>>> This patch is not acceptable because it adds support for tag tracing
>>> to the legacy block layer only. Any patch that adds a new feature to
>>> the legacy block layer must also add it to blk-mq.
>>>
>> To be honest, I don't understand your point, can you give me more
>explanation?
>
>The legacy block layer will be removed as soon as blk-mq provides all the
>functionality of the legacy block layer and as soon as it performs at least as
>well as the legacy block layer for all use cases. If new features are added to
>the legacy block layer but not to blk-mq that prevents removal of the legacy
>block layer. Hence the requirement I explained in my previous e-mail.
>
>Bart.
Thanks for your information.
I have several questions again.
When the legacy block layer will be replaced by blk-mq?
And "include/trece/event/block.h .. scsi.h" will also be changed?
Do you have the related git rep or mail list about this topic?
Maybe this is great big change, I am very interested in that. And want to have a look at.
By the way, these patches are not to add new feature, they are just to add print tag along with the other exist
Printed request parameters. The blk-mq is now still using "include/trace/evet/block.h" defined trace events.
For example:
void blk_mq_start_request(struct request *rq)
{
...
...
trace_block_rq_issue(q, rq);
...
...
}
Do you mean that this will also be removed/replaced by someone else?
Bean Huo
