Hi Tom,
> On Fri, Oct 16, 2026 at 10:43:41AM +0900, Takahiro Kuwano wrote:
>
> > The spi-nor-core.c contains optional logics like SFDP and manufcaturer
> > specific fixups. The spi-nor-ids.c has single large ID database. There
> > are many #ifdef directives in those files and that negatively affects
> > readability and maintainability.
> >
> > This series, as the first step of cleanup, move SFDP logic, IDs, and
> > fixups to newly source files. Reworking, renaming, and reformating
> > existing code is not a scope of this series and further cleanup will be
> > taken place in follow-up patches.
> >
> > Here is the code size impact when applying this series.
> > Measured by my AMD Zynq-7000 (ARM32) environment with SFDP and all
> > manufcaturers are enabled.
> >
> > SPI-NOR:
> > text data bss dec hex filename
> > BEFORE 720685 14084 0 734769 b3631 u-boot
> > AFTER 721827 14156 0 735983 b3aef u-boot
> > DELTA +1142 +72
> >
> > TINY:
> > text data bss dec hex filename
> > BEFORE 77399 3128 248 80775 13b87 spl/u-boot-spl
> > AFTER 77589 3200 248 81037 13c8d spl/u-boot-spl
> > DELTA +190 +72
>
> So this leads to imx28_btt3 and imx28_xea overflowing in CI on link.
I don't see any errors in my local. Something different from CI config.
imx28_btt3_defconfig:
text data bss dec hex filename
BEFORE 37947 2896 0 40843 9f8b spl/u-boot-spl
AFTER 38003 2912 0 40915 9fd3 spl/u-boot-spl
+56 +16
imx28_xea_defconfig:
text data bss dec hex filename
BEFORE 38004 2816 0 40820 9f74 spl/u-boot-spl
AFTER 38060 2832 0 40892 9fbc spl/u-boot-spl
+56 +16
Could you share the exact CI error or size report?
> Maybe it needs the tiny flag set? I think the rest of the growth is
> reasonable enough, and boards can disable part vendors they don't need.
SPI_FLASH_TINY is set by default in SPL build and I confirmed
spi-nor-tiny.o in both builds.
>
> --
> Tom
Thanks,
Takahiro
