"Comm ABI mismatch" needs a bit more space.
Signed-off-by: Ralf Ramsauer
---
tools/jailhouse.c | 4 ++--
1 file changed, 2 insertions(+), 2 deletions(-)
diff --git a/tools/jailhouse.c b/tools/jailhouse.c
index 74029bac..58b02e23 100644
--- a/tools/jailhouse.c
Signed-off-by: Ralf Ramsauer
Signed-off-by: Jan Kiszka
---
inmates/lib/arm/include/arch/inmate.h | 5 +
inmates/lib/arm64/include/arch/inmate.h | 5 +
inmates/lib/x86/inmate.h| 5 +
3 files changed, 15
Signed-off-by: Ralf Ramsauer
Signed-off-by: Jan Kiszka
---
inmates/demos/arm/gic-demo.c | 3 +--
inmates/demos/x86/ioapic-demo.c | 3 +--
inmates/demos/x86/ivshmem-demo.c | 2 +-
inmates/demos/x86/pci-demo.c | 3 +--
4 files
A cell may set this state if its internal ABI of the communication
region is incompatibel to the one provided by the hypervisor.
Signed-off-by: Ralf Ramsauer
Signed-off-by: Jan Kiszka
---
driver/sysfs.c| 2 ++
We will never return from c_entry. Make it clear on x86 by moving the stop
label to the top of the assembly code.
Signed-off-by: Ralf Ramsauer
Signed-off-by: Jan Kiszka
---
inmates/lib/x86/header-32.S | 11 +--
Hi,
ABI changes of the communication region can not be detected by inmates at the
moment. Let's introduce some magic and revision field at the beginning of the
page to signal ABI changes.
This will help us if we change the ABI at a later point in time.
Ralf
since v4:
- Move the cell state
Don't jump directly to inmate_main from assembly, jump to c_entry.
At the moment, c_entry does nothing else than jumping to inmate_main and
stop the machine if inmate_main returns.
Later we can use c_entry to do platform independant setup/checks.
Signed-off-by: Ralf Ramsauer
comm_region will become mandatory for all of our inmates.
On x86, we already use the communication region to pass certain platform
specific information to guests. So let's do this on all ARM
architectures as well. This patch is a preparation for that.
Let's use guest paddr 0x8000 (2GiB) for
Or fail by setting the cell's state and halt.
Signed-off-by: Ralf Ramsauer
Signed-off-by: Jan Kiszka
---
inmates/lib/setup.c | 10 +-
1 file changed, 9 insertions(+), 1 deletion(-)
diff --git a/inmates/lib/setup.c
Freeze the fields "Signature", "ABI Revision" and "Cell State". Future
changes of the ABI must not modify their offsets.
Signed-off-by: Ralf Ramsauer
[Jan: Clarify value and semantic of ABI revision]
[Ralf: amend commit message]
---
Signed-off-by: Ralf Ramsauer
Signed-off-by: Jan Kiszka
---
inmates/lib/inmate_common.h | 1 +
inmates/lib/string.c| 10 ++
2 files changed, 11 insertions(+)
diff --git a/inmates/lib/inmate_common.h
We will enter this function if we hit unrecoverable state.
Signed-off-by: Ralf Ramsauer
Signed-off-by: Jan Kiszka
---
inmates/lib/arm-common/include/inmate.h | 6 ++
inmates/lib/x86/inmate.h| 6 ++
2 files
Changes to the ABI of the communication region might break inmates
without noticing. Add the magic JHCOMM together with the ABI version as
a preamble of the structure.
The location signature, revision and cell_state are frozen from now on: their
location will be stable in future. Put those three
On 2018-05-07 21:47, Ralf Ramsauer wrote:
> On 05/07/2018 07:50 PM, Jan Kiszka wrote:
>> On 2018-05-07 18:40, Ralf Ramsauer wrote:
>>> Signed-off-by: Ralf Ramsauer
>>> ---
>>> Documentation/hypervisor-interfaces.txt | 7 ++-
>>> 1 file changed, 6
On 05/07/2018 07:50 PM, Jan Kiszka wrote:
> On 2018-05-07 18:40, Ralf Ramsauer wrote:
>> Signed-off-by: Ralf Ramsauer
>> ---
>> Documentation/hypervisor-interfaces.txt | 7 ++-
>> 1 file changed, 6 insertions(+), 1 deletion(-)
>>
>> diff --git
Branch: refs/heads/next
Home: https://github.com/siemens/jailhouse
Commit: e7780fcfd08893775d60352adbc6f56d281d5ea3
https://github.com/siemens/jailhouse/commit/e7780fcfd08893775d60352adbc6f56d281d5ea3
Author: Ralf Ramsauer
Date: 2018-05-07
On 2018-05-07 18:40, Ralf Ramsauer wrote:
> Signed-off-by: Ralf Ramsauer
> ---
> Documentation/hypervisor-interfaces.txt | 7 ++-
> 1 file changed, 6 insertions(+), 1 deletion(-)
>
> diff --git a/Documentation/hypervisor-interfaces.txt
>
Am Mon, 7 May 2018 15:54:35 +0200
schrieb "[ext] Jan Kiszka" :
> On 2018-05-07 15:48, Assila Belhouichet wrote:
> > Hi all,
> >
> > i reproduced the patchs that Henning provided for testing the
> > intercell communication on the bananapi, everything was okay until
> > i
Signed-off-by: Ralf Ramsauer
---
Documentation/hypervisor-interfaces.txt | 7 ++-
1 file changed, 6 insertions(+), 1 deletion(-)
diff --git a/Documentation/hypervisor-interfaces.txt
b/Documentation/hypervisor-interfaces.txt
index 8dd58750..2dcb7d74 100644
On 05/07/2018 06:29 PM, Jan Kiszka wrote:
> On 2018-05-07 18:08, Ralf Ramsauer wrote:
>> Signed-off-by: Ralf Ramsauer
>> ---
>> Documentation/hypervisor-interfaces.txt | 17 +++--
>> 1 file changed, 11 insertions(+), 6 deletions(-)
>>
>> diff --git
On 2018-05-07 18:08, Ralf Ramsauer wrote:
> Hi,
>
> ABI changes of the communication region can not be detected by inmates at the
> moment. Let's introduce some magic and revision field at the beginning of the
> page to signal ABI changes.
>
> This will help us if we change the ABI at a later
On 2018-05-07 18:08, Ralf Ramsauer wrote:
> Signed-off-by: Ralf Ramsauer
> ---
> Documentation/hypervisor-interfaces.txt | 17 +++--
> 1 file changed, 11 insertions(+), 6 deletions(-)
>
> diff --git a/Documentation/hypervisor-interfaces.txt
>
Signed-off-by: Ralf Ramsauer
---
inmates/demos/arm/gic-demo.c | 3 +--
inmates/demos/x86/ioapic-demo.c | 3 +--
inmates/demos/x86/ivshmem-demo.c | 2 +-
inmates/demos/x86/pci-demo.c | 3 +--
4 files changed, 4 insertions(+), 7 deletions(-)
diff --git
stop() disables all interrupts and enters halt().
Signed-off-by: Ralf Ramsauer
---
inmates/lib/inmate_common.h | 6 ++
1 file changed, 6 insertions(+)
diff --git a/inmates/lib/inmate_common.h b/inmates/lib/inmate_common.h
index 6c779b53..dfe33347 100644
---
We will never return from c_entry. Make it clear on x86 by moving the stop
label to the top of the assembly code.
Signed-off-by: Ralf Ramsauer
---
inmates/lib/x86/header-32.S | 11 +--
inmates/lib/x86/header.S| 11 +--
2 files changed, 10
Hi,
ABI changes of the communication region can not be detected by inmates at the
moment. Let's introduce some magic and revision field at the beginning of the
page to signal ABI changes.
This will help us if we change the ABI at a later point in time.
Ralf
since v2:
- fix some indentation
Don't jump directly to inmate_main from assembly, jump to c_entry.
At the moment, c_entry does nothing else than jumping to inmate_main and
stop the machine if inmate_main returns.
Later we can use c_entry to do platform independant setup/checks.
Signed-off-by: Ralf Ramsauer
We will enter this function if we hit unrecoverable state.
Signed-off-by: Ralf Ramsauer
---
inmates/lib/arm-common/include/inmate.h | 6 ++
inmates/lib/x86/inmate.h| 6 ++
2 files changed, 12 insertions(+)
diff --git
Changes to the ABI of the communication region might break inmates
without noticing. Add the magic JHCOMM together with the ABI version as
a preamble of the structure.
Signed-off-by: Ralf Ramsauer
---
hypervisor/control.c | 7 ++-
A cell may set this state if its internal ABI of the communication
region is incompatibel to the one provided by the hypervisor.
Signed-off-by: Ralf Ramsauer
---
driver/sysfs.c| 2 ++
hypervisor/control.c | 1 +
Signed-off-by: Ralf Ramsauer
---
inmates/lib/inmate_common.h | 1 +
inmates/lib/string.c| 10 ++
2 files changed, 11 insertions(+)
diff --git a/inmates/lib/inmate_common.h b/inmates/lib/inmate_common.h
index dfe33347..ba39550e 100644
---
Signed-off-by: Ralf Ramsauer
---
Documentation/hypervisor-interfaces.txt | 17 +++--
1 file changed, 11 insertions(+), 6 deletions(-)
diff --git a/Documentation/hypervisor-interfaces.txt
b/Documentation/hypervisor-interfaces.txt
index
Signed-off-by: Ralf Ramsauer
---
inmates/lib/arm/include/arch/inmate.h | 5 +
inmates/lib/arm64/include/arch/inmate.h | 5 +
inmates/lib/x86/inmate.h| 5 +
3 files changed, 15 insertions(+)
diff --git
comm_region will become mandatory for all of our inmates.
On x86, we already use the communication region to pass certain platform
specific information to guests. So let's do this on all ARM
architectures as well. This patch is a preparation for that.
Let's use guest paddr 0x8000 (2GiB) for
On 2018-05-07 17:24, Ralf Ramsauer wrote:
> We will never return from c_entry. Make it clear on x86 by moving the stop
> label to the top of the assembly code.
>
> Signed-off-by: Ralf Ramsauer
> ---
> inmates/lib/x86/header-32.S | 9 -
>
On 2018-05-07 17:24, Ralf Ramsauer wrote:
> Or fail by setting the cell's state and halt.
>
> Signed-off-by: Ralf Ramsauer
> ---
> inmates/lib/setup.c | 11 ++-
> 1 file changed, 10 insertions(+), 1 deletion(-)
>
> diff --git a/inmates/lib/setup.c
On 2018-05-07 17:24, Ralf Ramsauer wrote:
> Changes to the ABI of the communication region might break inmates
> without noticing. Add the magic JHCOMM together with the ABI version as
> a preamble of the structure.
>
> Signed-off-by: Ralf Ramsauer
> ---
>
Hi,
ABI changes of the communication region can not be detected by inmates at the
moment. Let's introduce some magic and revision field at the beginning of the
page to signal ABI changes.
This will help us if we change the ABI at a later point in time.
Ralf
since v1:
- minor stylistic
stop() disables all interrupts and enters halt().
Signed-off-by: Ralf Ramsauer
---
inmates/lib/inmate_common.h | 6 ++
1 file changed, 6 insertions(+)
diff --git a/inmates/lib/inmate_common.h b/inmates/lib/inmate_common.h
index 6c779b53..dfe33347 100644
---
Signed-off-by: Ralf Ramsauer
---
inmates/lib/arm/include/arch/inmate.h | 5 +
inmates/lib/arm64/include/arch/inmate.h | 5 +
inmates/lib/x86/inmate.h| 5 +
3 files changed, 15 insertions(+)
diff --git
Don't jump directly to inmate_main from assembly, jump to c_entry.
At the moment, c_entry does nothing else than jumping to inmate_main and
stop the machine if inmate_main returns.
Later we can use c_entry to do platform independant setup/checks.
Signed-off-by: Ralf Ramsauer
Signed-off-by: Ralf Ramsauer
---
inmates/demos/arm/gic-demo.c | 3 +--
inmates/demos/x86/ioapic-demo.c | 3 +--
inmates/demos/x86/ivshmem-demo.c | 2 +-
inmates/demos/x86/pci-demo.c | 3 +--
4 files changed, 4 insertions(+), 7 deletions(-)
diff --git
We will never return from c_entry. Make it clear on x86 by moving the stop
label to the top of the assembly code.
Signed-off-by: Ralf Ramsauer
---
inmates/lib/x86/header-32.S | 9 -
inmates/lib/x86/header.S| 9 -
2 files changed, 8
Changes to the ABI of the communication region might break inmates
without noticing. Add the magic JHCOMM together with the ABI version as
a preamble of the structure.
Signed-off-by: Ralf Ramsauer
---
hypervisor/control.c | 7 ++-
We will enter this function if we hit unrecoverable state.
Signed-off-by: Ralf Ramsauer
---
inmates/lib/arm-common/include/inmate.h | 6 ++
inmates/lib/x86/inmate.h| 6 ++
2 files changed, 12 insertions(+)
diff --git
A cell may set this state if its internal ABI of the communication
region is incompatibel to the one provided by the hypervisor.
Signed-off-by: Ralf Ramsauer
---
driver/sysfs.c| 2 ++
hypervisor/control.c | 1 +
Signed-off-by: Ralf Ramsauer
---
inmates/lib/inmate_common.h | 1 +
inmates/lib/string.c| 10 ++
2 files changed, 11 insertions(+)
diff --git a/inmates/lib/inmate_common.h b/inmates/lib/inmate_common.h
index dfe33347..ba39550e 100644
---
Or fail by setting the cell's state and halt.
Signed-off-by: Ralf Ramsauer
---
inmates/lib/setup.c | 11 ++-
1 file changed, 10 insertions(+), 1 deletion(-)
diff --git a/inmates/lib/setup.c b/inmates/lib/setup.c
index 5af94fc8..87837741 100644
---
comm_region will become mandatory for all of our inmates.
On x86, we already use the communication region to pass certain platform
specific information to guests. So let's do this on all ARM
architectures as well. This patch is a preparation for that.
Let's use guest paddr 0x8000 (2GiB) for
On 2018-05-07 15:54, Ralf Ramsauer wrote:
> On 05/07/2018 03:47 PM, Jan Kiszka wrote:
>> On 2018-05-07 15:42, Ralf Ramsauer wrote:
>>> On 05/07/2018 03:37 PM, Jan Kiszka wrote:
On 2018-05-07 14:12, Ralf Ramsauer wrote:
> Or fail by setting the cell's state and halt.
>
>
On 2018-05-07 16:12, Ralf Ramsauer wrote:
> On 05/07/2018 03:25 PM, Jan Kiszka wrote:
>> On 2018-05-07 14:12, Ralf Ramsauer wrote:
>>> Don't jump directly to inmate_main from assembly, jump to c_entry.
>>>
>>> At the moment, c_entry does nothing else than jumping to inmate_main and
>>> ending up
On 05/07/2018 03:25 PM, Jan Kiszka wrote:
> On 2018-05-07 14:12, Ralf Ramsauer wrote:
>> Don't jump directly to inmate_main from assembly, jump to c_entry.
>>
>> At the moment, c_entry does nothing else than jumping to inmate_main and
>> ending up in an endless loop if inmate_main returns.
>>
>>
On 2018-05-07 15:48, Ralf Ramsauer wrote:
> On 05/07/2018 03:42 PM, Jan Kiszka wrote:
>> On 2018-05-07 15:35, Ralf Ramsauer wrote:
>>> On 05/07/2018 03:25 PM, Jan Kiszka wrote:
On 2018-05-07 14:12, Ralf Ramsauer wrote:
> Don't jump directly to inmate_main from assembly, jump to c_entry.
On 2018-05-07 15:48, Assila Belhouichet wrote:
> Hi all,
>
> i reproduced the patchs that Henning provided for testing the intercell
> communication on the bananapi, everything was okay until i executed the
> command:
>
> --> jailhouse cell create bananapi-linux-demo.cell
>
> then i got this
On 05/07/2018 03:47 PM, Jan Kiszka wrote:
> On 2018-05-07 15:42, Ralf Ramsauer wrote:
>> On 05/07/2018 03:37 PM, Jan Kiszka wrote:
>>> On 2018-05-07 14:12, Ralf Ramsauer wrote:
Or fail by setting the cell's state and halt.
Signed-off-by: Ralf Ramsauer
Hi all,
i reproduced the patchs that Henning provided for testing the intercell
communication on the bananapi, everything was okay until i executed the command:
--> jailhouse cell create bananapi-linux-demo.cell
then i got this error message:
Adding virtual PCI device 00:00.0 to cell
On 2018-05-07 15:42, Ralf Ramsauer wrote:
> On 05/07/2018 03:37 PM, Jan Kiszka wrote:
>> On 2018-05-07 14:12, Ralf Ramsauer wrote:
>>> Or fail by setting the cell's state and halt.
>>>
>>> Signed-off-by: Ralf Ramsauer
>>> ---
>>> inmates/lib/setup.c | 11
On 05/07/2018 03:37 PM, Jan Kiszka wrote:
> On 2018-05-07 14:12, Ralf Ramsauer wrote:
>> Or fail by setting the cell's state and halt.
>>
>> Signed-off-by: Ralf Ramsauer
>> ---
>> inmates/lib/setup.c | 11 +++
>> 1 file changed, 11 insertions(+)
>>
>>
On 2018-05-07 15:35, Ralf Ramsauer wrote:
> On 05/07/2018 03:25 PM, Jan Kiszka wrote:
>> On 2018-05-07 14:12, Ralf Ramsauer wrote:
>>> Don't jump directly to inmate_main from assembly, jump to c_entry.
>>>
>>> At the moment, c_entry does nothing else than jumping to inmate_main and
>>> ending up
On 2018-05-07 14:12, Ralf Ramsauer wrote:
> Or fail by setting the cell's state and halt.
>
> Signed-off-by: Ralf Ramsauer
> ---
> inmates/lib/setup.c | 11 +++
> 1 file changed, 11 insertions(+)
>
> diff --git a/inmates/lib/setup.c
On 05/07/2018 03:25 PM, Jan Kiszka wrote:
> On 2018-05-07 14:12, Ralf Ramsauer wrote:
>> Don't jump directly to inmate_main from assembly, jump to c_entry.
>>
>> At the moment, c_entry does nothing else than jumping to inmate_main and
>> ending up in an endless loop if inmate_main returns.
>>
>>
On 05/07/2018 03:20 PM, Jan Kiszka wrote:
> On 2018-05-07 14:12, Ralf Ramsauer wrote:
>> We will enter this function if we hit unrecoverable state.
>>
>> Signed-off-by: Ralf Ramsauer
>> ---
>> inmates/lib/arm-common/include/inmate.h | 6 ++
>>
On 2018-05-07 14:12, Ralf Ramsauer wrote:
> Don't jump directly to inmate_main from assembly, jump to c_entry.
>
> At the moment, c_entry does nothing else than jumping to inmate_main and
> ending up in an endless loop if inmate_main returns.
>
> Later we can use c_entry to do platform
On 2018-05-07 14:12, Ralf Ramsauer wrote:
> We will enter this function if we hit unrecoverable state.
>
> Signed-off-by: Ralf Ramsauer
> ---
> inmates/lib/arm-common/include/inmate.h | 6 ++
> inmates/lib/x86/inmate.h| 6 ++
> 2 files
On 2018-05-07 14:12, Ralf Ramsauer wrote:
> A cell may set this state if its internal ABI of the communication
> region is incompatibel to the one provided by the hypervisor.
>
> Signed-off-by: Ralf Ramsauer
> ---
> driver/sysfs.c| 2 ++
>
We will enter this function if we hit unrecoverable state.
Signed-off-by: Ralf Ramsauer
---
inmates/lib/arm-common/include/inmate.h | 6 ++
inmates/lib/x86/inmate.h| 6 ++
2 files changed, 12 insertions(+)
diff --git
Hi,
ABI changes of the communication region can not be detected by inmates at the
moment. Let's introduce some magic and revision field at the beginning of the
page to signal ABI changes.
This will help us if we change the ABI at a later point in time.
Ralf
Ralf Ramsauer (7):
core:
Signed-off-by: Ralf Ramsauer
---
inmates/lib/inmate_common.h | 1 +
inmates/lib/string.c| 10 ++
2 files changed, 11 insertions(+)
diff --git a/inmates/lib/inmate_common.h b/inmates/lib/inmate_common.h
index 6c779b53..8d1c53e5 100644
---
Changes to the ABI of the communication region might break inmates
without noticing. Add the magic JHCOMM together with the ABI version as
a preamble of the structure.
Signed-off-by: Ralf Ramsauer
---
hypervisor/control.c | 7 ++-
A cell may set this state if its internal ABI of the communication
region is incompatibel to the one provided by the hypervisor.
Signed-off-by: Ralf Ramsauer
---
driver/sysfs.c| 2 ++
hypervisor/control.c | 1 +
Or fail by setting the cell's state and halt.
Signed-off-by: Ralf Ramsauer
---
inmates/lib/setup.c | 11 +++
1 file changed, 11 insertions(+)
diff --git a/inmates/lib/setup.c b/inmates/lib/setup.c
index ed35b6f8..be651696 100644
---
Don't jump directly to inmate_main from assembly, jump to c_entry.
At the moment, c_entry does nothing else than jumping to inmate_main and
ending up in an endless loop if inmate_main returns.
Later we can use c_entry to do platform independant setup/checks.
Signed-off-by: Ralf Ramsauer
Branch: refs/heads/next
Home: https://github.com/siemens/jailhouse
Commit: 97df80071a5392b958504084a992c6264c380716
https://github.com/siemens/jailhouse/commit/97df80071a5392b958504084a992c6264c380716
Author: Jan Kiszka
Date: 2018-05-06 (Sun, 06 May
From: Jan Kiszka
We already support movb-to-mem, adding the reverse direction is simple.
Accompany this with tests for both 8-bit accesses.
Signed-off-by: Jan Kiszka
---
hypervisor/arch/x86/include/asm/processor.h | 1 +
From: Jan Kiszka
If the guest is currently running in long mode does not only depend on
EFER having it enabled. We also need to check if CS.L is set. If L is 0,
we are actually in compatibility mode.
Rework vcpu_vendor_get_cs() to vcpu_vendor_get_cs_attr() so that it
From: Jan Kiszka
Mostly the 32-bit equivalent of the existing 64-bit case, just removing
long-mode-only tests and converting to 32-bit data/address size where
needed.
Signed-off-by: Jan Kiszka
---
inmates/tests/x86/Makefile | 5 +-
From: Jan Kiszka
As we are claiming to support the affected instructions, we should get
this completely right: When not in long mode, the combination of the
address-size prefix and the D/B bit from CS defines if we are using 16
or 32 bit addresses.
A similar pattern may
On 2018-05-04 22:56, Giovani Gracioli wrote:
> Thanks Jan.
>
> By changing the bar_mask and having the two vpci devices mapped in different
> pages, I was able to run it.
>
> Can you please explain how the bar mask is used to define the address in
> which the device is mapped to?
That's
78 matches
Mail list logo