Re: [PATCH v2 2/3] mm, page_owner: decouple freeing stack trace from debug_pagealloc

[Vlastimil Babka]

On 9/30/19 2:49 PM, Qian Cai wrote:
>> --- a/Documentation/admin-guide/kernel-parameters.txt
>> +++ b/Documentation/admin-guide/kernel-parameters.txt
>> @@ -3237,6 +3237,14 @@
>>  			we can turn it on.
>>  			on: enable the feature
>>  
>> +	page_owner_free=
>> +			[KNL] When enabled together with page_owner, store also
>> +			the stack of who frees a page, for error page dump
>> +			purposes. This is also implicitly enabled by
>> +			debug_pagealloc=on or KASAN, so only page_owner=on is
>> +			sufficient in those cases.
>> +			on: enable the feature
>> +
> 
> If users are willing to set page_owner=on, what prevent them from enabling KASAN
> as well? That way, we don't need this additional parameter.

Well, my use case is shipping production kernels with CONFIG_PAGE_OWNER
and CONFIG_DEBUG_PAGEALLOC enabled, and instructing users to boot-time
enable only for troubleshooting a crash or memory leak, without a need
to install a debug kernel. Things like static keys and page_ext
allocations makes this possible without CPU and memory overhead when not
boot-time enabled. I don't know too much about KASAN internals, but I
assume it's not possible to use it that way on production kernels yet?

> I read that KASAN
> supposes to be semi-production use ready, so the overhead is relatively low.
> There is even a choice to have KASAN_SW_TAGS on arm64 to work better with small
> devices.