On Mon, 2018-06-04 at 08:28 +0000, Bart Van Assche wrote: > On Sun, 2018-06-03 at 22:09 -0700, Himanshu Madhani wrote: > > This patch prevents driver from setting lower default speed > > of 1 GB/sec, if the switch does not support Get Port Speed > > Capabilities (GPSC) command. Setting this default speed results > > into much lower write performance for large sequential WRITE. > > This patch modifies driver to check for gpsc_supported flags and > > prevents driver from issuing MBC_SET_PORT_PARAM (001Ah) to set > > default speed of 1 GB/sec. If driver does not send this mailbox > > command, firmware assumes maximum supported link speed and will > > operate at the max speed. > > > > Cc: stable@xxxxxxxxxxxxxxx > > Signed-off-by: Himanshu Madhani <himanshu.madhani@xxxxxxxxxx> > > Is this a regression? In other words, does this patch need a "Fixes:" tag? It is, it appears to have been introduced by the series which updated the driver to 10.00.00.04-k. It seems to have been introduced somewhere between: 82abdcaf3e "scsi: qla2xxx: Allow target mode to accept PRLI in dual mode" and f352eeb754 "scsi: qla2xxx: Add ability to use GPNFT/GNNFT for RSCN handling" The regression is definitely present in: 6a2cf8d366 "scsi: qla2xxx: Fix crashes in qla2x00_probe_one on probe failure" Unfortunately I was not able to bisect further due to crashes on boot and a problem with LUN discovery that was not fixed until a later patch that does not apply easily without the many intervening patches. -Ewan > > Thanks, > > Bart. > > >
