Short Bump:
HPE released a new BIOS for the Gen10 Microserver. The changelog:
Zitat- Correct memory error of SMBIOS HCT testing
- Update for Red Hat Enterprise Linux 7.4 certification.
- Add ACPI table BERT and HEST to report memory error
looks like it will fix some of the dmesg acpi warnings / better linux support.
Also, I have some power consumption figures for my System:
- Microserver Gen10, 8GB RAM, X3216 CPU, headless, no keyboard/usb/etc.
- Debian Stretch, Mainline Kernel 4.13.x, OMV4.x
- Disks:
- Root/Cache: SSD Crucial_CT275MX3 (AMD AHCI)
- RAID1 1.0TB: 2xTOSHIBA MK1059GS (Marvell AHCI)
- RAID1 3.0TB: 2xTOSHIBA DT01ACA3 (Marvell AHCI)
- RAID1 0.5TB: Hitachi HTS54755, SAMSUNG HM500JI (vi Highpoint HPT RocketRaid 2300 SATA, taped inside case)
- idle, disks standby: ~20W
- idle, disks idle/active: ~32W
- measured at 230V AC via power-meter with impulse counter in a dedicated wall socket
The power figures might seem a bit high, but the DT01ACA3 disks are real power hogs - even in standby they use more than 1W each. The old HPT controller also does no power management whatsoever. But since all disks and the controller where unused and essentially free...
For reference: My internet/WiFi router with battery backup alone uses 12W more or less permanently.
Update:
- installed BIOS/UEFI ZA10320 from 2017-09-20, works fine, kept all settings.
- Update works by extracting EFI-Script and Updater from *.zip to /boot/efi, which is fs0: in the efi shell - no USB drive needed. Afterwards follow flash instructions
- installed NIC firmware 20.6.41
- download for RedHat; extract RPM to /
- run /usr/lib/x86_64-linux-gnu/firmware-nic-broadcom-2.18.15-1.1/setup
- version-bumped the NIC firmware in EFI quite a bit, but no changelog
- IOMMU
- manually enabled it in BIOS
- added kernel boot parameter iommu=pt (via /etc/default/grub and sudo update-grub
- you must modify both GRUB_CMDLINE_LINUX_DEFAULT and GRUB_CMDLINE_LINUX, otherwise the controller will not work in recovery mode
- also, this obviously does not work for OMV's SystemrescueCD-addon
- the marvell controller (still) works with all disks.
Currently, dmesg looks like this:
$ dmesg --level=warn,err
[ 0.000000] ACPI BIOS Warning (bug): Optional FADT field Pm2ControlBlock has valid Length but zero Address: 0x0000000000000000/0x1 (20170531/tbfadt-658)
[ 0.152777] PCCT header not found.
[ 0.152777] pmd_set_huge: Cannot satisfy [mem 0xf8000000-0xf8200000] with a huge-page mapping due to MTRR override.
[ 0.174800] [Firmware Bug]: HEST: Table contents overflow for hardware error source: 2.
[ 0.282204] pnp 00:04: disabling [mem 0xfeb00000-0xfeb00fff] because it overlaps 0000:00:01.0 BAR 5 [mem 0xfeb00000-0xfeb3ffff]
[ 1.274028] pci 0000:00:00.2: can't derive routing for PCI INT A
[ 1.274031] pci 0000:00:00.2: PCI INT A: not connected
[ 1.276556] PPR NX GT IA GA PC GA_vAPIC
[ 1.324824] usb: port power management may be unreliable
[ 1.371336] [Firmware Warn]: valid bits set for fields beyond structure
[ 10.195494] Error: Driver 'pcspkr' is already registered, aborting...
Alles anzeigen
I think this is quite an improvement because most warnings are gone now.