Re: [PATCH 0/3] argparse: improve handling of multi-instance args

[Bruce Richardson <bruce.richardson@intel.com>]

On Mon, Jun 16, 2025 at 11:49:40AM +0100, Bruce Richardson wrote:
> Coverity (correctly) identified an issue[1] where, after the recent
> rework[2], the internal flag, used by argparse to track what arguments
> were previously encountered or not, was out of range for the type and no
> longer having any effect. Fixing this flag to be back into range then,
> somewhat surprisingly, caused a number of unit test failures to occur.
> 
> The reason for these failures is that the tracking of args encountered
> is done via setting an internal flag on the user-passed arguments
> object. In the unit tests, this caused issues where the flags field was
> not getting properly reset between calls to the parse operation. [This
> is only an issue after the rework, because previously information like
> param type and optionality was encoded in the flags, so they were more
> often reset during testing].
> 
> Rather than fixing the tests directly to always reset the flags, which
> is simply working around the issue IMHO, this patchset instead fixes the
> issue in a more user-friendly way by changing the library to never
> modify the user-passed structure - making it completely safe to reuse
> across multiple calls. This is done in the first two patches.
> 
> The final, third patch, adds an additional unit test to check that the
> tracking of flags being seen or not, and the handling of the
> "RTE_ARGPARSE_FLAG_SUPPORT_MULTI" flag is correct. This closes a gap in
> testing, since the original issue of the flag being out-of-range should
> have been caught in testing, rather than having to rely on coverity.
> 
> [1] Coverity Issue: 470190
> [2] https://github.com/DPDK/dpdk/commit/04acc21beeeb78477b15a3f497d3628fd70a6a9f
> 
> Bruce Richardson (3):
>   argparse: track parsed arguments internally
>   argparse: mark parameter struct as const
>   test/argparse: add test for repeated arguments
> 
Hi Chengwen,

ping for review. I think this bug should be fixed for RC2.

Regards,
/Bruce