PC Engines apu2 experiences
-
I just wanted to check, does anyone else have these messages about needing to accept intel licenses in their /var/log/dmesg.boot?
I updated the firmware yesterday to 4.6.1, now I’m not sure if it’s related.
Of course, after doing what it says, the errors/warnings go away.
-
-
What would the LED's do when those drivers are installed?
I suppose you would be able to configure the LED’s to show gateway status etc.
There are packages that you can install from the pfSense software repository (blinked and gwled) for this if you want to see what the options are. But apparently there’s some extra steps required to actually make it work.I haven’t bothered, I never look at it anyway :P
-
Well, I decided to have a look at these LED's. Installed the driver, LED's work.
During boot, the LED's all dance, and if I install the gwled package, they show the gateway status. Wonderful :)Only, I noticed that gwled has a service, but doesn't start… it's annoying. Is this normal? See attachment. (On a sidenote, does anyone else 2x haproxy services?).
Secondly, I noticed the following in the my /var/log/dmesg.boot:
ada0: <samsung 850="" ssd="" evo="" msata="" 250gb="" emt41b6q="">ACS-2 ATA SATA 3.x device
ada0: 600.000MB/s transfers (SATA 3.x, UDMA6, PIO 512bytes)
ada0: Command Queueing enabled
ada0: 238475MB (488397168 512 byte sectors)
ada0: quirks=0x3<4K,NCQ_TRIM_BROKEN>What does this 4K,NCQ_TRIM_BROKEN mean exactly in normal English?
I also note that they have a solution below, but again, not quite sure what the difference is between the two….
if you're on 11.1-RELEASE or 11-RELEASE, you should add below on /boot/loader.conf.
If no other quirks is required:
kern.cam.ada.0.quirks="0x0"If you need 4k quirks but want to drop NCQ_TRIM one:
kern.cam.ada.0.quirks="0x1"*4k one is bit0 (0x1), and NCQ_TRIM one is bit1 (0x2).
The example above assumes the affected drive is recognized as ada0.
You should change "ada.0" to whatever appropreate.
</samsung> -
What does this 4K,NCQ_TRIM_BROKEN mean exactly in normal English?
4k means "TRIM only works on 4096 byte requests that are 4096 byte aligned".
NCQ_TRIM means TRIM doesn't work if you send it as a queued request. For some drives, the SSD stops working with first trim request is issued from the filesystem (UFS or ZFS).
There is a bug that is fixed in 11.1-RELEASE. These quirks keep your drive working.
If you're running pfSense 2.4.x you can re-enable 4K sectors and TRIM by clearing the quirks with:
kern.cam.ada.0.quirks="0x0"
in /boot/loader.conf
or, if you need 4k quirks but want to drop NCQ_TRIM one:
kern.cam.ada.0.quirks="0x1"4k one is bit0 (0x1), and NCQ_TRIM one is bit1 (0x2).
Or you can wait for pfSense 2.4.3, which will contain a software fix.
-
On a sidenote, does anyone else have 2x haproxy services?
I havn't seen that before..you should only have the lowercase 'haproxy' service.. Probably need to edit the config.xml to remove the wrong service tag.. (backup>edit>restore)or the more tricky:(edit /conf/config.xml,delete /tmp/config.cache) just make sure to keep the xml format valid..
-
I am using an APU2C4 with BIOS version 4.0.7.
It looks like PC Engines no longer maintains their web pages under
https://www.pcengines.ch/apu2c4.htmI am unable to find any binaries for latest versions 4.0.x and 4.6.x and appreciate any hints on that.
All I can find are related release infos/changelogs
https://github.com/pcengines/release_manifests/blob/coreboot-4.6.x/CHANGELOG.mdand source codes of coreboot
https://github.com/pcengines/coreboot/releasesPeter
-
I see some ROMs on the github:
https://github.com/pcengines/apu2-documentationAPU2 ROM should be there in legacy and mainline mode
-
I see some ROMs on the github:
https://github.com/pcengines/apu2-documentationAPU2 ROM should be there in legacy and mainline mode
Yeah, thanks a lot. Nevertheless, these are not the latest versions. I would have never searched for binaries in the documentation folder ;-).
-
Wow, I'm running 4.07. Any reason to go higher for pfSense? What's the recommended BIOS?
-
I thought 4.07 is the latest?
-
I thought 4.07 is the latest?
No, development is going on. Unfortunately, version schema is confusing and PC Engines does not update their corresponding web page. Moreover, binaries are at least for me, difficult to find.
What I have understood so far: There are two actively developed branches: 4.0.x denoted as "legacy" and a "mainline" 4.5.x/4.6.x. Latest versions are 4.0.14 and 4.6.6, respectively. Latest binary downloads for APU2 are available as 4.0.11 and 4.6.1.
One thing that I have just become aware of: coreboot determines the version number of the APU2 BIOS ROM but the APU2 ROM consists of several other components with its own version numbers like e.g. seabios and ipxe.
-
I guess I should have read outside of the PCEngines website! Thanks for that info.
OK, so has everyone successfully run 4.6.6 or should I just go to 4.0.11?
-
I thought 4.07 is the latest?
No, development is going on. Unfortunately, version schema is confusing and PC Engines does not update their corresponding web page. Moreover, binaries are at least for me, difficult to find.
What I have understood so far: There are two actively developed branches: 4.0.x denoted as "legacy" and a "mainline" 4.5.x/4.6.x. Latest versions are 4.0.14 and 4.6.6, respectively. Latest binary downloads for APU2 are available as 4.0.11 and 4.6.1.
One thing that I have just become aware of: coreboot determines the version number of the APU2 BIOS ROM but the APU2 ROM consists of several other components with its own version numbers like e.g. seabios and ipxe.
Oh I see. So everything is on github? I remember choosing between 4.0.x and 4.5.x before and many people we're having problems with 4.5.x for some reason. And this is why I chose to go with 4.0.7. Did this change now? Would it be recommended to go with the mainline this time?
I guess I should have read outside of the PCEngines website! Thanks for that info.
OK, so has everyone successfully run 4.6.6 or should I just go to 4.0.11?
Me too! All along I thought their website was updated.
-
I just went to 4.0.11 and it's working fine. I didn't see anything listed in the bios's after 4.0.11 that was relevant to APU2 boards.
I also noticed that PC Engines recommends the 4.0x track here:
http://pcengines.ch/howto.htm#bios -
FYI, anyone updating to the 4.5.x or 4.6.x mainline firmware (https://github.com/pcengines/apu2-documentation), you need to edit the /boot/loader.conf and add the following:
hint.ahci.0.msi="0"
Otherwise it reboots every 4-5 hours.
The rest of the items that they mention here were all already added on my system by default.
I'm now running 4.6.1 without any problems.
-
Mine still says PC Engines APU2 after the update to 2.3.4.
Yeah, mine did too. I more meant after the latest firmware update.
Can you post the output of
/bin/kenv -q smbios.system.product /bin/kenv -q smbios.system.maker
with the 4.0.7 FW.
FYI, This has changed again…. below the output from firmware 4.6.1:
/bin/kenv -q smbios.system.product -> PC Engines apu2
/bin/kenv -q smbios.system.maker -> PC Engines -
On a sidenote, does anyone else have 2x haproxy services?
I havn't seen that before..you should only have the lowercase 'haproxy' service.. Probably need to edit the config.xml to remove the wrong service tag.. (backup>edit>restore)or the more tricky:(edit /conf/config.xml,delete /tmp/config.cache) just make sure to keep the xml format valid..
Thanks! :)
-
Question, should I see the serial number of my PC Engines on the pfSense dashboard?
Reason I ask is because one of the PC Engines firmwares made the serial number show random characters. However, in doing so, the serial number field all of a sudden showed up on the dashboard
However, once I got the next firmware, the field was gone again from the dashboard.
The below command gives an actual output:
/bin/kenv -q smbios.system.serial
However, it’s not shown on the dashboard.
Is this intentional? Or should it be shown?
-
It's not expected to show in 2.4.3p1.
The input validation on that field was improved since it was first introduced. It should only appear now on devices that expose a real serial number via ACPI in the correct field.
I think the new forum code may have cut-off your kenv output.
Looking on an APU here it does seem to be present so perhaps the validation there could be tweaked.
It's not an issue with your board though, that's the expected behaviour currently.
Steve