Re: [PATCH v4 2/4] RISCV: Fix incorrect virtual address translation in? crashdump load

[luo.haiyang at zte.com.cn]

SimonHorman <horms at kernel.org> wrote:
>On Thu, May 21, 2026 at 12:09:46PM +0800, luo.haiyang at zte.com.cn wrote:
>> From: Luo Haiyang <luo.haiyang at zte.com.cn>
>> 
>> The kexec shows the following crashdump log:
>> 
>> Elf header: p_type = 1, p_offset = 0x80100000 p_paddr = 0x80100000 p_vaddr = 0xcff000 ...
>> Elf header: p_type = 1, p_offset = 0x100000000 p_paddr = 0x100000000 p_vaddr = 0x80bff000 ...
>> Elf header: p_type = 1, p_offset = 0x87ffff000 p_paddr = 0x87ffff000 p_vaddr = 0x800bfe000 ...
>> 
>> Obviously, phys_to_virt returns an incorrect virtual address (vaddr). On the RISC-V architecture,
>> the linear address mapping is defined as:
>> va = pa + kernel_map.va_pa_offset (kernel_map.va_pa_offset = PAGE_OFFSET - phys_ram_base)
>> Both PAGE_OFFSET and phys_ram_base can be read from /proc/kcore.
>> 
>> Signed-off-by: Luo Haiyang <luo.haiyang at zte.com.cn>
>
>Hi,
>
>Unfortunately this patch seems to be somewhat mangled
>and does not apply. Could you look into that?

Seems something got broken, I've resent the patch. Could you please see if it works?

Thanks
Luo haiyang