On 01/09/2013 03:38:22 PM, Tom Rini wrote:
On Wed, Jan 09, 2013 at 01:53:21PM -0600, Scott Wood wrote:
On 01/08/2013 04:57:20 PM, Albert ARIBAUD wrote:
Signed-off-by: Albert ARIBAUD albert.u.boot@aribaud.net
drivers/mtd/nand/Makefile | 4 ++++ 1 file changed, 4 insertions(+)
diff --git a/drivers/mtd/nand/Makefile b/drivers/mtd/nand/Makefile index 2c3812c..c77c0c4 100644 --- a/drivers/mtd/nand/Makefile +++ b/drivers/mtd/nand/Makefile @@ -79,6 +79,10 @@ COBJS-$(CONFIG_TEGRA_NAND) += tegra_nand.o COBJS-$(CONFIG_NAND_OMAP_GPMC) += omap_gpmc.o COBJS-$(CONFIG_NAND_PLAT) += nand_plat.o
+else # minimal SPL drivers
+COBJS-$(CONFIG_NAND_FSL_ELBC) += fsl_elbc_spl.o
endif # drivers endif # nand
So, it looks like this is repairing breakage that came in through a manual merge resolution. Should such merge resolutions not be posted to the list for review? Or was it posted and I missed it?
None of the above. That powerpc was broken twice (once by this, and once by the arm head.S changes) was missed in my build testing. We don't have spelled out rules (that I'm aware of) for manual merges other than asking that someone check that X still works (in this case, am335x NAND). It did, but I didn't read the merge myself was the problem.
The NAND Makefile breakage came from commit 79f38777947ac7685e2cef8bd977f954ab198c0e, which is a manual merge by Albert. Why should manual merges be exempt from the rule that all changes get posted to the list? What if next time it's a functional breakage rather than a broken build?
I tried repeating the merge between 96764df and 9bd5c1a and the only conflict marker was this:
ifdef CONFIG_SPL_BUILD <<<<<<< HEAD ifdef CONFIG_SPL_NAND_SIMPLE COBJS-y += nand_spl_simple.o endif COBJS-$(CONFIG_SPL_NAND_AM33XX_BCH) += am335x_spl_bch.o ifdef CONFIG_SPL_NAND_LOAD COBJS-y += nand_spl_load.o ||||||| merged common ancestors ifdef CONFIG_SPL_NAND_SIMPLE COBJS-y += nand_spl_simple.o endif ifdef CONFIG_SPL_NAND_LOAD COBJS-y += nand_spl_load.o =======
ifdef CONFIG_SPL_NAND_DRIVERS NORMAL_DRIVERS=y
> 96764df
endif
The fsl_elbc_spl.o part was still there, so it wasn't the automatic part of the merge that removed it.
If this was simply due to a bad patch in the ARM tree, which specific patch was it?
-Scott