Skip to contentSkip to conversation

qemu needs target kernel sets CONFIG_SCSI_VIRTIO=y to boot vmdk qcow2 and vdi

DISCUSSION Robert Yang 3 participants 7 messages Feb 25, 2016 – Feb 26, 2016

lore
Share & tools
lore
#0 lore raw

Hi Bruce and Darren,

We need enable target kernel's CONFIG_SCSI_VIRTIO=y to make native qemu can boot hdddirect, vmdk, qcow2 and vdi, otherwise, the scsi drive can't work in qemu since our vm image's root default to /dev/sda2, or we can change it default to /dev/hda2, then we don't have to make any changes on kernel, but I prefer scsi since ide is going to be out of date.

I've tried to enable it by menuconfig, it works well, would you please enable it by default ?

$ diff .config.old .config -Nur
--- .config.old 2016-02-25 01:04:25.816381913 -0800
+++ .config     2016-02-25 01:16:31.332232759 -0800
@@ -1479,7 +1479,7 @@
 # CONFIG_SCSI_DEBUG is not set
 # CONFIG_SCSI_PMCRAID is not set
 # CONFIG_SCSI_PM8001 is not set
-# CONFIG_SCSI_VIRTIO is not set
+CONFIG_SCSI_VIRTIO=y
 # CONFIG_SCSI_DH is not set
 # CONFIG_SCSI_OSD_INITIATOR is not set
 CONFIG_ATA=y
$ du bzImage bzImage_new
6752    bzImage
6756    bzImage_new

It nearly has no impact on the size.

Signature · 3 lines
Thanks

Robert
#1 lore raw
Quoted 6 lines · Robert Yang

We need enable target kernel's CONFIG_SCSI_VIRTIO=y to make native qemu can boot hdddirect, vmdk, qcow2 and vdi, otherwise, the scsi drive can't work in qemu since our vm image's root default to /dev/sda2, or we can change it default to /dev/hda2, then we don't have to make any changes on kernel, but I prefer scsi since ide is going to be out of date.

Are there any other virtio drivers that we really should be enabling in the qemu kernels?

Ross

#4 lore raw
In reply to Robert Yang
Quoted 8 lines · Robert Yang

Hi Bruce and Darren,

We need enable target kernel's CONFIG_SCSI_VIRTIO=y to make native qemu can boot hdddirect, vmdk, qcow2 and vdi, otherwise, the scsi drive can't work in qemu since our vm image's root default to /dev/sda2, or we can change it default to /dev/hda2, then we don't have to make any changes on kernel, but I prefer scsi since ide is going to be out of date.

The IDE fragment was on for compatibility reasons, but we dropped that compatibility about a year ago. I was under the impression that we had fully dropped all the IDE fragments.

Quoted 4 lines

I've tried to enable it by menuconfig, it works well, would you please enable it by default ?

Send a patch to the linux-yocto mailing list to change the virtio configuration
fragment.

You are already set up to test it out, so it is much easier.

I don't see why this shouldn't follow the normal patch flow for configuration
frag changes.

Cheers,

Bruce

Quoted 31 lines

$ diff .config.old .config -Nur
.config 2016-02-25 01:16:31.332232759 -0800 renamed from .config.old 2016-02-25 01:04:25.816381913 -0800 +1 -1
@@ -1479,7 +1479,7 @@
 # CONFIG_SCSI_DEBUG is not set
 # CONFIG_SCSI_PMCRAID is not set
 # CONFIG_SCSI_PM8001 is not set
-# CONFIG_SCSI_VIRTIO is not set
+CONFIG_SCSI_VIRTIO=y
 # CONFIG_SCSI_DH is not set
 # CONFIG_SCSI_OSD_INITIATOR is not set
 CONFIG_ATA=y
$ du bzImage bzImage_new
6752    bzImage
6756    bzImage_new

It nearly has no impact on the size.

--
Thanks

Robert
--
_______________________________________________
Openembedded-core mailing list
[email protected]
http://lists.openembedded.org/mailman/listinfo/openembedded-core

Signature · 2 lines
"Thou shalt not follow the NULL pointer, for chaos and madness await thee
at its end"
#2 lore raw
In reply to Burton, Ross
Quoted 11 lines · Ross
Quoted 6 lines · Robert Yang

We need enable target kernel's CONFIG_SCSI_VIRTIO=y to make native qemu can boot hdddirect, vmdk, qcow2 and vdi, otherwise, the scsi drive can't work in qemu since our vm image's root default to /dev/sda2, or we can change it default to /dev/hda2, then we don't have to make any changes on kernel, but I prefer scsi since ide is going to be out of date.

Quoted 3 lines

Are there any other virtio drivers that we really should be enabling in the qemu kernels?

We already have a fragment for this, and a set of virtio drivers enabled .. which
is why I find this entire thread confusing :)

Bruce

Quoted 10 lines

Ross

--
_______________________________________________
Openembedded-core mailing list
[email protected]
http://lists.openembedded.org/mailman/listinfo/openembedded-core

Signature · 2 lines
"Thou shalt not follow the NULL pointer, for chaos and madness await thee
at its end"
#5 lore raw
In reply to Bruce Ashfield
Quoted 14 lines · Bruce Ashfield
Quoted 8 lines · Robert Yang

Hi Bruce and Darren,

We need enable target kernel's CONFIG_SCSI_VIRTIO=y to make native qemu can boot hdddirect, vmdk, qcow2 and vdi, otherwise, the scsi drive can't work in qemu since our vm image's root default to /dev/sda2, or we can change it default to /dev/hda2, then we don't have to make any changes on kernel, but I prefer scsi since ide is going to be out of date.

Quoted 4 lines

The IDE fragment was on for compatibility reasons, but we dropped that compatibility about a year ago. I was under the impression that we had fully dropped all the IDE fragments.

Adding Cal and Saul, since we are working through all these configs in another effort, so I wanted to get their thoughts as well.

I had a chance to go back and look at the configs, and even though I can find the email where it has been proposed, and I've agreed, to shoot CONFIG_IDE
in the head, and use generic SCSI instead .. config IDE is still in
common-pc-drivers.cfg
which means that x86 board, and the x86 kvm guest are picking it up.

Cal/Saul: does this match what you are seeing ?

Quoted 11 lines
Quoted 5 lines

I've tried to enable it by menuconfig, it works well, would you please enable it by default ?

Send a patch to the linux-yocto mailing list to change the virtio configuration
fragment.

Quoted 6 lines

You are already set up to test it out, so it is much easier.

I don't see why this shouldn't follow the normal patch flow for configuration
frag changes.

Back to this. We just need a patch to the linux-yocto list, for the configuration
change, and we can evaluate it like any other patch!

Cheers,

Bruce

Quoted 44 lines

Cheers,

Bruce

Quoted 24 lines

$ diff .config.old .config -Nur
.config 2016-02-25 01:16:31.332232759 -0800 renamed from .config.old 2016-02-25 01:04:25.816381913 -0800 +1 -1
@@ -1479,7 +1479,7 @@
 # CONFIG_SCSI_DEBUG is not set
 # CONFIG_SCSI_PMCRAID is not set
 # CONFIG_SCSI_PM8001 is not set
-# CONFIG_SCSI_VIRTIO is not set
+CONFIG_SCSI_VIRTIO=y
 # CONFIG_SCSI_DH is not set
 # CONFIG_SCSI_OSD_INITIATOR is not set
 CONFIG_ATA=y
$ du bzImage bzImage_new
6752    bzImage
6756    bzImage_new

It nearly has no impact on the size.

--
Thanks

Quoted 7 lines

Robert
--
_______________________________________________
Openembedded-core mailing list
[email protected]
http://lists.openembedded.org/mailman/listinfo/openembedded-core

--
"Thou shalt not follow the NULL pointer, for chaos and madness await thee at its end"

Signature · 2 lines
"Thou shalt not follow the NULL pointer, for chaos and madness await thee
at its end"
#6 lore raw
In reply to Bruce Ashfield
Quoted 24 lines · Bruce Ashfield

On Thu, Feb 25, 2016 at 5:36 AM, Robert Yang <[email protected] <mailto:[email protected]>> wrote:

Hi Bruce and Darren,

We need enable target kernel's CONFIG_SCSI_VIRTIO=y to make native qemu can boot hdddirect, vmdk, qcow2 and vdi, otherwise, the scsi drive can't work in qemu since our vm image's root default to /dev/sda2, or we can change it default to /dev/hda2, then we don't have to make any changes on kernel, but I prefer scsi since ide is going to be out of date.

The IDE fragment was on for compatibility reasons, but we dropped that compatibility about a year ago. I was under the impression that we had fully dropped all the IDE fragments.

I've tried to enable it by menuconfig, it works well, would you please enable it by default ?

Quoted 7 lines

Send a patch to the linux-yocto mailing list to change the virtio configuration fragment.

You are already set up to test it out, so it is much easier.

I don't see why this shouldn't follow the normal patch flow for configuration frag changes.

Sorry, the problem was that I didn't know how to send patch for yocto kernel:-). I will send the patch to linux-yocto.

// Robert

Quoted 43 lines

Cheers,

Bruce

    $ diff .config.old .config -Nur
    --- .config.old 2016-02-25 01:04:25.816381913 -0800
    +++ .config     2016-02-25 01:16:31.332232759 -0800
    @@ -1479,7 +1479,7 @@
      # CONFIG_SCSI_DEBUG is not set
      # CONFIG_SCSI_PMCRAID is not set
      # CONFIG_SCSI_PM8001 is not set
    -# CONFIG_SCSI_VIRTIO is not set
    +CONFIG_SCSI_VIRTIO=y
      # CONFIG_SCSI_DH is not set
      # CONFIG_SCSI_OSD_INITIATOR is not set
      CONFIG_ATA=y
    $ du bzImage bzImage_new
    6752    bzImage
    6756    bzImage_new
    It nearly has no impact on the size.

    --
    Thanks

    Robert
    --
    _______________________________________________
    Openembedded-core mailing list
    [email protected]
    <mailto:[email protected]>
    http://lists.openembedded.org/mailman/listinfo/openembedded-core

--
"Thou shalt not follow the NULL pointer, for chaos and madness await thee at its end"

#3 lore raw
In reply to Bruce Ashfield
Quoted 16 lines · Bruce Ashfield

On Thu, Feb 25, 2016 at 7:10 AM, Burton, Ross <[email protected] <mailto:[email protected]>> wrote:

On 25 February 2016 at 10:36, Robert Yang <[email protected] <mailto:[email protected]>> wrote:

We need enable target kernel's CONFIG_SCSI_VIRTIO=y to make native qemu can boot hdddirect, vmdk, qcow2 and vdi, otherwise, the scsi drive can't work in qemu since our vm image's root default to /dev/sda2, or we can change it default to /dev/hda2, then we don't have to make any changes on kernel, but I prefer scsi since ide is going to be out of date.

Quoted 5 lines

Are there any other virtio drivers that we really should be enabling in the qemu kernels?

We already have a fragment for this, and a set of virtio drivers enabled .. which

Yes, others are fine AFAIK.

// Robert

Quoted 20 lines

is why I find this entire thread confusing :)

Bruce

    Ross

    --
    _______________________________________________
    Openembedded-core mailing list
    [email protected]
    <mailto:[email protected]>
    http://lists.openembedded.org/mailman/listinfo/openembedded-core

--
"Thou shalt not follow the NULL pointer, for chaos and madness await thee at its end"