Hi Michal, On 29 July 2016 at 05:29, Michal Simek <mon...@monstr.eu> wrote: > Hi Simon, > > > 2016-07-17 22:59 GMT+02:00 Simon Glass <s...@chromium.org>: >> >> On 5 July 2016 at 17:10, Simon Glass <s...@chromium.org> wrote: >> > This new function is more convenient for callers, and handles >> > pre-relocation >> > situations automatically. >> > >> > Signed-off-by: Simon Glass <s...@chromium.org> >> > --- >> > >> > Changes in v2: >> > - Add new patch to make use of dm_scan_fdt_dev() >> > >> > arch/x86/lib/lpc-uclass.c | 4 +--- >> > common/usb_hub.c | 3 +-- >> > drivers/core/simple-bus.c | 3 +-- >> > drivers/i2c/i2c-uclass.c | 3 +-- >> > drivers/i2c/sandbox_i2c.c | 4 +--- >> > drivers/misc/cros_ec.c | 3 +-- >> > drivers/pch/pch-uclass.c | 4 +--- >> > drivers/pci/pci-uclass.c | 11 +---------- >> > drivers/pci/pci_sandbox.c | 3 +-- >> > drivers/pinctrl/pinctrl_pic32.c | 3 +-- >> > drivers/pinctrl/rockchip/pinctrl_rk3036.c | 3 +-- >> > drivers/pinctrl/rockchip/pinctrl_rk3288.c | 3 +-- >> > drivers/power/pmic/pm8916.c | 3 +-- >> > drivers/power/regulator/Kconfig | 2 +- >> > drivers/spi/spi-uclass.c | 3 +-- >> > drivers/spmi/spmi-uclass.c | 3 +-- >> > drivers/usb/emul/usb-emul-uclass.c | 3 +-- >> > drivers/usb/host/usb-uclass.c | 4 +--- >> > include/power/regulator.h | 2 +- >> > test/dm/bus.c | 3 +-- >> > 20 files changed, 20 insertions(+), 50 deletions(-) >> >> Applied to u-boot-dm. > > > This is breaking zynqmp support - to be specific serial driver which gets > information about clock > from DT. > On zcu102 with early debug this is an output. > failed to get clock > failed to get clock > failed to get clock > No serial driver found > resetting ... > > I looked at it and the reason for that is that there is different handling > for pre_reloc_only. > dm_scan_fdt_node(dev, gd->fdt_blob, dev->of_offset, > gd->flags & GD_FLG_RELOC ? false : true); > > Called from simple-bus.c. > The reason for it is that my clk node has no u-boot,dm-pre-reloc; property. > I expect that this is the right fix to add u-boot,dm-pre-reloc; property. > > I will send a patch to fix this but I wanted to let you know.
Sorry about the breakage, that's one reason I wanted to get it in early. But yes, the fix is to add that property. The old behaviour was wrong and chewed up a lot of pre-reloc memory on one platform. Regards, Simon _______________________________________________ U-Boot mailing list U-Boot@lists.denx.de http://lists.denx.de/mailman/listinfo/u-boot
