On Fri, Oct 02, 2026 at 04:44:48AM +0800, Qing Li wrote: > U-Boot adds a ramoops reserved-memory node to the OS device tree when > CONFIG_CMD_PSTORE is enabled. Some kernels register the same region > themselves, for example Qualcomm downstream kernels through the > ramoops_memreserve=4M command line parameter; a second ramoops device > in the device tree then makes the kernel hang while populating > platform devices. > > Add CONFIG_CMD_PSTORE_FDT, defaulting to y to keep the existing > behaviour, and return early from fdt_fixup_pstore() when it is > disabled. The pstore command remains available to display or save > records.
This is not a NAK. But, I don't like doing something just to work-around downstream (in other words, not to be upstreamed) modifications in the kernel. Is there a really compelling reason to not just fix the downstream tree here instead? -- Tom
signature.asc
Description: PGP signature
