> This bug connects with kernel commit 7bc1a0f9e176 ("arm64: mm: use
> single quantity to represent the PA to VA translation"), memstart_addr
> can be negative, which makes it different from real phys_offset. If
> using memstart_addr to calculate the real paddr, the unreasonable paddr
> will be got.

> Furthermore, in crash utility, PTOV() needs memstart_addr to calculate
> VA from PA, while getting PFN offset in a dumpfile, phys_offset is
> required.

As you mentioned above, the calculation formula has been changed, how to

deal with the backward compatibility issue? Should we use kernel version

to determine which code branch it should execute? Please correct me if I

was wrong.

Thanks.

Lianbo

 
> To serve the different purpose, using phys_offset_nominal and
> phys_offset to store them.