On 08/30/2016 01:43 PM, Dan Williams wrote:
I tried running this, but it seems I'm failing to configure my test
environment correctly [1], but I'm worried that this "re-parenting the
scsi-disk" approach, even if the above warning is addressed, may not
be backwards compatible. We now have an ordering difference where the
link to the "block" attribute group is established after the disk's
KOBJ_ADD event which seems in the same class of problems that Fam
Zheng's patchset [2] is trying to solve.
Bart, if you can help me get the test case running I can take a look
at finishing off the disk_devt approach I proposed earlier.
[1]: https://gist.github.com/djbw/1c72526c90d1ea8fe2a05dcfbfc73dda
[2]: https://lkml.org/lkml/2016/8/17/63
Hello Dan,
Today running the srp-test software is only possible on a system
equipped with at least one InfiniBand port. Since the soft-RoCE driver
has been accepted in kernel v4.8 (drivers/infiniband/sw/rxe) it should
be possible to modify the srp-test software such that it runs on any
system equipped with at least one Ethernet port. However, that will
require adding RoCE (RDMA over Ethernet) support to the ib_srp and
ib_srpt drivers.
Bart.
--
To unsubscribe from this list: send the line "unsubscribe linux-scsi" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
