On Mon, Feb 04, 2019 at 04:21:37PM +0100, Simon Goldschmidt wrote:
Am Mo., 4. Feb. 2019, 16:16 hat Abel Vesa abel.vesa@nxp.com geschrieben:
On 19-02-04 08:19:23, Tom Rini wrote:
On Mon, Feb 04, 2019 at 09:03:43AM -0200, Fabio Estevam wrote:
On Mon, Feb 4, 2019 at 7:55 AM Abel Vesa abel.vesa@nxp.com wrote:
If the SPL size (without the dtb appended) is larger then yes, the
build fails.
Trouble is if the SPL (without the dtb appended) is, lets say, 63kB
and
then the dtb is larger than 1kB. Then there is no mechanism in place
to check that
and it will just fail to boot without giving any clues why. But this
is a totally
unrelated problem from this patchset's point of view and I think it
impacts all
the platforms that support SPL with DM.
Yes, it is unrelated to this series and should be treated separately.
I also agree. But, don't we have a mechanism for that? It sounds like some additional targets need to call $(BOARD_SIZE_CHECK) at the end. Or am I missing something?
Hmm, I believe that is true. I haven't looked deeper into it but it seems that's the thing the SPL is missing.
Haven't we talked about that just some days or weeks ago? That's exactly what I am missing for socfpga. I keep getting broken SPL images when adding functionality. And you don't even get an error message because the dtb is broken and cannot provide the serial output mode...
Yes, I believe it is. I guess what I'm waiting / hoping for is someone to patch the top-level Makefile to call $(BOARD_SIZE_CHECK) on a number of make targets that look reasonable, throw it at travis and see what if anything fails (and if stuff fails, ask for help to see if it's right or investigate a bit more). More or less, if it's a final target for U-Boot, we should probably give BOARD_SIZE_CHECK a try. Most of the rules start with u-boot, but I bet the %.imx target should also do it, and then we can drop all of that from arch/arm/mach-imx/Makefile.