Hi,
This patch fixes an issue with the routine that converts STRING_CST to a
StringExp for the dmd front-end to use during the semantic pass.
The null terminator gets re-added by the ExprVisitor when lowering
StringExp back into a STRING_CST during the code generator pass.
Bootstrapped and regression tested on x86_64-linux-gnu/-m32/-mx32,
committed to mainline and backported to the releases/gcc-11 branch.
Regards,
Iain.
---
gcc/d/ChangeLog:
PR d/102185
* d-builtins.cc (d_eval_constant_expression): Don't include
terminating null pointer in string expression conversion.
gcc/testsuite/ChangeLog:
PR d/102185
* gdc.dg/pr102185.d: New test.
---
gcc/d/d-builtins.cc | 2 +-
gcc/testsuite/gdc.dg/pr102185.d | 7 +++++++
2 files changed, 8 insertions(+), 1 deletion(-)
create mode 100644 gcc/testsuite/gdc.dg/pr102185.d
diff --git a/gcc/d/d-builtins.cc b/gcc/d/d-builtins.cc
index ab39d69c294..33347a14c67 100644
--- a/gcc/d/d-builtins.cc
+++ b/gcc/d/d-builtins.cc
@@ -380,7 +380,7 @@ d_eval_constant_expression (const Loc &loc, tree cst)
else if (code == STRING_CST)
{
const void *string = TREE_STRING_POINTER (cst);
- size_t len = TREE_STRING_LENGTH (cst);
+ size_t len = TREE_STRING_LENGTH (cst) - 1;
return StringExp::create (loc, CONST_CAST (void *, string), len);
}
else if (code == VECTOR_CST)
diff --git a/gcc/testsuite/gdc.dg/pr102185.d b/gcc/testsuite/gdc.dg/pr102185.d
new file mode 100644
index 00000000000..39823a3c556
--- /dev/null
+++ b/gcc/testsuite/gdc.dg/pr102185.d
@@ -0,0 +1,7 @@
+// https://gcc.gnu.org/bugzilla/show_bug.cgi?id=102185
+// { dg-do compile }
+
+static assert(__traits(getTargetInfo, "floatAbi").length == 0 ||
+ __traits(getTargetInfo, "floatAbi") == "hard" ||
+ __traits(getTargetInfo, "floatAbi") == "soft" ||
+ __traits(getTargetInfo, "floatAbi") == "softfp");
--
2.30.2
