On 2020-02-07 01:09, Martin Wilck wrote:
> Sorry for insisting - in your original submission "Revert "qla2xxx: Fix
> Nport ID display value" from 11/09/2019, you'd identified this as a
> regression caused by 0aabb6b699f7 ("scsi: qla2xxx: Fix Nport ID display
> value"). Are you saying that that wasn't the whole truth? Had N2N been
> broken before already? Did N2N with "old" adapters ever work with
> previous versions of the driver? Which ones?
Hi Martin,
My conclusion is indeed that N2N is broken since a long time for at
least the 8 Gb/s QLogic FC adapter. The test I run is to set the
qlini_mode kernel module parameter to dual, to connect two ports of a
dual port adapter to each other and to make both ports log in to each
other. If I run this test against a 32 Gb/s dual port adapter then I
always see that both ports log in to each other. If I run this test
against an 8 Gb/s dual port adapter then I see that most of the time
only one port logs in to another port. Sometimes no login happens at all.
Bart.
