Incompatible SSD
#1
Exclamation 
I've tried a few SSDs (via USB) with the Pine A64+. I hadn't run into any trouble until I tried a Kingston HyperX. It's a little dismaying to find an SSD that doesn't work, as that is an important component to me.

The model is SH103S3. I'm disappointed to find that I can't get it to work, even after trying three different SATA-USB interfaces.

The SH103S3 drive does work in any laptop/PC I try and it works on a different SBC.

Any insight into why this wouldn't work? Do these models have something different in their mode of I/O perhaps?
  Reply
#2
More than a little dismaying, considering Kingston is one of the biggest memory manufacturers (actually, I think they're actually the largest independent flash memory manufacturer now). 

All I can say is thank you for the report, which will hopefully help someone else not run into this issue. I don't know what the cause is... it could be a simply incompatibility between the USB-SATA controller and SSD controller, or some thing about that particular SSD controller and the pine64. I know I had some issues a few years ago with USB HDD docks and compatibility... especially the USB3 ones... paid quite a bit of money to get a particular chipset USB HDD dock from the US as it was the most compatible USB3 dock at the time. 

Just some questions to try and gather some more information...

Did the Kingston drive work on those USB-SATA controllers on the other SBC? Did this happen when you tried mounting it... or when you were trying to boot from it? And no indication from the dmesg output as to what was going on? Did it report the Kingston drive detected at all? Maybe compare the dmesg output when connecting the USB-SATA controller with another SSD, and see what difference this is in the messages, if any?
  Reply
#3
many drives have proprietary software that is used by the O/S to achieve the performance ratings that are given for the device. it's not so much a driver as it is a helper app, at least on windows, not sure how linux handles it if in fact linux supports it at all but i have had usb drives that would not run at all unless the drive software was first installed on windows units and i'm wondering if this may be a drive that operates via installed software?
  Reply
#4
(04-02-2017, 01:05 PM)dkryder Wrote: many drives have proprietary software that is used by the O/S to achieve the performance ratings that are given for the device. it's not so much a driver as it is a helper app, at least on windows, not sure how linux handles it if in fact linux supports it at all but i have had usb drives that would not run at all unless the drive software was first installed on windows units and i'm wondering if this may be a drive that operates via installed software?

The drive works in another SBC, using a similar Linux kernel and SoC. I also tried the Android Pine image and that failed, as well. It also works in any laptop/desktop I've tried. I don't think drivers are the issue. I also checked the SSD's firmware and updated it to the latest.

(04-01-2017, 09:01 PM)pfeerick Wrote: Just some questions to try and gather some more information...

Did the Kingston drive work on those USB-SATA controllers on the other SBC? Did this happen when you tried mounting it... or when you were trying to boot from it? And no indication from the dmesg output as to what was going on? Did it report the Kingston drive detected at all? Maybe compare the dmesg output when connecting the USB-SATA controller with another SSD, and see what difference this is in the messages, if any?

Other SBC:

  1. The drive works, using any of the USB-SATA adapters.
  2. I did a full test, partitioning, copying and verifying data.
Pine:

I tried three different USB-SATA interfaces.

  1. The drive is never available as a device. It is detected (according to dmesg) by never "ready".
  2. One interface times out and the other two continually disconnect and reconnect the device.
dmesg output (this will time-out):

[  116.755832] usb-storage 1-1:1.0: USB Mass Storage device detected
[  116.758762] scsi0 : usb-storage 1-1:1.0
[  117.738320] scsi 0:0:0:0: Direct-Access     SABRENT                   4101 PQ: 0 ANSI: 6
[  123.009510] sd 0:0:0:0: [sda] Unit Not Ready

[  123.009524] sd 0:0:0:0: [sda]  
[  123.009529] Sense Key : 0x4 [current]
[  123.009537] sd 0:0:0:0: [sda]  
[  123.009542] ASC=0x44 <<vendor>> ASCQ=0x8

The repeated I/O detect and drop looks like this (from dmesg):

[  284.691383] ehci_irq: highspeed device connect
[  284.911185] usb 1-1: new high-speed USB device number 35 using sunxi-ehci
[  285.061345] usb-storage 1-1:1.0: USB Mass Storage device detected
[  285.061619] scsi33 : usb-storage 1-1:1.0
[  286.043989] scsi 33:0:0:0: Direct-Access     Generic  External         0205 PQ: 0 ANSI: 6
[  286.048014] sd 33:0:0:0: [sda] Attached SCSI removable disk
[  286.348514] ehci_irq: highspeed device disconnect
[  286.348636] usb 1-1: USB disconnect, device number 35
[  287.082449] ehci_irq: highspeed device connect
[  287.302291] usb 1-1: new high-speed USB device number 36 using sunxi-ehci



Working state, dmesg output from two USB-SATA interfaces on the other SBC:

Interface 1:
[   31.092689] usb-storage 1-1.3:1.0: USB Mass Storage device detected
[   31.092905] scsi1 : usb-storage 1-1.3:1.0
[   32.088598] scsi 1:0:0:0: Direct-Access     KINGSTON  SH103S3240G     526A PQ: 0 ANSI: 6
[   32.089867] sd 1:0:0:0: [sda] 468862124 512-byte logical blocks: (240 GB/223 GiB)
[   32.090462] sd 1:0:0:0: [sda] Write Protect is off
[   32.090477] sd 1:0:0:0: [sda] Mode Sense: 31 00 00 00
[   32.091102] sd 1:0:0:0: [sda] Write cache: enabled, read cache: enabled, doesn't support DPO or FUA
[   32.097106]  sda: sda1
[   32.099836] sd 1:0:0:0: [sda] Attached SCSI disk

Interface 2:
[   82.707754] usb 1-1.3: new high-speed USB device number 8 using dwc_otg
[   82.889240] usb-storage 1-1.3:1.0: USB Mass Storage device detected
[   82.889435] scsi2 : usb-storage 1-1.3:1.0
[   83.888146] scsi 2:0:0:0: Direct-Access     SABRENT                   4101 PQ: 0 ANSI: 6
[   84.599067] sd 2:0:0:0: [sda] 468862128 512-byte logical blocks: (240 GB/223 GiB)
[   84.599680] sd 2:0:0:0: [sda] Write Protect is off
[   84.599686] sd 2:0:0:0: [sda] Mode Sense: 47 00 00 08
[   84.600301] sd 2:0:0:0: [sda] Write cache: enabled, read cache: enabled, doesn't support DPO or FUA
[   84.603480]  sda: sda1
[   84.606342] sd 2:0:0:0: [sda] Attached SCSI disk
  Reply
#5
might try a powered hub if not already and try both usb ports if you have not already.
  Reply
#6
(04-02-2017, 09:11 PM)dkryder Wrote: might try a powered hub if not already and try both usb ports if you have not already.

Yeah, the port doesn't matter. I've actually tried it on two, separate Pine boards. I'll see if I can dig out a powered USB hub but that's stretching it. This drive likely draws more power than others but it's dismaying if it can't be attached to a Pine with a minimal OS (no windowing / graphics system). Unfortunately, the Pine board does seem to be bad at handling variation in the power (e.g. 1A from an iPhone charge cube) but I am using the stock power adapter from Pine.
  Reply
#7
one of the spec sheets i read said it uses ~ 2 watts for read/writes. 2.5 watts for startup. that's roughly .400 - .500 amps. that is not an insignificant number considering it is using the ~.4 amps at the same time the board itself has reached peak states for current consumption. there have been several reports in the forum that powered hubs improve performance of hard disks. it's not really all that far fetched to consider using a power hub although it is by no means a sure bet to help in all situations.
  Reply
#8
(04-03-2017, 07:11 PM)dkryder Wrote: one of the spec sheets i read said it uses ~ 2 watts for read/writes. 2.5 watts for startup. that's roughly .400 - .500 amps. that is not an insignificant number considering it is using the ~.4 amps at the same time the board itself has reached peak states for current consumption. there have been several reports in the forum that powered hubs improve performance of hard disks. it's not really all that far fetched to consider using a power hub although it is by no means a sure bet to help in all situations.

Agreed. The power usage is significant but with 2A from the power adapter, we can theoretically get up to 10w. To provide a safety margin, I'd target 8w as the max. According to the wiki, the board uses 2.5w (500mA). That leaves plenty for the SSD  As this is a minimal, console installation, I'd expect the power draw to be the lowest possible. If there are any optimizations to disable unused hardware, or reduce the draw, please let me know.

I say that a powered hub is "stretching it", not because I doubt its effectiveness. I just think it puts the Pine in a bad light; my problem is that a 2A power supply ought to sufficiently accommodate the board and a single SSD. Keep in mind that I've tried another SSD without any issues *and* I've tried the Kingston SSD on another SBC (also a quad-core ARM, with a 2A input) and it had no trouble with the drive.

If this is an issue with power draw, then I think the 2.5w spec is inaccurate or the power supply isn't really capable of drawing 2A. Even if the SSD draws a full 1A, the total power draw would still be under 8w.
  Reply
#9
Keep in mind that 2.5W are the idle power requirements. My boards need about 500-600mA (mostly depending on GbE) at 5V idle current. Under load the A64 and on-board peripherals need 1.5-2A ... the USB ports have compliant 500mA integrated current limiters which would also cut off any load spikes. I always use external power supplies for my external HDD/SSD.
Come have a chat in the Pine IRC channel >>
  Reply


Forum Jump:


Users browsing this thread: 2 Guest(s)