USB data write failure

[Autoboot]

Hello,

running autoboot with coreboot as of commit
a3e41c089602c58409e8dfd4aceecbdd7d4f4a5b being the latest not breaking
USB and PCI-E on my ThinkPad X201.

Writing data to USB storage appears bugged.

I tested this using different disks on three different machines:

- ThinkPad X201, autoboot, latest Parabola
    - ThinkPad T43, stock firmware, Debian 8
    - Asus EeePC 901, stock firmware, latest Arch

In all combinations, I created a btrfs filesystem inside a LUKS
container on the whole disk, using both USB and internal SATA (except
for the 901):

$ crytpsetup luksFormat /dev/sdb
    $ cryptsetup open /dev/sdb test
    $ mkfs.btrfs /dev/mapper/test
    $ mount -o compress=lzo /dev/mapper/test /mnt
    $ dd if=/dev/urandom of=/mnt/20M bs=1M count=20 iflag=fullblock
    $ umount /mnt
    $ btrfs check /dev/mapper/test

Only when using my X201 *with coreboot and USB*, this procedure resulted
in the following error:

$ btrfs check /dev/mapper/test
    Checking filesystem on /dev/mapper/test
    UUID: 33c88ec2-2c4e-4844-905f-c0fbebc00a1f
    checking extents
    bad metadata [29605888, 29622272) crossing stripe boundary
    bad metadata [29736960, 29753344) crossing stripe boundary
    Errors found in extent allocation tree or chunk allocation
    checking free space cache
    checking fs roots
    checking csums
    checking root refs
    found 21135362 bytes used err is 0
    total csum bytes: 20480
    total tree bytes: 163840
    total fs tree bytes: 32768
    total extent tree bytes: 16384
    btree space waste bytes: 129312
    file data blocks allocated: 21495808
     referenced 21495808
    btrfs-progs v4.1.2-32-g6054018

Using the same disk with my internal SATA port and testing from a live
USB stick showed no errors at all.

Every other setup worked fine and didn't show any error either. I also
tested this with ext4, but as expected, with ext4 everything seems to be
ok since it doesn't even provide this type of error detection.

This clearly seems to be a coreboot related issue. I used to run the
latest coreboot with autoboot, but booting from USB somehow didn't work
and someone in #coreboot told me USB an PCI-E was broken as of commit
456f495d, so I reset to the above mentioned. USB boot worked again, but
then I noticed this faulty behaviour.

I attached the coreboot config used to build my current autoboot image with.

Hopefully this report is useful to find the bug and eventually fix it asap.