Am Montag, den 11.02.2019, 20:21 +0100 schrieb Sascha Hauer: > Hi Teresa, > > On Mon, Feb 11, 2019 at 03:44:28PM +0100, Teresa Remmet wrote: > > Hello, > > > > AM335x is broken on current master in multiple ways. > > NAND and and SPI boot are not working right now as the image size > > can > > not be found: > > > > barebox 2019.01.0-00305-g96abbe544e19 #1682 Mon Feb 11 15:20:00 CET > > 2019 > > > > > > Board: Phytec phyCORE AM335x > > nand: ONFI flash detected > > nand: NAND device: Manufacturer ID: 0x2c, Chip ID: 0xdc (Micron > > MT29F4G08ABADAH4), 512Mi > > B, page size: 2048, OOB size: 64 > > booting from NAND > > failed to get image size > > trying to load image from backup partition. > > failed to get image size > > booting failed > > > > I bisected the issue down to: > > commit 5c0e16591d9471a345b77a41fde76de34f301f6b > > Author: Sascha Hauer <s.hauer@xxxxxxxxxxxxxx> > > Date: Wed Jan 16 10:15:55 2019 +0100 > > > > images: Drop unnecessary fix_size > > > > Now that we compile the compressed binary into the decompressor > > we > > no longer need fix_size but can use the linker to fill in the > > image > > size into the binary. > > > > Signed-off-by: Sascha Hauer <s.hauer@xxxxxxxxxxxxxx> > > > > Reverting this patch helps. But what is the correct solution? > > Update to a newer toolchain ;) This helped :) Thanks for the hint. Teresa > > This problem seems to exist with gcc-5 and older where the > barebox_image_size linker variable is generated as something which > needs > runtime relocation. > > We can just revert 5c0e16591d94. This brings us the problem that > fix-size will be used by the Mips multiimage builds. It fails there > as fix-size tests for the input being a valid Arm barebox image. We > would have to relax that constraint. > > Sascha > _______________________________________________ barebox mailing list barebox@xxxxxxxxxxxxxxxxxxx http://lists.infradead.org/mailman/listinfo/barebox
