GDB 10 branching - 2020-07-26 update

classic Classic list List threaded Threaded
2 messages Options
Reply | Threaded
Open this post in threaded view
|

GDB 10 branching - 2020-07-26 update

Joel Brobecker
Hi everyone,

Here is a quick recap of where we are with respect to the GDB 10
release. We did well this week, with getting 5 entries out of the way,
and with only 2 being added (knowing that patches for those have
already been sent).

I'm concerned about two entries:

  * [TomT/HannesD] <PR win32/25302>
    Mismatching fstat() function calls in gdb_bfd_open() and cache_bstat()
    https://sourceware.org/bugzilla/show_bug.cgi?id=25302

        I'm wondering if we should be setting up a live discussion
        to first recap, and then answer Pedro's questions about
        Tom's comments on his current patch.

        This one triggers a crash, so it's hard to consider releasing
        without it :-(.

  * [ChristianB (reported by Eli)]
    Update gnulib to import fixes needed for mingw.org's MinGW

        The concern for this one is less, but there hasn't been
        any message being exchanged on this topic that I could see
        since Pedro's reply on the 18th.

What do people think of our chances for those two?

Fixed Since the Previous Update:
--------------------------------

  * [KevinB] <PR gdb/25631>
    GDB cannot access unwritten-to mmap'd buffer from core file
    https://sourceware.org/bugzilla/show_bug.cgi?id=25631

  * [EliZ]
    gdbserver: incorrect socklen_t configure test on mingw.org's MinGW
    https://sourceware.org/pipermail/gdb-patches/2020-July/170019.html

  * [LudovicC (reported by Simon)]
    Add support for Guile 2.2
    https://sourceware.org/pipermail/gdb-patches/2020-June/169936.html

  * [EliZ]
    binutils: libctf build failure on Windows with mingw.org's MinGW
    https://sourceware.org/pipermail/gdb-patches/2020-July/170042.html

  * [SandraL]
    testsuite: gdb_wrapper handling polluting source directory with .o files
    https://sourceware.org/pipermail/gdb-patches/2020-July/170224.html

Added Since the Last Update:
----------------------------

  * [TomT (reported by AndrewB)] <PR symtab/26270>
    Disassembler no longer showing demangled symbol names
    https://sourceware.org/bugzilla/show_bug.cgi?id=26270

    Regression introduced by Apr 24th path, which itself comes
    from the "physname" patch series...

    Patch sent for review (Jul 23rd):25631
    https://sourceware.org/pipermail/gdb-patches/2020-July/170791.html

  * [TomT] <PR rust/26197>
    rust FAILs with rustc 1.36.0 and llvm 7
    https://sourceware.org/bugzilla/show_bug.cgi?id=26197

    Patch sent for review on Jul 22nd:
    https://sourceware.org/pipermail/gdb-patches/2020-July/170764.html

Other Ongoing Items:
--------------------

  * [TomT/HannesD] <PR win32/25302>
    Mismatching fstat() function calls in gdb_bfd_open() and cache_bstat()
    https://sourceware.org/bugzilla/show_bug.cgi?id=25302

    Latest discussion at:
    https://sourceware.org/pipermail/gdb-patches/2020-June/169670.html

  * [RainerO]
    Unify Solaris procfs and largefile handling
    https://sourceware.org/pipermail/gdb-patches/2020-June/169977.html

    binutils approved that part of the patch. It leaves us with the GDB
    part (the part in config/, although used by both GCC and GDB, primarily
    benefits GDB).

  * [ChristianB (reported by Eli)]
    Update gnulib to import fixes needed for mingw.org's MinGW
    https://sourceware.org/pipermail/gdb-patches/2020-June/169987.html
    https://sourceware.org/pipermail/gdb-patches/2020-July/170059.html

    Waiting for answer to Pedro's comments sent 2020-07-16.

Not Critical, but Requested:
----------------------------

  <none>

Thank you!
--
Joel
Reply | Threaded
Open this post in threaded view
|

Re: GDB 10 branching - 2020-07-26 update

Sourceware - gdb-patches mailing list
 Am Montag, 27. Juli 2020, 01:45:58 MESZ hat Joel Brobecker <[hidden email]> Folgendes geschrieben:

> Here is a quick recap of where we are with respect to the GDB 10
> release. We did well this week, with getting 5 entries out of the way,
> and with only 2 being added (knowing that patches for those have
> already been sent).
>
> I'm concerned about two entries:
>
>   * [TomT/HannesD] <PR win32/25302>
>     Mismatching fstat() function calls in gdb_bfd_open() and cache_bstat()
>     https://sourceware.org/bugzilla/show_bug.cgi?id=25302
>
>         I'm wondering if we should be setting up a live discussion
>         to first recap, and then answer Pedro's questions about
>         Tom's comments on his current patch.
>
>         This one triggers a crash, so it's hard to consider releasing
>         without it :-(.

If all else fails, you might want to consider the workaround I use, and
which I described in my initial bug report:

--- a/gdb/gdb_bfd.c
+++ b/gdb/gdb_bfd.c
@@ -384,6 +384,11 @@ gdb_bfd_iovec_fileio_fstat (struct bfd *abfd, void *stream,
 
 /* See gdb_bfd.h.  */
 
+#pragma push_macro("fstat")
+#pragma push_macro("stat")
+#undef fstat
+#undef stat
+
 gdb_bfd_ref_ptr
 gdb_bfd_open (const char *name, const char *target, int fd,
           bool warn_if_slow)
@@ -480,6 +485,9 @@ gdb_bfd_open (const char *name, const char *target, int fd,
   return gdb_bfd_ref_ptr::new_reference (abfd);
 }
 
+#pragma pop_macro("fstat")
+#pragma pop_macro("stat")
+
 /* A helper function that releases any section data attached to the
    BFD.  */
 

This solves at least the crash problem.


Hannes