[PATCH 09 of 10] tools: new tool to get gcov data out of dirty firmware

link to its gcov lib. Signed-off-by: Henning Schild <henning.sch...@siemens.com> diff --git a/.gitignore b/.gitignore --- a/.gitignore +++ b/.gitignore @@ -4,6 +4,7 @@ *.cmd *.bin *.gcno +*.gcda .tmp_versions *.dtb *.dtb.S @@ -14,6 +15,7 @@ hypervisor/include/jailhouse/config.h hype

[PATCH 08 of 10] hypervisor: introcude GCOV support

. Signed-off-by: Henning Schild <henning.sch...@siemens.com> diff --git a/driver/main.c b/driver/main.c --- a/driver/main.c +++ b/driver/main.c @@ -419,6 +419,7 @@ header = (struct jailhouse_header *)hypervisor_mem; header->max_cpus = max_cpus; + header->jailhouse_

[PATCH 03 of 10] make: do not support GCOV in out of kernel components

If the root-cell kernel is built with GCOV (CONFIG_GCOV_PROFILE_ALL) kbuild will build everything with GCOV support. For the driver it is ok to inherit that, but for out of kernel parts like hypervisor, tools, inmates it is not. Signed-off-by: Henning Schild <henning.sch...@siemens.com>

[PATCH 04 of 10] driver: create new function to undo firmware mapping

Signed-off-by: Henning Schild <henning.sch...@siemens.com> diff --git a/driver/main.c b/driver/main.c --- a/driver/main.c +++ b/driver/main.c @@ -274,6 +274,14 @@ return ret; } +static void jailhouse_firmware_free(void) +{ + if (hypervisor_mem) { +

[PATCH 05 of 10] driver: keep the firmware mapped after "disable"

We used to unmap the firmware image right after "jailhouse disable". Change that and keep the firmware mapped until we exit the module or restart jailhouse. This way we can look into the firmware that just ran. This will be used by later patches. Signed-off-by: Henning Schild &l

[PATCH 02 of 10] make: include Makefile.lib on x86 as well

For consistency and to cater for a following patch. Signed-off-by: Henning Schild <henning.sch...@siemens.com> diff --git a/inmates/lib/x86/Makefile b/inmates/lib/x86/Makefile --- a/inmates/lib/x86/Makefile +++ b/inmates/lib/x86/Makefile @@ -10,6 +10,8 @@ # the COPYING file in the top

[PATCH 06 of 10] driver: make the firware visisble in sysfs

Make the "used, dirty" firmware image visible in sysfs. The data segment contains data that could be useful for all sorts of debugging tools. This feature is introduced to extract code coverage information (gcov). Signed-off-by: Henning Schild <henning.sch...@siemens.com> dif

[RFC] using gcov for code coverage in jailhouse

Hi, for safety certification we need code coverage information from the hypervisor. The first open tool/format of choice in this area is gcov. The way gcov works is that you simply compile and link your code with "--coverage". In the compile step gcc will automatically add instrumentation for

[PATCH] core: remove useless statement

fixes: f0c99e2a5e8f Signed-off-by: Henning Schild <henning.sch...@siemens.com> diff --git a/hypervisor/setup.c b/hypervisor/setup.c --- a/hypervisor/setup.c +++ b/hypervisor/setup.c @@ -76,7 +76,6 @@ hyp_phys_start = system_config->hypervisor_memory.phys_start; hyp

Re: [PATCH 2 of 2] make: do not support GCOV in out of kernel components

On Tue, 7 Feb 2017 18:11:05 +0100 Jan Kiszka <jan.kis...@siemens.com> wrote: > On 2017-02-07 13:32, Henning Schild wrote: > > If the root-cell kernel is built with GCOV (CONFIG_GCOV_PROFILE_ALL) > > kbuild will build everything with GCOV support. For the driver it

Re: [PATCH 1 of 2] gitignore: ignore GCOV output

fferent tools). Well this one would be gcc itself depending on your kernel-config. If you have CONFIG_GCOV_PROFILE_ALL activated a simple "make" will create such files for you. > On 07.02.2017 21:32, Henning Schild wrote: > > Signed-off-by: Henning Schild <henning.sch...@sie

[PATCH 2 of 2] make: do not support GCOV in out of kernel components

If the root-cell kernel is built with GCOV (CONFIG_GCOV_PROFILE_ALL) kbuild will build everything with GCOV support. For the driver it is ok to inherit that, but for out of kernel parts like hypervisor, tools, inmates it is not. Signed-off-by: Henning Schild <henning.sch...@siemens.com>

[PATCH 1 of 2] gitignore: ignore GCOV output

Signed-off-by: Henning Schild <henning.sch...@siemens.com> diff --git a/.gitignore b/.gitignore --- a/.gitignore +++ b/.gitignore @@ -3,6 +3,7 @@ *.mod.[co] *.cmd *.bin +*.gcno .tmp_versions *.dtb *.dtb.S -- You received this message because you are subscribed to the Google

Re: Linux: ioremap_page_range is no longer exported

On Thu, 2 Feb 2017 18:36:59 +0100 Jan Kiszka wrote: > On 2017-02-02 18:17, Ralf Ramsauer wrote: > > Hi, > > > > upstream Linux commit 3277953 (mm: do not export ioremap_page_range > > symbol for external module) unexports the ioremap_page_range symbol, > > which we need

[PATCH v2] tools: cell-linux clean up and rewrite, no functional change

- introduce BASE_OFFSET and replace some magic values - add comments to point people to related docs - introduce setters for rw values of x86 boot header - easier to read/maintain than struct.pack_into - only rw values can be changed, for ro we do not have code Signed-off-by: Henning Schild

Re: [PATCH] tools: cell-linux: place ramdisk behind kernel like on arm

Am Thu, 5 Jan 2017 10:46:42 +0100 schrieb Jan Kiszka <jan.kis...@siemens.com>: > On 2017-01-05 10:09, Henning Schild wrote: > > Am Thu, 5 Jan 2017 00:06:10 +0100 > > schrieb Jan Kiszka <jan.kis...@siemens.com>: > > > >> On 2017-01-04 19:03, Henning

Re: [PATCH] tools: cell-linux: place ramdisk behind kernel like on arm

Am Thu, 5 Jan 2017 00:06:10 +0100 schrieb Jan Kiszka <jan.kis...@siemens.com>: > On 2017-01-04 19:03, Henning Schild wrote: > > The kernel is copied to its preferred address which seems to > > be 16MB in most cases. With the comm region in low memory > > we can

Re: [PATCH] tools: cell-linux always load kernel to preferred address

; > > > Btw.: Your cell config is missing the comm region > > > > I've updated the cell config. Thanks ! > So your non-root Linux comes up as expected? Henning > > > From: Henning Schild <henning.sch...@siemens.com&g

Re: [PATCH] tools: cell-linux always load kernel to preferred address

t; > That's good to know, Jan. > > > > > > From: Jan Kiszka <jan.kis...@siemens.com> > Sent: Wednesday, January 4, 2017 12:23:37 PM > To: Vaidyanath, Rajiv; Henning Schild; jailhouse-dev@googlegroups.com > Subject: Re:

[PATCH] tools: cell-linux: place ramdisk behind kernel like on arm

The kernel is copied to its preferred address which seems to be 16MB in most cases. With the comm region in low memory we can had a limit on 14MB initrds. Placing the ramdisk behind the kernel adds consistency with ARM and gives us address space for bigger ramdisks. Signed-off-by: Henning Schild

[PATCH 1 of 2] tools: cell-linux fix loader path when printing commands

Signed-off-by: Henning Schild <henning.sch...@siemens.com> diff --git a/tools/jailhouse-cell-linux b/tools/jailhouse-cell-linux --- a/tools/jailhouse-cell-linux +++ b/tools/jailhouse-cell-linux @@ -258,7 +258,7 @@ self.params = self._zero_page.get_data() + setu

[PATCH 1 of 2 v3] inmates: x86: match pci device class in ivshmem-demo

Match the device class and "protocol" while probing and skip unknow devices. Signed-off-by: Henning Schild <henning.sch...@siemens.com> diff --git a/inmates/demos/x86/ivshmem-demo.c b/inmates/demos/x86/ivshmem-demo.c --- a/inmates/demos/x86/ivshmem-demo.c +++ b/inmates/demos/x8

[PATCH 1 of 2 v2] inmates: x86: match pci device class in ivshmem-demo

Match the device class and "protocol" while probing and skip unknow devices. Signed-off-by: Henning Schild <henning.sch...@siemens.com> diff --git a/inmates/demos/x86/ivshmem-demo.c b/inmates/demos/x86/ivshmem-demo.c --- a/inmates/demos/x86/ivshmem-demo.c +++ b/inmates/demos/x8

Re: [PATCH 2 of 2] core: ivshmem: match device class to connect

Am Wed, 30 Nov 2016 18:19:27 +0100 schrieb Jan Kiszka <jan.kis...@siemens.com>: > On 2016-11-30 17:50, Henning Schild wrote: > > Since 3e970ab533c4 we can specify a protocol to run on the ivshmem > > channel. This patch considers a protocol mismatch between the two > >

[PATCH 2 of 2] core: ivshmem: match device class to connect

Since 3e970ab533c4 we can specify a protocol to run on the ivshmem channel. This patch considers a protocol mismatch between the two endpoints a configuration error and does not connect them to each other. Signed-off-by: Henning Schild <henning.sch...@siemens.com> diff --git a/Documen

[PATCH 1 of 2] inmates: x86: match pci device class in ivshmem-demo

Match the device class and "protocol" while probing and skip unknow devices. Signed-off-by: Henning Schild <henning.sch...@siemens.com> diff --git a/inmates/demos/x86/ivshmem-demo.c b/inmates/demos/x86/ivshmem-demo.c --- a/inmates/demos/x86/ivshmem-demo.c +++ b/inmates/demos/x8

Re: Issues with multiple instances of a PCIe board

Am Mon, 28 Nov 2016 03:24:50 -0800 schrieb Josiah Payn <josiah.p...@gmail.com>: > On Monday, November 28, 2016 at 5:22:47 PM UTC+8, Henning Schild > wrote: > > Am Fri, 25 Nov 2016 09:47:36 -0800 > > schrieb Josiah Payn <josiah.p...@gmail.com>: > > > >

Re: Issues with multiple instances of a PCIe board

Am Fri, 25 Nov 2016 09:47:36 -0800 schrieb Josiah Payn <josiah.p...@gmail.com>: > On Friday, November 25, 2016 at 11:16:30 PM UTC+8, Henning Schild > wrote: > > Hi, > > > > linux-x86-demo.c line 26 ...: > > > struct jailhouse_pci_device pci_devices[

Re: [PATCH 02/20] tooling: Install jailhouse bash completion script

Am Tue, 22 Nov 2016 16:34:03 +0100 schrieb Jan Kiszka <jan.kis...@siemens.com>: > On 2016-11-22 16:22, Henning Schild wrote: > > Am Mon, 21 Nov 2016 13:17:25 +0100 > > schrieb Jan Kiszka <jan.kis...@siemens.com>: > > > >> Those scripts should go

Re: [PATCH v2] Documentation: Fix Bananapi tutorial

Am Fri, 18 Nov 2016 14:28:13 +0100 schrieb Ralf Ramsauer <r...@ramses-pyramidenbau.de>: > On 11/18/2016 02:17 PM, Henning Schild wrote: > > Am Fri, 18 Nov 2016 13:32:09 +0100 > > schrieb <andreas.koe...@st.oth-regensburg.de>: > > > >> From: Andreas Köl

Re: [PATCH 1/1] FAQ: minimal kernel versions

Am Thu, 17 Nov 2016 11:59:04 +0100 schrieb Claudio Scordino : > This patch adds a section to the FAQ to specify the minimal version of > the Linux kernel required under each architecture. > > Signed-off-by: Claudio Scordino > --- > FAQ.md | 9

[PATCH v4] pci: avoid implicit device claiming with pci_add_dynid

-function. Reported-by: Koehrer Mathias (ETAS/ESW5) <mathias.koeh...@etas.com> Signed-off-by: Henning Schild <henning.sch...@siemens.com> diff --git a/driver/pci.c b/driver/pci.c --- a/driver/pci.c +++ b/driver/pci.c @@ -11,16 +11,38 @@ * the COPYING file in the top-lev

[PATCH v3] pci: avoid implicit device claiming with pci_add_dynid

-function. Reported-by: Koehrer Mathias (ETAS/ESW5) <mathias.koeh...@etas.com> Signed-off-by: Henning Schild <henning.sch...@siemens.com> diff --git a/driver/pci.c b/driver/pci.c --- a/driver/pci.c +++ b/driver/pci.c @@ -11,16 +11,45 @@ * the COPYING file in the top-lev

[PATCH v2] pci: avoid implicit device claiming with pci_add_dynid

-function. Reported-by: Koehrer Mathias (ETAS/ESW5) <mathias.koeh...@etas.com> Signed-off-by: Henning Schild <henning.sch...@siemens.com> diff --git a/driver/pci.c b/driver/pci.c --- a/driver/pci.c +++ b/driver/pci.c @@ -11,16 +11,32 @@ * the COPYING file in the top-lev

Re: Issues with multiple instances of a PCIe board

Am Fri, 11 Nov 2016 13:34:12 + schrieb "Koehrer Mathias (ETAS/ESW5)" : > Hi all, > > I am doing some experiments using jailhouse... > In my x86 PC I have a dual network adapter (PCIe, Intel i350, driver: > igb). This is reflected as two PCIe devices. > For test

Re: How to use shared memory between host and non-root celll

Am Thu, 10 Nov 2016 09:29:05 + schrieb "Koehrer Mathias (ETAS/ESW5)" : > Viele Grüße > > Mathias > > Hi Henning, > > > > A general questions on the root cell configuration: > > > If I understand this correctly, this configuration has to reflect > > > the full

Re: [PATCH 3/5] core: ivshmem: Mark device as different via subsystem vendor ID

othing official, > > just a natural local choice. > > > > CC: Henning Schild <henning.sch...@siemens.com> > > CC: Måns Rullgård <m...@denx.de> > > Signed-off-by: Jan Kiszka <jan.kis...@siemens.com> > > --- > > hypervisor/pci_ivshmem.

Re: [RESEND][PATCH] pci-ivshmem: add local/remote state registers

Am Wed, 9 Nov 2016 15:10:36 +0100 schrieb Jan Kiszka <jan.kis...@siemens.com>: > On 2016-11-09 15:02, Henning Schild wrote: > > Am Wed, 9 Nov 2016 13:07:17 + > > schrieb Mans Rullgard <m...@mansr.com>: > > > >> This adds a register p

Re: How to use shared memory between host and non-root celll

Am Wed, 9 Nov 2016 15:39:39 +0100 schrieb Henning Schild <henning.sch...@siemens.com>: > Am Wed, 9 Nov 2016 14:19:16 + > schrieb "Koehrer Mathias (ETAS/ESW5)" <mathias.koeh...@etas.com>: > > > Hi Henning! > > > > > > # grep ivs

Re: How to use shared memory between host and non-root celll

Am Wed, 9 Nov 2016 14:19:16 + schrieb "Koehrer Mathias (ETAS/ESW5)" : > Hi Henning! > > > > # grep ivshmem /proc/interrupts > > > 61: 0 0 0 0 0 > > > 0 0 IR-PCI-MSI 245760-edge ivshmem-config > > > > > > No

Re: How to use shared memory between host and non-root celll

Am Wed, 9 Nov 2016 13:37:18 + schrieb "Koehrer Mathias (ETAS/ESW5)" : > Hi all! > > Sorry for asking many questions... > > > > Did you check out the jailhouse branch? > No - I used the master branch. > > Now I switched to the jailhouse branch. > This looks

Re: [RESEND][PATCH] pci-ivshmem: add local/remote state registers

Am Wed, 9 Nov 2016 13:07:17 + schrieb Mans Rullgard : > This adds a register pair allowing inmate drivers to communicate their > status, for instance to trigger a protocol reset when (re)loading the > inmate in one of the cells. > > Signed-off-by: Mans Rullgard

Re: How to use shared memory between host and non-root celll

Hey, you basically put a memory region with the same phys_start and size in the root-cell and non-root cell. If the sharing should happen between root and non-root you need the JAILHOUSE_MEM_ROOTSHARED memory flag, for two non-root cells that is not required. I suggest you base your work on the

Re: [siemens/jailhouse] Unhandled MSR write

Am Mon, 7 Nov 2016 16:04:07 +0100 schrieb Jan Kiszka : > On 2016-11-07 13:35, josiah.p...@gmail.com wrote: > > Hi Jan, > > > > Thank you for your advice. I was able to run a non-root linux cell > > using v4.8.0 as host kernel and using a v4.8.6 kernel for the guest > >

Re: Minimal kernel release ?

Am Wed, 26 Oct 2016 17:32:56 +0200 schrieb Claudio Scordino : > Hi all, > > is there any minimal kernel release for building/running Jailhouse ? > > The README is quite generic, talking about 3.9 (for Qemu) and 3.14. > > In the source code, however, I noticed a call to

Re: [PATCH] config: fix address overlap in TK1 configs

Am Tue, 4 Oct 2016 15:10:31 +0200 schrieb Ralf Ramsauer : > Hi, > > On 10/04/2016 02:49 PM, Jan Kiszka wrote: > > On 2016-10-04 14:43, Ralf Ramsauer wrote: > >> Hi, > >> > >> this patch gives me an idea: > >> > >> Why don't we just mark a dedicated memory region as

[PATCH] tools: config-create: fix iterator index manipulation in PCIBARs

The n += 1, to skip a line in the 64bit case, never had an effect. Change the iterator so it can be manipulated like that. Fixes: d1f763162ce6 ("core, tools: Add BAR masks to jailhouse_pci_device") Signed-off-by: Henning Schild <henning.sch...@siemens.com> --- tools/jailhouse-c

Re: [PATCH V9 1/3] tools: Add the hierarchy of the IOPortRegion.

Am Tue, 9 Aug 2016 04:10:13 -0400 schrieb charles king <huddy1...@gmail.com>: > Hi, > Henning Schild. > > 2016-07-25 4:28 GMT-04:00 Henning Schild <henning.sch...@siemens.com>: > > > Am Sun, 24 Jul 2016 09:58:05 +0800 > > schrieb charles kin

Re: [PATCH 3/7] tools: jailhouse: add additional sanity checks to read_file()

Am Thu, 4 Aug 2016 21:28:14 +0200 schrieb Ralf Ramsauer : > Check if file size is not zero and check the return value of close() > as it might fail, though it's very unlikely. > > Signed-off-by: Ralf Ramsauer > --- > tools/jailhouse.c |

Re: [PATCH 00/12] Assorted fixes and simplifications

Am Sun, 31 Jul 2016 10:39:39 +0200 schrieb Jan Kiszka : > Besides config updates and some cleanups, this series comes with two > major changes. > > The most user visible one is demanding VT-d support. QEMU 2.7 is > currently in -rc and will deliver all features we need to run

Re: [PATCH V9 1/3] tools: Add the hierarchy of the IOPortRegion.

Am Sun, 24 Jul 2016 09:58:05 +0800 schrieb charles king <huddy1...@gmail.com>: > Hi, > Henning Schild, you mean the bitmask should defined like: > > ALL_ALIGNED = ALIGNED_T and ALIGNED_H "or" not "and" > ALIGNED_T = 1 > ALIGNED_H = 2 &

Re: [PATCH V7 1/3] tools: Add the hierarchy of the IOPortRegion.

Good to see that work on these patches is still ongoing! The result looks good, i did not test it yet. But the code still needs some work. For now just one comment after looking at the patches. Am Wed, 20 Jul 2016 16:29:20 +0800 schrieb Xuguo Wang : > Abstract the common

Re: [PATCH V7 0/3] *** pio_bitmap generator ***

Your patches apply but introduce pep8 violations again. Am Wed, 20 Jul 2016 16:29:19 +0800 schrieb Xuguo Wang : > *** > v7: > - fix merge conflicts. > - remove cosmetic fixes > - refactor docstrings style > *** > > Xuguo Wang

<    1   2   3   4