Hi, Jiaxun,
On 11/13/2020 10:30 AM, Jiaxun Yang wrote:
Hi all,
在 2020/11/11 22:52, Serge Semin 写道:
Hello Alexander
On Tue, Nov 10, 2020 at 11:29:50AM +0100, Alexander Sverdlin wrote: 2) The check_kernel_sections_mem() method can be removed. But it should be done carefully. We at least need to try to find all the platforms, which rely on its functionality.
It have been more than 10 years since introducing this this check, but I believe there must be a reason at the time.
Also currently we have some unmaintained platform and it's impossible to test on them for us.
For Cavium's issue, I believe removing the page rounding can help. I'd suggest keep this method but remove the rounding part, also leave a warning or kernel taint when out of boundary kernel image is detected.
Thanks.
- Jiaxun
I found the first submission. Commit: d3ff93380232 (mips: Make sure kernel memory is in iomem) As I thought, this check is related to kdump. More directly, it is related to the "mem=" parameter. Kexec-tools provide a "mem=" parameter to limit the kernel location in kdump. But sometimes the memory may be not enough and this check gives a way to ensure the capture kernel can start rightly. Although "mem=" is not in kexec-tools now, I thought it is also useful if someone use "mem=" to do other things.
Thanks, Jinyang
-Sergey
-- Best regards, Alexander Sverdlin.