- Maintaining the DualDMA and BlueDMA codepaths is tedious
    - DualDMA did not receive any maintenance for a long time and is
    lacking features such as work queues and parallel reads/write
    - Performance was also worse overall
    - Only advantage is a somewhat smaller LUT footprint, but that seems
    unimportant on the large devices we use for PCIe right now

    - Currently only one buffer per direction is used as there is a
    problem with lost interrupts
    - The interrupts are lost when there is concurrent DMA and MSIx PCIe
    traffic

    - Otherwise the Platform can not match interrupt and PE
    - Only one interrupt per PE is allowed
    - Mapping could be moved into status core as suggested in #166

Squashed commit of the following:

commit 65214142
Author: Jaco Hofmann <hofmann@esa.tu-darmstadt.de>
Date:   Wed Jul 25 18:44:22 2018 +0200

    Remove trusty support

commit 76f0bf53
Author: Jaco Hofmann <hofmann@esa.tu-darmstadt.de>
Date:   Wed Jul 25 18:40:03 2018 +0200

    Fix GCC8 strncpy errors

commit f28aa33a
Author: Jaco Hofmann <hofmann@esa.tu-darmstadt.de>
Date:   Wed Jul 25 18:30:31 2018 +0200

    Add ubuntu trusty

commit 3ab103d5
Author: Jaco Hofmann <hofmann@esa.tu-darmstadt.de>
Date:   Wed Jul 25 18:26:00 2018 +0200

    Remove multiline commands

commit 3e638602
Author: Jaco Hofmann <hofmann@esa.tu-darmstadt.de>
Date:   Wed Jul 25 18:20:10 2018 +0200

    Use proper kernel headers

commit e17f2fe8
Author: Jaco Hofmann <hofmann@esa.tu-darmstadt.de>
Date:   Wed Jul 25 17:21:07 2018 +0200

    Fix syntax

commit 472c0796
Author: Jaco Hofmann <hofmann@esa.tu-darmstadt.de>
Date:   Wed Jul 25 17:18:28 2018 +0200

    Add kernel and tapasco-lib CI steps