Re: [PATCH bpf-next v1 0/1] Using the right format specifiers for bpftool

[Jiayuan Chen]

On Fri, Feb 07, 2025 at 01:22:19PM +0000, Quentin Monnet wrote:
> On 07/02/2025 12:37, Jiayuan Chen wrote:
> > Fixed some incorrect formatting specifiers that were exposed when I added
> > the "-Wformat" flag to the compiler options.
> > 
> > This patch doesn't include "-Wformat" in the Makefile for now, as I've
> > only addressed some obvious semantic issues with the compiler warnings.
> > There are still other warnings that need to be tackled.
> > 
> > For example, there's an ifindex that's sometimes defined as a signed type
> > and sometimes as an unsigned type, which makes formatting a real pain
> > - sometimes it needs %d and sometimes %u. This might require a more
> > fundamental fix from the variable definition side.
> > 
> > If the maintainer is okay with adding "-Wformat" to the
> > tools/bpf/bpftool/Makefile, please let us know, and we can follow up with
> > further fixes.
> 
> No objection from the maintainer, thanks for looking into this. Did you
> catch these issues with just "-Wformat"? I'm asking because I need to
> use an additional flag, "-Wformat-signedness", to have my compiler
> display the %d/%u reports.
> 
> Thanks,
> Quentin
Yes, I previously added '-Wformat -Wformat-signedness', but I just tried
again and it turns out that only '-Wformat-signedness' takes effect.