Debian Beagle-X15 integration status

BeagleBoard-X15
#1

Spent some time since with the beagle-x15 getting integration into
Debian, figured I'd let folks know how it was going.

If there's anything mentioned below (particularly upstreaming
linux/u-boot support and features) that you would be interested in
working on, please let me know!

So far, I've managed to forward-port the u-boot-ti patches for
beagle-x15 to u-boot 2016.01, but haven't yet uploaded.; will do that
soon. Would be good to get those patches integrated into mainline u-boot
at some point:

  https://anonscm.debian.org/cgit/collab-maint/u-boot.git/tree/debian/patches/am57xx

Some needed modules and kernel options are now enabled in the Debian
provided linux kernel. The biggest problem I've encountered is that
eSATA doesn't work (need to get that fixed in mainline), and
unfortunately the version is only present in Debian's experimental
repository so far:

  https://bugs.debian.org/807624
  https://bugs.debian.org/810048

I've also enabled support in flash-kernel for Beagle-x15, and it is
present in both Debian unstable and testing:

  https://anonscm.debian.org/cgit/d-i/flash-kernel.git/commit/?id=940eeb8c4938b7a25d2281e0a3ac0513b5cae3cb

Once the linux kernel 4.4.x and u-boot are uploaded to unstable, I can
start working more on debian-installer integration, but the hardest
parts are basically done so far. It's pretty likely the Beagle-X15 will
have decent out of the box support in Debian's future release!

I haven't yet added it to the reproducible-builds.org infrastructure, as
I haven't managed to get a fan set up for it yet, and I'm guessing that
would generate a fair bit of heat... (and also wanted to wait till
debian-installer integration is in a working state).

Many thanks to Robert C Nelson and all the other folks in #beagle who've
been available to answer questions and help troubleshoot!

live well,
  vagrant

#2

Here you go: :wink:

https://www.digikey.com/product-detail/en/X15FANKIT/X15FANKIT-ND/5822502

It is still weird on the cold -> warm phase... but, once warmed up
it'll go full tilt..

Regards,

#3

I don’t see any specs on that fan. I tried several fans and the one that is quiet and doesn’t heat up the fan power regulator is:

Digikey Part No: 563-1112-ND

I installed the fan to blow upwards so that it doesn’t adversely affect the temp sensor. The fan is only good for 1.2CFM, but it does seem enough when the processor is running at 100% CPU utilization.

Another fan I used was 259-1569-ND which is good for 3CFM, but with a noise level of 16dBA and a current of 0.43A, the noise is annoying and the fan power regulator gets very hot.

Regards,
John

#4

So we used: 563-1109-ND, also blowing upwards.

15dB vs 9dB of 563-1112-ND

Regards,

#5

On the X15 with a 5V fan, the power dissipated in the linear voltage regulator powering the fan

563-1112-ND = 0.48W
563-1109-ND = 0.72W

For a small voltage regulator, 0.72W dissipation is going to get pretty hot. From what Gerald said, he is changing to a 12V fan with a FET controller to eliminate this issue.

Regards,
John

#6

New design changed this area of the board. There is no longer a regulator there. But, yes, it will get warm on the current design.

Gerald

#7

From the TL5209 datasheet, the maximum power dissipation at ambient temperature of 25C is Pd (max) = (Tj - Ta)/Rja = (125 -25)/116.1 = 0.86W. At 0.72W, this is getting very close to the maximum rating of the regulator.

Regards,
John

#8

Okay we should be good in v4.5.0-rc3:

debian@BeagleBoard-X15:~$ dmesg | grep ata
[ 0.000000] CPU: PIPT / VIPT nonaliasing data cache, PIPT instruction cache
[ 0.000000] Memory policy: Data cache writealloc
[ 0.000000] Memory: 1979492K/2094080K available (12019K kernel
code, 1118K rwdata, 4044K rodata, 908K init, 920K bss, 49052K
reserved, 65536K cma-reserved, 1242112K highmem)
      .data : 0xc109c000 - 0xc11b390c (1119 kB)
[ 0.248419] omap_hwmod: l3_main_2 using broken dt data from ocp
[ 0.453739] omap4_sram_init:Unable to allocate sram needed to
handle errata I688
[ 0.453748] omap4_sram_init:Unable to get sram pool needed to
handle errata I688
[ 1.579076] libata version 3.00 loaded.
[ 3.193932] ahci 4a140000.sata: SSS flag set, parallel bus scan disabled
[ 3.200704] ahci 4a140000.sata: AHCI 0001.0300 32 slots 1 ports 3
Gbps 0x1 impl platform mode
[ 3.209288] ahci 4a140000.sata: flags: 64bit ncq sntf stag pm led
clo only pmp pio slum part ccc apst
[ 3.223007] ata1: SATA max UDMA/133 mmio [mem
0x4a140000-0x4a1410ff] port 0x100 irq 321
[ 3.541763] ata1: SATA link down (SStatus 0 SControl 300)
[ 4.433757] EXT4-fs (mmcblk0p1): mounted filesystem with ordered
data mode. Opts: (null)
[ 5.497885] systemd[1]: Starting Increase datagram queue length...
[ 5.712580] systemd[1]: Started Increase datagram queue length.
[ 7.713840] ata1: exception Emask 0x10 SAct 0x0 SErr 0x4050002
action 0xe frozen
[ 7.713845] ata1: irq_stat 0x00000040, connection status changed
[ 7.713851] ata1: SError: { RecovComm PHYRdyChg CommWake DevExch }
[ 7.713870] ata1: hard resetting link
[ 8.587328] ata1: SATA link up 3.0 Gbps (SStatus 123 SControl 300)
[ 8.599340] ata1.00: ATA-8: OCZ-AGILITY3, 2.15, max UDMA/133
[ 8.605061] ata1.00: 117231408 sectors, multi 16: LBA48 NCQ (depth 31/32)
[ 8.619339] ata1.00: configured for UDMA/133
[ 8.623697] ata1: EH complete
debian@BeagleBoard-X15:~$ lsblk
NAME MAJ:MIN RM SIZE RO TYPE MOUNTPOINT
sda 8:0 0 55.9G 0 disk
mmcblk1boot0 179:16 0 4M 1 disk
mmcblk1boot1 179:24 0 4M 1 disk
mmcblk0 179:0 0 59.5G 0 disk
└─mmcblk0p1 179:1 0 3.3G 0 part /
mmcblk1 179:8 0 3.6G 0 disk
debian@BeagleBoard-X15:~$ uname -r
4.5.0-rc3-armv7-lpae-x0

my "lpae" target contains only 6 patches away from pure mainline:
(none are touch sata)

https://github.com/RobertCNelson/armv7-lpae-multiplatform/blob/v4.5.x/patch.sh

Regards,

#9

So far, I've managed to forward-port the u-boot-ti patches for
beagle-x15 to u-boot 2016.01, but haven't yet uploaded.; will do that
soon. Would be good to get those patches integrated into mainline u-boot
at some point:

  https://anonscm.debian.org/cgit/collab-maint/u-boot.git/tree/debian/patches/am57xx

Just uploaded, so should be in the debian.org repositories in the next
day or so.

Some needed modules and kernel options are now enabled in the Debian
provided linux kernel. The biggest problem I've encountered is that
eSATA doesn't work (need to get that fixed in mainline), and
unfortunately the version is only present in Debian's experimental
repository so far:

Okay we should be good in v4.5.0-rc3:

...

[ 8.587328] ata1: SATA link up 3.0 Gbps (SStatus 123 SControl 300)
[ 8.599340] ata1.00: ATA-8: OCZ-AGILITY3, 2.15, max UDMA/133
[ 8.605061] ata1.00: 117231408 sectors, multi 16: LBA48 NCQ (depth 31/32)
[ 8.619339] ata1.00: configured for UDMA/133
[ 8.623697] ata1: EH complete
debian@BeagleBoard-X15:~$ lsblk
NAME MAJ:MIN RM SIZE RO TYPE MOUNTPOINT
sda 8:0 0 55.9G 0 disk
mmcblk1boot0 179:16 0 4M 1 disk
mmcblk1boot1 179:24 0 4M 1 disk
mmcblk0 179:0 0 59.5G 0 disk
└─mmcblk0p1 179:1 0 3.3G 0 part /
mmcblk1 179:8 0 3.6G 0 disk
debian@BeagleBoard-X15:~$ uname -r
4.5.0-rc3-armv7-lpae-x0

my "lpae" target contains only 6 patches away from pure mainline:
(none are touch sata)

Great news!

Would be good to figure out exactly what fixed eSATA upstream, so it can
be backported to 4.4.x, as 4.4 should hit debian sid/unstable in a week
or so. Then I can work on ensuring debian-installer support
works...

In the end, I suspect Debian will use a newer kernel for stretch (since
the freeze is scheduled for November), but exactly which is uncertain at
this point:

  https://lists.debian.org/debian-kernel/2016/02/threads.html#00067

live well,
  vagrant

#10

...

Some needed modules and kernel options are now enabled in the Debian
provided linux kernel. The biggest problem I've encountered is that
eSATA doesn't work (need to get that fixed in mainline), and
unfortunately the version is only present in Debian's experimental
repository so far:

Okay we should be good in v4.5.0-rc3:

...

[ 7.713840] ata1: exception Emask 0x10 SAct 0x0 SErr 0x4050002
action 0xe frozen
[ 7.713845] ata1: irq_stat 0x00000040, connection status changed
[ 7.713851] ata1: SError: { RecovComm PHYRdyChg CommWake DevExch }
[ 7.713870] ata1: hard resetting link
[ 8.587328] ata1: SATA link up 3.0 Gbps (SStatus 123 SControl 300)
[ 8.599340] ata1.00: ATA-8: OCZ-AGILITY3, 2.15, max UDMA/133
[ 8.605061] ata1.00: 117231408 sectors, multi 16: LBA48 NCQ (depth 31/32)
[ 8.619339] ata1.00: configured for UDMA/133
[ 8.623697] ata1: EH complete

No such luck for me.

Grabbed the config from 4.4 in debian experimental, copied to .config in
a 4.5-rc3 kernel tree, ran make olddefconfig on it, and make deb-pkg...

[ 1.572683] ahci 4a140000.sata: couldn't get PHY in node sata: -517
[ 1.654258] ahci 4a140000.sata: SSS flag set, parallel bus scan
disabled
[ 1.654298] ahci 4a140000.sata: AHCI 0001.0300 32 slots 1 ports 3
Gbps 0x1 impl platform mode
[ 1.654312] ahci 4a140000.sata: flags: 64bit ncq sntf stag pm led clo
only pmp pio slum part ccc apst
[ 1.664180] ata1: SATA max UDMA/133 mmio [mem 0x4a140000-0x4a1410ff]
port 0x100 irq 321
[ 11.663900] ata1: softreset failed (1st FIS failed)
[ 21.668102] ata1: softreset failed (1st FIS failed)
[ 56.672744] ata1: softreset failed (1st FIS failed)
[ 56.677881] ata1: limiting SATA link speed to 1.5 Gbps
[ 61.843807] ata1: softreset failed (device not ready)
[ 61.849119] ata1: reset failed, giving up

Makes me think I'm still missing some config option, or something is a
module that needs to be built-in, etc.

  https://www.aikidev.net/~vagrant/bbx15/config-4.5.0-rc3

live well,
  vagrant