On March 29, 2021 8:11:51 PM GMT+02:00, Jan Hubicka <[email protected]> wrote:
>Hi,
>this patch fixes if (deref) that is reversed in merge_call_lhs_flags.
>I re-run stats on cc1plus that seems comparable to last run.
>
>Bootstrapped/regtested x86_64-linux.
>
>Alias oracle query stats:
>
>refs_may_alias_p: 66575703 disambiguations, 76462658 queries
>
>ref_maybe_used_by_call_p: 364568 disambiguations, 67280898 queries
>
>call_may_clobber_ref_p: 217587 disambiguations, 221304 queries
>
>nonoverlapping_component_refs_p: 0 disambiguations, 36514 queries
>
>nonoverlapping_refs_since_match_p: 17351 disambiguations, 61127 must
>overlaps, 79253 queries
>aliasing_component_refs_p: 64537 disambiguations, 2076259 queries
>
>TBAA oracle: 24082735 disambiguations 57590168 queries
>
> 11651231 are in alias set 0
> 10865812 queries asked about the same object
> 132 queries asked about the same alias set
> 0 access volatile
> 9473619 are dependent in the DAG
> 1516639 are aritificially in conflict with void *
>
>Modref stats:
>
>modref use: 13892 disambiguations, 409262 queries
>
>modref clobber: 1358267 disambiguations, 15950909 queries
>
>3287388 tbaa queries (0.206094 per modref query)
>
>519328 base compares (0.032558 per modref query)
>
>
>PTA query stats:
>
>pt_solution_includes: 11809203 disambiguations, 19340135 queries
>
>pt_solutions_intersect: 1217666 disambiguations, 13770732 queries
>
>
>gcc/ChangeLog:
>
>2021-03-29 Jan Hubicka <[email protected]>
>
> * ipa-modref.c (merge_call_lhs_flags): Correct handling of deref.
> (analyze_ssa_name_flags): Fix typo in comment.
>
>gcc/testsuite/ChangeLog:
>
>2021-03-29 Jan Hubicka <[email protected]>
>
> * gcc.c-torture/compile/pr99751.c: New test.
Why compile torture?
>diff --git a/gcc/ipa-modref.c b/gcc/ipa-modref.c
>index 7aaf53be8f4..5f33bb5b410 100644
>--- a/gcc/ipa-modref.c
>+++ b/gcc/ipa-modref.c
>@@ -1545,9 +1545,9 @@ merge_call_lhs_flags (gcall *call, int arg, int
>index, bool deref,
> tree lhs = gimple_call_lhs (call);
> analyze_ssa_name_flags (lhs, lattice, depth + 1, ipa);
> if (deref)
>- lattice[index].merge (lattice[SSA_NAME_VERSION (lhs)]);
>- else
> lattice[index].merge_deref (lattice[SSA_NAME_VERSION (lhs)], false);
>+ else
>+ lattice[index].merge (lattice[SSA_NAME_VERSION (lhs)]);
> }
> /* In the case of memory store we can do nothing. */
> else
>@@ -1621,7 +1621,7 @@ analyze_ssa_name_flags (tree name,
>vec<modref_lattice> &lattice, int depth,
> else if (gcall *call = dyn_cast <gcall *> (use_stmt))
> {
> tree callee = gimple_call_fndecl (call);
>- /* Return slot optiomization would require bit of propagation;
>+ /* Return slot optimization would require bit of propagation;
> give up for now. */
> if (gimple_call_return_slot_opt_p (call)
> && gimple_call_lhs (call) != NULL_TREE
>diff --git a/gcc/testsuite/gcc.c-torture/compile/pr99751.c
>b/gcc/testsuite/gcc.c-torture/compile/pr99751.c
>new file mode 100644
>index 00000000000..f1d5101c359
>--- /dev/null
>+++ b/gcc/testsuite/gcc.c-torture/compile/pr99751.c
>@@ -0,0 +1,20 @@
>+int *ptr1 = 0, **ptr2 = &ptr1;
>+
>+int *identity(int *p)
>+{
>+ return p;
>+}
>+
>+void store_to_c(int *p)
>+{
>+ *ptr2 = identity(p);
>+}
>+
>+int main()
>+{
>+ int f;
>+ store_to_c(&f);
>+ if (ptr1 != &f)
>+ __builtin_abort();
>+ return 0;
>+}
