Incorrect drive detection on ASUS ROG Zenith Extreme

[g051051]

I have an ASUS ROG Zenith Extreme (latest firmware rev. 2001) with 2 RAID arrays. One array is detected fine. It has 5 drives in a RAID 10 configuration (4 main drives and a hot spare). The second array is actually my boot drive. It has 2 M.2 SATA drives in RAID 1 configuration. In RAIDXpert2, the drives and arrays are identified and listed correctly. But in HWiNFO 611_3890, it's not identifying the second array, lumping the 2 RAID 1 drives under the other array, and displaying garbage for one of the drives:

Žd [1:5]

[General Information]
	Drive Controller:	IDE
	Host Controller:	AMD-RAID Controller [storport]
	Drive Channel:	Tertiary, Slave
	Drive Address:	RAID Set #1, Disk #5
	Drive Model:	Žd
	Drive Firmware Revision:
	Drive Serial Number:
	Drive Capacity:	520,893,633,658,880 MBytes
	Drive Capacity [MB]:	520893633658880
	ATA Major Version Supported:	ACS-3

Note the corrupted drive description and incorrect information, that it's in the wrong RAID set, etc.

The other M.2 SATA drive is identified correctly, but is in the wrong RAID set as well:

Crucial_CT1050MX300SSD4 [1:6]

[General Information]
	Drive Controller:	Serial ATA 6Gb/s @ 6Gb/s
	Host Controller:	AMD-RAID Controller [storport]
	Drive Address:	RAID Set #1, Disk #6
	Drive Model:	Crucial_CT1050MX300SSD4
	Drive Firmware Revision:	M0CR060
	Drive Serial Number:	18021A88C6BA
	World Wide Name:	500A7511A88C6BA
	Drive Capacity:	1,001,562 MBytes (1050 GB)
	Drive Capacity [MB]:	1001562
	Media Rotation Rate:	SSD Drive (Non-rotating)
	Nominal Form Factor:	<1.8"
	ATA Major Version Supported:	ATA/ATAPI-5, ATA/ATAPI-6, ATA/ATAPI-7, ATA8-ACS, ACS-2, ACS-3
	ATA Minor Version Supported:	ACS-3 Revision 5
	ATA Transport Version Supported:	SATA 3.2

 

[Martin]

I have analyzed the data provided, but I believe the problem is on the side of the RAID driver. It provides invalid data regarding RAID configuration and that's confusing HWiNFO.
Unfortunately I haven't found a way how could I work-around this, it needs to be fixed in the RAID driver.

 

[g051051]

That's odd because the RAIDxpert2 software for managing the arrays shows everything correctly, per the attached screenshot. I guess it's getting it from somewhere else? Can you tell me from the debug info what driver is involved? AMD provides 3: rcraid, rccfg, rcbottom. Anything else I might use in a bug report to AMD?

 

[Martin]

Wait a bit before reporting... I noticed something interesting on the RAIDxpert2 screenshot. It shows the disk addresses, the bootable array disks as 2 0 and 2 4. Does the tool perhaps show more details about that, especially what those numbers mean ? Can it show more details about the config ?

 

[g051051]

I pulled this from the firmware config:

3 controllers listed:
0: AMD-RAID 00:01:00:01 VID 0x1022 DID 0x43BD
Segment:Bus: Device:Function 00:01:00:01
PCI ID: 1022:43BD:1b21:1062
Controller Physical Disk Count: 6
Total Physical Disk Count: 8
Array Count: 2

1: AMD-RAID 00:08:00:02 VID 0x1022 DID 0x7917
Segment:Bus: Device:Function 00:08:00:02
PCI ID: 1022:7917:1043:8761
Controller Physical Disk Count: 0
Total Physical Disk Count: 8
Array Count: 2

2: AMD-RAID 00:43:00:02 VID 0x1022 DID 0x7917
Segment:Bus: Device:Function 00:43:00:02
PCI ID: 1022:7917:1043:8761
Controller Physical Disk Count: 2
Total Physical Disk Count: 8
Array Count: 2

Physical Disk info:
0:1:0, SATA/SSD, 2.0 TB, Online
0:1:1, SATA/SSD, 2.0 TB, Online
0:1:2, SATA/SSD, 2.0 TB, Online
0:1:3, SATA/SSD, 2.0 TB, Online
0:1:4, SATA/SSD, 2.0 TB, Dedicated Spare
0:1:5, DVD CD, ATAPI
2:1:0, SATA/SSD, 1.0 TB, Online
2:1:4, SATA/SSD, 1.0 TB, Online

Controller 0 is the ports on the motherboard, of which there are 6. I have 5 2TB drives, plus my DVD drive. That's almost certainly what's throwing things off.

Controller 2 is a bit weird. ASUS has this thing called a DIMM.2 riser card. You can put 2 M.2 SATA drives on it. It has a special slot in the motherboard. You can also switch the functionality to put NVMe drives on the DIMM.2 card. I don't know why it's numbering them with 0 and 4.

I don't know for sure what Controller 1 is. It might be the NVMe device.

 

[Martin]

Well, I don't know what interface RAIDxpert2 is using to retrieve that information, but it might be something private/proprietary.
HWiNFO uses an industry-wide standard called CSMI-SAS, which seems to be supported on this controller (its drivers), but...
Only Controller 0 responds here with all devices and RAID sets, but it gives wrong information. For example it describes 2 RAID sets, which is correct, but since the driver doesn't utilize internal SAS addresses, it's impossible to determine to which RAID set a drive belongs to (based on RAID_CONFIG information). Moreover, the Controller returns Phy information for all 7 drives in the system: 0, 1, 2, 3, 4, 5, 6. The last two values belong to the Controller 2 above. When HWiNFO requests drive details for a given phy# (drive) it receives correct data for all except phy #5, which returns some garbage. Not sure if this is because of the DVD drive, but if you want to make a test, try to detach that device and see what's reported then (let me know the result).
Sorry, I was too technical, but if you're going to submit a bug report to AMD, this might be useful for them.

 

[g051051]

No, that's not too technical for me. I'm 100% sure the DVD drive is the problem and it's being treated as one of the 7 devices being reported. I've attached the new report and debug info with the DVD removed. It's better, but still wrong, as in all drives are being reported under one RAID array. I see what you mean about enumerating the devices under the wrong SCSI interface. Well, not sure how I'd write this up for AMD.

 

[Martin]

Indeed, the presence of DVD caused the driver not to provide correct CSMI-SAS details.
The inability to display correct RAID set assignment is because the driver doesn't use bSASAddress values for drives. And because of that I cannot find a link between a RAID set and particular drive - the only way to do this is by looking up which bSASAddress is assigned to a particular RAID set.
If you file a bug report to AMD, you can copy my words, point to this thread or give them my contact and I'll explain them in detail.

 

[g051051]

I have submitted a bug report to AMD, referencing this thread.

Interestingly, I noticed that the device enumeration you're seeing from the driver seems to match what I see in the firmware POST screen...it lists the 5 SATA port drives, then the DVD drive, and then the 2 M.2 drives.

 

[g051051]

Indeed, the presence of DVD caused the driver not to provide correct CSMI-SAS details.
The inability to display correct RAID set assignment is because the driver doesn't use bSASAddress values for drives. And because of that I cannot find a link between a RAID set and particular drive - the only way to do this is by looking up which bSASAddress is assigned to a particular RAID set.
If you file a bug report to AMD, you can copy my words, point to this thread or give them my contact and I'll explain them in detail.

I did file a bug report with AMD, and I have no idea if it made a difference, but with the latest driver (9.03.00.206 released on 10/09/2020) the behavior changed. Now all versions of HWiNFO that I tested with including going back to (611) report the RAID volumes, but don't show any of the physical drives that compose the arrays. The DVD now shows correctly as well.

According to HWiNFO.DBG, the CSMI calls are all failing:

10880 PhysDrvMap:
10881 \\.\PHYSICALDRIVE2: SCSIAddr=0:0:0:0; ID="USBSTOR\DISK&VEN_APPLE&PROD_IPOD&REV_1.62\000A27002242D3FD&0"; Parent="USB\VID_05AC&PID_120A\000A27002242D3FD"
10882 \\.\PHYSICALDRIVE1: SCSIAddr=1:8:3:0; ID="SCSI\DISK&VEN_AMD-RAID&PROD_ARRAY_2_\6&1C3E9326&0&080300"; Parent="{54CB850D-A731-8590-0628-1992592BD448}\RCBOTTOM\5&37459AA7&0&196607"
10883 \\.\PHYSICALDRIVE0: SCSIAddr=0:8:2:0; ID="SCSI\DISK&VEN_AMD-RAID&PROD_ARRAY_1_\6&181055A5&0&080200"; Parent="{54CB850D-A731-8590-0628-1992592BD448}\RCBOTTOM\5&3423556F&0&131071"
10884
10885 [SCSI0] Get CSMI_SAS_DRIVER_INFO - Failed (1117)
10886 [SCSI0] Get CSMI_SAS_RAID_INFO - Failed (1117,0)
10887 [SCSI0] Get CSMI_SAS_PHY_INFO - Failed (1117,0)
10888 [SCSI1] Get CSMI_SAS_DRIVER_INFO - Failed (31)
10889 [SCSI1] Get CSMI_SAS_RAID_INFO - Failed (31,0)
10890 [SCSI1] Get CSMI_SAS_PHY_INFO - Failed (31,0)
10891 [SCSI2] Get CSMI_SAS_DRIVER_INFO - Failed (31)
10892 [SCSI2] Get CSMI_SAS_RAID_INFO - Failed (31,0)
10893 [SCSI2] Get CSMI_SAS_PHY_INFO - Failed (31,0)
10894 [SCSI3] Get CSMI_SAS_DRIVER_INFO - Failed (1117)
10895 [SCSI3] Get CSMI_SAS_RAID_INFO - Failed (1117,0)
10896 [SCSI3] Get CSMI_SAS_PHY_INFO - Failed (1117,0)
10897
10898 [SCSI0] Get NVMe_ID (0) - Failed (1117)
10899 [SCSI1] Get NVMe_ID (0) - Failed (31)
10900 [SCSI2] Get NVMe_ID (0) - Failed (31)
10901 [SCSI3] Get NVMe_ID (0) - Failed (1117)
10902
10903 [PhDr0] Get SMART Ver - Error 1117
10904 [PhDr0] SCSI Adr = 0:8:2:0
10905 [PhDr0] Get SMART IDE ID - Failed (1117, 1369936)
10906 [PhDr1] SCSI Adr = 1:8:3:0
10907 [PhDr1] Get SMART IDE ID - Failed (1117, 1369936)
10908 [PhDr2] SCSI Adr = 0:0:0:0
10909 [PhDr2] Get SMART IDE ID - Failed (50, 1369936)
10910
10911
10912 [SCSI0] Get MiniportID - Failed (1117)
10913 [SCSI0] Get MiniportID - Failed (1117)
10914 [SCSI1] Get MiniportID - Failed (31)
10915 [SCSI1] Get MiniportID - Failed (31)
10916 [SCSI2] Get MiniportID - Failed (31)
10917 [SCSI2] Get MiniportID - Failed (31)
10918 [SCSI3] Get MiniportID - Failed (1117)
10919 [SCSI3] Get MiniportID - Failed (1117)

One other caveat is I now have Hyper-V enabled on my machine, but I don't know if it would cause a change here. Edit: I turned off all Hyper-V and VM features, no difference.

 

[Martin]

Thanks for the feedback, but actually this is even worse than before.
Looks like AMD has now disabled even the few methods that in the past worked at least partially CSMI-SAS support seems now completely removed in drivers.

 

[g051051]

Yeah, I assumed that as well, but I figured you'd at least be interested in the latest development. My guess is that the "support" from before was left in by mistake (hence the weird behavior), and is now removed. They're terrible at providing release notes:

Revision History

9.3.0.00206
- Fix for hibernate resume time on RAID-5 HDD.
- Fix 0x9f issue during S3 when multiple creates are active.

9.3.0.00198
- Added Clear metadata feature to clear the RAID metadata.
- updated user guide with clear metadata changes.

9.3.0.00177
-Fix for CLRF Line ending.

9.3.0.00167
- Unified driver version for all the Platforms.
- Change the folder structure

The only thing that stands out is the "Unified driver version for all the Platforms" comment. I checked back in the previous drivers (9.3.0.38 and 9.2.0.127) and they have no release notes at all.