On Thu, 7 Jun 2018 13:27:59 +0200
Christoph Hellwig [off-list ref] wrote:

On Thu, Jun 07, 2018@10:38:47AM +0200, Hannes Reinecke wrote:

quoted

Adding a tracepoint to trace bio remapping for native nvme
multipath.  

I guess we should just move the trace point into direct_make_request?

Otherwise this looks sane to me.

That would run afoul with device-mapper usage:

	case DM_MAPIO_REMAPPED:
		/* the bio has been remapped so dispatch it */
		trace_block_bio_remap(clone->bi_disk->queue, clone,
				      bio_dev(io->orig_bio), sector);
		if (md->type == DM_TYPE_NVME_BIO_BASED)
			ret = direct_make_request(clone);
		else
			ret = generic_make_request(clone);
		break;

and we can't move the call into generic_make_request().
So I fear it need to stay there.

Cheers,

Hannes