[PATCH] [gdb/build] Reimplement Wstringop-overread workaround

[PATCH] [gdb/build] Reimplement Wstringop-overread workaround

[Tom de Vries]

While working on commit 391c4026573 ("[gdb] Simplify debuginfod_is_enabled") I
noticed this Wstringop-overread workaround:
...
      url_view = url_view.substr (off);
      /* g++ 11.2.1 on s390x, g++ 11.3.1 on ppc64le and g++ 11 on
	 hppa seem convinced url_view might be of SIZE_MAX length.
	 And so complains because the length of an array can only
	 be PTRDIFF_MAX.  */
      DIAGNOSTIC_PUSH
      DIAGNOSTIC_IGNORE_STRINGOP_OVERREAD
      off = url_view.find_first_of (' ');
      DIAGNOSTIC_POP
...

I had difficulty understanding how the warning got triggered, and why it was
ok to ignore it, so I investigated this and ended up filing a gcc PR [1].

While doing so, I realized that this:
...
-      url_view = url_view.substr (off);
+      url_view = url_view.substr (off, PTRDIFF_MAX);
...
is a simpler workaround, that:
- is not specific to the warning and also
- states explicitly what the assumption is we're making.

I ended up using this instead to make the workaround part more minimal:
...
       url_view = url_view.substr (off);
+#if defined (__GNUC__) && !defined (__clang__)
+      url_view = url_view.substr (0, PTRDIFF_MAX);
+#endif
       off = url_view.find_first_of (' ');
...

The gcc PR got closed (because there's no known reproducer with gcc 12 and
later), but I use the workaround without version limit, because there's no
actual proof that the problem is fixed.

Tested on x86_64-linux.

[1] https://gcc.gnu.org/bugzilla/show_bug.cgi?id=124879
---
 gdb/debuginfod-support.c | 15 ++++++++-------
 1 file changed, 8 insertions(+), 7 deletions(-)

diff --git a/gdb/debuginfod-support.c b/gdb/debuginfod-support.c
index 40b816c5a40..b5e790c3b95 100644
--- a/gdb/debuginfod-support.c
+++ b/gdb/debuginfod-support.c
@@ -249,14 +249,15 @@ debuginfod_is_enabled ()
       if (off == std::string_view::npos)
 	break;
       url_view = url_view.substr (off);
-      /* g++ 11.2.1 on s390x, g++ 11.3.1 on ppc64le and g++ 11 on
-	 hppa seem convinced url_view might be of SIZE_MAX length.
-	 And so complains because the length of an array can only
-	 be PTRDIFF_MAX.  */
-      DIAGNOSTIC_PUSH
-      DIAGNOSTIC_IGNORE_STRINGOP_OVERREAD
+#if defined (__GNUC__) && !defined (__clang__)
+      /* With g++ 11, we encounter a Wstringop-overread in
+	 url_view.find_first_of.  G++ seems convinced url_view might be of
+	 SIZE_MAX length here.  And so complains because the length of an
+	 array can only be PTRDIFF_MAX.  Work around this by explicitly
+	 limiting the size of url_view to PTRDIFF_MAX.  See PR gcc/124879.  */
+      url_view = url_view.substr (0, PTRDIFF_MAX);
+#endif
       off = url_view.find_first_of (' ');
-      DIAGNOSTIC_POP
       gdb_printf
 	(_("  <%ps>\n"),
 	 styled_string (file_name_style.style (),

base-commit: 391c4026573aeb1e6e09f36816901f2ade537e1a
-- 
2.51.0

Re: [PATCH] [gdb/build] Reimplement Wstringop-overread workaround

[Tom Tromey]

>>>>> "Tom" == Tom de Vries <tdevries@suse.de> writes:

Tom> The gcc PR got closed (because there's no known reproducer with gcc 12 and
Tom> later), but I use the workaround without version limit, because there's no
Tom> actual proof that the problem is fixed.

IIUC from the bug, it is believed to have been fixed though right?

Anyway this seems fine to me.
Approved-By: Tom Tromey <tom@tromey.com>

Tom

Re: [PATCH] [gdb/build] Reimplement Wstringop-overread workaround

[Tom de Vries]

On 4/14/26 7:38 PM, Tom Tromey wrote:
>>>>>> "Tom" == Tom de Vries <tdevries@suse.de> writes:
> 
Hi Tom,

thanks for the review.

> Tom> The gcc PR got closed (because there's no known reproducer with gcc 12 and
> Tom> later), but I use the workaround without version limit, because there's no
> Tom> actual proof that the problem is fixed.
> 
> IIUC from the bug, it is believed to have been fixed though right?

Yes, so I've updated the workaround to trigger only for g++ < 12.1.

> Anyway this seems fine to me.
> Approved-By: Tom Tromey <tom@tromey.com>

Thanks, pushed.
- Tom