[siemens/jailhouse] 504d35: README: Add ARM requirements description

13 views
Skip to first unread message

GitHub

unread,
Feb 23, 2015, 10:56:20 AM2/23/15
to jailho...@googlegroups.com
Branch: refs/heads/coverity_scan
Home: https://github.com/siemens/jailhouse
Commit: 504d35d1755e9b629065b980f29ec8e88432d1fb
https://github.com/siemens/jailhouse/commit/504d35d1755e9b629065b980f29ec8e88432d1fb
Author: Jan Kiszka <jan.k...@siemens.com>
Date: 2015-02-09 (Mon, 09 Feb 2015)

Changed paths:
M README.md

Log Message:
-----------
README: Add ARM requirements description

Signed-off-by: Jan Kiszka <jan.k...@siemens.com>


Commit: a7984ce57d5cec8ee7b9c0ae7dd8f30fe9e71a2e
https://github.com/siemens/jailhouse/commit/a7984ce57d5cec8ee7b9c0ae7dd8f30fe9e71a2e
Author: Jan Kiszka <jan.k...@siemens.com>
Date: 2015-02-12 (Thu, 12 Feb 2015)

Changed paths:
M ci/jailhouse-config-banana-pi.h
M ci/jailhouse-config-vexpress.h
M hypervisor/arch/arm/Makefile
M hypervisor/arch/arm/include/asm/platform.h
M inmates/demos/arm/gic-demo.c
M inmates/lib/arm/Makefile.lib

Log Message:
-----------
arm: Rename CONFIG_ARCH_* to CONFIG_MACH_*

Based on proposal by Benedikt Spranger: CONFIG_ARCH_SUN7I and
CONFIG_ARCH_VEXPRESS select a machine type, not an architecture. Rename
them to clarify this.

Signed-off-by: Jan Kiszka <jan.k...@siemens.com>


Commit: df1fa5bb6debfbf6d68a855215652af5726214ef
https://github.com/siemens/jailhouse/commit/df1fa5bb6debfbf6d68a855215652af5726214ef
Author: Jan Kiszka <jan.k...@siemens.com>
Date: 2015-02-12 (Thu, 12 Feb 2015)

Changed paths:
M hypervisor/arch/arm/include/asm/psci.h
M hypervisor/arch/arm/psci.c

Log Message:
-----------
arm: Add PSCI v0.1 support as provided by U-Boot

Upstream U-Boot will continue to offer only a PSCI v0.1 interface. To
make us independent of patches, add the IDs for CPU_OFF and CPU_ON to
the dispatcher. Those functions are practically compatible between 0.1
and 0.2.

Signed-off-by: Jan Kiszka <jan.k...@siemens.com>


Commit: 1443281b5d10167f0c4878ff43b24985d1e8a6a4
https://github.com/siemens/jailhouse/commit/1443281b5d10167f0c4878ff43b24985d1e8a6a4
Author: Jan Kiszka <jan.k...@siemens.com>
Date: 2015-02-12 (Thu, 12 Feb 2015)

Changed paths:
M README.md

Log Message:
-----------
README: Update U-Boot requirement for Banana Pi

We no longer depend on PSCI v0.2 patches, just on a recent upstream
version.

Signed-off-by: Jan Kiszka <jan.k...@siemens.com>


Commit: 5a639deb32f6d452437a305d1e2bae1df44b2c12
https://github.com/siemens/jailhouse/commit/5a639deb32f6d452437a305d1e2bae1df44b2c12
Author: Jan Kiszka <jan.k...@siemens.com>
Date: 2015-02-18 (Wed, 18 Feb 2015)

Changed paths:
M TODO.md

Log Message:
-----------
TODO: Add items about MMIO dispatching enhancements

Signed-off-by: Jan Kiszka <jan.k...@siemens.com>


Commit: f94e9e685980e26eebd19010801d3f3f594b508b
https://github.com/siemens/jailhouse/commit/f94e9e685980e26eebd19010801d3f3f594b508b
Author: Jan Kiszka <jan.k...@siemens.com>
Date: 2015-02-23 (Mon, 23 Feb 2015)

Changed paths:
M driver.c

Log Message:
-----------
driver: Account for removal of cpumask_scnprintf

4.0 provides cpumask formatting via the %*pb format specifier.
cpumask_scnprintf was removed. So we need to implement both variants
for staying compatible with older kernels.

Signed-off-by: Jan Kiszka <jan.k...@siemens.com>


Commit: 36e4a857269dc1dbb9912582e1cfbc4b0a88cacb
https://github.com/siemens/jailhouse/commit/36e4a857269dc1dbb9912582e1cfbc4b0a88cacb
Author: Jan Kiszka <jan.k...@siemens.com>
Date: 2015-02-23 (Mon, 23 Feb 2015)

Changed paths:
M driver.c

Log Message:
-----------
driver: Update CR4 on x86 after hypervisor enabling/disabling

Linux gained CR4 shadowing in 4.0: it now keeps a software copy of the
mm's CR4 state and only write updates on changes. But this also means it
will miss changes to this register when enabling or disabling Jailhouse.
That will cause sporadic access violations or false reporting of
virtualization-busy CPUs on Jailhouse enable.

Fix it by re-initializing the shadow on x86 after returning from the
hypervisor entry or the disable command. We are still protected from
interrupts at that point, so nothing can leak to other tasks.

Signed-off-by: Jan Kiszka <jan.k...@siemens.com>


Commit: e060138dd255b3e7860b13ac8757c5bb19ef7772
https://github.com/siemens/jailhouse/commit/e060138dd255b3e7860b13ac8757c5bb19ef7772
Author: Jan Kiszka <jan.k...@siemens.com>
Date: 2015-02-23 (Mon, 23 Feb 2015)

Changed paths:
A configs/jetson-tk1-demo.c
A configs/jetson-tk1.c
M hypervisor/arch/arm/Makefile
M hypervisor/arch/arm/include/asm/platform.h
A hypervisor/arch/arm/include/asm/uart-tegra.h
A hypervisor/arch/arm/smp-tegra124.c
A hypervisor/arch/arm/uart-tegra.c
M inmates/lib/arm/Makefile.lib
A inmates/lib/arm/include/mach-tegra124/mach/gic_v2.h
A inmates/lib/arm/include/mach-tegra124/mach/timer.h
A inmates/lib/arm/include/mach-tegra124/mach/uart.h
A inmates/lib/arm/uart-tegra.c

Log Message:
-----------
arm: Add support for Jetson TK1 board

The Jetson TK1 board uses the NVIDIA K1 processor, namely the Tegra124.
The CPU has 4 Cortex-A15 cores and a low-power companion core which we
do not support, though. This SoC apparently contains all virtualization
features we need, specifically full SMMU coverage of I/O devices.

For now, board support is similar to the Banana Pi: We replicate UART
and SMP/PSCI features and add some configs (only one for both demos as
only one UART is accessible).

Open issues, besides adding SMMU support, are managing the access to the
so-called legacy Interrupt controller of the K1 (an additional stage
between devices and GIC) and sub-page access control to various devices
(same issue as with the Allwinner A20). Also, it becomes clearer than
ever that we urgently need to refactor the UART layer as well as
SMP/PSCI support to reduce duplications.

Signed-off-by: Jan Kiszka <jan.k...@siemens.com>


Commit: 0de30494b8967f8e24d012150ff0afa8fdede7c6
https://github.com/siemens/jailhouse/commit/0de30494b8967f8e24d012150ff0afa8fdede7c6
Author: Jan Kiszka <jan.k...@siemens.com>
Date: 2015-02-23 (Mon, 23 Feb 2015)

Changed paths:
M hypervisor/arch/arm/control.c

Log Message:
-----------
arm: Power off non-root CPUs before shutting down the hypervisor

When handing back a CPU previously used by a non-root cell directly to
Linux during hypervisor shutdown, we have to power it off when Linux is
using PSCI. Otherwise, the CPU_ON command issued by Linux later on to
gain control over the CPU again may not work as expected.

All our supported boards except for the Versatile Express come with PSCI
support. Try both v0.2 and the v0.1 encoding of U-Boot for CPU_OFF. At
least one of them must work.

The GICC is not reset by the power-down, thus we have to reset it shut
it down explicitly.

Signed-off-by: Jan Kiszka <jan.k...@siemens.com>


Compare: https://github.com/siemens/jailhouse/compare/dd3ef73c56c3...0de30494b896
Reply all
Reply to author
Forward
0 new messages