diff options
author | Hao Liu <hliu@os.amperecomputing.com> | 2023-12-06 14:52:19 +0800 |
---|---|---|
committer | Hao Liu <hliu@os.amperecomputing.com> | 2023-12-08 11:18:03 +0800 |
commit | 2efe3a7de0107618397264017fb045f237764cc7 (patch) | |
tree | ee0a1dd559c20e6e9fa0b038ad49e2f16564974c /gcc/tree-ssa-loop-niter.cc | |
parent | 9f7ad5eff3bf1e42aac0825b37d2c9ab43eaafd2 (diff) | |
download | gcc-2efe3a7de0107618397264017fb045f237764cc7.zip gcc-2efe3a7de0107618397264017fb045f237764cc7.tar.gz gcc-2efe3a7de0107618397264017fb045f237764cc7.tar.bz2 |
tree-optimization/112774: extend the SCEV CHREC tree with a nonwrapping flag
The flag is defined as CHREC_NOWRAP(tree), and will be dumped from
"{offset, +, 1}_1" to "{offset, +, 1}<nw>_1" (nw is short for nonwrapping).
Two SCEV interfaces record_nonwrapping_chrec and nonwrapping_chrec_p are
added to set and check the flag respectively.
As resetting the SCEV cache (i.e., the chrec trees) may not reset the
loop->estimate_state, free_numbers_of_iterations_estimates is called
explicitly in loop vectorization to make sure the flag can be
calculated propriately by niter.
gcc/ChangeLog:
PR tree-optimization/112774
* tree-pretty-print.cc: if nonwrapping flag is set, chrec will be
printed with additional <nw> info.
* tree-scalar-evolution.cc: add record_nonwrapping_chrec and
nonwrapping_chrec_p to set and check the new flag respectively.
* tree-scalar-evolution.h: Likewise.
* tree-ssa-loop-niter.cc (idx_infer_loop_bounds,
infer_loop_bounds_from_pointer_arith, infer_loop_bounds_from_signedness,
scev_probably_wraps_p): call record_nonwrapping_chrec before
record_nonwrapping_iv, call nonwrapping_chrec_p to check the flag is
set and return false from scev_probably_wraps_p.
* tree-vect-loop.cc (vect_analyze_loop): call
free_numbers_of_iterations_estimates explicitly.
* tree-core.h: document the nothrow_flag usage in CHREC_NOWRAP
* tree.h: add CHREC_NOWRAP(NODE), base.nothrow_flag is used to
represent the nonwrapping info.
gcc/testsuite/ChangeLog:
* gcc.dg/tree-ssa/scev-16.c: New test.
Diffstat (limited to 'gcc/tree-ssa-loop-niter.cc')
-rw-r--r-- | gcc/tree-ssa-loop-niter.cc | 21 |
1 files changed, 16 insertions, 5 deletions
diff --git a/gcc/tree-ssa-loop-niter.cc b/gcc/tree-ssa-loop-niter.cc index 2098bef..d465e0e 100644 --- a/gcc/tree-ssa-loop-niter.cc +++ b/gcc/tree-ssa-loop-niter.cc @@ -4206,11 +4206,15 @@ idx_infer_loop_bounds (tree base, tree *idx, void *dta) /* If access is not executed on every iteration, we must ensure that overlow may not make the access valid later. */ - if (!dominated_by_p (CDI_DOMINATORS, loop->latch, gimple_bb (data->stmt)) - && scev_probably_wraps_p (NULL_TREE, - initial_condition_in_loop_num (ev, loop->num), - step, data->stmt, loop, true)) - upper = false; + if (!dominated_by_p (CDI_DOMINATORS, loop->latch, gimple_bb (data->stmt))) + { + if (scev_probably_wraps_p (NULL_TREE, + initial_condition_in_loop_num (ev, loop->num), + step, data->stmt, loop, true)) + upper = false; + } + else + record_nonwrapping_chrec (ev); record_nonwrapping_iv (loop, init, step, data->stmt, low, high, false, upper); return true; @@ -4324,6 +4328,7 @@ infer_loop_bounds_from_pointer_arith (class loop *loop, gimple *stmt) if (flag_delete_null_pointer_checks && int_cst_value (low) == 0) low = build_int_cstu (TREE_TYPE (low), TYPE_ALIGN_UNIT (TREE_TYPE (type))); + record_nonwrapping_chrec (scev); record_nonwrapping_iv (loop, base, step, stmt, low, high, false, true); } @@ -4371,6 +4376,7 @@ infer_loop_bounds_from_signedness (class loop *loop, gimple *stmt) high = wide_int_to_tree (type, r.upper_bound ()); } + record_nonwrapping_chrec (scev); record_nonwrapping_iv (loop, base, step, stmt, low, high, false, true); } @@ -5505,6 +5511,11 @@ scev_probably_wraps_p (tree var, tree base, tree step, if (loop_exits_before_overflow (base, step, at_stmt, loop)) return false; + /* Check the nonwrapping flag, which may be set by niter analysis (e.g., the + above loop exits before overflow). */ + if (var && nonwrapping_chrec_p (analyze_scalar_evolution (loop, var))) + return false; + /* At this point we still don't have a proof that the iv does not overflow: give up. */ return true; |