Re: [PATCH v4 2/2] RISC-V: Implement sparsemem

[Logan Gunthorpe]

On 2019-08-15 3:31 a.m., Greentime Hu wrote:
> Hi Logan,
> 
> On Thu, Aug 15, 2019 at 6:21 AM Logan Gunthorpe <logang@xxxxxxxxxxxx> wrote:
>>
>> Hey,
>>
>> On 2019-08-14 7:35 a.m., Greentime Hu wrote:
>>> How about this fix? Not sure if it is good for everyone.
>>
>> I applied your fix to the patch and it seems ok. But it doesn't seem to
>> work on a recent version of the kernel. Have you got it working on v5.3?
>> It seems the following patch breaks things:
>>
>> 671f9a3e2e24 ("RISC-V: Setup initial page tables in two stages")
>>
>> I don't really have time right now to debug this any further.
>>
> 
> I just tried v5.3-rc4 and it failed. I try to debug this case.
> I found it failed might be because of an unmapped virtual address is used
> in memblocks_present() -> memblock_alloc ().
> 
> In this commit 671f9a3e2e24 ("RISC-V: Setup initial page tables in two
> stages"), it finishes all the VA/PA mapping after setup_vm_final() is
> called.
> So we have to call memblocks_present() and sparse_init() right after
> setup_vm_final().
> 
> Here is my patch and I tested with memory-with-hole.
> It can boot normally in Unleashed board after applying this patch.

Great, thanks! I'll roll this into my patch and send v5 out when I have
a moment. Can I add your Signed-off-by for the bits you've contributed
to give you credit for your work?

Logan