HC4 doesn't see second SSD, again, no politics please

[ChrisO]

As the original thread, which was closed by moderator for obvious reason, did attract attention of few peoples having the same problem as me, it would be a waste not to follow up on this, I think.

Let me sum up:

I have two HC4, one bought in 8.2021 and other one 11.2022, both have two SSDs each.

The older one after a kernel upgrade didn't see a second disk any more.

I've just tried few kernels I could find with the following results:

 

OK: Linux odroidhc4 5.12.10-meson64 #21.05.4 SMP PREEMPT Wed Jun 16 08:19:35 UTC 2021 aarch64 aarch64 aarch64 GNU/Linux

OK: Linux odroidhc4 5.13.12-meson64 #21.08.1 SMP PREEMPT Wed Aug 25 19:35:13 UTC 2021 aarch64 aarch64 aarch64 GNU/Linux

Not OK:Linux odroidhc4 5.15.4-meson64 #21.08.6 SMP PREEMPT Mon Nov 22 11:46:10 UTC 2021 aarch64 aarch64 aarch64 GNU/Linux

 

Couldn't find any kernel between 5.13.12 and 5.15.4, so it's not clear which is exact the version where the problem started.

Here is what I found in /var/log/syslog:

 

Nov 22 14:40:46 odroidhc4 kernel: [ 2.355049] ata1: SATA max UDMA/133 abar m512@0xfc700000 port 0xfc700100 irq 38

Nov 22 14:40:46 odroidhc4 kernel: [ 2.355060] ata2: SATA max UDMA/133 abar m512@0xfc700000 port 0xfc700180 irq 38

Nov 22 14:40:46 odroidhc4 kernel: [ 2.358363] libphy: mdio_mux: probed

Nov 22 14:40:46 odroidhc4 kernel: [ 2.564017] libphy: mdio_mux: probed

Nov 22 14:40:46 odroidhc4 kernel: [ 2.830521] ata1: SATA link up 6.0 Gbps (SStatus 133 SControl 300)

Nov 22 14:40:46 odroidhc4 kernel: [ 2.830799] ata1.00: FORCE: horkage modified (noncq)

Nov 22 14:40:46 odroidhc4 kernel: [ 2.830854] ata1.00: supports DRM functions and may not be fully accessible

Nov 22 14:40:46 odroidhc4 kernel: [ 2.830860] ata1.00: ATA-11: Samsung SSD 870 QVO 1TB, SVQ02B6Q, max UDMA/133

Nov 22 14:40:46 odroidhc4 kernel: [ 2.830865] ata1.00: 1953525168 sectors, multi 1: LBA48 NCQ (not used)

Nov 22 14:40:46 odroidhc4 kernel: [ 2.832778] ata1.00: Features: Trust Dev-Sleep

Nov 22 14:40:46 odroidhc4 kernel: [ 2.833128] ata1.00: supports DRM functions and may not be fully accessible

Nov 22 14:40:46 odroidhc4 kernel: [ 2.835140] ata1.00: configured for UDMA/133

Nov 22 14:40:46 odroidhc4 kernel: [ 2.835408] scsi 0:0:0:0: Direct-Access ATA Samsung SSD 870 2B6Q PQ: 0 ANSI: 5

Nov 22 14:40:46 odroidhc4 kernel: [ 2.835890] ata1.00: Enabling discard_zeroes_data

Nov 22 14:40:46 odroidhc4 kernel: [ 2.836066] sd 0:0:0:0: [sda] 1953525168 512-byte logical blocks: (1.00 TB/932 GiB)

Nov 22 14:40:46 odroidhc4 kernel: [ 2.836102] sd 0:0:0:0: [sda] Write Protect is off

Nov 22 14:40:46 odroidhc4 kernel: [ 2.836108] sd 0:0:0:0: [sda] Mode Sense: 00 3a 00 00

Nov 22 14:40:46 odroidhc4 kernel: [ 2.836159] sd 0:0:0:0: [sda] Write cache: enabled, read cache: enabled, doesn't support DPO or FUA

Nov 22 14:40:46 odroidhc4 kernel: [ 2.850916] ata1.00: Enabling discard_zeroes_data

Nov 22 14:40:46 odroidhc4 kernel: [ 2.872432] sda: sda1 sda9

Nov 22 14:40:46 odroidhc4 kernel: [ 2.872842] ata1.00: Enabling discard_zeroes_data

Nov 22 14:40:46 odroidhc4 kernel: [ 2.886688] sd 0:0:0:0: [sda] Attached SCSI disk

Nov 22 14:40:46 odroidhc4 kernel: [ 3.148738] ata2: SATA link down (SStatus 0 SControl 300)

 

The newer HC4 doesn't show this problem which leads me to the conclusion: their hardware (firmware?) must differ. Asked about it on the HardKernel Forum, didn't get a clear answer (yet?).

I don't know how to proceed i.e. where to look for further help. Any idea anyone?

I'm to old for an advanced GIT class 😉

 

Thanks for your time.

Chris

[SteeMan]

I don't have this hardware, so I'm just going to make some basic debugging suggestions.

 

Have you tried a 5.19 kernel from the most current release?  Or even a 6.x nightly edge build?

 

Do you have the same boot loader installed on both boards (uart output of each boot)

 

Also armbianmonitor output is always helpful.

[ChrisO]

Again, I have two HC4 boards, the newer one runs now: Linux 5.19.17-meson64

As for the older one: every kernel newer then 5.13.12 incl. 6.x I tried gives me the same problem: only one disk visible.

As for the boot loader, both boots from a SD card.

Yesterday I put Armbian_21.08.6_Odroidhc4_jammy_edge_5.15.4.img on the SD card, newer HC4 OK, older one bootet from the _same_ SD card: only one disk there.

If a NON self modifying software behaves differently on two pieces of HW, they must, IMHO, differ, no?

HardKernel means they don't. I guess I'm stuck 😞

[nettings]

@ChrisO, have you figured out a way to determine the HC4 board revision?

And I wonder: is there any part of the firmware that gets flashed into some persistent memory rather than uploaded at boot? Which might account for the difference?

 

[ChrisO]

There is no BIOS or such there, I couldn't find nothing like a serial number using lshw/hwinfo. On the boards themselves there some numbers.

Both boards are Rev 10.020200807, however another numbers I found on them differ, on the old one there is a E257384 2034, on the newer one E257384 2214.

Go figure out. From HK didn't get any clear answer, even if people there in the forum seam nice and helpful.

As for what you wonder about: don't think so, I might be wrong, though 🙂

[ChrisO]

Haven't got any help in HardKernel Forum: https://forum.odroid.com/viewtopic.php?f=205&t=45941&p=366587#p366587

As I said, I just want to find out what is the cause of the problem, the few $ for the now useless hc4 is not the problem. I don't have the skills to do necessary diffs on some kernel versions to find out what and when originated the problem.

If anybody feels like challenged, let me/us know 🙂