Re: [PATCH v5 0/2] PR gdb/33747: gdb/ser-unix: fix musl build failure and add POSIX custom baud rate support

[Kevin Buettner <kevinb@redhat.com>]

On Tue, 24 Mar 2026 09:45:25 -0700
sunilkumar.dora@windriver.com wrote:

> From: Sunil Dora <sunilkumar.dora@windriver.com>
> 
> On musl-based systems <asm/termbits.h> exposes BOTHER, so the
> Linux-specific custom baud rate path was being compiled.  However,
> musl's struct termios does not provide the public members c_ispeed
> and c_ospeed, causing a build failure.
> 
> This series fixes the build issue at the macro level (patch 1).
> It then adds the preferred POSIX cfsetispeed/cfsetospeed path as the
> first choice on platforms where the host libc supports arbitrary baud
> rates, such as glibc 2.42 and later, GNU Hurd, and potentially other
> libc implementations (patch 2).  The existing Linux (termios2/BOTHER)
> and Darwin (IOSSIOSPEED) paths remain as fallbacks.
> 
> The two patches are independent:
>   - Patch 1 is a pure build fix with no functional change.
>   - Patch 2 adds the new POSIX feature on top of the fixed guard.
> 
> set_custom_baudrate_linux is left unchanged.
> 
> Changes since V4:
>   - Split into two patches, with patch 1 as a pure build fix
>     and patch 2 adding the POSIX interface.
>   - Fixed #if guard formatting and added full stop to the
>     AC_DEFINE description.
>   - Removed parentheses from function name references in
>     commit messages and comments.
>   - Updated NEWS entry to describe which configurations are
>     affected.
> 
> v4: https://sourceware.org/pipermail/gdb-patches/2026-March/226133.html
> v3: https://sourceware.org/pipermail/gdb-patches/2026-March/225952.html
> v2: https://sourceware.org/pipermail/gdb-patches/2026-February/225251.html
> v1: https://sourceware.org/pipermail/gdb-patches/2026-February/224968.html
> 
> Bug: https://sourceware.org/bugzilla/show_bug.cgi?id=33747

Both parts are approved.  Thanks for persevering through the many
requests for changes.

One thing I noticed - Some error messages in ser-unix.c still use
"Can not" whereas you uniformly (and correctly IMO) use "Cannot".
If you're up for doing a follow-on patch, one which does
s/Can not/Cannot/ on the remaining occurrences would be appreciated.

Approved-by: Kevin Buettner <kevinb@redhat.com>