[Bug preprocessor/88974] [9 Regression] ICE: Segmentation fault (in linemap_resolve_location)

[Bug preprocessor/88974] [9 Regression] ICE: Segmentation fault (in linemap_resolve_location)

[dcb314 at hotmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=88974

David Binderman <dcb314 at hotmail dot com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
                 CC|                            |dcb314 at hotmail dot com

--- Comment #4 from David Binderman <dcb314 at hotmail dot com> ---
Interestingly, a valgrind enabled version of recent gcc trunk does this:

./c-c++-common/cpp/pr88974.c:4:31: error: missing terminating > character
    4 | #if __has_include (<pr88974.h)
      |                               ^
./c-c++-common/cpp/pr88974.c:4:31: error: missing ')' after "__has_include"
operand
==48268== Conditional jump or move depends on uninitialised value(s)
==48268==    at 0x165057A: cpp_get_token_1(cpp_reader*, unsigned int*)
(macro.c:2918)
==48268==    by 0x163CA9C: _cpp_parse_expr (expr.c:1308)
==48268==    by 0x1635B41: do_if(cpp_reader*) (directives.c:2018)
==48268==    by 0x163780A: _cpp_handle_directive (directives.c:541)


Maybe the fix could be tidied up a bit ?

[Bug preprocessor/88974] [9 Regression] ICE: Segmentation fault (in linemap_resolve_location)

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=88974

--- Comment #5 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Created attachment 48042
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=48042&action=edit
gcc10-pr88974.patch

Untested fix for that.  Or instead set result->src_loc to result->src_loc =
pfile->line_table->highest_line; ?
Dunno what is better.

[Bug preprocessor/88974] [9 Regression] ICE: Segmentation fault (in linemap_resolve_location)

[jakub at gcc dot gnu.org]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=88974

--- Comment #6 from Jakub Jelinek <jakub at gcc dot gnu.org> ---
Created attachment 48049
  --> https://gcc.gnu.org/bugzilla/attachment.cgi?id=48049&action=edit
gcc10-pr88974.patch

Unfortunately, that patch failed bootstrap miserably (e.g. the compiler
generated __static_initialization_and_destruction_0 then has UNKNOWN_LOCATION
and ICEs because of that).
And this updated patch, while it bootstrapped, had quite a few testsuite
regressions.  So, I'm afraid I don't know where would in some cases the src_loc
be preinitialized from earlier and where it wouldn't.  Mystery.

[Bug preprocessor/88974] [9 Regression] ICE: Segmentation fault (in linemap_resolve_location)

[dcb314 at hotmail dot com]

https://gcc.gnu.org/bugzilla/show_bug.cgi?id=88974

--- Comment #7 from David Binderman <dcb314 at hotmail dot com> ---
(In reply to Jakub Jelinek from comment #6)
> And this updated patch, while it bootstrapped, had quite a few testsuite
> regressions.  So, I'm afraid I don't know where would in some cases the
> src_loc be preinitialized from earlier and where it wouldn't.  Mystery.

Thanks for looking into this. It doesn't look all that important to me,
but, for the record, it is still producing valgrind errors over 2 years later:

./c-c++-common/cpp/pr88974.c:4:31: error: missing terminating > character
    4 | #if __has_include (<pr88974.h)
      |                               ^
./c-c++-common/cpp/pr88974.c:4:31: error: missing ')' after "__has_include"
operand
==24548== Conditional jump or move depends on uninitialised value(s)
==24548==    at 0x1D21FD4: cpp_get_token_1(cpp_reader*, unsigned int*)
(macro.cc:3044)