23.01 upgrade no longer sees ada1

ohmantics

I have two M.2 drives in a StarTech dual SATA adapter. 22.05 happily sees both and they act just fine as a ZFS mirror.

23.01 has some errors and never sees ada1, thus leaving the pool degraded.

ns8250: UART FCR is broken
...
(aprobe0:ahcich4:0:0:0): SET_MULTI. ACB: c6 00 00 00 00 40 00 00 00 00 01 00
(aprobe0:ahcich4:0:0:0): CAM status: ATA Status Error
(aprobe0:ahcich4:0:0:0): ATA status: 51 (DRDY SERV ERR), error: 04 (ABRT )
(aprobe0:ahcich4:0:0:0): RES: 51 04 00 00 00 00 00 00 00 00 00
(aprobe0:ahcich4:0:0:0): Retrying command, 0 more tries remain
(aprobe0:ahcich4:0:0:0): SET_MULTI. ACB: c6 00 00 00 00 40 00 00 00 00 01 00
(aprobe0:ahcich4:0:0:0): CAM status: ATA Status Error
(aprobe0:ahcich4:0:0:0): ATA status: 51 (DRDY SERV ERR), error: 04 (ABRT )
(aprobe0:ahcich4:0:0:0): RES: 51 04 00 00 00 00 00 00 00 00 00
(aprobe0:ahcich4:0:0:0): Error 5, Retries exhausted
...
ses0 at ahciem0 bus 0 scbus1 target 0 lun 0
ses0: <AHCI SGPIO Enclosure 2.00 0001> SEMB S-E-S 2.00 device
ses0: SEMB SES Device
ada0 at ahcich4 bus 0 scbus0 target 0 lun 0
ada0: <WDC WDS500G1R0B-xxx 411000WR> ACS-4 ATA SATA 3.x device
ada0: Serial Number xxx
ada0: 300.000MB/s transfers (SATA 2.x, UDMA6, PIO 512bytes)
ada0: Command Queueing enabled
ada0: 476940MB (976773168 512 byte sectors)
ses0: pass0,ada0 in 'Slot 04', SATA Slot: scbus0 target 0

In addition to seeing ada1, 22.05 was happier with the adapter:

ada0: 600.000MB/s transfers (SATA 3.x, UDMA6, PIO 512bytes)

Reverting to prior boot environment works fine.

Presumably this is an upstream 14-CURRENT issue, but we should start here.

ohmantics

I was able to reproduce this with a stock 14-CURRENT on different hardware.

https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=270459 is tracking it upstream.

tedquade

@ohmantics Thanks for posting this. I was about to head down the same road with a similar dual drive adapter.

Ted Quade

ohmantics

The fix for this is in review upstream at https://reviews.freebsd.org/D39572.

I'd love to apply that patch to 23.01 and rebuild the kernel to test this out, but "instructions unclear."

I expected to set KERNCONF=pfSense, but the commit that seems to correspond to 23.01 doesn't have sys/amd64/conf/pfSense at that point in history.

Setting up the whole poudriere config for the build.sh script family seems... overkill for this task.

Any help welcome.

stephenw10

Can you test it in a FreeBSD 14 snap?

It's unlikely to make it into 23.05 at this point though.

ohmantics

@stephenw10 I already know this fix works in plain FreeBSD 14-CURRENT; I tested that already. I'd like to upgrade my pfSense box with the fix sooner than whenever it'll be integrated, which perhaps would be late Summer or early Fall if it will miss 23.05.

I wonder if the issues I encountered in building are because of something in the plus branches I don't have access to. uname for 23.01 is (FreeBSD 14.0-CURRENT #0 plus-RELENG_23_01-n256037-6e914874a5e: Fri Feb 10 20:30:29 UTC 2023) but the expected KERNCONF file is absent. Certainly plus isn't running off GENERIC, so I'm at a loss as to what to build. I've browsed through the build.sh family of scripts, but don't see what I should be using.

stephenw10

Yeah, I wouldn't even attempt to build a kernel for something like that.

Let me see what what we can do here. Might be possible to cherry pick it.

tedquade

@stephenw10 Would be great if you could.

Ted Quade

ohmantics

@stephenw10 I've confirmed that the snapshot pfSense-CE-2.7.0-DEVELOPMENT-amd64-20230418-0600.iso worked. I did notice that there were a few small follow-up commits upstream but they don't seem critical.

I assume this means that when the 23.05 Plus build happens, these changes will be included?

stephenw10

Yes that same fix is in the current 23.05 code.