Rockpro64 Sata Card kills itself

[Derisis13]

This is a thermal problem.
I had the same issue with the Pine store SATA card, and I was to buy a new one, when I noticed that all other SATA cards have some kind of a heat sink. I touched the IC on my (running) card... It was hot. 
I shut down the system, took out the card and glued a tiny 2cm*2cm heatsink on it with thermal glue. I plugged it back, and although it was still a bit too hot, I managed to run it with a single HDD for days. 
I also sat a fan blowing air at the card and the Rockpro64 and that reduced the temperatures to a comfortable level (airflow isn't the best, it's a 12V fan running on 3.3V to be quiet) and my NAS is online for weeks now without disk issues.

[AnujDatar]

I have the pine store SATA card (ASM1062) too, and it has always been really slow. I never bothered because I only use my nas box as a backup and all the media on it was copied with the HDD in a USB enclosure.

But I have been looking at adding another drive and using the NAS more actively, so I tried looking into the issue. I saw the same `Buffer I/O error` randomly pop up. I have been getting only like 3-4 MB/s over gigabit ethernet. I tried using inserting a sata ssd today, I get like 7MB/s. The SSD is slow even when I test it with `hdparm`.

Here `sda` is the HDD and `sdb` is the SSD.

Code:

/dev/sda:
Timing buffered disk reads: 354 MB in  3.01 seconds = 117.49 MB/sec

/dev/sdb:
Timing buffered disk reads: 586 MB in  3.01 seconds = 194.85 MB/sec

Then I removed the drives from the box inserted them in a USB enclosure and ran `hdparm` again

Code:

/dev/sda:
Timing buffered disk reads: 352 MB in  3.00 seconds = 117.23 MB/sec

/dev/sdb:
Timing buffered disk reads: 1148 MB in  3.09 seconds = 371.95 MB/sec

Marvell cards seem to have fixed some issues for many of you, so I ordered a Marvell 9215 based card last week. It came in the mail today, and the seller sent me an ASM1062 .

Just checking if anyone's got their ASM1061/1062 working fine? Trying to see if there's anything I can do before sending this back.

I have the RP64 v2.1 board, that has the 4 PCIe resistors on it. A couple of weeks ago, I read in one of the threads that those shouldn't be there, and someone tried removing theirs and that fixed some issue for them. And these resistors are not on newer revisions of the board either. But I can't seem to find that thread anymore. I was wondering if I should try to remove them, but I don't have a hot air reflow station here, so I'd have to use my decade old soldering iron and hope it doesn't burn the board.

[AnujDatar]

So I just checked the dmesg for the new card and there are no errors.


The old/original pine64 sata card ASM1061 card had the same errors as everyone else

Code:

ata2: SATA link up 6.0 Gbps (SStatus 133 SControl 300)
[    8.597283] ata2.00: ATA-8: TOSHIBA MQ01ABD100, AX0P3D, max UDMA/100
[    8.597297] ata2.00: 1953525168 sectors, multi 16: LBA48 NCQ (depth 32), AA
[    8.603786] ata2.00: configured for UDMA/100
[    8.604497] scsi 1:0:0:0: Direct-Access     ATA      TOSHIBA MQ01ABD1 3D   PQ: 0 ANSI: 5
[    8.605912] sd 1:0:0:0: [sda] 1953525168 512-byte logical blocks: (1.00 TB/932 GiB)
[    8.605930] sd 1:0:0:0: [sda] 4096-byte physical blocks
[    8.606013] sd 1:0:0:0: [sda] Write Protect is off
[    8.606029] sd 1:0:0:0: [sda] Mode Sense: 00 3a 00 00
[    8.606168] sd 1:0:0:0: [sda] Write cache: enabled, read cache: enabled, doesn't support DPO or FUA
[    8.920089] ata2.00: exception Emask 0x10 SAct 0x1 SErr 0x280100 action 0x6 frozen
[    8.920103] ata2.00: irq_stat 0x08000000, interface fatal error
[    8.920118] ata2: SError: { UnrecovData 10B8B BadCRC }
[    8.920133] ata2.00: failed command: READ FPDMA QUEUED
[    8.920159] ata2.00: cmd 60/08:00:00:00:00/00:00:00:00:00/40 tag 0 ncq dma 4096 in
                        res 40/00:00:00:00:00/00:00:00:00:00/40 Emask 0x10 (ATA bus error)                                                                                       
[    8.920170] ata2.00: status: { DRDY }
[    8.920190] ata2: hard resetting link
[    9.396077] ata2: SATA link up 6.0 Gbps (SStatus 133 SControl 300)
[    9.407542] ata2.00: configured for UDMA/100
[    9.407585] ata2: EH complete
[    9.540073] ata2.00: exception Emask 0x10 SAct 0x4 SErr 0x280100 action 0x6 frozen
[    9.540085] ata2.00: irq_stat 0x08000000, interface fatal error
[    9.540099] ata2: SError: { UnrecovData 10B8B BadCRC }
[    9.540112] ata2.00: failed command: READ FPDMA QUEUED
[    9.540138] ata2.00: cmd 60/08:10:08:00:00/00:00:00:00:00/40 tag 2 ncq dma 4096 in
                        res 40/00:10:08:00:00/00:00:00:00:00/40 Emask 0x10 (ATA bus error)                                                                                       
[    9.540148] ata2.00: status: { DRDY }
[    9.540164] ata2: hard resetting link
[   10.016101] ata2: SATA link up 6.0 Gbps (SStatus 133 SControl 300)
[   10.027560] ata2.00: configured for UDMA/100
[   10.027595] ata2: EH complete
[   10.188071] ata2.00: exception Emask 0x10 SAct 0x80 SErr 0x280100 action 0x6 frozen
[   10.188083] ata2.00: irq_stat 0x08000000, interface fatal error
[   10.188096] ata2: SError: { UnrecovData 10B8B BadCRC }
[   10.188110] ata2.00: failed command: READ FPDMA QUEUED
[   10.188136] ata2.00: cmd 60/08:38:a8:6d:70/00:00:74:00:00/40 tag 7 ncq dma 4096 in
                        res 40/00:38:a8:6d:70/00:00:74:00:00/40 Emask 0x10 (ATA bus error)                                                                                       
[   10.188146] ata2.00: status: { DRDY }
[   10.188162] ata2: hard resetting link
[   10.664075] ata2: SATA link up 6.0 Gbps (SStatus 133 SControl 300)
[   10.675536] ata2.00: configured for UDMA/100
[   10.675570] ata2: EH complete
[   10.836071] ata2: limiting SATA link speed to 3.0 Gbps
[   10.836088] ata2.00: exception Emask 0x10 SAct 0x8000 SErr 0x280100 action 0x6 frozen
[   10.836098] ata2.00: irq_stat 0x08000000, interface fatal error
[   10.836111] ata2: SError: { UnrecovData 10B8B BadCRC }
[   10.836125] ata2.00: failed command: READ FPDMA QUEUED
[   10.836151] ata2.00: cmd 60/08:78:a0:6d:70/00:00:74:00:00/40 tag 15 ncq dma 4096 in
                        res 40/00:78:a0:6d:70/00:00:74:00:00/40 Emask 0x10 (ATA bus error)                                                                                       
[   10.836162] ata2.00: status: { DRDY }
[   10.836178] ata2: hard resetting link
[   11.312078] ata2: SATA link up 3.0 Gbps (SStatus 123 SControl 320)
[   11.323543] ata2.00: configured for UDMA/100
[   11.323577] ata2: EH complete
[   11.443552]  sda: sda1 sda2
[   11.446493] sd 1:0:0:0: [sda] Attached SCSI disk
[   11.447787] sda: detected capacity change from 0 to 1000204886016
[   11.448945] sda: detected capacity change from 0 to 1000204886016

SATA related dmesg for the new ASM1062 card is:

Code:

[    2.763071] ata1: SATA max UDMA/133 abar m512@0xfa010000 port 0xfa010100 irq 79
[    2.763083] ata2: SATA max UDMA/133 abar m512@0xfa010000 port 0xfa010180 irq 79
[    3.078772] ata1: SATA link down (SStatus 0 SControl 300)
[    4.092099] ata2: SATA link up 6.0 Gbps (SStatus 133 SControl 300)
[    4.092621] ata2.15: Port Multiplier 1.2, 0x1b21:0x1092 r0, 3 ports, feat 0x5/0xf
[    4.098161] ata2.00: hard resetting link
[    4.576483] ata2.00: SATA link up 6.0 Gbps (SStatus 133 SControl 330)
[    4.576559] ata2.01: hard resetting link
[    4.892568] ata2.01: SATA link down (SStatus 0 SControl 330)
[    4.892701] ata2.02: hard resetting link
[    5.208603] ata2.02: SATA link down (SStatus 0 SControl 330)
[    5.514408] ata2.00: ATA-8: TOSHIBA MQ01ABD100, AX0P3D, max UDMA/100
[    5.514422] ata2.00: 1953525168 sectors, multi 16: LBA48 NCQ (depth 32), AA
[    5.516231] ata2.00: configured for UDMA/100
[    5.516374] ata2: EH complete
[    5.516782] scsi 1:0:0:0: Direct-Access     ATA      TOSHIBA MQ01ABD1 3D   PQ: 0 ANSI: 5
[    5.517750] sd 1:0:0:0: [sda] 1953525168 512-byte logical blocks: (1.00 TB/932 GiB)
[    5.517762] sd 1:0:0:0: [sda] 4096-byte physical blocks
[    5.517822] sd 1:0:0:0: [sda] Write Protect is off
[    5.517833] sd 1:0:0:0: [sda] Mode Sense: 00 3a 00 00
[    5.517931] sd 1:0:0:0: [sda] Write cache: enabled, read cache: enabled, doesn't support DPO or FUA
[    5.640460]  sda: sda1 sda2
[    5.643385] sd 1:0:0:0: [sda] Attached SCSI disk

It does a couple of hard resets, but seems to work fine in the end. And I'm still stuck with excruciatingly slow network transfer speeds. I just tried to transfer a 3GB video file. It took 15 minutes at like 3 MB/s. Yes, this is over gigabit ethernet.

PS: I am running Armbian Focal minimal, kernel 5.10.y

[AnujDatar]

So, I just tried different OS versions.

1. Armbian Focal, 5.10.35-rockchip  -> avg transfer speeds -> 3-4 MB/s

2. Armbian Buster, 5.10.35-rockchip -> avg transfer speed -> 10 MB/s

3. Armbian Buster, 4.4.y                  -> avg transfer speed -> 14-15 MB/s

4. Ayufan Bionic, 4.4.190                 -> avg transfer speed -> 9-10 MB/s


I get the same speeds on an Toshiba 5400rpm HDD and a Kioxia sata SSD. I pulled my WD Red out for testing cuz I didn't want to accidentally corrupt my data.

And I know I am not saturating my ethernet connection. I tried transferring the same file to both drives at the same time. They transferred at exactly the speeds I mentioned above, both simultaneously. So there's plenty of network bandwidth left over.

iperf shows 950+Mbps if I connect both devices right to the main router. But, I don't typically use that setting. I have a couple of hubs spread over the house with wired ethernet between them (cableing I installed myself). So, not the same as plugging directly into same switch/router. But even with that, the worst case scenario is `837 Mbits/sec`. So that's not the issue.

I have no clue what is going on. Anyone else run into this? What are the best transfer speeds you're getting?

[LMM]

Hi,
I made some tests on my network compound of two loops. The router for the second loop is a rockpro64 with additional Gigabit eth on USB3. A rockpro64 is linked on each loop. One rockpro64 has a SATA card with a 5400 rpm HDD attached. All cards are on Armbian Focal
- iperf3 :  926 Mbits/sec
- rsync :  50,541,366.40 bytes/sec

You could eventually check :
- the load average
- the retransmission rate

[AnujDatar]

I just installing OMV5 on my Armbian buster. Not sure what settings that changed, but my transfer speeds just went up to 18-20mbps.

I should have also tried this sooner, but I booted my laptop into windows and tried file transfer. I get 50MB/s there. I knew smb is slow on linux, but didn't realize it could be over 50% lower. I'm going to set up an NFS and check my speeds again.

Will report back.

[rantoie]

I have the same issue as reported here, with two HDD, the system cannot handle things.

Either the SATA card in the Pine64 store is not very good, or the power-supply sold there (5A) is not powerful enough. Can someone definitively say if they got some other thing to work?

I can lower the SATA transfer speeds to 1.5G (add extraargs=libata.force=1.5G to /boot/armbianEnv.txt) or disable ncq (add extraargs=libata.force=noncq), and then things work "fine", but this has significantly degraded the performance of the disks.

I do not think Pine64 should be selling this kit with this problem. There are other threads about this issue, and they have been around for several *years* now, with no reaction. @Luke or anyone out there listening? The other thread where this was discussed has been closed: https://forum.pine64.org/showthread.php?tid=6932

[varac]

After recieving and assembling my Pine NASCase (with 2x 3,5" HDDs, Western
Digital WD40EFAX-68JH4N1, certified for NAS) I immediately got a lot of these
errors when I created some IO (i.e. running badblocks on both disks, or creating
an md softraid):


ata1.00: exception Emask 0x10 SAct 0x200 SErr 0x400000 action 0x6 frozen
ata1.00: irq_stat 0x08000000, interface fatal error
ata1: SError: { Handshk }
ata1.00: failed command: WRITE FPDMA QUEUED
ata1.00: cmd 61/90:48:70:f6:af/09:00:05:00:00/40 tag 9 ncq dma 1253376 ou
res 40/00:48:70:f6:af/00:00:05:00:00/40 Emask 0x10 (ATA bus error)
ata1.00: status: { DRDY }
ata1: hard resetting link

see https://forum.pine64.org/showthread.php?tid=6511,
https://forum.pine64.org/showthread.php?tid=6932&page=3,
and https://wiki.archlinux.org/index.php/Sol...NCQ_errors

Booting with the "libata.force=noncq" kernel parameter resolved those issues.

echo 'extraargs=libata.force=noncq' >> /boot/armbianEnv.txt

When doing high IO on the disks (badblocks in parallel on both disks,
syncing an md softraid etc) I frequently get `WRITE DMA EXT` errors,
which fortunatly don't corrupt the write operations:


ata2.00: irq_stat 0x08000000, interface fatal error
ata2: SError: { Handshk }
ata2.00: failed command: WRITE DMA EXT
ata2.00: cmd 35/00:00:00:14:a8/00:0a:47:00:00/e0 tag 9 dma 1310720 out
res 50/00:00:00:14:a8/00:00:47:00:00/e0 Emask 0x10 (ATA bus error)
ata2.00: status: { DRDY }
ata2: hard resetting link
ata1: SATA link up 3.0 Gbps (SStatus 123 SControl 320)
ata1.00: configured for UDMA/133
ata1: EH complete
ata2: SATA link up 1.5 Gbps (SStatus 113 SControl 310)
ata2.00: configured for UDMA/133
ata2: EH complete
ata1: limiting SATA link speed to 1.5 Gbps
ata1.00: exception Emask 0x10 SAct 0x0 SErr 0x400000 action 0x6 frozen
ata1.00: irq_stat 0x08000000, interface fatal error
ata1: SError: { Handshk }
ata1.00: failed command: WRITE DMA EXT
ata1.00: cmd 35/00:00:00:28:bd/00:0a:47:00:00/e0 tag 0 dma 1310720 out
res 50/00:00:00:28:bd/00:00:47:00:00/e0 Emask 0x10 (ATA bus error)
ata1.00: status: { DRDY }
ata1: hard resetting link
ata1: SATA link up 1.5 Gbps (SStatus 113 SControl 310)
ata1.00: configured for UDMA/133
ata1: EH complete

Still, those errors are a mess in the syslog and downgrade the SATA link spped
from 6 Gbit/s to 1,5 Gbit/s (back and forth).

I even tried glueing some heatsink onto the SATA controller's main chip which
got hot under load (just enough that it hurt my fingertip), but this doesn't
resolve the issue.

I also replaced both SATA cables with ones that I had lying around,
without any difference.

I must say I'm unhappy by the lack of proper support by Pine64 here in this
thread, which started 2018, and I didn't see any investigations/solutions by Pine64.

[hyc]

I just received my Rockpro64 with NAS case last week, and got 2 fresh 4TB Samsung 970 EVO SSDs today to install in it.
Running freshly installed Armbian Focal. I wasn't even able to mkfs any ext4 filesystems on the drives, got I/O errors all over the place.
Same as in other peoples' dmesg excerpts. All of the parts were from the pine64 store: SATA card, 12V/5A EU power supply,
case fan.

I tried attaching only 1 drive, same problem. So it's not a power supply issue. I also swapped in an older 500GB 850 EVO SSD,
and it's working perfectly. I also plugged in one of the 870 EVOs using a USB - SATA adapater, and that's working perfectly too.

Considering that it works fine at 1.5Gbps, maybe it's just bad SATA cables? Or do I really need to junk this PCIe card and get a better one.

[rantoie]

Considering that it works fine at 1.5Gbps, maybe it's just bad SATA cables? Or do I really need to junk this PCIe card and get a better one.

Did try to change the SATA cables here, but no difference.

It really seems like the PCIe card is just garbage.

I tried to look around for a Marvell card to try instead, but it will easily cost an additional 80euros to get one, and then I'm not sure if it will make a difference.