.
|
132 |
|
|
|
133 |
|
|
We keep track of those temporary equivalences in a stack
|
134 |
|
|
structure so that we can unwind them when we're done processing
|
135 |
|
|
a particular edge. This routine handles unwinding the data
|
136 |
|
|
structures. */
|
137 |
|
|
|
138 |
|
|
static void
|
139 |
|
|
remove_temporary_equivalences (VEC(tree, heap) **stack)
|
140 |
|
|
{
|
141 |
|
|
while (VEC_length (tree, *stack) > 0)
|
142 |
|
|
{
|
143 |
|
|
tree prev_value, dest;
|
144 |
|
|
|
145 |
|
|
dest = VEC_pop (tree, *stack);
|
146 |
|
|
|
147 |
|
|
/* A NULL value indicates we should stop unwinding, otherwise
|
148 |
|
|
pop off the next entry as they're recorded in pairs. */
|
149 |
|
|
if (dest == NULL)
|
150 |
|
|
break;
|
151 |
|
|
|
152 |
|
|
prev_value = VEC_pop (tree, *stack);
|
153 |
|
|
set_ssa_name_value (dest, prev_value);
|
154 |
|
|
}
|
155 |
|
|
}
|
156 |
|
|
|
157 |
|
|
/* Record a temporary equivalence, saving enough information so that
|
158 |
|
|
we can restore the state of recorded equivalences when we're
|
159 |
|
|
done processing the current edge. */
|
160 |
|
|
|
161 |
|
|
static void
|
162 |
|
|
record_temporary_equivalence (tree x, tree y, VEC(tree, heap) **stack)
|
163 |
|
|
{
|
164 |
|
|
tree prev_x = SSA_NAME_VALUE (x);
|
165 |
|
|
|
166 |
|
|
if (TREE_CODE (y) == SSA_NAME)
|
167 |
|
|
{
|
168 |
|
|
tree tmp = SSA_NAME_VALUE (y);
|
169 |
|
|
y = tmp ? tmp : y;
|
170 |
|
|
}
|
171 |
|
|
|
172 |
|
|
set_ssa_name_value (x, y);
|
173 |
|
|
VEC_reserve (tree, heap, *stack, 2);
|
174 |
|
|
VEC_quick_push (tree, *stack, prev_x);
|
175 |
|
|
VEC_quick_push (tree, *stack, x);
|
176 |
|
|
}
|
177 |
|
|
|
178 |
|
|
/* Record temporary equivalences created by PHIs at the target of the
|
179 |
|
|
edge E. Record unwind information for the equivalences onto STACK.
|
180 |
|
|
|
181 |
|
|
If a PHI which prevents threading is encountered, then return FALSE
|
182 |
|
|
indicating we should not thread this edge, else return TRUE. */
|
183 |
|
|
|
184 |
|
|
static bool
|
185 |
|
|
record_temporary_equivalences_from_phis (edge e, VEC(tree, heap) **stack)
|
186 |
|
|
{
|
187 |
|
|
gimple_stmt_iterator gsi;
|
188 |
|
|
|
189 |
|
|
/* Each PHI creates a temporary equivalence, record them.
|
190 |
|
|
These are context sensitive equivalences and will be removed
|
191 |
|
|
later. */
|
192 |
|
|
for (gsi = gsi_start_phis (e->dest); !gsi_end_p (gsi); gsi_next (&gsi))
|
193 |
|
|
{
|
194 |
|
|
gimple phi = gsi_stmt (gsi);
|
195 |
|
|
tree src = PHI_ARG_DEF_FROM_EDGE (phi, e);
|
196 |
|
|
tree dst = gimple_phi_result (phi);
|
197 |
|
|
|
198 |
|
|
/* If the desired argument is not the same as this PHI's result
|
199 |
|
|
and it is set by a PHI in E->dest, then we can not thread
|
200 |
|
|
through E->dest. */
|
201 |
|
|
if (src != dst
|
202 |
|
|
&& TREE_CODE (src) == SSA_NAME
|
203 |
|
|
&& gimple_code (SSA_NAME_DEF_STMT (src)) == GIMPLE_PHI
|
204 |
|
|
&& gimple_bb (SSA_NAME_DEF_STMT (src)) == e->dest)
|
205 |
|
|
return false;
|
206 |
|
|
|
207 |
|
|
/* We consider any non-virtual PHI as a statement since it
|
208 |
|
|
count result in a constant assignment or copy operation. */
|
209 |
|
|
if (is_gimple_reg (dst))
|
210 |
|
|
stmt_count++;
|
211 |
|
|
|
212 |
|
|
record_temporary_equivalence (dst, src, stack);
|
213 |
|
|
}
|
214 |
|
|
return true;
|
215 |
|
|
}
|
216 |
|
|
|
217 |
|
|
/* Fold the RHS of an assignment statement and return it as a tree.
|
218 |
|
|
May return NULL_TREE if no simplification is possible. */
|
219 |
|
|
|
220 |
|
|
static tree
|
221 |
|
|
fold_assignment_stmt (gimple stmt)
|
222 |
|
|
{
|
223 |
|
|
enum tree_code subcode = gimple_assign_rhs_code (stmt);
|
224 |
|
|
|
225 |
|
|
switch (get_gimple_rhs_class (subcode))
|
226 |
|
|
{
|
227 |
|
|
case GIMPLE_SINGLE_RHS:
|
228 |
|
|
return fold (gimple_assign_rhs1 (stmt));
|
229 |
|
|
|
230 |
|
|
case GIMPLE_UNARY_RHS:
|
231 |
|
|
{
|
232 |
|
|
tree lhs = gimple_assign_lhs (stmt);
|
233 |
|
|
tree op0 = gimple_assign_rhs1 (stmt);
|
234 |
|
|
return fold_unary (subcode, TREE_TYPE (lhs), op0);
|
235 |
|
|
}
|
236 |
|
|
|
237 |
|
|
case GIMPLE_BINARY_RHS:
|
238 |
|
|
{
|
239 |
|
|
tree lhs = gimple_assign_lhs (stmt);
|
240 |
|
|
tree op0 = gimple_assign_rhs1 (stmt);
|
241 |
|
|
tree op1 = gimple_assign_rhs2 (stmt);
|
242 |
|
|
return fold_binary (subcode, TREE_TYPE (lhs), op0, op1);
|
243 |
|
|
}
|
244 |
|
|
|
245 |
|
|
case GIMPLE_TERNARY_RHS:
|
246 |
|
|
{
|
247 |
|
|
tree lhs = gimple_assign_lhs (stmt);
|
248 |
|
|
tree op0 = gimple_assign_rhs1 (stmt);
|
249 |
|
|
tree op1 = gimple_assign_rhs2 (stmt);
|
250 |
|
|
tree op2 = gimple_assign_rhs3 (stmt);
|
251 |
|
|
|
252 |
|
|
/* Sadly, we have to handle conditional assignments specially
|
253 |
|
|
here, because fold expects all the operands of an expression
|
254 |
|
|
to be folded before the expression itself is folded, but we
|
255 |
|
|
can't just substitute the folded condition here. */
|
256 |
|
|
if (gimple_assign_rhs_code (stmt) == COND_EXPR)
|
257 |
|
|
op0 = fold (op0);
|
258 |
|
|
|
259 |
|
|
return fold_ternary (subcode, TREE_TYPE (lhs), op0, op1, op2);
|
260 |
|
|
}
|
261 |
|
|
|
262 |
|
|
default:
|
263 |
|
|
gcc_unreachable ();
|
264 |
|
|
}
|
265 |
|
|
}
|
266 |
|
|
|
267 |
|
|
/* Try to simplify each statement in E->dest, ultimately leading to
|
268 |
|
|
a simplification of the COND_EXPR at the end of E->dest.
|
269 |
|
|
|
270 |
|
|
Record unwind information for temporary equivalences onto STACK.
|
271 |
|
|
|
272 |
|
|
Use SIMPLIFY (a pointer to a callback function) to further simplify
|
273 |
|
|
statements using pass specific information.
|
274 |
|
|
|
275 |
|
|
We might consider marking just those statements which ultimately
|
276 |
|
|
feed the COND_EXPR. It's not clear if the overhead of bookkeeping
|
277 |
|
|
would be recovered by trying to simplify fewer statements.
|
278 |
|
|
|
279 |
|
|
If we are able to simplify a statement into the form
|
280 |
|
|
SSA_NAME = (SSA_NAME | gimple invariant), then we can record
|
281 |
|
|
a context sensitive equivalence which may help us simplify
|
282 |
|
|
later statements in E->dest. */
|
283 |
|
|
|
284 |
|
|
static gimple
|
285 |
|
|
record_temporary_equivalences_from_stmts_at_dest (edge e,
|
286 |
|
|
VEC(tree, heap) **stack,
|
287 |
|
|
tree (*simplify) (gimple,
|
288 |
|
|
gimple))
|
289 |
|
|
{
|
290 |
|
|
gimple stmt = NULL;
|
291 |
|
|
gimple_stmt_iterator gsi;
|
292 |
|
|
int max_stmt_count;
|
293 |
|
|
|
294 |
|
|
max_stmt_count = PARAM_VALUE (PARAM_MAX_JUMP_THREAD_DUPLICATION_STMTS);
|
295 |
|
|
|
296 |
|
|
/* Walk through each statement in the block recording equivalences
|
297 |
|
|
we discover. Note any equivalences we discover are context
|
298 |
|
|
sensitive (ie, are dependent on traversing E) and must be unwound
|
299 |
|
|
when we're finished processing E. */
|
300 |
|
|
for (gsi = gsi_start_bb (e->dest); !gsi_end_p (gsi); gsi_next (&gsi))
|
301 |
|
|
{
|
302 |
|
|
tree cached_lhs = NULL;
|
303 |
|
|
|
304 |
|
|
stmt = gsi_stmt (gsi);
|
305 |
|
|
|
306 |
|
|
/* Ignore empty statements and labels. */
|
307 |
|
|
if (gimple_code (stmt) == GIMPLE_NOP
|
308 |
|
|
|| gimple_code (stmt) == GIMPLE_LABEL
|
309 |
|
|
|| is_gimple_debug (stmt))
|
310 |
|
|
continue;
|
311 |
|
|
|
312 |
|
|
/* If the statement has volatile operands, then we assume we
|
313 |
|
|
can not thread through this block. This is overly
|
314 |
|
|
conservative in some ways. */
|
315 |
|
|
if (gimple_code (stmt) == GIMPLE_ASM && gimple_asm_volatile_p (stmt))
|
316 |
|
|
return NULL;
|
317 |
|
|
|
318 |
|
|
/* If duplicating this block is going to cause too much code
|
319 |
|
|
expansion, then do not thread through this block. */
|
320 |
|
|
stmt_count++;
|
321 |
|
|
if (stmt_count > max_stmt_count)
|
322 |
|
|
return NULL;
|
323 |
|
|
|
324 |
|
|
/* If this is not a statement that sets an SSA_NAME to a new
|
325 |
|
|
value, then do not try to simplify this statement as it will
|
326 |
|
|
not simplify in any way that is helpful for jump threading. */
|
327 |
|
|
if ((gimple_code (stmt) != GIMPLE_ASSIGN
|
328 |
|
|
|| TREE_CODE (gimple_assign_lhs (stmt)) != SSA_NAME)
|
329 |
|
|
&& (gimple_code (stmt) != GIMPLE_CALL
|
330 |
|
|
|| gimple_call_lhs (stmt) == NULL_TREE
|
331 |
|
|
|| TREE_CODE (gimple_call_lhs (stmt)) != SSA_NAME))
|
332 |
|
|
continue;
|
333 |
|
|
|
334 |
|
|
/* The result of __builtin_object_size depends on all the arguments
|
335 |
|
|
of a phi node. Temporarily using only one edge produces invalid
|
336 |
|
|
results. For example
|
337 |
|
|
|
338 |
|
|
if (x < 6)
|
339 |
|
|
goto l;
|
340 |
|
|
else
|
341 |
|
|
goto l;
|
342 |
|
|
|
343 |
|
|
l:
|
344 |
|
|
r = PHI <&w[2].a[1](2), &a.a[6](3)>
|
345 |
|
|
__builtin_object_size (r, 0)
|
346 |
|
|
|
347 |
|
|
The result of __builtin_object_size is defined to be the maximum of
|
348 |
|
|
remaining bytes. If we use only one edge on the phi, the result will
|
349 |
|
|
change to be the remaining bytes for the corresponding phi argument.
|
350 |
|
|
|
351 |
|
|
Similarly for __builtin_constant_p:
|
352 |
|
|
|
353 |
|
|
r = PHI <1(2), 2(3)>
|
354 |
|
|
__builtin_constant_p (r)
|
355 |
|
|
|
356 |
|
|
Both PHI arguments are constant, but x ? 1 : 2 is still not
|
357 |
|
|
constant. */
|
358 |
|
|
|
359 |
|
|
if (is_gimple_call (stmt))
|
360 |
|
|
{
|
361 |
|
|
tree fndecl = gimple_call_fndecl (stmt);
|
362 |
|
|
if (fndecl
|
363 |
|
|
&& (DECL_FUNCTION_CODE (fndecl) == BUILT_IN_OBJECT_SIZE
|
364 |
|
|
|| DECL_FUNCTION_CODE (fndecl) == BUILT_IN_CONSTANT_P))
|
365 |
|
|
continue;
|
366 |
|
|
}
|
367 |
|
|
|
368 |
|
|
/* At this point we have a statement which assigns an RHS to an
|
369 |
|
|
SSA_VAR on the LHS. We want to try and simplify this statement
|
370 |
|
|
to expose more context sensitive equivalences which in turn may
|
371 |
|
|
allow us to simplify the condition at the end of the loop.
|
372 |
|
|
|
373 |
|
|
Handle simple copy operations as well as implied copies from
|
374 |
|
|
ASSERT_EXPRs. */
|
375 |
|
|
if (gimple_assign_single_p (stmt)
|
376 |
|
|
&& TREE_CODE (gimple_assign_rhs1 (stmt)) == SSA_NAME)
|
377 |
|
|
cached_lhs = gimple_assign_rhs1 (stmt);
|
378 |
|
|
else if (gimple_assign_single_p (stmt)
|
379 |
|
|
&& TREE_CODE (gimple_assign_rhs1 (stmt)) == ASSERT_EXPR)
|
380 |
|
|
cached_lhs = TREE_OPERAND (gimple_assign_rhs1 (stmt), 0);
|
381 |
|
|
else
|
382 |
|
|
{
|
383 |
|
|
/* A statement that is not a trivial copy or ASSERT_EXPR.
|
384 |
|
|
We're going to temporarily copy propagate the operands
|
385 |
|
|
and see if that allows us to simplify this statement. */
|
386 |
|
|
tree *copy;
|
387 |
|
|
ssa_op_iter iter;
|
388 |
|
|
use_operand_p use_p;
|
389 |
|
|
unsigned int num, i = 0;
|
390 |
|
|
|
391 |
|
|
num = NUM_SSA_OPERANDS (stmt, (SSA_OP_USE | SSA_OP_VUSE));
|
392 |
|
|
copy = XCNEWVEC (tree, num);
|
393 |
|
|
|
394 |
|
|
/* Make a copy of the uses & vuses into USES_COPY, then cprop into
|
395 |
|
|
the operands. */
|
396 |
|
|
FOR_EACH_SSA_USE_OPERAND (use_p, stmt, iter, SSA_OP_USE | SSA_OP_VUSE)
|
397 |
|
|
{
|
398 |
|
|
tree tmp = NULL;
|
399 |
|
|
tree use = USE_FROM_PTR (use_p);
|
400 |
|
|
|
401 |
|
|
copy[i++] = use;
|
402 |
|
|
if (TREE_CODE (use) == SSA_NAME)
|
403 |
|
|
tmp = SSA_NAME_VALUE (use);
|
404 |
|
|
if (tmp)
|
405 |
|
|
SET_USE (use_p, tmp);
|
406 |
|
|
}
|
407 |
|
|
|
408 |
|
|
/* Try to fold/lookup the new expression. Inserting the
|
409 |
|
|
expression into the hash table is unlikely to help. */
|
410 |
|
|
if (is_gimple_call (stmt))
|
411 |
|
|
cached_lhs = fold_call_stmt (stmt, false);
|
412 |
|
|
else
|
413 |
|
|
cached_lhs = fold_assignment_stmt (stmt);
|
414 |
|
|
|
415 |
|
|
if (!cached_lhs
|
416 |
|
|
|| (TREE_CODE (cached_lhs) != SSA_NAME
|
417 |
|
|
&& !is_gimple_min_invariant (cached_lhs)))
|
418 |
|
|
cached_lhs = (*simplify) (stmt, stmt);
|
419 |
|
|
|
420 |
|
|
/* Restore the statement's original uses/defs. */
|
421 |
|
|
i = 0;
|
422 |
|
|
FOR_EACH_SSA_USE_OPERAND (use_p, stmt, iter, SSA_OP_USE | SSA_OP_VUSE)
|
423 |
|
|
SET_USE (use_p, copy[i++]);
|
424 |
|
|
|
425 |
|
|
free (copy);
|
426 |
|
|
}
|
427 |
|
|
|
428 |
|
|
/* Record the context sensitive equivalence if we were able
|
429 |
|
|
to simplify this statement. */
|
430 |
|
|
if (cached_lhs
|
431 |
|
|
&& (TREE_CODE (cached_lhs) == SSA_NAME
|
432 |
|
|
|| is_gimple_min_invariant (cached_lhs)))
|
433 |
|
|
record_temporary_equivalence (gimple_get_lhs (stmt), cached_lhs, stack);
|
434 |
|
|
}
|
435 |
|
|
return stmt;
|
436 |
|
|
}
|
437 |
|
|
|
438 |
|
|
/* Simplify the control statement at the end of the block E->dest.
|
439 |
|
|
|
440 |
|
|
To avoid allocating memory unnecessarily, a scratch GIMPLE_COND
|
441 |
|
|
is available to use/clobber in DUMMY_COND.
|
442 |
|
|
|
443 |
|
|
Use SIMPLIFY (a pointer to a callback function) to further simplify
|
444 |
|
|
a condition using pass specific information.
|
445 |
|
|
|
446 |
|
|
Return the simplified condition or NULL if simplification could
|
447 |
|
|
not be performed. */
|
448 |
|
|
|
449 |
|
|
static tree
|
450 |
|
|
simplify_control_stmt_condition (edge e,
|
451 |
|
|
gimple stmt,
|
452 |
|
|
gimple dummy_cond,
|
453 |
|
|
tree (*simplify) (gimple, gimple),
|
454 |
|
|
bool handle_dominating_asserts)
|
455 |
|
|
{
|
456 |
|
|
tree cond, cached_lhs;
|
457 |
|
|
enum gimple_code code = gimple_code (stmt);
|
458 |
|
|
|
459 |
|
|
/* For comparisons, we have to update both operands, then try
|
460 |
|
|
to simplify the comparison. */
|
461 |
|
|
if (code == GIMPLE_COND)
|
462 |
|
|
{
|
463 |
|
|
tree op0, op1;
|
464 |
|
|
enum tree_code cond_code;
|
465 |
|
|
|
466 |
|
|
op0 = gimple_cond_lhs (stmt);
|
467 |
|
|
op1 = gimple_cond_rhs (stmt);
|
468 |
|
|
cond_code = gimple_cond_code (stmt);
|
469 |
|
|
|
470 |
|
|
/* Get the current value of both operands. */
|
471 |
|
|
if (TREE_CODE (op0) == SSA_NAME)
|
472 |
|
|
{
|
473 |
|
|
tree tmp = SSA_NAME_VALUE (op0);
|
474 |
|
|
if (tmp)
|
475 |
|
|
op0 = tmp;
|
476 |
|
|
}
|
477 |
|
|
|
478 |
|
|
if (TREE_CODE (op1) == SSA_NAME)
|
479 |
|
|
{
|
480 |
|
|
tree tmp = SSA_NAME_VALUE (op1);
|
481 |
|
|
if (tmp)
|
482 |
|
|
op1 = tmp;
|
483 |
|
|
}
|
484 |
|
|
|
485 |
|
|
if (handle_dominating_asserts)
|
486 |
|
|
{
|
487 |
|
|
/* Now see if the operand was consumed by an ASSERT_EXPR
|
488 |
|
|
which dominates E->src. If so, we want to replace the
|
489 |
|
|
operand with the LHS of the ASSERT_EXPR. */
|
490 |
|
|
if (TREE_CODE (op0) == SSA_NAME)
|
491 |
|
|
op0 = lhs_of_dominating_assert (op0, e->src, stmt);
|
492 |
|
|
|
493 |
|
|
if (TREE_CODE (op1) == SSA_NAME)
|
494 |
|
|
op1 = lhs_of_dominating_assert (op1, e->src, stmt);
|
495 |
|
|
}
|
496 |
|
|
|
497 |
|
|
/* We may need to canonicalize the comparison. For
|
498 |
|
|
example, op0 might be a constant while op1 is an
|
499 |
|
|
SSA_NAME. Failure to canonicalize will cause us to
|
500 |
|
|
miss threading opportunities. */
|
501 |
|
|
if (tree_swap_operands_p (op0, op1, false))
|
502 |
|
|
{
|
503 |
|
|
tree tmp;
|
504 |
|
|
cond_code = swap_tree_comparison (cond_code);
|
505 |
|
|
tmp = op0;
|
506 |
|
|
op0 = op1;
|
507 |
|
|
op1 = tmp;
|
508 |
|
|
}
|
509 |
|
|
|
510 |
|
|
/* Stuff the operator and operands into our dummy conditional
|
511 |
|
|
expression. */
|
512 |
|
|
gimple_cond_set_code (dummy_cond, cond_code);
|
513 |
|
|
gimple_cond_set_lhs (dummy_cond, op0);
|
514 |
|
|
gimple_cond_set_rhs (dummy_cond, op1);
|
515 |
|
|
|
516 |
|
|
/* We absolutely do not care about any type conversions
|
517 |
|
|
we only care about a zero/nonzero value. */
|
518 |
|
|
fold_defer_overflow_warnings ();
|
519 |
|
|
|
520 |
|
|
cached_lhs = fold_binary (cond_code, boolean_type_node, op0, op1);
|
521 |
|
|
if (cached_lhs)
|
522 |
|
|
while (CONVERT_EXPR_P (cached_lhs))
|
523 |
|
|
cached_lhs = TREE_OPERAND (cached_lhs, 0);
|
524 |
|
|
|
525 |
|
|
fold_undefer_overflow_warnings ((cached_lhs
|
526 |
|
|
&& is_gimple_min_invariant (cached_lhs)),
|
527 |
|
|
stmt, WARN_STRICT_OVERFLOW_CONDITIONAL);
|
528 |
|
|
|
529 |
|
|
/* If we have not simplified the condition down to an invariant,
|
530 |
|
|
then use the pass specific callback to simplify the condition. */
|
531 |
|
|
if (!cached_lhs
|
532 |
|
|
|| !is_gimple_min_invariant (cached_lhs))
|
533 |
|
|
cached_lhs = (*simplify) (dummy_cond, stmt);
|
534 |
|
|
|
535 |
|
|
return cached_lhs;
|
536 |
|
|
}
|
537 |
|
|
|
538 |
|
|
if (code == GIMPLE_SWITCH)
|
539 |
|
|
cond = gimple_switch_index (stmt);
|
540 |
|
|
else if (code == GIMPLE_GOTO)
|
541 |
|
|
cond = gimple_goto_dest (stmt);
|
542 |
|
|
else
|
543 |
|
|
gcc_unreachable ();
|
544 |
|
|
|
545 |
|
|
/* We can have conditionals which just test the state of a variable
|
546 |
|
|
rather than use a relational operator. These are simpler to handle. */
|
547 |
|
|
if (TREE_CODE (cond) == SSA_NAME)
|
548 |
|
|
{
|
549 |
|
|
cached_lhs = cond;
|
550 |
|
|
|
551 |
|
|
/* Get the variable's current value from the equivalence chains.
|
552 |
|
|
|
553 |
|
|
It is possible to get loops in the SSA_NAME_VALUE chains
|
554 |
|
|
(consider threading the backedge of a loop where we have
|
555 |
|
|
a loop invariant SSA_NAME used in the condition. */
|
556 |
|
|
if (cached_lhs
|
557 |
|
|
&& TREE_CODE (cached_lhs) == SSA_NAME
|
558 |
|
|
&& SSA_NAME_VALUE (cached_lhs))
|
559 |
|
|
cached_lhs = SSA_NAME_VALUE (cached_lhs);
|
560 |
|
|
|
561 |
|
|
/* If we're dominated by a suitable ASSERT_EXPR, then
|
562 |
|
|
update CACHED_LHS appropriately. */
|
563 |
|
|
if (handle_dominating_asserts && TREE_CODE (cached_lhs) == SSA_NAME)
|
564 |
|
|
cached_lhs = lhs_of_dominating_assert (cached_lhs, e->src, stmt);
|
565 |
|
|
|
566 |
|
|
/* If we haven't simplified to an invariant yet, then use the
|
567 |
|
|
pass specific callback to try and simplify it further. */
|
568 |
|
|
if (cached_lhs && ! is_gimple_min_invariant (cached_lhs))
|
569 |
|
|
cached_lhs = (*simplify) (stmt, stmt);
|
570 |
|
|
}
|
571 |
|
|
else
|
572 |
|
|
cached_lhs = NULL;
|
573 |
|
|
|
574 |
|
|
return cached_lhs;
|
575 |
|
|
}
|
576 |
|
|
|
577 |
|
|
/* TAKEN_EDGE represents the an edge taken as a result of jump threading.
|
578 |
|
|
See if we can thread around TAKEN_EDGE->dest as well. If so, return
|
579 |
|
|
the edge out of TAKEN_EDGE->dest that we can statically compute will be
|
580 |
|
|
traversed.
|
581 |
|
|
|
582 |
|
|
We are much more restrictive as to the contents of TAKEN_EDGE->dest
|
583 |
|
|
as the path isolation code in tree-ssa-threadupdate.c isn't prepared
|
584 |
|
|
to handle copying intermediate blocks on a threaded path.
|
585 |
|
|
|
586 |
|
|
Long term a more consistent and structured approach to path isolation
|
587 |
|
|
would be a huge help. */
|
588 |
|
|
static edge
|
589 |
|
|
thread_around_empty_block (edge taken_edge,
|
590 |
|
|
gimple dummy_cond,
|
591 |
|
|
bool handle_dominating_asserts,
|
592 |
|
|
tree (*simplify) (gimple, gimple),
|
593 |
|
|
bitmap visited)
|
594 |
|
|
{
|
595 |
|
|
basic_block bb = taken_edge->dest;
|
596 |
|
|
gimple_stmt_iterator gsi;
|
597 |
|
|
gimple stmt;
|
598 |
|
|
tree cond;
|
599 |
|
|
|
600 |
|
|
/* This block must have a single predecessor (E->dest). */
|
601 |
|
|
if (!single_pred_p (bb))
|
602 |
|
|
return NULL;
|
603 |
|
|
|
604 |
|
|
/* This block must have more than one successor. */
|
605 |
|
|
if (single_succ_p (bb))
|
606 |
|
|
return NULL;
|
607 |
|
|
|
608 |
|
|
/* This block can have no PHI nodes. This is overly conservative. */
|
609 |
|
|
if (!gsi_end_p (gsi_start_phis (bb)))
|
610 |
|
|
return NULL;
|
611 |
|
|
|
612 |
|
|
/* Skip over DEBUG statements at the start of the block. */
|
613 |
|
|
gsi = gsi_start_nondebug_bb (bb);
|
614 |
|
|
|
615 |
|
|
if (gsi_end_p (gsi))
|
616 |
|
|
return NULL;
|
617 |
|
|
|
618 |
|
|
/* This block can have no statements other than its control altering
|
619 |
|
|
statement. This is overly conservative. */
|
620 |
|
|
stmt = gsi_stmt (gsi);
|
621 |
|
|
if (gimple_code (stmt) != GIMPLE_COND
|
622 |
|
|
&& gimple_code (stmt) != GIMPLE_GOTO
|
623 |
|
|
&& gimple_code (stmt) != GIMPLE_SWITCH)
|
624 |
|
|
return NULL;
|
625 |
|
|
|
626 |
|
|
/* Extract and simplify the condition. */
|
627 |
|
|
cond = simplify_control_stmt_condition (taken_edge, stmt, dummy_cond,
|
628 |
|
|
simplify, handle_dominating_asserts);
|
629 |
|
|
|
630 |
|
|
/* If the condition can be statically computed and we have not already
|
631 |
|
|
visited the destination edge, then add the taken edge to our thread
|
632 |
|
|
path. */
|
633 |
|
|
if (cond && is_gimple_min_invariant (cond))
|
634 |
|
|
{
|
635 |
|
|
edge taken_edge = find_taken_edge (bb, cond);
|
636 |
|
|
|
637 |
|
|
if (bitmap_bit_p (visited, taken_edge->dest->index))
|
638 |
|
|
return NULL;
|
639 |
|
|
bitmap_set_bit (visited, taken_edge->dest->index);
|
640 |
|
|
return taken_edge;
|
641 |
|
|
}
|
642 |
|
|
|
643 |
|
|
return NULL;
|
644 |
|
|
}
|
645 |
|
|
|
646 |
|
|
/* E1 and E2 are edges into the same basic block. Return TRUE if the
|
647 |
|
|
PHI arguments associated with those edges are equal or there are no
|
648 |
|
|
PHI arguments, otherwise return FALSE. */
|
649 |
|
|
|
650 |
|
|
static bool
|
651 |
|
|
phi_args_equal_on_edges (edge e1, edge e2)
|
652 |
|
|
{
|
653 |
|
|
gimple_stmt_iterator gsi;
|
654 |
|
|
int indx1 = e1->dest_idx;
|
655 |
|
|
int indx2 = e2->dest_idx;
|
656 |
|
|
|
657 |
|
|
for (gsi = gsi_start_phis (e1->dest); !gsi_end_p (gsi); gsi_next (&gsi))
|
658 |
|
|
{
|
659 |
|
|
gimple phi = gsi_stmt (gsi);
|
660 |
|
|
|
661 |
|
|
if (!operand_equal_p (gimple_phi_arg_def (phi, indx1),
|
662 |
|
|
gimple_phi_arg_def (phi, indx2), 0))
|
663 |
|
|
return false;
|
664 |
|
|
}
|
665 |
|
|
return true;
|
666 |
|
|
}
|
667 |
|
|
|
668 |
|
|
/* We are exiting E->src, see if E->dest ends with a conditional
|
669 |
|
|
jump which has a known value when reached via E.
|
670 |
|
|
|
671 |
|
|
Special care is necessary if E is a back edge in the CFG as we
|
672 |
|
|
may have already recorded equivalences for E->dest into our
|
673 |
|
|
various tables, including the result of the conditional at
|
674 |
|
|
the end of E->dest. Threading opportunities are severely
|
675 |
|
|
limited in that case to avoid short-circuiting the loop
|
676 |
|
|
incorrectly.
|
677 |
|
|
|
678 |
|
|
Note it is quite common for the first block inside a loop to
|
679 |
|
|
end with a conditional which is either always true or always
|
680 |
|
|
false when reached via the loop backedge. Thus we do not want
|
681 |
|
|
to blindly disable threading across a loop backedge.
|
682 |
|
|
|
683 |
|
|
DUMMY_COND is a shared cond_expr used by condition simplification as scratch,
|
684 |
|
|
to avoid allocating memory.
|
685 |
|
|
|
686 |
|
|
HANDLE_DOMINATING_ASSERTS is true if we should try to replace operands of
|
687 |
|
|
the simplified condition with left-hand sides of ASSERT_EXPRs they are
|
688 |
|
|
used in.
|
689 |
|
|
|
690 |
|
|
STACK is used to undo temporary equivalences created during the walk of
|
691 |
|
|
E->dest.
|
692 |
|
|
|
693 |
|
|
SIMPLIFY is a pass-specific function used to simplify statements. */
|
694 |
|
|
|
695 |
|
|
void
|
696 |
|
|
thread_across_edge (gimple dummy_cond,
|
697 |
|
|
edge e,
|
698 |
|
|
bool handle_dominating_asserts,
|
699 |
|
|
VEC(tree, heap) **stack,
|
700 |
|
|
tree (*simplify) (gimple, gimple))
|
701 |
|
|
{
|
702 |
|
|
gimple stmt;
|
703 |
|
|
|
704 |
|
|
/* If E is a backedge, then we want to verify that the COND_EXPR,
|
705 |
|
|
SWITCH_EXPR or GOTO_EXPR at the end of e->dest is not affected
|
706 |
|
|
by any statements in e->dest. If it is affected, then it is not
|
707 |
|
|
safe to thread this edge. */
|
708 |
|
|
if (e->flags & EDGE_DFS_BACK)
|
709 |
|
|
{
|
710 |
|
|
ssa_op_iter iter;
|
711 |
|
|
use_operand_p use_p;
|
712 |
|
|
gimple last = gsi_stmt (gsi_last_bb (e->dest));
|
713 |
|
|
|
714 |
|
|
FOR_EACH_SSA_USE_OPERAND (use_p, last, iter, SSA_OP_USE | SSA_OP_VUSE)
|
715 |
|
|
{
|
716 |
|
|
tree use = USE_FROM_PTR (use_p);
|
717 |
|
|
|
718 |
|
|
if (TREE_CODE (use) == SSA_NAME
|
719 |
|
|
&& gimple_code (SSA_NAME_DEF_STMT (use)) != GIMPLE_PHI
|
720 |
|
|
&& gimple_bb (SSA_NAME_DEF_STMT (use)) == e->dest)
|
721 |
|
|
goto fail;
|
722 |
|
|
}
|
723 |
|
|
}
|
724 |
|
|
|
725 |
|
|
stmt_count = 0;
|
726 |
|
|
|
727 |
|
|
/* PHIs create temporary equivalences. */
|
728 |
|
|
if (!record_temporary_equivalences_from_phis (e, stack))
|
729 |
|
|
goto fail;
|
730 |
|
|
|
731 |
|
|
/* Now walk each statement recording any context sensitive
|
732 |
|
|
temporary equivalences we can detect. */
|
733 |
|
|
stmt = record_temporary_equivalences_from_stmts_at_dest (e, stack, simplify);
|
734 |
|
|
if (!stmt)
|
735 |
|
|
goto fail;
|
736 |
|
|
|
737 |
|
|
/* If we stopped at a COND_EXPR or SWITCH_EXPR, see if we know which arm
|
738 |
|
|
will be taken. */
|
739 |
|
|
if (gimple_code (stmt) == GIMPLE_COND
|
740 |
|
|
|| gimple_code (stmt) == GIMPLE_GOTO
|
741 |
|
|
|| gimple_code (stmt) == GIMPLE_SWITCH)
|
742 |
|
|
{
|
743 |
|
|
tree cond;
|
744 |
|
|
|
745 |
|
|
/* Extract and simplify the condition. */
|
746 |
|
|
cond = simplify_control_stmt_condition (e, stmt, dummy_cond, simplify,
|
747 |
|
|
handle_dominating_asserts);
|
748 |
|
|
|
749 |
|
|
if (cond && is_gimple_min_invariant (cond))
|
750 |
|
|
{
|
751 |
|
|
edge taken_edge = find_taken_edge (e->dest, cond);
|
752 |
|
|
basic_block dest = (taken_edge ? taken_edge->dest : NULL);
|
753 |
|
|
bitmap visited;
|
754 |
|
|
edge e2;
|
755 |
|
|
|
756 |
|
|
if (dest == e->dest)
|
757 |
|
|
goto fail;
|
758 |
|
|
|
759 |
|
|
/* DEST could be null for a computed jump to an absolute
|
760 |
|
|
address. If DEST is not null, then see if we can thread
|
761 |
|
|
through it as well, this helps capture secondary effects
|
762 |
|
|
of threading without having to re-run DOM or VRP. */
|
763 |
|
|
if (dest)
|
764 |
|
|
{
|
765 |
|
|
/* We don't want to thread back to a block we have already
|
766 |
|
|
visited. This may be overly conservative. */
|
767 |
|
|
visited = BITMAP_ALLOC (NULL);
|
768 |
|
|
bitmap_set_bit (visited, dest->index);
|
769 |
|
|
bitmap_set_bit (visited, e->dest->index);
|
770 |
|
|
do
|
771 |
|
|
{
|
772 |
|
|
e2 = thread_around_empty_block (taken_edge,
|
773 |
|
|
dummy_cond,
|
774 |
|
|
handle_dominating_asserts,
|
775 |
|
|
simplify,
|
776 |
|
|
visited);
|
777 |
|
|
if (e2)
|
778 |
|
|
taken_edge = e2;
|
779 |
|
|
}
|
780 |
|
|
while (e2);
|
781 |
|
|
BITMAP_FREE (visited);
|
782 |
|
|
}
|
783 |
|
|
|
784 |
|
|
remove_temporary_equivalences (stack);
|
785 |
|
|
register_jump_thread (e, taken_edge, NULL);
|
786 |
|
|
return;
|
787 |
|
|
}
|
788 |
|
|
}
|
789 |
|
|
|
790 |
|
|
/* We were unable to determine what out edge from E->dest is taken. However,
|
791 |
|
|
we might still be able to thread through successors of E->dest. This
|
792 |
|
|
often occurs when E->dest is a joiner block which then fans back out
|
793 |
|
|
based on redundant tests.
|
794 |
|
|
|
795 |
|
|
If so, we'll copy E->dest and redirect the appropriate predecessor to
|
796 |
|
|
the copy. Within the copy of E->dest, we'll thread one or more edges
|
797 |
|
|
to points deeper in the CFG.
|
798 |
|
|
|
799 |
|
|
This is a stopgap until we have a more structured approach to path
|
800 |
|
|
isolation. */
|
801 |
|
|
{
|
802 |
|
|
edge e2, e3, taken_edge;
|
803 |
|
|
edge_iterator ei;
|
804 |
|
|
bool found = false;
|
805 |
|
|
bitmap visited = BITMAP_ALLOC (NULL);
|
806 |
|
|
|
807 |
|
|
/* Look at each successor of E->dest to see if we can thread through it. */
|
808 |
|
|
FOR_EACH_EDGE (taken_edge, ei, e->dest->succs)
|
809 |
|
|
{
|
810 |
|
|
/* Avoid threading to any block we have already visited. */
|
811 |
|
|
bitmap_clear (visited);
|
812 |
|
|
bitmap_set_bit (visited, taken_edge->dest->index);
|
813 |
|
|
bitmap_set_bit (visited, e->dest->index);
|
814 |
|
|
|
815 |
|
|
/* Record whether or not we were able to thread through a successor
|
816 |
|
|
of E->dest. */
|
817 |
|
|
found = false;
|
818 |
|
|
e3 = taken_edge;
|
819 |
|
|
do
|
820 |
|
|
{
|
821 |
|
|
e2 = thread_around_empty_block (e3,
|
822 |
|
|
dummy_cond,
|
823 |
|
|
handle_dominating_asserts,
|
824 |
|
|
simplify,
|
825 |
|
|
visited);
|
826 |
|
|
if (e2)
|
827 |
|
|
{
|
828 |
|
|
e3 = e2;
|
829 |
|
|
found = true;
|
830 |
|
|
}
|
831 |
|
|
}
|
832 |
|
|
while (e2);
|
833 |
|
|
|
834 |
|
|
/* If we were able to thread through a successor of E->dest, then
|
835 |
|
|
record the jump threading opportunity. */
|
836 |
|
|
if (found)
|
837 |
|
|
{
|
838 |
|
|
edge tmp;
|
839 |
|
|
/* If there is already an edge from the block to be duplicated
|
840 |
|
|
(E2->src) to the final target (E3->dest), then make sure that
|
841 |
|
|
the PHI args associated with the edges E2 and E3 are the
|
842 |
|
|
same. */
|
843 |
|
|
tmp = find_edge (taken_edge->src, e3->dest);
|
844 |
|
|
if (!tmp || phi_args_equal_on_edges (tmp, e3))
|
845 |
|
|
register_jump_thread (e, taken_edge, e3);
|
846 |
|
|
}
|
847 |
|
|
|
848 |
|
|
}
|
849 |
|
|
BITMAP_FREE (visited);
|
850 |
|
|
}
|
851 |
|
|
|
852 |
|
|
fail:
|
853 |
|
|
remove_temporary_equivalences (stack);
|
854 |
|
|
}
|