[PATCH] ata: libata-pmp: disable the SEMB port on Marvell 88SM9705

[Marco Ziliani <rylos78@gmail.com>]

The Marvell 88SM9705 is a 1-to-5 SATA port multiplier with one
additional SEMB (SATA Enclosure Management Bridge) port [1].  Ports
0-4 are device ports and port 5 is the SEMB.  The number of fan-out
ports in GSCR[2] is 6 when the SEMB is enabled and 5 when it is not,
so on boards without the SEMB there is no link 5 and this quirk does
nothing.  That includes the common 8-port PCIe cards that pair an
88SE9215 with an 88SM9705, which report 5 ports.

The SEMB port never answers SRST.  libata probes it like a device
port, and every EH round on the PMP ends in a retry loop that lasts
about a minute before EH gives up, then starts over:

  ata19.15: Port Multiplier 1.2, 0x1b4b:0x9705 r160, 6 ports, feat 0x1/0x1f
  ata19.05: hard resetting link
  ata19.05: failed to resume link (SControl 0)
  ata19.05: softreset failed (1st FIS failed)
  ata19.05: hard resetting link
  ...
  ata20.05: reset failed, giving up

Disk discovery waits for this, including disks on other controllers.

libata.force cannot work around it: the "disable" horkage only takes
effect after IDENTIFY, which the SEMB link never reaches.

Treat port 5 the same way as port 4 of the Marvell 88SM4140 and mark
the link ATA_LFLAG_DISABLED.

Tested by the user who found the problem (see the synodeb link below)
on a Synology DX1215 expansion unit (four 88SM9705 r160 behind a
Marvell 88SE9235), one disk in the DX1215 and one on a separate AHCI
controller, one boot each, Debian 6.12 libata with and without this
change:

                                      without     with
  "softreset failed" lines            60          0
  disk on the separate controller     200.6 s     18.6 s
  disk behind an 88SM9705             203.3 s     21.3 s

Each PMP still logs one "failed to resume link (SControl 0)" for port 5
at probe time.

Link: https://www.marvell.com/content/dam/marvell/en/public-collateral/storage/marvell-storage-88sm9705-functional-specifications-2015-06.pdf [1]
Link: https://github.com/rylos/synodeb/discussions/8
Cc: stable@vger.kernel.org
Signed-off-by: Marco Ziliani <rylos78@gmail.com>
---
I don't have this hardware myself: the problem was found and measured
on a DX1215 by a user of synodeb (Debian on Synology NAS units), who
preferred not to send the patch. It is carried in the synodeb kernels.

Kyrylo, your DX1222 has the same 88SE9235 + 88SM9705 chain: if it shows
the same port 5 retry loop, a Tested-by would be very welcome.
---
 drivers/ata/libata-pmp.c | 13 +++++++++++++
 1 file changed, 13 insertions(+)

diff --git a/drivers/ata/libata-pmp.c b/drivers/ata/libata-pmp.c
index 904f418387..72bc458568 100644
--- a/drivers/ata/libata-pmp.c
+++ b/drivers/ata/libata-pmp.c
@@ -482,6 +482,19 @@ static void sata_pmp_quirks(struct ata_port *ap)
 			if (link->pmp == 4)
 				link->flags |= ATA_LFLAG_DISABLED;
 		}
+	} else if (vendor == 0x1b4b && devid == 0x9705) {
+		/*
+		 * Marvell 88SM9705 quirks
+		 *
+		 * 1-to-5 PMP with an extra SEMB port.  It reports 6 ports and
+		 * port 5 is the SEMB, which never answers SRST: EH spends about
+		 * a minute per round retrying it, and disk discovery waits for
+		 * that.  Seen behind a Marvell 88SE9235 in the Synology DX1215.
+		 */
+		ata_for_each_link(link, ap, EDGE) {
+			if (link->pmp == 5)
+				link->flags |= ATA_LFLAG_DISABLED;
+		}
 	}
 }
 

---
base-commit: fb2c3f44a7e14627b9dca8e6b568e6cdc234b40b
change-id: 20261028-pmp-88sm9705-2f741b435f16

Best regards,
--  
Marco Ziliani <rylos78@gmail.com>