On 2021-03-09 9:26 a.m., Jakub Jelinek wrote:
On Tue, Mar 09, 2021 at 09:12:36AM -0500, Vladimir Makarov via Gcc-patches
wrote:
diff --git a/gcc/lra-constraints.c b/gcc/lra-constraints.c
index 76e3ff7efe6..feff766c590 100644
--- a/gcc/lra-constraints.c
+++ b/gcc/lra-constraints.c
@@ -3452,6 +3452,10 @@ process_address_1 (int nop, bool check_only_p,
constraint
= skip_contraint_modifiers (curr_static_id->operand[nop].constraint);
+ if ('0' <= constraint[0] && constraint[0] <= '9')
Yoda conditions?
Anyway, more importantly, the docs say:
If multiple
digits are encountered consecutively, they are interpreted as a
single decimal integer. There is scant chance for ambiguity, since
to-date it has never been desirable that '10' be interpreted as
matching either operand 1 _or_ operand 0. Should this be desired,
one can use multiple alternatives instead.
so, doesn't this need to use
{
char **end;
unsigned long dup = strtoul (constraint, &end, 10);
constraint
= skip_constraint_modifiers (curr_static_id_operand[dup].constraint);
}
instead?
Yes, you are right. Thank you for pointing this out. I'll fix it soon.
There are places where these constraints are treated the same which
should be fixed too. They are not critical as they are used in cost
calculation.
Otherwise, if you have (even large) testcase that misbehaves with 0-9
dups, it should be possible to create one with 10-30?
