Re: [PATCH v2] nvmem: u-boot-env: align endianness of crc32 values

[INAGAKI Hiroshi]

Hi Rafał,

thank you for letting me know.

And for all, sorry for my incomplete patch.

On 2023/01/09 18:52, Rafał Miłecki wrote:
> On 2023-01-09 10:48, Srinivas Kandagatla wrote:
> > On 06/01/2023 19:15, Rafał Miłecki wrote:
> > > On 2022-11-11 18:41, Srinivas Kandagatla wrote:
> > > > On 12/10/2022 16:51, INAGAKI Hiroshi wrote:
> > > > > This patch fixes crc32 error on Big-Endianness system by 
> > > > > conversion of
> > > > > calculated crc32 value.
> > > > >
> > > > > Little-Endianness system:
> > > > >
> > > > >    obtained crc32: Little
> > > > > calculated crc32: Little
> > > > >
> > > > > Big-Endianness system:
> > > > >
> > > > >    obtained crc32: Little
> > > > > calculated crc32: Big
> > > > >
> > > > > log (APRESIA ApresiaLightGS120GT-SS, RTL8382M, Big-Endianness):
> > > > >
> > > > > [    8.570000] u_boot_env 
> > > > > 18001200.spi:flash@0:partitions:partition@c0000: Invalid 
> > > > > calculated CRC32: 0x88cd6f09 (expected: 0x096fcd88)
> > > > > [    8.580000] u_boot_env: probe of 
> > > > > 18001200.spi:flash@0:partitions:partition@c0000 failed with error -22
> > > > >
> > > > > Fixes: f955dc1445069 ("nvmem: add driver handling U-Boot 
> > > > > environment variables")
> > > > >
> > > > > Signed-off-by: INAGAKI Hiroshi <musashino.open@xxxxxxxxx>
> > > > > ---
> > > >
> > > > Applied thanks,
> > >
> > > has this patch been lost somewhere in the process?
> > >
> > > I'm quite sure I saw it in linux-next.git and probably in nvmem.git. 
> > > Now
> > > it seems to be gone.
> > Yes, I had to revert this one as next reported sparse warnings [1]
> > with this patch which were not addressed.
>
> I missed that, thank you.
>
> INAGAKI: could you take a look at this, please?

To be honest, I'm a newbie in Linux & driver development and don't fully 
understand the essence of the warning by "kernel test robot", sorry.
As far as I have seen some similar cases in the past, does it mean that 
the type of calculated crc variable("calc") should be changed to __le32? 
(Maybe I'm saying something off the mark...)

what I assumed:

__le32 calc;
calc = crc32(~0, buf + crc32_data_offset, crc32_data_len) ^ ~0L;
calc = le32_to_cpu(calc);