On 08/03/18 23:36, Jeff Law wrote:
> On 08/01/2018 05:35 AM, Bernd Edlinger wrote:
>> Hi,
>>
>> this completes the previous patches, and adds a check in varasm.c
>> that ensures that all string constants are NUL terminated,
>> And that varasm does not strip anything but _exactly_ one NUL
>> character.
>>
>> Bootstrapped and reg-tested on x86_64-pc-linux-gnu.
>> Is it OK for trunk?
>>
>>
>> Thanks
>> Bernd.
>>
>>
>> patch-varasm.diff
>>
>>
>> 2018-08-01 Bernd Edlinger <bernd.edlin...@hotmail.de>
>>
>> * varasm.c (check_string_literal): New checking function.
>> (output_constant): Use it.
> My biggest concern here is that we've explicitly defined STRING_CST
> nodes as not needing NUL termination. See generic.texi. That explicit
> definition is almost certainly derived from existing practice that long
> pre-dates the introduction of gimple/generic.
>
> Even so I'm generally OK with the concept of tightening the rules here.
> If we need to fault in more fixes, that's fine with me. I'm assuming
> that you and Ian have sorted out what to do WRT Go.
>
> If we decide to go forward, you'll definitely want to update the
> STRING_CST documentation in generic.texi.
>
This is the same patch with STRING_CST documentation updated.
Bernd.
2018-08-01 Bernd Edlinger <bernd.edlin...@hotmail.de>
* doc/generic.texi (STRING_CST): Update.
* varasm.c (check_string_literal): New checking function.
(output_constant): Use it.
diff -pur gcc/varasm.c gcc/varasm.c
--- gcc/varasm.c 2018-07-17 11:19:27.000000000 +0200
+++ gcc/varasm.c 2018-07-31 10:16:12.058827505 +0200
@@ -4774,6 +4774,29 @@ initializer_constant_valid_for_bitfield_
return false;
}
+/* Check if a STRING_CST fits into the field.
+ Tolerate only the case when the NUL termination
+ does not fit into the field. */
+
+bool
+check_string_literal (tree string, unsigned HOST_WIDE_INT size)
+{
+ tree eltype = TREE_TYPE (TREE_TYPE (string));
+ unsigned HOST_WIDE_INT elts = tree_to_uhwi (TYPE_SIZE_UNIT (eltype));
+ const char *p = TREE_STRING_POINTER (string);
+ int len = TREE_STRING_LENGTH (string);
+
+ if (elts != 1 && elts != 2 && elts != 4)
+ return false;
+ if (len <= 0 || len % elts != 0)
+ return false;
+ if ((unsigned)len != size && (unsigned)len != size + elts)
+ return false;
+ if (memcmp (p + len - elts, "\0\0\0\0", elts) != 0)
+ return false;
+ return true;
+}
+
/* output_constructor outer state of relevance in recursive calls, typically
for nested aggregate bitfields. */
@@ -4942,6 +4965,7 @@ output_constant (tree exp, unsigned HOST
case STRING_CST:
thissize
= MIN ((unsigned HOST_WIDE_INT)TREE_STRING_LENGTH (exp), size);
+ gcc_checking_assert (check_string_literal (exp, thissize));
assemble_string (TREE_STRING_POINTER (exp), thissize);
break;
case VECTOR_CST:
Index: gcc/doc/generic.texi
===================================================================
--- gcc/doc/generic.texi (revision 263072)
+++ gcc/doc/generic.texi (working copy)
@@ -1162,9 +1162,13 @@ These nodes represent string-constants. The @code
returns the length of the string, as an @code{int}. The
@code{TREE_STRING_POINTER} is a @code{char*} containing the string
itself. The string may not be @code{NUL}-terminated, and it may contain
-embedded @code{NUL} characters. Therefore, the
-@code{TREE_STRING_LENGTH} includes the trailing @code{NUL} if it is
-present.
+embedded @code{NUL} characters. However, the
+@code{TREE_STRING_LENGTH} always includes a trailing @code{NUL} that
+is not part of the language string literal but appended by the front end.
+If the string shall not be @code{NUL}-terminated the @code{TREE_TYPE}
+is one character shorter than @code{TREE_STRING_LENGTH}.
+Excess caracters other than one trailing @code{NUL} character are not
+permitted.
For wide string constants, the @code{TREE_STRING_LENGTH} is the number
of bytes in the string, and the @code{TREE_STRING_POINTER}
@@ -1173,6 +1177,11 @@ target system (that is, as integers in the target
non-wide string constants are distinguished only by the @code{TREE_TYPE}
of the @code{STRING_CST}.
+String constants may also be used for other purpose like assember
+constraints or attributes. These have no @code{TREE_TYPE} and
+need no explicit @code{NUL}-termination. Note there is always
+another @code{NUL}-byte beyond @code{TREE_STRING_LENGTH}.
+
FIXME: The formats of string constants are not well-defined when the
target system bytes are not the same width as host system bytes.
