[Bug preprocessor/110558] New: __has_include argument expansion results in unexpected filename

[Bug preprocessor/110558] New: __has_include argument expansion results in unexpected filename

[provisorisch at online dot de]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=110558

            Bug ID: 110558
           Summary: __has_include argument expansion results in unexpected
                    filename
           Product: gcc
           Version: unknown
            Status: UNCONFIRMED
          Severity: normal
          Priority: P3
         Component: preprocessor
          Assignee: unassigned at gcc dot gnu.org
          Reporter: provisorisch at online dot de
  Target Milestone: ---

Hello!
I am building a path to an include file using a function like macro and then
checking if the file exists using __has_include.
However, if the argument to that macro is padded with spaces it will fail to
find the file.

Here an example with following 2 files in the same directory:


my.header.h:
--------
#pragma message("included my.header.h")
--------


test.c:
--------
#ifndef USE_IMPL
#   define USE_IMPL my
#endif

#define STRINGIFY_(X) #X
#define STRINGIFY(X) STRINGIFY_(X)
#define MAKE_INCLUDE_PATH_(IMPL, NAME) STRINGIFY(IMPL.NAME)
#define MAKE_INCLUDE_PATH(NAME) MAKE_INCLUDE_PATH_(USE_IMPL, NAME)

// checking __has_include with spaces
#if __has_include(MAKE_INCLUDE_PATH( header.h )) // this will return false!
#   pragma message("\""MAKE_INCLUDE_PATH( header.h )"\" exists")
#else
#   pragma message("\""MAKE_INCLUDE_PATH( header.h )"\" does NOT exist!") //
except it does
#endif
#include MAKE_INCLUDE_PATH( header.h ) // works nonetheless

// checking __has_include without spaces
#if __has_include(MAKE_INCLUDE_PATH(header.h)) // works
#   pragma message("\""MAKE_INCLUDE_PATH(header.h)"\" exists")
#else
#   pragma message("\""MAKE_INCLUDE_PATH(header.h)"\" does NOT exist!")
#endif
#include MAKE_INCLUDE_PATH(header.h)


int main() { return 0; }
--------


Then to compile:

# gcc -Wall -Wextra test.c                                                     
                                                                               
                                                                               
          test.c:14:12: note: ‘#pragma message: "my.header.h" does NOT exist!’
   14 | #   pragma message("\""MAKE_INCLUDE_PATH( header.h )"\" does NOT
exist!") // except it does
      |            ^~~~~~~
In file included from test.c:16:
my.header.h:1:9: note: ‘#pragma message: included my.header.h’
    1 | #pragma message("included my.header.h")
      |         ^~~~~~~
test.c:20:12: note: ‘#pragma message: "my.header.h" exists’
   20 | #   pragma message("\""MAKE_INCLUDE_PATH(header.h)"\" exists")
      |            ^~~~~~~
In file included from test.c:24:
my.header.h:1:9: note: ‘#pragma message: included my.header.h’
    1 | #pragma message("included my.header.h")
      |         ^~~~~~~


The first __has_include fails to find "my.header.h" but the subsequent #include
works.
The second __has_include works as expected.

Using strace I noticed that an extra space character ended up in the filename:
openat(AT_FDCWD, "my. header.h", O_RDONLY|O_NOCTTY) = -1 ENOENT (No such file
or directory)

I would have expected the first __has_include to behave like the second one.

[Bug preprocessor/110558] __has_include argument expansion results in unexpected filename

[pinskia at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=110558

Andrew Pinski <pinskia at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           See Also|                            |https://gcc.gnu.org/bugzill
                   |                            |a/show_bug.cgi?id=80753

--- Comment #1 from Andrew Pinski <pinskia at gcc dot gnu.org> ---
I think this is a dup of bug 80753 .

[Bug preprocessor/110558] __has_include argument expansion results in unexpected filename

[provisorisch at online dot de]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=110558

--- Comment #2 from provisorisch at online dot de ---
Not quite, but you will run into bug 80753 if the header file does not exist:
The second #include will not cause an error in that case.
The first #include however will cause an error as expected - presumably only
because the first __has_include ends up with a different file name.

[Bug preprocessor/110558] __has_include argument expansion results in unexpected filename

[fw at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=110558

--- Comment #3 from Florian Weimer <fw at gcc dot gnu.org> ---
It tries to read "my. header.h" for some reason, even though the
MAKE_INCLUDE_PATH macro produces "my.header.h" in other contexts (not just in
#include directives). I doubt this is related to bug 80753.

[Bug preprocessor/110558] __has_include argument expansion results in unexpected filename

[lhyatt at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=110558

Lewis Hyatt <lhyatt at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |NEW
     Ever confirmed|0                           |1
   Last reconfirmed|                            |2023-12-11
                 CC|                            |lhyatt at gcc dot gnu.org

--- Comment #4 from Lewis Hyatt <lhyatt at gcc dot gnu.org> ---
__has_include was added I think in GCC 5, and re-implemented in GCC 10, but
this issue with padding in the macro expansion was never handled correctly it
seems. I am testing the below which will fix it, will submit it soon with a
testcase. Not sure if it would be eligible for backport or not, but it would
apply cleanly to all active branches also.

diff --git a/libcpp/macro.cc b/libcpp/macro.cc
index 6f24a9d6f3a..15140c60023 100644
--- a/libcpp/macro.cc
+++ b/libcpp/macro.cc
@@ -398,6 +398,8 @@ builtin_has_include (cpp_reader *pfile, cpp_hashnode *op,
bool has_next)
               NODE_NAME (op));

   pfile->state.angled_headers = true;
+  const auto sav_padding = pfile->state.directive_wants_padding;
+  pfile->state.directive_wants_padding = true;
   const cpp_token *token = cpp_get_token_no_padding (pfile);
   bool paren = token->type == CPP_OPEN_PAREN;
   if (paren)
@@ -406,6 +408,7 @@ builtin_has_include (cpp_reader *pfile, cpp_hashnode *op,
bool has_next)
     cpp_error (pfile, CPP_DL_ERROR,
               "missing '(' before \"%s\" operand", NODE_NAME (op));
   pfile->state.angled_headers = false;
+  pfile->state.directive_wants_padding = sav_padding;

   bool bracket = token->type != CPP_STRING;
   char *fname = NULL;

[Bug preprocessor/110558] __has_include argument expansion results in unexpected filename

[lhyatt at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=110558

Lewis Hyatt <lhyatt at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
           Keywords|                            |patch
                URL|                            |https://gcc.gnu.org/piperma
                   |                            |il/gcc-patches/2023-Decembe
                   |                            |r/640386.html

--- Comment #5 from Lewis Hyatt <lhyatt at gcc dot gnu.org> ---
Patch submitted for review:
https://gcc.gnu.org/pipermail/gcc-patches/2023-December/640386.html

[Bug preprocessor/110558] __has_include argument expansion results in unexpected filename

[cvs-commit at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=110558

--- Comment #6 from GCC Commits <cvs-commit at gcc dot gnu.org> ---
The master branch has been updated by Lewis Hyatt <lhyatt@gcc.gnu.org>:

https://gcc.gnu.org/g:942497ad74272e0ef16020d628e471c5f21474b0

commit r14-9465-g942497ad74272e0ef16020d628e471c5f21474b0
Author: Lewis Hyatt <lhyatt@gmail.com>
Date:   Tue Dec 12 17:46:36 2023 -0500

    libcpp: Fix macro expansion for argument of __has_include [PR110558]

    When the file name for a #include directive is the result of stringifying a
    macro argument, libcpp needs to take some care to get the whitespace
    correct; in particular stringify_arg() needs to see a CPP_PADDING token
    between macro tokens so that it can figure out when to output space between
    tokens. The CPP_PADDING tokens are not normally generated when handling a
    preprocessor directive, but for #include-like directives, libcpp sets the
    state variable pfile->state.directive_wants_padding to TRUE so that the
    CPP_PADDING tokens will be output, and then everything works fine for
    computed includes.

    As the PR points out, things do not work fine for __has_include. Fix that
by
    setting the state variable the same as is done for #include.

    libcpp/ChangeLog:

            PR preprocessor/110558
            * macro.cc (builtin_has_include): Set
            pfile->state.directive_wants_padding prior to lexing the
            file name, in case it comes from macro expansion.

    gcc/testsuite/ChangeLog:

            PR preprocessor/110558
            * c-c++-common/cpp/has-include-2.c: New test.
            * c-c++-common/cpp/has-include-2.h: New test.

[Bug preprocessor/110558] __has_include argument expansion results in unexpected filename

[lhyatt at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=110558

Lewis Hyatt <lhyatt at gcc dot gnu.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
   Target Milestone|---                         |14.0
         Resolution|---                         |FIXED
             Status|NEW                         |RESOLVED

--- Comment #7 from Lewis Hyatt <lhyatt at gcc dot gnu.org> ---
Fixed for GCC 14.