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

[Quentin Monnet <qmo@kernel.org>]

2025-02-07 22:27 UTC+0800 ~ Jiayuan Chen <mrpre@163.com>
> 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.


I rememeber now that -Wformat is already included in bpftool's Makefile
via tools/scripts/Makefile.include (variable $(EXTRA_WARNINGS)), so it
wouldn't make a difference anyway whether you add it again in bpftool's
Makefile or not.