This is the mail archive of the
gdb-patches@sourceware.org
mailing list for the GDB project.
Re: [PATCH] Fix implicit conversion warning produced by clang
- From: John Baldwin <jhb at freebsd dot org>
- To: gdb-patches at sourceware dot org
- Cc: Taras Tsugrii <ttsugrii at fb dot com>
- Date: Tue, 14 Jun 2016 12:07:51 -0700
- Subject: Re: [PATCH] Fix implicit conversion warning produced by clang
- Authentication-results: sourceware.org; auth=none
- References: <1465926942-37759-1-git-send-email-ttsugrii at fb dot com>
On Tuesday, June 14, 2016 10:55:42 AM Taras Tsugrii wrote:
> When compiling with clang, which is part of Xcode 7.3, following warnings
> are emitted:
> objcopy.c:224:50: error: implicit conversion from enumeration type 'enum bfd_link_elf_stt_common' to different enumeration type 'enum bfd_link_discard' [-Werror,-Wenum-conversion]
> static enum bfd_link_discard do_elf_stt_common = unchanged;
> ~~~~~~~~~~~~~~~~~ ^~~~~~~~~
> objcopy.c:4290:26: error: implicit conversion from enumeration type 'enum bfd_link_elf_stt_common' to different enumeration type 'enum bfd_link_discard' [-Werror,-Wenum-conversion]
> do_elf_stt_common = elf_stt_common;
> ~ ^~~~~~~~~~~~~~
> objcopy.c:4292:26: error: implicit conversion from enumeration type 'enum bfd_link_elf_stt_common' to different enumeration type 'enum bfd_link_discard' [-Werror,-Wenum-conversion]
> do_elf_stt_common = no_elf_stt_common;
> ~ ^~~~~~~~~~~~~~~~~
> 3 errors generated.
>
> This is risky, since essentially there is a coupling between 2 different
> enum orderings. Also since by default warnings are promoted to errors
> it breaks default build.
I had the same patch in queue and just pushed it. I have a few more patches
from clang warnings pending as well. (I am testing with clang 3.8.0 in
FreeBSD.) Some of clang's warnings are a bit noisy though and I haven't tried
to address the -Wunused-function spam from VEC() (I presume that GDB will
eventually switch the VEC stuff to templates which will fix that). Also,
Some of the -Wtautological-compare warnings for things that do:
if (foo >= 0 && foo <= bar)
where 'foo' is unsigned I haven't posted patches to "fix" as I'm not sure that
the existing code isn't more readable in its current form.
--
John Baldwin