On 14. 02. 19 13:37, Alexander Graf wrote:
On 14.02.19 13:33, Michal Simek wrote:
On 14. 02. 19 13:27, Alexander Graf wrote:
On 14.02.19 13:18, Michal Simek wrote:
distro boot expects that fdtfile name is setup for alternative DTB. Create this file based on the first platform compatible string. This should ensure that one rootfs can store multiple DTBs for different boards.
Signed-off-by: Michal Simek michal.simek@xilinx.com
How good/bad idea is this?
board/xilinx/zynqmp/zynqmp.c | 30 ++++++++++++++++++++++++++++++ 1 file changed, 30 insertions(+)
diff --git a/board/xilinx/zynqmp/zynqmp.c b/board/xilinx/zynqmp/zynqmp.c index a49f2f3e9ebe..64800d869819 100644 --- a/board/xilinx/zynqmp/zynqmp.c +++ b/board/xilinx/zynqmp/zynqmp.c @@ -514,6 +514,32 @@ static u32 reset_reason(void) return ret; }
+static int set_fdtfile(void) +{
- char *compatible, *fdtfile;
- if (env_get("fdtfile"))
return 0;- compatible = (char *)fdt_getprop(gd->fdt_blob, 0, "compatible", NULL);
- if (compatible) {
debug("Compatible: %s\n", compatible);strsep(&compatible, ",");fdtfile = calloc(1, strlen(compatible) + 4);if (!fdtfile)return -ENOMEM;sprintf(fdtfile, "%s%s", compatible, ".dtb");AArch64 dtbs are usually stored in a directory structure which you want to match again. Is this the case for the compatible string? From a quick glimpse, it looks like it - but please make sure that something is in place to ensure it stays that way (upstream review for example).
I have actually remove xlnx prefix from compatible string but just doing simple conversion xlnx,zynqmp-zcu102-revA to xlnx/zynqmp-zcu102-revA.dtb looks good to me too.
That won't work, as the kernel will install the dtb in a directory called "xilinx".
Vendor prefixes should be stable enough to ensures that.
So the only change I would make here is that this should be "xilinx/%s.dtb".
Any issue with xlnx prefix?
Yup :). Please stick to exactly where the kernel installs the dtbs to.
Ah ok. I see. Didn't realized that xilinx is coming from arch/arm64/boot/dts structures.
Thanks, Michal