Project

General

Profile

Feature #8947

Support the Marvell 88SE9230 PCIe to SATA controller

Added by Martin Clemons over 1 year ago. Updated about 1 year ago.

Status:
New
Priority:
Normal
Assignee:
-
Category:
driver - device drivers
Start date:
2018-01-03
Due date:
% Done:

0%

Estimated time:
Difficulty:
Medium
Tags:
needs-triage

Description

It seems that currently the Marvell 88SE9230 PCIe to SATA controller is not supported. When booting my Proliant Microserver gen10 in SmartOS, I get the following error on the console and in dmesg roughly 2x per second:
2018-01-03T17:40:15.877739+00:00 polaris ahci: [ID 432157 kern.warning] WARNING: ahci0: Cannot allocate ports structure
2018-01-03T17:40:15.881598+00:00 polaris ahci: [ID 405770 kern.info] NOTICE: ahci0: hba AHCI version = 1.30
2018-01-03T17:40:15.881625+00:00 polaris pcplusmp: [ID 805372 kern.info] pcplusmp: pciclass,010601 (ahci) instance 0 irq 0x13 vector 0x43 ioapic 0x0 intin 0x13 is bound to cpu 0

Marvell 88SE9230 does not seem to be in the HCL.

SmartOS runs fine on the gen7 microserver, it would be awesome if the gen10 were also supported.


Related issues

Related to illumos gate - Bug #3865: AHCI: 88SE9230 watchdog timeouts, "hung" panicNew2013-07-03

Actions

History

#1

Updated by Martin Clemons over 1 year ago

The full error message is:
2018-01-03T17:40:16.379668+00:00 polaris ahci: [ID 432157 kern.warning] WARNING: ahci0: Cannot allocate ports structure
2018-01-03T17:40:16.383529+00:00 polaris ahci: [ID 405770 kern.info] NOTICE: ahci0: hba AHCI version = 1.30
2018-01-03T17:40:16.383553+00:00 polaris pcplusmp: [ID 805372 kern.info] pcplusmp: pciclass,010601 (ahci) instance 0 irq 0x13 vector 0x43 ioapic 0x0 intin 0x13 is bound to cpu 1
2018-01-03T17:40:16.881893+00:00 polaris ahci: [ID 432157 kern.warning] WARNING: ahci0: Cannot allocate ports structure
2018-01-03T17:40:16.885836+00:00 polaris ahci: [ID 405770 kern.info] NOTICE: ahci0: hba AHCI version = 1.30
2018-01-03T17:40:16.885860+00:00 polaris pcplusmp: [ID 805372 kern.info] pcplusmp: pciclass,010601 (ahci) instance 0 irq 0x13 vector 0x43 ioapic 0x0 intin 0x13 is bound to cpu 0

#2

Updated by Marcel Telka over 1 year ago

  • Related to Bug #3865: AHCI: 88SE9230 watchdog timeouts, "hung" panic added
#3

Updated by Martin Clemons over 1 year ago

I tried installing OI Hipster 2017.10 (5.11 illumos-2727bb055f i86pc i386 i86pc) and the 88SE9230 is recognized properly. The output from Device Driver Utility is included below. I've tried attaching 2 SATA HDDs and they are recognized in Gparted.
So this seems to be a problem with SmartOS only?


node name: pci1b4b,9230
Vendor: Marvell Technology Group Ltd.
Device: 88SE9230 PCIe SATA 6Gb/s Controller
Sub-Vendor: Marvell Technology Group Ltd.
binding name: pciclass,010601
devfs path: /pci@0,0/pci1022,157c@2,2/pci1b4b,9230@0
bus addr: 0
pci path: 1,0,0
compatible name: (pciex1b4b,9230.1b4b.9230.11)(pciex1b4b,9230.1b4b.9230)(pciex1b4b,9230.11)(pciex1b4b,9230)(pciexclass,010601)(pciexclass,0106)(pci1b4b,9230.1b4b.9230.11)(pci1b4b,9230.1b4b.9230)(pci1b4b,9230)(pci1b4b,9230.11)(pci1b4b,9230)(pciclass,010601)(pciclass,0106)
driver name: ahci
instance: 1
driver state: Attached
sata: 1
fm-accchk-capable: TRUE
fm-dmachk-capable: TRUE
fm-errcb-capable: TRUE
fm-ereport-capable: TRUE
scsi-binding-set: spi
initiator-interconnect-type: SPI
scsi-tag-age-limit: 2
scsi-selection-timeout: fa
scsi-watchdog-tick: a
scsi-reset-delay: bb8
scsi-options: 107ff8
scsi-enumeration: 0
pci-msi-capid-pointer: 50
acpi-namespace: _SB_.PCI0.GPP1.D02F
assigned-addresses: 81010010
reg: 10000
compatible: pciex1b4b,9230.1b4b.9230.11
model: SATA AHCI Interface
power-consumption: 1
devsel-speed: 0
interrupts: 1
subsystem-vendor-id: 1b4b
subsystem-id: 9230
unit-address: 0
class-code: 10601
revision-id: 11
vendor-id: 1b4b
device-id: 9230
DISK : WDC WD40EFRX-68N32N0
Capacity :
Driver :
Path : /dev/rdsk/c4t0d0s2 WDC WD40EFRX-68N32N0

#4

Updated by Sebastian Wiedenroth over 1 year ago

just a bit of additional information: the messages go away when all four drive bays have disks in them.

#5

Updated by Jürgen Bereuter about 1 year ago

Sebastian Wiedenroth wrote:

just a bit of additional information: the messages go away when all four drive bays have disks in them.

i had to attach 5 disks (i have no cd tray), than the errors disappear. Thanks to Till, he mentioned to disable fmd (svcadm disable fmd), because it fills up the logfiles.

#6

Updated by Martin Clemons about 1 year ago

I've done some additional testing and found that the critical missing disk in my case is the upper bay. As long as I have a disk in that bay the errors go away. Thanks Sebastian and Jürgen!

#7

Updated by Jürgen Bereuter about 1 year ago

some info for others, who tried this Marvell chip: i installed 4 Exos Seagate with 6 TB and they are not recognized. The server refuses to boot. When i install one of them at the internal SATA port it is recognized. BTW, this is independent from the OS, the problem begins at startup, around BIOS activation.

Also available in: Atom PDF