Re: [PATCH v9 0/2] lib: checksum: Fix issues with checksum tests

[Charlie Jenkins <charlie@rivosinc.com>]

On Fri, Feb 23, 2024 at 09:22:58AM +0000, Christophe Leroy wrote:
> 
> 
> Le 23/02/2024 à 08:22, Michael Ellerman a écrit :
> > Palmer Dabbelt <palmer@dabbelt.com> writes:
> >> On Wed, 21 Feb 2024 18:55:48 PST (-0800), Charlie Jenkins wrote:
> >>> The ip_fast_csum and csum_ipv6_magic tests did not work on all
> >>> architectures due to differences in endianness and misaligned access
> >>> support. Fix those issues by changing endianness of data and aligning
> >>> the data.
> >>>
> >>> Signed-off-by: Charlie Jenkins <charlie@rivosinc.com>
> >>> ---
> >>> Changes in v9:
> >>> - Revert back to v7, the changes to v8 were not needed
> >>> - Link to v8: https://lore.kernel.org/r/20240214-fix_sparse_errors_checksum_tests-v8-0-36b60e673593@rivosinc.com
> > ...
> >>>
> >>> ---
> >>> Charlie Jenkins (2):
> >>>        lib: checksum: Fix type casting in checksum kunits
> >>>        lib: checksum: Use aligned accesses for ip_fast_csum and csum_ipv6_magic tests
> >>>
> >>>   lib/checksum_kunit.c | 396 ++++++++++++++++++---------------------------------
> >>>   1 file changed, 136 insertions(+), 260 deletions(-)
> >>> ---
> >>> base-commit: 6613476e225e090cc9aad49be7fa504e290dd33d
> >>> change-id: 20240119-fix_sparse_errors_checksum_tests-26b86b34d784
> >>
> >> I put a
> >>
> >> Acked-by: Palmer Dabbelt <palmer@rivosinc.com>
> >>
> >> on the v4, but looks like it got lost.  I'm happy to take this via the
> >> RISC-V tree, as that's how I merged the broken patches in the first
> >> place, but no big deal if someone else wants to pick it up.
> >>
> >> It looks like the issues are all resolved and such, but there's been a
> >> long tail of them so I'm not 100% sure here...
> > 
> > I tested v9 on ppc32/64 BE, and it fixes the test failures and the
> > sparse errors, so LGTM.
> > 
> > Tested-by: Michael Ellerman <mpe@ellerman.id.au> (powerpc)
> 
> It magically works, but patch 1 is wrong and patch 2 subject is 
> unrelated to the problem on powerpc.
> 
> This fix series needs rework.
> 
> Christophe

I will rearrange the patches so that patch 1 fixes the endianness and
patch 2 addresses alignment. The second patch is subject to endianness
and alignment and the title only including alignment (but the
body including both) is the source of confusion.

- Charlie