Login

pello.heriz · 04-06-2017, 07:17 AM

Hi all,

-Host: Ubuntu 16.04 (in Virtual Box)
-Petalinux and SDK version: 2016.3
-XZD version: XZD20161231

I have followed the steps shown in the last XZD user manual (january 6, 2017) to launch Xen on QEMU, but I'm having some problems. When I execute the command to launch the image on QEMU, sometimes (1 / 10) the launching seems to be good, but in the rest of the cases a kernel panic occurs. This is how the kernel panic looks like:

[ 49.994179] Call trace:
[ 49.996072] [<ffffff8008088840>] dump_backtrace+0x0/0x198
[ 49.996982] [<ffffff80080889ec>] show_stack+0x14/0x20
[ 49.997567] [<ffffff80080bf910>] sched_show_task+0x98/0xf8
[ 49.998254] [<ffffff80080e3d00>] rcu_check_callbacks+0x740/0x748
[ 49.998755] [<ffffff80080e6a4c>] update_process_times+0x3c/0x68
[ 49.999230] [<ffffff80080f5b3c>] tick_sched_handle.isra.5+0x3c/0x50
[ 50.001318] [<ffffff80080f5b94>] tick_sched_timer+0x44/0x90
[ 50.001933] [<ffffff80080e7400>] __hrtimer_run_queues+0xf0/0x178
[ 50.002557] [<ffffff80080e7790>] hrtimer_interrupt+0x98/0x1c8
[ 50.003283] [<ffffff8008633af0>] arch_timer_handler_virt+0x30/0x40
[ 50.003966] [<ffffff80080dab68>] handle_percpu_devid_irq+0x78/0xa0
[ 50.005726] [<ffffff80080d635c>] generic_handle_irq+0x24/0x38
[ 50.006402] [<ffffff80080d6694>] __handle_domain_irq+0x5c/0xb8
[ 50.007023] [<ffffff80080814bc>] gic_handle_irq+0x64/0xc0
[ 50.007603] Exception stack(0xffffffc0328b7e40 to 0xffffffc0328b7f60)
[ 50.008420] 7e40: 0000000000000000 0000000000000000 0000000000000001 0000000000000001
[ 50.009126] 7e60: 00000000000001c0 0100000000000000 00297c1e00000000 00000000ffff0918
[ 50.009731] 7e80: ffffffc0328ac9e0 ffffffc0328b4000 0000000000000780 0000000000000001
[ 50.010877] 7ea0: 0000000000000001 dead000000000200 dead000000000100 ffffffbe00000000
[ 50.011673] 7ec0: ffffff8008ad9000 ffffffc032809d28 000000000000000c 0000000000000000
[ 50.013353] 7ee0: ffffffc0328b4000 ffffff8008ad9000 ffffffc0328b4000 ffffff8008ad9000
[ 50.014122] 7f00: ffffff8008ab7ff0 ffffff8008b58000 ffffffc0328b4000 0000000000000000
[ 50.014904] 7f20: 0000000000000000 ffffffc0328b7f60 ffffff800808567c ffffffc0328b7f60
[ 50.015693] 7f40: ffffff8008085680 0000000060000145 ffffffc0328b7f70 7fffffffffffffff
[ 50.016527] [<ffffff8008084720>] el1_irq+0xa0/0x100
[ 50.017201] [<ffffff8008085680>] arch_cpu_idle+0x10/0x18
[ 50.017919] [<ffffff80080cf36c>] cpu_startup_entry+0x144/0x1d8
[ 50.019187] [<ffffff800808df90>] secondary_start_kernel+0x170/0x1c8
[ 50.020968] [<000000002008184c>] 0x2008184c
[ 50.022011] rcu_sched kthread starved for 5247 jiffies! g583 c582 f0x2 RCU_GP_WAIT_FQS(3) ->state=0x100
[ 50.023441] rcu_sched W ffffff8008085b1c 0 7 2 0x00000000
[ 50.024350] Call trace:
[ 50.024641] [<ffffff8008085b1c>] __switch_to+0xc4/0xd0
[ 50.025238] [<ffffff80087d6734>] __schedule+0x19c/0x588
[ 50.025721] [<ffffff80087d6b50>] schedule+0x30/0x90
[ 50.026034] [<ffffff80087d93e8>] schedule_timeout+0x108/0x1b0
[ 50.028872] [<ffffff80080e2e10>] rcu_gp_kthread+0x368/0x7a0
[ 50.029466] [<ffffff80080b60e8>] kthread+0xd0/0xe8
[ 50.029889] [<ffffff8008084dd0>] ret_from_fork+0x10/0x40

The command that I'm using to launch QEMU is the one shown in the user manual.

Anyway, I have observe that even when the image booting seems to be correct and I arrive into Dom0 terminal, the system never mounts /proc, /sys and /debug directories, and even if the kernel panic hasn't happened yet, it can happens also when I try to run a xl command or simply when Dom0 or the guest Dom1 is running.

Here you have how the system never mounts the mentioned directories:

Starting network...
Starting Xilinx Xen Boot Scripts: mount: mounting none on /proc failed: Device or resource busy
mount: mounting none on /sys/ failed: Device or resource busy
mount: mounting none on /debug/ failed: No such file or directory

Finally, I can also observe that sometimes the launching process get stuck in some step and doesn't appear any kernel panic as for example in the next case:

tarting network...
Starting Xilinx Xen Boot Scripts: mount: mounting none on /proc failed: Device or resource busy
mount: mounting none on /sys/ failed: Device or resource busy
mount: mounting none on /debug/ failed: No such file or directory
[ 42.827299] IPv6: ADDRCONF(NETDEV_UP): eth0: link is not ready
[ 44.620198] macb ff0e0000.ethernet eth0: unable to generate target frequency: 25000000 Hz
[ 44.621754] macb ff0e0000.ethernet eth0: link up (100/Half)
[ 44.679022] IPv6: ADDRCONF(NETDEV_CHANGE): eth0: link becomes ready
[ 50.693458] xenbr0: port 1(eth0) entered blocking state
[ 50.696538] xenbr0: port 1(eth0) entered disabled state
[ 50.725696] device eth0 entered promiscuous mode
udhcpc (v1.23.1) started
[ 51.648543] xenbr0: port 1(eth0) entered blocking state
[ 51.649553] xenbr0: port 1(eth0) entered forwarding state
Sending discover...
Sending select for 10.0.2.15...
Lease of 10.0.2.15 obtained, lease time 86400
deleting routers

Any answer will be welcome,
Best regards,
Pello

Nathan.Studer · 04-06-2017, 06:01 PM

(04-06-2017, 07:17 AM)pello.heriz Wrote: When I execute the command to launch the image on QEMU, sometimes (1 / 10) the launching seems to be good, but in the rest of the cases a kernel panic occurs.

...

Finally, I can also observe that sometimes the launching process get stuck in some step and doesn't appear any kernel panic as for example in the next case:

There are some issues with how QEMU emulates interrupts, which seem to strike certain host configurations worse than others. (It rarely happens on my laptop, but other developers with similar laptop configurations frequently report this issue.) We've been looking for a work-around with mixed success. You could try running your Virtual Machine in VMWare instead of VirtualBox, or try running on a different computer, but in general it's best to run on an actual zcu102 board.

Quote:pello.heriz

Anyway, I have observe that even when the image booting seems to be correct and I arrive into Dom0 terminal, the system never mounts /proc, /sys and /debug directories, and even if the kernel panic hasn't happened yet, it can happens also when I try to run a xl command or simply when Dom0 or the guest Dom1 is running.

Here you have how the system never mounts the mentioned directories:

Starting network...
Starting Xilinx Xen Boot Scripts: mount: mounting none on /proc failed: Device or resource busy
mount: mounting none on /sys/ failed: Device or resource busy
mount: mounting none on /debug/ failed: No such file or directory

Those messages are due to one of the bootscripts trying to mount devices that are already mounted due to their inclusion in the fstab. You should safely be able to ignore these errors.

These errors will be eliminated in future XZD releases, which will be using a Yocto based rootfs instead of a buildroot one.

Nate

pello.heriz · 04-07-2017, 09:28 AM

Thank you very much for your reply Nathan,

I would like to ask you another question. Is it possible to add an image created myself with Yocto as a dom1 or dom2 guest? If the answer is yes, which are the steps I need to follow to add this image as a guest?

Best regards,
Pello

Nathan.Studer

You can adapt the instructions in "Building and Installing the Guest" in the User Manual to copy the Yocto Kernel Image and Yocto RootFS to the Dom0 filesystem and configure the virtual machine. Let me know if you have specific questions about these instructions.

When building the guest you need to make sure you have Xen kernel support enabled and that the Xen front end modules are included in the RootFS. If you have the meta-virtualization layer as part of your Yocto setup this should already be taken care of for you.

You may also want to adjust your kernel to add "console=hvc0" to the default kernel bootline. As an example you can look at the DomU image recipe in our soon to be announced Yocto layer. It's the same as meta-virtualization's Xen DomU recipe but with x86 specific functionality removed.

Nate

pello.heriz · 04-11-2017, 07:23 AM

Hi Nathan,

Thank you once again for your reply, I would try to do these steps and if I have any problem I will let you know.

Best regards,
Pello

pello.heriz · 05-03-2017, 09:03 AM

Hi Nathan,

I have some questions related to the previous post I sent. You told me that including meta-virtualization layer in my Yocto project would be enough in order to create an image that supports Xen and use it as a guest. Anyway, I have seen your post about your new layer called meta-xzd where it is said that it depends on lot of more layers apart of meta-virtualization.

So in order to create an image and copy the kernel and rootfs to the Dom0, is necessary to include all the layers that are mentioned in the meta-xzd readme file on the Yocto project bblayers?

Another question, when you say "You may also want to adjust your kernel to add "console=hvc0" to the default kernel bootline", what are you referring to? Where do I need to change/add "console=hvc0"? in the local.conf file?

Best regards,
Pello

Nathan.Studer · 05-03-2017, 03:35 PM

(05-03-2017, 09:03 AM)pello.heriz Wrote: Hi Nathan,

I have some questions related to the previous post I sent. You told me that including meta-virtualization layer in my Yocto project would be enough in order to create an image that supports Xen and use it as a guest. Anyway, I have seen your post about your new layer called meta-xzd where it is said that it depends on lot of more layers apart of meta-virtualization.

I might not have been clear but you do need more than the meta-virtualization layer; primarily whatever layers Petalinux depends on. I was assuming that you already had all those layers, but if you don't you probably want to get them.

Quote:pello.heriz

So in order to create an image and copy the kernel and rootfs to the Dom0, is necessary to include all the layers that are mentioned in the meta-xzd readme file on the Yocto project bblayers?

The Yocto recipes will enable you to create a Dom0 image/rootfs (e.g. bitbake xzd-image-minimal) and a DomU image/rootfs (e.g. bitbake xzd-domu-image-minimal). There is currently no Yocto support for integrating DomU images/rootfs images into a Dom0, so this needs to be done manually. Either add it to the Dom0 rootfs produced by Yocto or copy it to the SD card that the Dom0 rootfs was extracted to.

Our Yocto setup currently uses an initramfs for the main rootfs and overlays the first SD card partition on this rootfs to include guests images from the SD card. If you want to do something different, you will likely need to disable the initramfs in your local.conf file by removing these lines:

Code:

# Build the xzd initramfs

INITRAMFS_IMAGE = "xzd-image-minimal-initramfs"

# Bundle the initramfs with the kernel

INITRAMFS_IMAGE_BUNDLE = "1"

# The initramfs is kind of big

INITRAMFS_MAXSIZE = "262144"

Quote:pello.heriz

Another question, when you say "You may also want to adjust your kernel to add "console=hvc0" to the default kernel bootline", what are you referring to? Where do I need to change/add "console=hvc0"? in the local.conf file?

Xen guests use the "hvc0" console for terminal output, which needs to be specified on the kernel bootline. See the xzd-domu-image-minimal or xen-domu-image-minimal for how this can be done with your image.

Nate

pello.heriz · 05-17-2017, 07:53 AM

(05-03-2017, 03:35 PM)Nathan.Studer Wrote:
(05-03-2017, 09:03 AM)pello.heriz Wrote: Hi Nathan,

I have some questions related to the previous post I sent. You told me that including meta-virtualization layer in my Yocto project would be enough in order to create an image that supports Xen and use it as a guest. Anyway, I have seen your post about your new layer called meta-xzd where it is said that it depends on lot of more layers apart of meta-virtualization.

I might not have been clear but you do need more than the meta-virtualization layer; primarily whatever layers Petalinux depends on. I was assuming that you already had all those layers, but if you don't you probably want to get them.

Quote:pello.heriz

So in order to create an image and copy the kernel and rootfs to the Dom0, is necessary to include all the layers that are mentioned in the meta-xzd readme file on the Yocto project bblayers?

The Yocto recipes will enable you to create a Dom0 image/rootfs (e.g. bitbake xzd-image-minimal) and a DomU image/rootfs (e.g. bitbake xzd-domu-image-minimal). There is currently no Yocto support for integrating DomU images/rootfs images into a Dom0, so this needs to be done manually. Either add it to the Dom0 rootfs produced by Yocto or copy it to the SD card that the Dom0 rootfs was extracted to.

Our Yocto setup currently uses an initramfs for the main rootfs and overlays the first SD card partition on this rootfs to include guests images from the SD card. If you want to do something different, you will likely need to disable the initramfs in your local.conf file by removing these lines:

Code:

# Build the xzd initramfs INITRAMFS_IMAGE = "xzd-image-minimal-initramfs" # Bundle the initramfs with the kernel INITRAMFS_IMAGE_BUNDLE = "1" # The initramfs is kind of big INITRAMFS_MAXSIZE = "262144"

Quote:pello.heriz

Another question, when you say "You may also want to adjust your kernel to add "console=hvc0" to the default kernel bootline", what are you referring to? Where do I need to change/add "console=hvc0"? in the local.conf file?

Xen guests use the "hvc0" console for terminal output, which needs to be specified on the kernel bootline. See the xzd-domu-image-minimal or xen-domu-image-minimal for how this can be done with your image.

Nate

Hi Nathan,

I'm trying to build a xen-domu-image-minimal using meta-xzd layer. Anyway, I have try to clone this layer from https://github.com/dornerworks/meta-xzd web page but when I look into the cloned folder, there is nothing inside. This is what I get when I clone the layer:

generalelectric@mlan11214m117linux:/opt/yocto_GE/poky/master$ sudo git clone https://github.com/dornerworks/meta-xzd
Cloning into 'meta-xzd'...
remote: Counting objects: 416, done.
remote: Total 416 (delta 0), reused 0 (delta 0), pack-reused 416
Receiving objects: 100% (416/416), 856.59 KiB | 542.00 KiB/s, done.
Resolving deltas: 100% (182/182), done.
Checking connectivity... done.
warning: remote HEAD refers to nonexistent ref, unable to checkout.

generalelectric@mlan11214m117linux:/opt/yocto_GE/poky/master$
generalelectric@mlan11214m117linux:/opt/yocto_GE/poky/master$ ls meta-xzd/
generalelectric@mlan11214m117linux:/opt/yocto_GE/poky/master$

Other thing is that even that you have told me to use xen-domu-image-minimal command, I have the next error:

generalelectric@mlan11214m117linux:/opt/yocto_GE/poky/master/build$ bitbake xen-domu-image-minimal
Loading cache: 100% |############################################| Time: 0:00:00
Loaded 2865 entries from dependency cache.
Parsing recipes: 100% |##########################################| Time: 0:00:01
Parsing of 2097 .bb files complete (2080 cached, 17 parsed). 2881 targets, 198 skipped, 0 masked, 0 errors.
WARNING: No recipes available for:
/opt/yocto_GE/poky/master/meta-petalinux/recipes-devtools/python/python-smartpm_%.bbappend
ERROR: Nothing PROVIDES 'xen-domu-image-minimal'. Close matches:
xen-image-minimal
xen-guest-image-minimal
kvm-image-minimal

So I supose that I need to execute the xen-guest-image-minimal option. Is this true?

Finally if I execute the second option, I have the next error:

generalelectric@mlan11214m117linux:/opt/yocto_GE/poky/master/build$ bitbake xen-guest-image-minimal
Loading cache: 100% |############################################| Time: 0:00:00
Loaded 2865 entries from dependency cache.
Parsing recipes: 100% |##########################################| Time: 0:00:01
Parsing of 2097 .bb files complete (2080 cached, 17 parsed). 2881 targets, 198 skipped, 0 masked, 0 errors.
WARNING: No recipes available for:
/opt/yocto_GE/poky/master/meta-petalinux/recipes-devtools/python/python-smartpm_%.bbappend
NOTE: Resolving any missing task queue dependencies
ERROR: Nothing RPROVIDES 'xf86-video-vesa' (but /opt/yocto_GE/poky/master/meta-virtualization/recipes-extended/images/xen-guest-image-minimal.bb RDEPENDS on or otherwise requires it)
ERROR: xf86-video-vesa was skipped: incompatible with host aarch64-poky-linux (not in COMPATIBLE_HOST)
NOTE: Runtime target 'xf86-video-vesa' is unbuildable, removing...
Missing or unbuildable dependency chain was: ['xf86-video-vesa']
ERROR: Required build target 'xen-guest-image-minimal' has no buildable providers.
Missing or unbuildable dependency chain was: ['xen-guest-image-minimal', 'xf86-video-vesa']

Summary: There was 1 WARNING message shown.
Summary: There were 2 ERROR messages shown, returning a non-zero exit code.

I don't know if you can help me, but anywise any answer will be welcome.

Best regards,
Pello

Nathan.Studer · 05-17-2017, 12:14 PM

(05-17-2017, 07:53 AM)pello.heriz Wrote: I'm trying to build a xen-domu-image-minimal using meta-xzd layer. Anyway, I have try to clone this layer from https://github.com/dornerworks/meta-xzd web page but when I look into the cloned folder, there is nothing inside. This is what I get when I clone the layer:

generalelectric@mlan11214m117linux:/opt/yocto_GE/poky/master$ sudo git clone https://github.com/dornerworks/meta-xzd
Cloning into 'meta-xzd'...
remote: Counting objects: 416, done.
remote: Total 416 (delta 0), reused 0 (delta 0), pack-reused 416
Receiving objects: 100% (416/416), 856.59 KiB | 542.00 KiB/s, done.
Resolving deltas: 100% (182/182), done.
Checking connectivity... done.
warning: remote HEAD refers to nonexistent ref, unable to checkout.

generalelectric@mlan11214m117linux:/opt/yocto_GE/poky/master$
generalelectric@mlan11214m117linux:/opt/yocto_GE/poky/master$ ls meta-xzd/
generalelectric@mlan11214m117linux:/opt/yocto_GE/poky/master$

It looks like you are in the master branch, which doesn't have anything in it. Switch to the `dw-v2016.4` branch and you should see all the layer files.

Quote:pello.heriz

Other thing is that even that you have told me to use xen-domu-image-minimal command, I have the next error:

generalelectric@mlan11214m117linux:/opt/yocto_GE/poky/master/build$ bitbake xen-domu-image-minimal
Loading cache: 100% |############################################| Time: 0:00:00
Loaded 2865 entries from dependency cache.
Parsing recipes: 100% |##########################################| Time: 0:00:01
Parsing of 2097 .bb files complete (2080 cached, 17 parsed). 2881 targets, 198 skipped, 0 masked, 0 errors.
WARNING: No recipes available for:
/opt/yocto_GE/poky/master/meta-petalinux/recipes-devtools/python/python-smartpm_%.bbappend
ERROR: Nothing PROVIDES 'xen-domu-image-minimal'. Close matches:
xen-image-minimal
xen-guest-image-minimal
kvm-image-minimal

So I supose that I need to execute the xen-guest-image-minimal option. Is this true?

Once you are in the correct branch you can run 'bitbake xzd-domu-image-minimal', which should work without issue.

You can instead use meta-virtualization's xen-guest-image-minimal, if you wish.

Quote:pello.heriz

Finally if I execute the second option, I have the next error:

generalelectric@mlan11214m117linux:/opt/yocto_GE/poky/master/build$ bitbake xen-guest-image-minimal
Loading cache: 100% |############################################| Time: 0:00:00
Loaded 2865 entries from dependency cache.
Parsing recipes: 100% |##########################################| Time: 0:00:01
Parsing of 2097 .bb files complete (2080 cached, 17 parsed). 2881 targets, 198 skipped, 0 masked, 0 errors.
WARNING: No recipes available for:
/opt/yocto_GE/poky/master/meta-petalinux/recipes-devtools/python/python-smartpm_%.bbappend
NOTE: Resolving any missing task queue dependencies
ERROR: Nothing RPROVIDES 'xf86-video-vesa' (but /opt/yocto_GE/poky/master/meta-virtualization/recipes-extended/images/xen-guest-image-minimal.bb RDEPENDS on or otherwise requires it)
ERROR: xf86-video-vesa was skipped: incompatible with host aarch64-poky-linux (not in COMPATIBLE_HOST)
NOTE: Runtime target 'xf86-video-vesa' is unbuildable, removing...
Missing or unbuildable dependency chain was: ['xf86-video-vesa']
ERROR: Required build target 'xen-guest-image-minimal' has no buildable providers.
Missing or unbuildable dependency chain was: ['xen-guest-image-minimal', 'xf86-video-vesa']

Summary: There was 1 WARNING message shown.
Summary: There were 2 ERROR messages shown, returning a non-zero exit code.

I don't know if you can help me, but anywise any answer will be welcome.

Best regards,
Pello

However, xen-guest-image-minimal will attempt to include some x86 specific packages since x11 is an included feature of the distro, which will result in the errors you posted. Either create your own image without these packages or remove the packages from the xen-guest-image-minimal. The former is what xzd-domu-image-minimal is: xen-guest-image-minimal with x86 specific packages removed.

'xf86-video-vesa' and 'xf86-video-fbdev' are the packages in question.

Nate

pello.heriz · 05-18-2017, 07:29 AM

Hi Nathan,

I have switch to the dw-v2016.4 branch and cloned the meta-xzd layer correctly. Anyway, I have tried to execute 'bitbake xzd-domu-image-minimal' command and I have a problem related with a license. Here it is:

ou can also run generated qemu images with a command like 'runqemu qemux86'
generalelectric@mlan11214m117linux:/opt/yocto_GE/poky/master/build$ bitbake xzd-domu-image-minimal
NOTE: /opt/yocto_GE/poky/master/meta-qt5/recipes-qt/qt5/qtbase_git.bb: base_contains is deprecated, please use bb.utils.contains instead.
NOTE: /opt/yocto_GE/poky/master/meta-qt5/recipes-qt/qt5/qtbase_git.bb: base_contains is deprecated, please use bb.utils.contains instead.
ERROR: /opt/yocto_GE/poky/master/meta-xzd/recipes-bsp/u-boot/u-boot-xlnx_xzd.bb: This recipe does not have the LICENSE field set (u-boot-xlnx)
ERROR: Failed to parse recipe: /opt/yocto_GE/poky/master/meta-xzd/recipes-bsp/u-boot/u-boot-xlnx_xzd.bb

What do I need to do in order to get that license or avoid this issue? Do I need to clone this https://github.com/Xilinx/u-boot-xlnx/tr...r/Licenses and add u-boot-xlnx to my bblayers or something similar to that?

Best regards,
Pello

Login
Username:
Password:	Lost Password?
	Remember me