[U-Boot] [PATCH v3 16/20] mtd: spi: Add lightweight SPI flash stack for SPL

[Jagan Teki]

On Fri, Feb 1, 2019 at 10:35 PM Vignesh R <vigneshr at ti.com> wrote:
>
>
>
> On 01-Feb-19 9:18 PM, Jagan Teki wrote:
> > On Thu, Jan 31, 2019 at 11:20 PM Vignesh R <vigneshr at ti.com> wrote:
> >>
> [...]
> >>>
> >>> This doesn't look good to me, this change is part of 08/20 and now it
> >>> removed. better do the same change in 08/20 by adding new file
> >>> spi-nor-ids.c
> >>>
> >>
> >> This is intentional. Patch 8-11 clearly shows what all is being sync'd
> >> from Kernel and I would like to keep that as is.
> >> Merging U-Boot specific changes with those patches does not provide a
> >> clean history
> >> spi-nor-ids table is moved out of spi-nor-core.c in this patch because
> >> we need it for two separate compilation units (spi-nor-tiny and
> >> spi-nor-core).
> >
> > Understand this point, but since it's not a direct commit sync and we
> > even add changes wrt u-boot and remove unneeded changes related to
> > Linux.
>
> This is a direct sync. I removed code not related to U-Boot on your
> insistence. But, code organization is same as Linux.
> I had no intention of splitting ID table into a separate file (see RFC
> v1 where it was still in the original file) as there was no other user
> of ID table and keeping table in same file allows compiler to carry out
> compile time optimizations.
>
> It's fine to create -ids.c file in the same commit, otherwise
> > it is simply adding code and removing the same in following commit
> > doesn't suit for bisectable.
> >
>
> Sorry, I do not agree on this.. This patch clearly captures the _need_
> to move ID table out of spi-nor-core.c. Code is being moved out of
> spi-nor-core.c to _support_ tiny stack. At no point in this series
> bisectablility is broken.
>
> Redoing patch 8/20 is non trivial and would be painful rework with no
> gain. I will have to rework patch 8/20 and 14/20 so as to not break
> compilation.

OK.