https://bugzilla.kernel.org/show_bug.cgi?id=196543 --- Comment #1 from James.Bottomley@xxxxxxxxxxxxxxxxxxxxx --- On Mon, 2017-07-31 at 02:26 +0000, bugzilla-daemon@xxxxxxxxxxxxxxxxxxx wrote: > https://bugzilla.kernel.org/show_bug.cgi?id=196543 > > Bug ID: 196543 > Summary: Adaptec 6405H can not understand Queue full > message > from Disks. > Product: IO/Storage > Version: 2.5 > Kernel Version: 4.11.11(Fedora26 Kernel)/3.10.0(RHEL 7.3) > Hardware: x86-64 > OS: Linux > Tree: Mainline > Status: NEW > Severity: normal > Priority: P1 > Component: SCSI > Assignee: linux-scsi@xxxxxxxxxxxxxxx > Reporter: kiyomi.kakitsubata@xxxxxxxxx > Regression: No > > Hi > > I happened something different caused by SAS HBA. > Adaptec 6405H can not understand Queue full message send by Disks. > > Perhaps, normally, any other HBA receive Queue full message from SAS > disks, it will wait send some command to disks. But Adaptec 6405H is > not wait send command to disks, so, happen disk time out. > > I think driver or firmware's bug,because other HBA like using > mpt3sas, is not happened. > > If you need,perhaps I will give SAS bus trace. A bus trace might help. The problem with the analysis above is that QUEUE_FULL isn't handled in the driver or the transport, it's handled at the scsi mid-layer level since it's a SCSI status return. There doesn't seem to be anywhere in the 94xx driver where it intercepts queue full, so for what you theorise to be the problem, the intercept would either have to be in firmware, or there would have to be some generic problem in the mid-layer. We might be able to track it down, but we'd need more data. James -- You are receiving this mail because: You are the assignee for the bug.
