Most up to date blktap2 kernel driver

Most up to date blktap2 kernel driver

[Roger Pau Monné]

Hello,

I'm trying to port the blktap2 driver to NetBSD using a pure
user-space solution. Since blktap is not included in the kernel, I
would like to know where I can find the most up to date version of the
code, I've seen several repositories that contain blktap2 code, like:

http://xenbits.xen.org/linux-2.6.18-xen.hg
http://xenbits.xensource.com/gitweb/?p=people/dstodden/blktap.git;a=summary
http://xenbits.xensource.com/gitweb/?p=people/dstodden/linux.git;a=summary
(I cannot access the tree of this one, I get a 404 error)

Whats the best one to fetch to get a current version of the blktap2 code?

Thanks, Roger.

Re: Most up to date blktap2 kernel driver

[Pasi Kärkkäinen]

On Wed, Aug 24, 2011 at 10:09:10AM +0200, Roger Pau Monné wrote:
> Hello,
> 
> I'm trying to port the blktap2 driver to NetBSD using a pure
> user-space solution. Since blktap is not included in the kernel, I
> would like to know where I can find the most up to date version of the
> code, I've seen several repositories that contain blktap2 code, like:
> 
> http://xenbits.xen.org/linux-2.6.18-xen.hg
> http://xenbits.xensource.com/gitweb/?p=people/dstodden/blktap.git;a=summary
> http://xenbits.xensource.com/gitweb/?p=people/dstodden/linux.git;a=summary
> (I cannot access the tree of this one, I get a 404 error)
> 
> Whats the best one to fetch to get a current version of the blktap2 code?
> 

I'm not sure which one is the latest but blktap2 can also be found from:

Jeremy's xen.git branch "xen/stable-2.6.32.x"
http://git.eu.kernel.org/?p=linux/kernel/git/jeremy/xen.git;a=shortlog;h=xen/stable-2.6.32.x

and from XenServer/XCP 2.6.32 xenlinux kernel:
http://xenbits.xen.org/XCP/linux-2.6.32.pq.hg

Daniel (cc) can probably comment..

-- Pasi

Re: Most up to date blktap2 kernel driver

[Ian Campbell]

On Wed, 2011-08-24 at 15:15 +0100, Pasi Kärkkäinen wrote:
> On Wed, Aug 24, 2011 at 10:09:10AM +0200, Roger Pau Monné wrote:
> > Hello,
> > 
> > I'm trying to port the blktap2 driver to NetBSD using a pure
> > user-space solution.

Thanks! This would be awesome to have. We're interested in taking this
same approach for Linux in the future since (as you've noticed) blktap
is not in the upstream kernel

>  Since blktap is not included in the kernel, I
> > would like to know where I can find the most up to date version of the
> > code, I've seen several repositories that contain blktap2 code, like:
> > 
> > http://xenbits.xen.org/linux-2.6.18-xen.hg
> > http://xenbits.xensource.com/gitweb/?p=people/dstodden/blktap.git;a=summary
> > http://xenbits.xensource.com/gitweb/?p=people/dstodden/linux.git;a=summary
> > (I cannot access the tree of this one, I get a 404 error)
> > 
> > Whats the best one to fetch to get a current version of the blktap2 code?
> > 
> 
> I'm not sure which one is the latest but blktap2 can also be found from:
> 
> Jeremy's xen.git branch "xen/stable-2.6.32.x"
> http://git.eu.kernel.org/?p=linux/kernel/git/jeremy/xen.git;a=shortlog;h=xen/stable-2.6.32.x
> 
> and from XenServer/XCP 2.6.32 xenlinux kernel:
> http://xenbits.xen.org/XCP/linux-2.6.32.pq.hg
> 
> Daniel (cc) can probably comment..

FWIW I'd probably lean towards the xen.git one but either is probably a
good start.

The Xen backend framework in upstream qemu might also be interesting in
this context?

Does NetBSD have an equivalent of (at least) /dev/xen/gntdev
and /dev/xen/evtchn? I think they are a requirement to make this work.

Ian.

Re: Most up to date blktap2 kernel driver

[Stefano Stabellini]

On Wed, 24 Aug 2011, Ian Campbell wrote:
> The Xen backend framework in upstream qemu might also be interesting in
> this context?

Give a look at hw/xen_backend.c and hw/xen_disk.c in upstream qemu: they
provide the basic infrustructure to run xen backends in userspace and
the blkback userspace code.


> Does NetBSD have an equivalent of (at least) /dev/xen/gntdev
> and /dev/xen/evtchn? I think they are a requirement to make this work.

You'll need a way to receive event channel notifications and a way to map
grant references in userspace.

Re: Most up to date blktap2 kernel driver

[Roger Pau Monné]

2011/8/24 Stefano Stabellini <stefano.stabellini@eu.citrix.com>:
> On Wed, 24 Aug 2011, Ian Campbell wrote:
>> The Xen backend framework in upstream qemu might also be interesting in
>> this context?
>
> Give a look at hw/xen_backend.c and hw/xen_disk.c in upstream qemu: they
> provide the basic infrustructure to run xen backends in userspace and
> the blkback userspace code.

This looks interesting, it provides an example of how to interact with
grant tables from user space.

>> Does NetBSD have an equivalent of (at least) /dev/xen/gntdev
>> and /dev/xen/evtchn? I think they are a requirement to make this work.
>
> You'll need a way to receive event channel notifications and a way to map
> grant references in userspace.


NetBSD has a device related to Xen events I think (/dev/xenevt), but
hasn't got any gntdev, I'm not sure if that's necessary, I saw that
hw/xen_block.c uses xc_gnttab functions to interact with grant tables,
and they work on NetBSD event though it has no gntdev (but has the
necessary hypercall HYPERVISOR_grant_table_op, I still have to figure
out how is that called from user space).

Regarding kernels, xen.git seems to be missing the blktap2 driver:

http://git.eu.kernel.org/?p=linux/kernel/git/jeremy/xen.git;a=tree;f=drivers/xen;h=60ce8f2fa68504158843caefdbb8c6f149a37edf;hb=xen/stable-2.6.32.x

I only see a blktap folder, and no blktap2. Now I'm reading the code from:

http://xenbits.xen.org/hg/linux-2.6.18-xen.hg/file/0b8a9b22b70a/drivers/xen/blktap2

I have a couple of books regarding Xen, the most helpful one being
"The Definitive Guide to the Xen Hypervisor" by David Chisnall, it has
been very helpful in understanding grant tables, but it doesn't have
any information regarding blktap/blktap2 driver internals, and I
cannot find any information apart from the code itself. Is there any
book that contains helpful information about the blktap
implementation?

Thanks, Roger.

Re: Most up to date blktap2 kernel driver

[Ian Campbell]

On Thu, 2011-08-25 at 12:45 +0100, Roger Pau Monné wrote:
> Regarding kernels, xen.git seems to be missing the blktap2 driver:
> 
> http://git.eu.kernel.org/?p=linux/kernel/git/jeremy/xen.git;a=tree;f=drivers/xen;h=60ce8f2fa68504158843caefdbb8c6f149a37edf;hb=xen/stable-2.6.32.x
> 
> I only see a blktap folder, and no blktap2. Now I'm reading the code from:

The contents of drivers/blktap in that tree is the blktap2 kernel side.

> http://xenbits.xen.org/hg/linux-2.6.18-xen.hg/file/0b8a9b22b70a/drivers/xen/blktap2
> 
> I have a couple of books regarding Xen, the most helpful one being
> "The Definitive Guide to the Xen Hypervisor" by David Chisnall, it has
> been very helpful in understanding grant tables, but it doesn't have
> any information regarding blktap/blktap2 driver internals, and I
> cannot find any information apart from the code itself. Is there any
> book that contains helpful information about the blktap
> implementation?

Sorry, not that I know of.

Ian.

Re: Most up to date blktap2 kernel driver

[Ian Campbell]

On Wed, 2011-08-24 at 09:09 +0100, Roger Pau Monné wrote:
> Hello,
> 
> I'm trying to port the blktap2 driver to NetBSD using a pure
> user-space solution.

I was just wondering about this, is your plan to bolt a block frontend
onto the existing tapdisk2 process or are you building a thing using a
NetBSD library for doing block devices in userspace (I think you are
doing the former but either you or Christoph mentioned that library
thing at one point too).

The former plan would be awesome since it would make this work useful on
all platforms and not just NetBSD. As you've no doubt noticed the
blktap2 kernel side isn't (and won't be) in Linux upstream either.

Ian.

>  Since blktap is not included in the kernel, I
> would like to know where I can find the most up to date version of the
> code, I've seen several repositories that contain blktap2 code, like:
> 
> http://xenbits.xen.org/linux-2.6.18-xen.hg
> http://xenbits.xensource.com/gitweb/?p=people/dstodden/blktap.git;a=summary
> http://xenbits.xensource.com/gitweb/?p=people/dstodden/linux.git;a=summary
> (I cannot access the tree of this one, I get a 404 error)
> 
> Whats the best one to fetch to get a current version of the blktap2 code?
> 
> Thanks, Roger.
> 
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@lists.xensource.com
> http://lists.xensource.com/xen-devel

Re: Most up to date blktap2 kernel driver

[Roger Pau Monné]

> I was just wondering about this, is your plan to bolt a block frontend
> onto the existing tapdisk2 process or are you building a thing using a
> NetBSD library for doing block devices in userspace (I think you are
> doing the former but either you or Christoph mentioned that library
> thing at one point too).

Well, my first try was to create a user-space implementation of the
block devices found in blktap, so there was no need to change the
current blktap tools. This is however not possible right now, because
the library for implementing block and char devices as user-space
daemons doesn't support mmap, and there's no clear way about how to
implement this.

Correct me if I'm wrong, but the second option would be to directly
implement the I/O ring from user-space without using any devices, I
haven't had much time to look into this, but I think this is the best
way, because it doesn't involve the constant context changes that the
implementation explained before has. I hope that looking into qemu
disk backend file can provide me with some information about how to do
this, possibly using libxc?

> The former plan would be awesome since it would make this work useful on
> all platforms and not just NetBSD. As you've no doubt noticed the
> blktap2 kernel side isn't (and won't be) in Linux upstream either.
>
> Ian.

Re: Most up to date blktap2 kernel driver

[Ian Campbell]

On Thu, 2011-09-22 at 09:59 +0100, Roger Pau Monné wrote:
> > I was just wondering about this, is your plan to bolt a block frontend
> > onto the existing tapdisk2 process or are you building a thing using a
> > NetBSD library for doing block devices in userspace (I think you are
> > doing the former but either you or Christoph mentioned that library
> > thing at one point too).
> 
> Well, my first try was to create a user-space implementation of the
> block devices found in blktap, so there was no need to change the
> current blktap tools. This is however not possible right now, because
> the library for implementing block and char devices as user-space
> daemons doesn't support mmap, and there's no clear way about how to
> implement this.

Right. And that would be very OS specific I guess?

> Correct me if I'm wrong, but the second option would be to directly
> implement the I/O ring from user-space without using any devices,

Yes, that is what I was suggesting. I think we generally have consensus
that this is the right direction to take blktap.

> I haven't had much time to look into this, but I think this is the best
> way, because it doesn't involve the constant context changes that the
> implementation explained before has. I hope that looking into qemu
> disk backend file can provide me with some information about how to do
> this, possibly using libxc?

Using libxc would be better than using the OS provided interfaces
directly since it abstracts out the differences between the
ioctls/devices on different OSs.

> 
> > The former plan would be awesome since it would make this work useful on
> > all platforms and not just NetBSD. As you've no doubt noticed the
> > blktap2 kernel side isn't (and won't be) in Linux upstream either.
> >
> > Ian.

Re: Most up to date blktap2 kernel driver

[Stefano Stabellini]

On Thu, 22 Sep 2011, Roger Pau Monn? wrote:
> > I was just wondering about this, is your plan to bolt a block frontend
> > onto the existing tapdisk2 process or are you building a thing using a
> > NetBSD library for doing block devices in userspace (I think you are
> > doing the former but either you or Christoph mentioned that library
> > thing at one point too).
> 
> Well, my first try was to create a user-space implementation of the
> block devices found in blktap, so there was no need to change the
> current blktap tools. This is however not possible right now, because
> the library for implementing block and char devices as user-space
> daemons doesn't support mmap, and there's no clear way about how to
> implement this.
> 
> Correct me if I'm wrong, but the second option would be to directly
> implement the I/O ring from user-space without using any devices, I
> haven't had much time to look into this, but I think this is the best
> way, because it doesn't involve the constant context changes that the
> implementation explained before has. I hope that looking into qemu
> disk backend file can provide me with some information about how to do
> this, possibly using libxc?

Give a look at hw/xen_backend.c and hw/xen_disk.c in upstream qemu,
they should give you an idea of what you need.
You would still need a way to access the grant table from userspace,
that on Linux is provided by the gntdev device, I hope there is
something similar in NetBSD.