All of lore.kernel.org
 help / color / mirror / Atom feed
* [uml-devel] Using UML in libguestfs
@ 2013-08-09 20:43 Richard W.M. Jones
  2013-08-09 20:57 ` richard -rw- weinberger
                   ` (3 more replies)
  0 siblings, 4 replies; 17+ messages in thread
From: Richard W.M. Jones @ 2013-08-09 20:43 UTC (permalink / raw)
  To: user-mode-linux-devel

Today I wrote some patches (for libguestfs) which add an experimental
UML backend:

  https://www.redhat.com/archives/libguestfs/2013-August/msg00005.html

There are some observations from this.

* 'ubd<N>=cow,original' is not parsed correctly.  This was reported a
while back and I'm seeing the same thing with the latest kernel from
git:

  http://thread.gmane.org/gmane.linux.uml.devel/13556

If I have some time I'll have a go at fixing this since it seems like
a simple regression.

* I'm trying to use a serial port to communicate between guest and
host, using ssl3=fd:<N>.  This causes the Linux process to crash soon
(but not immediately) after the guest has opened /dev/ttyS3:

  spurious interrupt in ubd_handler, err = -4

I've put the full call trace at the end of the email because it's
quite long[1].  This was also reported before, although the suggested
patch did not fix it:

  http://marc.info/?l=linux-kernel&m=123242364211797&w=2

I wonder if anyone has used the fd:<N> option in anger?  I will try
using a Unix domain socket and see if it makes a difference.

* UML is pretty fast!  It's certainly faster (by a factor of > 5) than
spinning up a lightweight KVM VM.

Rich.

[1] Call trace

spurious interrupt in ubd_handler, err = -4

Modules linked in:
Pid: 1614169860, comm:  Not tainted 3.11.0-rc4-00103-g6c2580c
RIP: 0033:[<000000006001db74>]
RSP: 0000000080ca9f98  EFLAGS: 00010206
RAX: 0000000000000008 RBX: 0000000000000008 RCX: 0000003f2cae6970
RDX: 0000000000000000 RSI: 0000000080ca9fb0 RDI: 0000000000000005
RBP: 0000000080ca9fe8 R08: 0000000080ca9ed8 R09: 0000000080ca9e40
R10: 0000000000000008 R11: 0000000000000246 R12: 0000000000001000
R13: 0000000080e867c4 R14: 0000000000000001 R15: 68c00000000080ea
Call Trace: 
80ca9538:  [<600151ac>]------------[ cut here ]------------
WARNING: CPU: 0 PID: 6 at block/blk.h:210 generic_make_request_checks+0x1aa/0x234()
BUG: recent printk recursion!
80ca9648:  [<60022207>] sig_handler_common+0x8c/0xa2
80ca96d0:  [<6001db74>] io_thread+0x63/0x243
80ca9988:  [<600222c2>] sig_handler+0x2e/0x39
80ca99a8:  [<60021f33>] hard_handler+0x83/0xab
80ca9aa8:  [<6001db74>] io_thread+0x63/0x243
80ca9d88:  [<60021eb0>] hard_handler+0x0/0xab
80ca9f70:  [<60020308>] os_seek_file+0x11/0x27
80ca9f80:  [<6002032a>] os_read_file+0xc/0x1f
80ca9f90:  [<6001db3b>] io_thread+0x2a/0x243

Kernel panic - not syncing: Segfault with no mm
CPU: 0 PID: 1614169860 Comm:  Not tainted 3.11.0-rc4-00103-g6c2580c #3

Modules linked in:
Pid: 1614169860, comm:  Not tainted 3.11.0-rc4-00103-g6c2580c
RIP: 0033:[<0000000060026715>]
RSP: 0000000080ca9328  EFLAGS: 00010246
RAX: 0000000080ca9398 RBX: 00000000602d2460 RCX: 0000000080ca8000
RDX: 0000000000000008 RSI: 636f6c6e6f6e5f88 RDI: 0000000080ca9398
RBP: 0000000080ca9350 R08: 636f6c6e6f6e5f70 R09: 636f6c6e6f6e5f70
R10: 0000003f2ca892da R11: 0000003f2ca892da R12: 000000006026b27c
R13: 000000006001db74 R14: 0000000000000004 R15: 0000000000000000
Call Trace: 
80ca88b8:  [<600151ac>] segv+0x66/0x241
80ca8988:  [<600153e9>] segv_handler+0x62/0x6c
80ca89c8:  [<60022207>] sig_handler_common+0x8c/0xa2
80ca89e0:  [<6001db74>] io_thread+0x63/0x243
80ca8a50:  [<60026715>] memcpy+0xb5/0x140
80ca8d08:  [<600222c2>] sig_handler+0x2e/0x39
80ca8d28:  [<60021f33>] hard_handler+0x83/0xab
80ca8d58:  [<6001db74>] io_thread+0x63/0x243
80ca8dd0:  [<6001db74>] io_thread+0x63/0x243
80ca8e28:  [<60026715>] memcpy+0xb5/0x140
80ca90e8:  [<60017226>] generic_write+0xf/0x38
80ca9108:  [<600172f9>] generic_console_write+0x3e/0x165
80ca9288:  [<6004c396>] vprintk_emit+0x2dd/0x2ee
80ca92d8:  [<6001db74>] io_thread+0x63/0x243
80ca92f8:  [<601edf38>] printk+0x97/0x99
80ca9308:  [<60015eff>] copy_from_user+0x36/0x8a
80ca9328:  [<60015eff>] copy_from_user+0x36/0x8a
80ca9358:  [<600699d4>] __probe_kernel_read+0x3c/0x6d
80ca9368:  [<6003f2ac>] probe_kthread_data+0x29/0x2f
80ca9388:  [<6003becc>] print_worker_info+0x66/0x124
80ca93f8:  [<6004d204>] dump_stack_print_info+0xa6/0xaf
80ca9428:  [<601efbeb>] dump_stack+0xe/0x19
80ca9438:  [<601ed8b7>] panic+0xf7/0x1e4
80ca9490:  [<60056982>] __module_text_address+0xd/0x4d
80ca94a8:  [<6005a007>] is_module_text_address+0x9/0x11
80ca94b8:  [<6003cd40>] __kernel_text_address+0x26/0x49
80ca94d8:  [<60014146>] show_trace+0x8e/0x95
80ca9508:  [<60026491>] show_regs+0x27/0x2a
80ca9520:  [<6001db74>] io_thread+0x63/0x243
80ca9538:  [<6001524f>] segv+0x109/0x241
80ca9608:  [<600153e9>] segv_handler+0x62/0x6c
80ca9628:  [<60022169>] unblock_signals+0x4a/0x5c
80ca9648:  [<60022207>] sig_handler_common+0x8c/0xa2
80ca96d0:  [<6001db74>] io_thread+0x63/0x243
80ca9988:  [<600222c2>] sig_handler+0x2e/0x39
80ca99a8:  [<60021f33>] hard_handler+0x83/0xab
80ca9aa8:  [<6001db74>] io_thread+0x63/0x243
80ca9d88:  [<60021eb0>] hard_handler+0x0/0xab
80ca9f70:  [<60020308>] os_seek_file+0x11/0x27
80ca9f80:  [<6002032a>] os_read_file+0xc/0x1f
80ca9f90:  [<6001db3b>] io_thread+0x2a/0x243

Kernel panic - not syncing: Segfault with no mm

Modules linked in:
Pid: 1614169860, comm:  Not tainted 3.11.0-rc4-00103-g6c2580c
RIP: 0000:[<000001a400000004>]
RSP: 0000000000000000  EFLAGS: 6002e720
RAX: 000000006002e770 RBX: 0000000000000000 RCX: 0000000000000000
RDX: 0000000000000000 RSI: 0000000000000000 RDI: 00000000602706ae
RBP: 0000000000000000 R08: 0000000000000000 R09: 000001a400000004
R10: 00000000602d04ac R11: 000000006027069b R12: 0000000000000000
R13: 000000006002e720 R14: 0000000000000000 R15: 000001a400000004
Call Trace: 
80ca8748:  [<6001564c>] panic_exit+0x2b/0x41
80ca8758:  [<6015a3ac>] vsnprintf+0xd0/0x431
80ca8768:  [<600427c7>] notifier_call_chain+0x39/0x5c
80ca8790:  [<60026715>] memcpy+0xb5/0x140
80ca87a8:  [<60042804>] atomic_notifier_call_chain+0xf/0x11
80ca87b8:  [<601ed8d2>] panic+0x112/0x1e4
80ca8810:  [<60056982>] __module_text_address+0xd/0x4d
80ca8828:  [<6005a007>] is_module_text_address+0x9/0x11
80ca8838:  [<6003cd40>] __kernel_text_address+0x26/0x49
80ca8858:  [<60014146>] show_trace+0x8e/0x95
80ca8888:  [<60026491>] show_regs+0x27/0x2a
80ca88a0:  [<60026715>] memcpy+0xb5/0x140
80ca88b8:  [<6001524f>] segv+0x109/0x241
80ca8988:  [<600153e9>] segv_handler+0x62/0x6c
80ca89c8:  [<60022207>] sig_handler_common+0x8c/0xa2
80ca89e0:  [<6001db74>] io_thread+0x63/0x243

-- 
Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones
virt-p2v converts physical machines to virtual machines.  Boot with a
live CD or over the network (PXE) and turn machines into KVM guests.
http://libguestfs.org/virt-v2v

------------------------------------------------------------------------------
Get 100% visibility into Java/.NET code with AppDynamics Lite!
It's a free troubleshooting tool designed for production.
Get down to code-level detail for bottlenecks, with <2% overhead. 
Download for free and get started troubleshooting in minutes. 
http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
_______________________________________________
User-mode-linux-devel mailing list
User-mode-linux-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/user-mode-linux-devel


^ permalink raw reply	[flat|nested] 17+ messages in thread
* Re: [uml-devel] Using UML in libguestfs
  2013-08-09 20:43 [uml-devel] Using UML in libguestfs Richard W.M. Jones
@ 2013-08-09 20:57 ` richard -rw- weinberger
  2013-08-09 21:31   ` Richard W.M. Jones
  2013-08-10  4:16   ` Mattia Dongili
  2013-08-09 21:15 ` Richard W.M. Jones
                   ` (2 subsequent siblings)
  3 siblings, 2 replies; 17+ messages in thread
From: richard -rw- weinberger @ 2013-08-09 20:57 UTC (permalink / raw)
  To: Richard W.M. Jones; +Cc: user-mode-linux-devel

On Fri, Aug 9, 2013 at 10:43 PM, Richard W.M. Jones <rjones@redhat.com> wrote:
> Today I wrote some patches (for libguestfs) which add an experimental
> UML backend:
>
>   https://www.redhat.com/archives/libguestfs/2013-August/msg00005.html
>
> There are some observations from this.
>
> * 'ubd<N>=cow,original' is not parsed correctly.  This was reported a
> while back and I'm seeing the same thing with the latest kernel from
> git:
>
>   http://thread.gmane.org/gmane.linux.uml.devel/13556
>
> If I have some time I'll have a go at fixing this since it seems like
> a simple regression.

Patches are very welcome!

> * I'm trying to use a serial port to communicate between guest and
> host, using ssl3=fd:<N>.  This causes the Linux process to crash soon
> (but not immediately) after the guest has opened /dev/ttyS3:
>
>   spurious interrupt in ubd_handler, err = -4
>
> I've put the full call trace at the end of the email because it's
> quite long[1].  This was also reported before, although the suggested
> patch did not fix it:
>
>   http://marc.info/?l=linux-kernel&m=123242364211797&w=2
>
> I wonder if anyone has used the fd:<N> option in anger?  I will try
> using a Unix domain socket and see if it makes a difference.

Hmmm. I *should* work.
Which fd are you using?
Can you also try pts?

> * UML is pretty fast!  It's certainly faster (by a factor of > 5) than
> spinning up a lightweight KVM VM.

Wow!
Does it only boot faster or is it also faster with reading data from a
filesystem?

Maybe it's time to exhume the libLinux project (UML as a normal
userspace library).
http://user-mode-linux.sourceforge.net/old/projects.html

Thanks,
//richard

> Rich.
>
> [1] Call trace
>
> spurious interrupt in ubd_handler, err = -4
>
> Modules linked in:
> Pid: 1614169860, comm:  Not tainted 3.11.0-rc4-00103-g6c2580c
> RIP: 0033:[<000000006001db74>]

Can you please decode that address using addr2line?

> RSP: 0000000080ca9f98  EFLAGS: 00010206
> RAX: 0000000000000008 RBX: 0000000000000008 RCX: 0000003f2cae6970
> RDX: 0000000000000000 RSI: 0000000080ca9fb0 RDI: 0000000000000005
> RBP: 0000000080ca9fe8 R08: 0000000080ca9ed8 R09: 0000000080ca9e40
> R10: 0000000000000008 R11: 0000000000000246 R12: 0000000000001000
> R13: 0000000080e867c4 R14: 0000000000000001 R15: 68c00000000080ea
> Call Trace:
> 80ca9538:  [<600151ac>]------------[ cut here ]------------
> WARNING: CPU: 0 PID: 6 at block/blk.h:210 generic_make_request_checks+0x1aa/0x234()
> BUG: recent printk recursion!
> 80ca9648:  [<60022207>] sig_handler_common+0x8c/0xa2
> 80ca96d0:  [<6001db74>] io_thread+0x63/0x243
> 80ca9988:  [<600222c2>] sig_handler+0x2e/0x39
> 80ca99a8:  [<60021f33>] hard_handler+0x83/0xab
> 80ca9aa8:  [<6001db74>] io_thread+0x63/0x243
> 80ca9d88:  [<60021eb0>] hard_handler+0x0/0xab
> 80ca9f70:  [<60020308>] os_seek_file+0x11/0x27
> 80ca9f80:  [<6002032a>] os_read_file+0xc/0x1f
> 80ca9f90:  [<6001db3b>] io_thread+0x2a/0x243
>
> Kernel panic - not syncing: Segfault with no mm
> CPU: 0 PID: 1614169860 Comm:  Not tainted 3.11.0-rc4-00103-g6c2580c #3
>
> Modules linked in:
> Pid: 1614169860, comm:  Not tainted 3.11.0-rc4-00103-g6c2580c
> RIP: 0033:[<0000000060026715>]

Ditto.

> RSP: 0000000080ca9328  EFLAGS: 00010246
> RAX: 0000000080ca9398 RBX: 00000000602d2460 RCX: 0000000080ca8000
> RDX: 0000000000000008 RSI: 636f6c6e6f6e5f88 RDI: 0000000080ca9398
> RBP: 0000000080ca9350 R08: 636f6c6e6f6e5f70 R09: 636f6c6e6f6e5f70
> R10: 0000003f2ca892da R11: 0000003f2ca892da R12: 000000006026b27c
> R13: 000000006001db74 R14: 0000000000000004 R15: 0000000000000000
> Call Trace:
> 80ca88b8:  [<600151ac>] segv+0x66/0x241
> 80ca8988:  [<600153e9>] segv_handler+0x62/0x6c
> 80ca89c8:  [<60022207>] sig_handler_common+0x8c/0xa2
> 80ca89e0:  [<6001db74>] io_thread+0x63/0x243
> 80ca8a50:  [<60026715>] memcpy+0xb5/0x140
> 80ca8d08:  [<600222c2>] sig_handler+0x2e/0x39
> 80ca8d28:  [<60021f33>] hard_handler+0x83/0xab
> 80ca8d58:  [<6001db74>] io_thread+0x63/0x243
> 80ca8dd0:  [<6001db74>] io_thread+0x63/0x243
> 80ca8e28:  [<60026715>] memcpy+0xb5/0x140
> 80ca90e8:  [<60017226>] generic_write+0xf/0x38
> 80ca9108:  [<600172f9>] generic_console_write+0x3e/0x165
> 80ca9288:  [<6004c396>] vprintk_emit+0x2dd/0x2ee
> 80ca92d8:  [<6001db74>] io_thread+0x63/0x243
> 80ca92f8:  [<601edf38>] printk+0x97/0x99
> 80ca9308:  [<60015eff>] copy_from_user+0x36/0x8a
> 80ca9328:  [<60015eff>] copy_from_user+0x36/0x8a
> 80ca9358:  [<600699d4>] __probe_kernel_read+0x3c/0x6d
> 80ca9368:  [<6003f2ac>] probe_kthread_data+0x29/0x2f
> 80ca9388:  [<6003becc>] print_worker_info+0x66/0x124
> 80ca93f8:  [<6004d204>] dump_stack_print_info+0xa6/0xaf
> 80ca9428:  [<601efbeb>] dump_stack+0xe/0x19
> 80ca9438:  [<601ed8b7>] panic+0xf7/0x1e4
> 80ca9490:  [<60056982>] __module_text_address+0xd/0x4d
> 80ca94a8:  [<6005a007>] is_module_text_address+0x9/0x11
> 80ca94b8:  [<6003cd40>] __kernel_text_address+0x26/0x49
> 80ca94d8:  [<60014146>] show_trace+0x8e/0x95
> 80ca9508:  [<60026491>] show_regs+0x27/0x2a
> 80ca9520:  [<6001db74>] io_thread+0x63/0x243
> 80ca9538:  [<6001524f>] segv+0x109/0x241
> 80ca9608:  [<600153e9>] segv_handler+0x62/0x6c
> 80ca9628:  [<60022169>] unblock_signals+0x4a/0x5c
> 80ca9648:  [<60022207>] sig_handler_common+0x8c/0xa2
> 80ca96d0:  [<6001db74>] io_thread+0x63/0x243
> 80ca9988:  [<600222c2>] sig_handler+0x2e/0x39
> 80ca99a8:  [<60021f33>] hard_handler+0x83/0xab
> 80ca9aa8:  [<6001db74>] io_thread+0x63/0x243
> 80ca9d88:  [<60021eb0>] hard_handler+0x0/0xab
> 80ca9f70:  [<60020308>] os_seek_file+0x11/0x27
> 80ca9f80:  [<6002032a>] os_read_file+0xc/0x1f
> 80ca9f90:  [<6001db3b>] io_thread+0x2a/0x243
>
> Kernel panic - not syncing: Segfault with no mm
>
> Modules linked in:
> Pid: 1614169860, comm:  Not tainted 3.11.0-rc4-00103-g6c2580c
> RIP: 0000:[<000001a400000004>]
> RSP: 0000000000000000  EFLAGS: 6002e720
> RAX: 000000006002e770 RBX: 0000000000000000 RCX: 0000000000000000
> RDX: 0000000000000000 RSI: 0000000000000000 RDI: 00000000602706ae
> RBP: 0000000000000000 R08: 0000000000000000 R09: 000001a400000004
> R10: 00000000602d04ac R11: 000000006027069b R12: 0000000000000000
> R13: 000000006002e720 R14: 0000000000000000 R15: 000001a400000004
> Call Trace:
> 80ca8748:  [<6001564c>] panic_exit+0x2b/0x41
> 80ca8758:  [<6015a3ac>] vsnprintf+0xd0/0x431
> 80ca8768:  [<600427c7>] notifier_call_chain+0x39/0x5c
> 80ca8790:  [<60026715>] memcpy+0xb5/0x140
> 80ca87a8:  [<60042804>] atomic_notifier_call_chain+0xf/0x11
> 80ca87b8:  [<601ed8d2>] panic+0x112/0x1e4
> 80ca8810:  [<60056982>] __module_text_address+0xd/0x4d
> 80ca8828:  [<6005a007>] is_module_text_address+0x9/0x11
> 80ca8838:  [<6003cd40>] __kernel_text_address+0x26/0x49
> 80ca8858:  [<60014146>] show_trace+0x8e/0x95
> 80ca8888:  [<60026491>] show_regs+0x27/0x2a
> 80ca88a0:  [<60026715>] memcpy+0xb5/0x140
> 80ca88b8:  [<6001524f>] segv+0x109/0x241
> 80ca8988:  [<600153e9>] segv_handler+0x62/0x6c
> 80ca89c8:  [<60022207>] sig_handler_common+0x8c/0xa2
> 80ca89e0:  [<6001db74>] io_thread+0x63/0x243
>
> --
> Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones
> virt-p2v converts physical machines to virtual machines.  Boot with a
> live CD or over the network (PXE) and turn machines into KVM guests.
> http://libguestfs.org/virt-v2v
>
> ------------------------------------------------------------------------------
> Get 100% visibility into Java/.NET code with AppDynamics Lite!
> It's a free troubleshooting tool designed for production.
> Get down to code-level detail for bottlenecks, with <2% overhead.
> Download for free and get started troubleshooting in minutes.
> http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
> _______________________________________________
> User-mode-linux-devel mailing list
> User-mode-linux-devel@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/user-mode-linux-devel



-- 
Thanks,
//richard

------------------------------------------------------------------------------
Get 100% visibility into Java/.NET code with AppDynamics Lite!
It's a free troubleshooting tool designed for production.
Get down to code-level detail for bottlenecks, with <2% overhead. 
Download for free and get started troubleshooting in minutes. 
http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
_______________________________________________
User-mode-linux-devel mailing list
User-mode-linux-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/user-mode-linux-devel


^ permalink raw reply	[flat|nested] 17+ messages in thread
* Re: [uml-devel] Using UML in libguestfs
  2013-08-09 20:43 [uml-devel] Using UML in libguestfs Richard W.M. Jones
  2013-08-09 20:57 ` richard -rw- weinberger
@ 2013-08-09 21:15 ` Richard W.M. Jones
  2013-08-09 21:18   ` Richard Weinberger
  2013-08-11 19:05 ` Geert Uytterhoeven
  2013-08-14 17:59 ` [uml-devel] Performance tests (was: Using UML in libguestfs) Richard W.M. Jones
  3 siblings, 1 reply; 17+ messages in thread
From: Richard W.M. Jones @ 2013-08-09 21:15 UTC (permalink / raw)
  To: user-mode-linux-devel

On Fri, Aug 09, 2013 at 09:43:55PM +0100, Richard W.M. Jones wrote:
> * I'm trying to use a serial port to communicate between guest and
> host, using ssl3=fd:<N>.  This causes the Linux process to crash soon
> (but not immediately) after the guest has opened /dev/ttyS3:
> 
>   spurious interrupt in ubd_handler, err = -4

So I realized this is caused by a coding mistake by me.  I had set to
O_CLOEXEC flag on the file descriptor, so it was being closed across
the exec of vmlinux.

It still doesn't quite work, but the error above has gone away.

Rich.

-- 
Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones
virt-df lists disk usage of guests without needing to install any
software inside the virtual machine.  Supports Linux and Windows.
http://people.redhat.com/~rjones/virt-df/

------------------------------------------------------------------------------
Get 100% visibility into Java/.NET code with AppDynamics Lite!
It's a free troubleshooting tool designed for production.
Get down to code-level detail for bottlenecks, with <2% overhead. 
Download for free and get started troubleshooting in minutes. 
http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
_______________________________________________
User-mode-linux-devel mailing list
User-mode-linux-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/user-mode-linux-devel


^ permalink raw reply	[flat|nested] 17+ messages in thread
* Re: [uml-devel] Using UML in libguestfs
  2013-08-09 21:15 ` Richard W.M. Jones
@ 2013-08-09 21:18   ` Richard Weinberger
  2013-08-09 21:24     ` Richard W.M. Jones
  0 siblings, 1 reply; 17+ messages in thread
From: Richard Weinberger @ 2013-08-09 21:18 UTC (permalink / raw)
  To: user-mode-linux-devel

Am 09.08.2013 23:15, schrieb Richard W.M. Jones:
> On Fri, Aug 09, 2013 at 09:43:55PM +0100, Richard W.M. Jones wrote:
>> * I'm trying to use a serial port to communicate between guest and
>> host, using ssl3=fd:<N>.  This causes the Linux process to crash soon
>> (but not immediately) after the guest has opened /dev/ttyS3:
>>
>>   spurious interrupt in ubd_handler, err = -4
> 
> So I realized this is caused by a coding mistake by me.  I had set to
> O_CLOEXEC flag on the file descriptor, so it was being closed across
> the exec of vmlinux.
> 
> It still doesn't quite work, but the error above has gone away.

So you are still facing the crash are reported in your first mail?

Thanks,
//richard


------------------------------------------------------------------------------
Get 100% visibility into Java/.NET code with AppDynamics Lite!
It's a free troubleshooting tool designed for production.
Get down to code-level detail for bottlenecks, with <2% overhead. 
Download for free and get started troubleshooting in minutes. 
http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
_______________________________________________
User-mode-linux-devel mailing list
User-mode-linux-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/user-mode-linux-devel


^ permalink raw reply	[flat|nested] 17+ messages in thread
* Re: [uml-devel] Using UML in libguestfs
  2013-08-09 21:18   ` Richard Weinberger
@ 2013-08-09 21:24     ` Richard W.M. Jones
  0 siblings, 0 replies; 17+ messages in thread
From: Richard W.M. Jones @ 2013-08-09 21:24 UTC (permalink / raw)
  To: Richard Weinberger; +Cc: user-mode-linux-devel

On Fri, Aug 09, 2013 at 11:18:24PM +0200, Richard Weinberger wrote:
> Am 09.08.2013 23:15, schrieb Richard W.M. Jones:
> > On Fri, Aug 09, 2013 at 09:43:55PM +0100, Richard W.M. Jones wrote:
> >> * I'm trying to use a serial port to communicate between guest and
> >> host, using ssl3=fd:<N>.  This causes the Linux process to crash soon
> >> (but not immediately) after the guest has opened /dev/ttyS3:
> >>
> >>   spurious interrupt in ubd_handler, err = -4
> > 
> > So I realized this is caused by a coding mistake by me.  I had set to
> > O_CLOEXEC flag on the file descriptor, so it was being closed across
> > the exec of vmlinux.
> > 
> > It still doesn't quite work, but the error above has gone away.
> 
> So you are still facing the crash are reported in your first mail?

No, the crash has gone away.  There's now a protocol error which I'm
trying to get to the bottom of.  Perhaps /dev/ttyS<N> isn't 8 bit clean..?

Rich.

-- 
Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones
Fedora Windows cross-compiler. Compile Windows programs, test, and
build Windows installers. Over 100 libraries supported.
http://fedoraproject.org/wiki/MinGW

------------------------------------------------------------------------------
Get 100% visibility into Java/.NET code with AppDynamics Lite!
It's a free troubleshooting tool designed for production.
Get down to code-level detail for bottlenecks, with <2% overhead. 
Download for free and get started troubleshooting in minutes. 
http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
_______________________________________________
User-mode-linux-devel mailing list
User-mode-linux-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/user-mode-linux-devel


^ permalink raw reply	[flat|nested] 17+ messages in thread
* Re: [uml-devel] Using UML in libguestfs
  2013-08-09 20:57 ` richard -rw- weinberger
@ 2013-08-09 21:31   ` Richard W.M. Jones
  2013-08-10  4:16   ` Mattia Dongili
  1 sibling, 0 replies; 17+ messages in thread
From: Richard W.M. Jones @ 2013-08-09 21:31 UTC (permalink / raw)
  To: richard -rw- weinberger; +Cc: user-mode-linux-devel

On Fri, Aug 09, 2013 at 10:57:20PM +0200, richard -rw- weinberger wrote:
> On Fri, Aug 9, 2013 at 10:43 PM, Richard W.M. Jones <rjones@redhat.com> wrote:
> > * UML is pretty fast!  It's certainly faster (by a factor of > 5) than
> > spinning up a lightweight KVM VM.
> 
> Wow!
> Does it only boot faster or is it also faster with reading data from a
> filesystem?

I've only got it booting so far, so I haven't been able to test file
transfer speeds.

Rich.

-- 
Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones
Read my programming blog: http://rwmj.wordpress.com
Fedora now supports 80 OCaml packages (the OPEN alternative to F#)

------------------------------------------------------------------------------
Get 100% visibility into Java/.NET code with AppDynamics Lite!
It's a free troubleshooting tool designed for production.
Get down to code-level detail for bottlenecks, with <2% overhead. 
Download for free and get started troubleshooting in minutes. 
http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
_______________________________________________
User-mode-linux-devel mailing list
User-mode-linux-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/user-mode-linux-devel


^ permalink raw reply	[flat|nested] 17+ messages in thread
* Re: [uml-devel] Using UML in libguestfs
  2013-08-09 20:57 ` richard -rw- weinberger
  2013-08-09 21:31   ` Richard W.M. Jones
@ 2013-08-10  4:16   ` Mattia Dongili
  1 sibling, 0 replies; 17+ messages in thread
From: Mattia Dongili @ 2013-08-10  4:16 UTC (permalink / raw)
  To: richard -rw- weinberger; +Cc: user-mode-linux-devel

On Fri, Aug 09, 2013 at 10:57:20PM +0200, richard -rw- weinberger wrote:
> On Fri, Aug 9, 2013 at 10:43 PM, Richard W.M. Jones <rjones@redhat.com> wrote:
> > Today I wrote some patches (for libguestfs) which add an experimental
> > UML backend:
> >
> >   https://www.redhat.com/archives/libguestfs/2013-August/msg00005.html
> >
> > There are some observations from this.
> >
> > * 'ubd<N>=cow,original' is not parsed correctly.  This was reported a
> > while back and I'm seeing the same thing with the latest kernel from
> > git:
> >
> >   http://thread.gmane.org/gmane.linux.uml.devel/13556
> >
> > If I have some time I'll have a go at fixing this since it seems like
> > a simple regression.
> 
> Patches are very welcome!
> 
> > * I'm trying to use a serial port to communicate between guest and
> > host, using ssl3=fd:<N>.  This causes the Linux process to crash soon
> > (but not immediately) after the guest has opened /dev/ttyS3:
> >
> >   spurious interrupt in ubd_handler, err = -4
> >
> > I've put the full call trace at the end of the email because it's
> > quite long[1].  This was also reported before, although the suggested
> > patch did not fix it:
> >
> >   http://marc.info/?l=linux-kernel&m=123242364211797&w=2
> >
> > I wonder if anyone has used the fd:<N> option in anger?  I will try
> > using a Unix domain socket and see if it makes a difference.
> 
> Hmmm. I *should* work.
> Which fd are you using?
> Can you also try pts?

just for the record, rootstrap in debian (script to create a debian FS)
uses that option
umlargs.extend(['con0=fd:0,fd:1',                                                                                                         
                'con=pts',                                                                                                                
                                'root=/dev/root',                                                                                         
                                'rootflags=/',                                                                                            
                                'rootfstype=hostfs',                                                                                      
                'ubd1=' + imagefile,                                                                                                      
                'init=/usr/lib/rootstrap/builder',                                                                                        
                'devfs=nomount',                                                                                                          
                'rsworkdir=' + os.getcwd()])                                                                                              

Last time I checked it worked.
-- 
mattia
:wq!

------------------------------------------------------------------------------
Get 100% visibility into Java/.NET code with AppDynamics Lite!
It's a free troubleshooting tool designed for production.
Get down to code-level detail for bottlenecks, with <2% overhead. 
Download for free and get started troubleshooting in minutes. 
http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
_______________________________________________
User-mode-linux-devel mailing list
User-mode-linux-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/user-mode-linux-devel


^ permalink raw reply	[flat|nested] 17+ messages in thread
* Re: [uml-devel] Using UML in libguestfs
  2013-08-09 20:43 [uml-devel] Using UML in libguestfs Richard W.M. Jones
  2013-08-09 20:57 ` richard -rw- weinberger
  2013-08-09 21:15 ` Richard W.M. Jones
@ 2013-08-11 19:05 ` Geert Uytterhoeven
  2013-08-11 19:27   ` Richard W.M. Jones
  2013-08-14 17:59 ` [uml-devel] Performance tests (was: Using UML in libguestfs) Richard W.M. Jones
  3 siblings, 1 reply; 17+ messages in thread
From: Geert Uytterhoeven @ 2013-08-11 19:05 UTC (permalink / raw)
  To: Richard W.M. Jones; +Cc: user-mode-linux-devel

On Fri, Aug 9, 2013 at 10:43 PM, Richard W.M. Jones <rjones@redhat.com> wrote:
> * UML is pretty fast!  It's certainly faster (by a factor of > 5) than
> spinning up a lightweight KVM VM.

This is interesting. Typically people say you have to use KVM or XEN for
good virtualization performance. But for that you need sufficient privileges.
With UML, all you need is to be able to compile and run your own
executables.

Gr{oetje,eeting}s,

                        Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds


^ permalink raw reply	[flat|nested] 17+ messages in thread
* Re: [uml-devel] Using UML in libguestfs
  2013-08-11 19:05 ` Geert Uytterhoeven
@ 2013-08-11 19:27   ` Richard W.M. Jones
  2013-08-13 11:16     ` Geert Uytterhoeven
  0 siblings, 1 reply; 17+ messages in thread
From: Richard W.M. Jones @ 2013-08-11 19:27 UTC (permalink / raw)
  To: Geert Uytterhoeven; +Cc: user-mode-linux-devel

On Sun, Aug 11, 2013 at 09:05:08PM +0200, Geert Uytterhoeven wrote:
> On Fri, Aug 9, 2013 at 10:43 PM, Richard W.M. Jones <rjones@redhat.com> wrote:
> > * UML is pretty fast!  It's certainly faster (by a factor of > 5) than
> > spinning up a lightweight KVM VM.

> This is interesting. Typically people say you have to use KVM or XEN
> for good virtualization performance. But for that you need
> sufficient privileges.

I think this is true of Xen, but to be fair to KVM it doesn't require
special privileges, with one exception.  If you want KVM to use a tunX
network interface then you have to start it as root (or use a setuid
helper).  I think this is probably true of UML too although I've not
explored UML's networking yet.  Basically it's a restriction in Linux
itself.

> With UML, all you need is to be able to compile and run your own
> executables.

With libguestfs we specifically don't want people to use root, so UML
not requiring root was one reason why it's possible to write a
UML-based libguestfs backend.

I now have a UML backend for libguestfs working, with a significant
part of the testsuite passing too.

Rich.

-- 
Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones
Read my programming blog: http://rwmj.wordpress.com
Fedora now supports 80 OCaml packages (the OPEN alternative to F#)


^ permalink raw reply	[flat|nested] 17+ messages in thread
* Re: [uml-devel] Using UML in libguestfs
  2013-08-11 19:27   ` Richard W.M. Jones
@ 2013-08-13 11:16     ` Geert Uytterhoeven
  0 siblings, 0 replies; 17+ messages in thread
From: Geert Uytterhoeven @ 2013-08-13 11:16 UTC (permalink / raw)
  To: Richard W.M. Jones; +Cc: uml-devel

On Sun, Aug 11, 2013 at 9:27 PM, Richard W.M. Jones <rjones@redhat.com> wrote:
> On Sun, Aug 11, 2013 at 09:05:08PM +0200, Geert Uytterhoeven wrote:
>> On Fri, Aug 9, 2013 at 10:43 PM, Richard W.M. Jones <rjones@redhat.com> wrote:
>> > * UML is pretty fast!  It's certainly faster (by a factor of > 5) than
>> > spinning up a lightweight KVM VM.
>
>> This is interesting. Typically people say you have to use KVM or XEN
>> for good virtualization performance. But for that you need
>> sufficient privileges.
>
> I think this is true of Xen, but to be fair to KVM it doesn't require
> special privileges, with one exception.  If you want KVM to use a tunX
> network interface then you have to start it as root (or use a setuid
> helper).  I think this is probably true of UML too although I've not

There you name it...

> explored UML's networking yet.  Basically it's a restriction in Linux
> itself.
>
>> With UML, all you need is to be able to compile and run your own
>> executables.

With UML, I can use slirp-fullbolt.

So if I want to run e.g. wireshark on the development machine at work (no root
for me), I just launch a UML with a Debian root fs that has wireshark and
run my app.

Gr{oetje,eeting}s,

                        Geert

--
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds


^ permalink raw reply	[flat|nested] 17+ messages in thread
* [uml-devel] Performance tests (was: Using UML in libguestfs)
  2013-08-09 20:43 [uml-devel] Using UML in libguestfs Richard W.M. Jones
                   ` (2 preceding siblings ...)
  2013-08-11 19:05 ` Geert Uytterhoeven
@ 2013-08-14 17:59 ` Richard W.M. Jones
  2013-08-15 17:37   ` richard -rw- weinberger
  3 siblings, 1 reply; 17+ messages in thread
From: Richard W.M. Jones @ 2013-08-14 17:59 UTC (permalink / raw)
  To: user-mode-linux-devel


I performed some benchmarks against KVM:

http://rwmj.wordpress.com/2013/08/14/performance-of-user-mode-linux-as-a-libguestfs-backend/#content

Notes for those that don't want to read all that, there are at least
two bugs I found:

(1) fsync /dev/ubda (in the guest) is not honoured by the user-mode
linux process.  It should do an fsync on the corresponding backing file.

(2) The ssl<N>=... serial port is really slow, at least 10 times
slower than virtio-serial which is the equivalent KVM feature.
libguestfs is chatty over the serial port, and so this has a
significant knock-on impact all over the place.

Rich.

-- 
Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones
Fedora Windows cross-compiler. Compile Windows programs, test, and
build Windows installers. Over 100 libraries supported.
http://fedoraproject.org/wiki/MinGW

------------------------------------------------------------------------------
Get 100% visibility into Java/.NET code with AppDynamics Lite!
It's a free troubleshooting tool designed for production.
Get down to code-level detail for bottlenecks, with <2% overhead. 
Download for free and get started troubleshooting in minutes. 
http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
_______________________________________________
User-mode-linux-devel mailing list
User-mode-linux-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/user-mode-linux-devel


^ permalink raw reply	[flat|nested] 17+ messages in thread
* Re: [uml-devel] Performance tests (was: Using UML in libguestfs)
  2013-08-14 17:59 ` [uml-devel] Performance tests (was: Using UML in libguestfs) Richard W.M. Jones
@ 2013-08-15 17:37   ` richard -rw- weinberger
  2013-08-15 17:59     ` Richard W.M. Jones
  0 siblings, 1 reply; 17+ messages in thread
From: richard -rw- weinberger @ 2013-08-15 17:37 UTC (permalink / raw)
  To: Richard W.M. Jones; +Cc: user-mode-linux-devel

On Wed, Aug 14, 2013 at 7:59 PM, Richard W.M. Jones <rjones@redhat.com> wrote:
>
> I performed some benchmarks against KVM:
>
> http://rwmj.wordpress.com/2013/08/14/performance-of-user-mode-linux-as-a-libguestfs-backend/#content
>
> Notes for those that don't want to read all that, there are at least
> two bugs I found:
>
> (1) fsync /dev/ubda (in the guest) is not honoured by the user-mode
> linux process.  It should do an fsync on the corresponding backing file.

What exactly does libguestfs do?

> (2) The ssl<N>=... serial port is really slow, at least 10 times
> slower than virtio-serial which is the equivalent KVM feature.
> libguestfs is chatty over the serial port, and so this has a
> significant knock-on impact all over the place.

So, libguestfs transfers all data over a serial line?!
We could hack a faster transport into UML.
As UML is only a process we can very easily access
the host.
Have looked at hostfs?

-- 
Thanks,
//richard

------------------------------------------------------------------------------
Get 100% visibility into Java/.NET code with AppDynamics Lite!
It's a free troubleshooting tool designed for production.
Get down to code-level detail for bottlenecks, with <2% overhead. 
Download for free and get started troubleshooting in minutes. 
http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
_______________________________________________
User-mode-linux-devel mailing list
User-mode-linux-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/user-mode-linux-devel


^ permalink raw reply	[flat|nested] 17+ messages in thread
* Re: [uml-devel] Performance tests (was: Using UML in libguestfs)
  2013-08-15 17:37   ` richard -rw- weinberger
@ 2013-08-15 17:59     ` Richard W.M. Jones
  2013-08-15 18:41       ` richard -rw- weinberger
  0 siblings, 1 reply; 17+ messages in thread
From: Richard W.M. Jones @ 2013-08-15 17:59 UTC (permalink / raw)
  To: richard -rw- weinberger; +Cc: user-mode-linux-devel

On Thu, Aug 15, 2013 at 07:37:06PM +0200, richard -rw- weinberger wrote:
> On Wed, Aug 14, 2013 at 7:59 PM, Richard W.M. Jones <rjones@redhat.com> wrote:
> >
> > I performed some benchmarks against KVM:
> >
> > http://rwmj.wordpress.com/2013/08/14/performance-of-user-mode-linux-as-a-libguestfs-backend/#content
> >
> > Notes for those that don't want to read all that, there are at least
> > two bugs I found:
> >
> > (1) fsync /dev/ubda (in the guest) is not honoured by the user-mode
> > linux process.  It should do an fsync on the corresponding backing file.
> 
> What exactly does libguestfs do?

http://libguestfs.org/ ...

> > (2) The ssl<N>=... serial port is really slow, at least 10 times
> > slower than virtio-serial which is the equivalent KVM feature.
> > libguestfs is chatty over the serial port, and so this has a
> > significant knock-on impact all over the place.
> 
> So, libguestfs transfers all data over a serial line?!

Well, it requires an 8-bit clean, reliable, bi-directional connection
into the guest.  We used to use TCP/IP but it's pretty slow compared
to virtio-serial and has a bunch of other problems.

For the UML backend I used the serial port.  There's no problem with
it -- it works and it's reliable, it's just slow.

> We could hack a faster transport into UML.
> As UML is only a process we can very easily access
> the host.
> Have looked at hostfs?

Not sure how hostfs would help?

Rich.

-- 
Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones
virt-df lists disk usage of guests without needing to install any
software inside the virtual machine.  Supports Linux and Windows.
http://people.redhat.com/~rjones/virt-df/

------------------------------------------------------------------------------
Get 100% visibility into Java/.NET code with AppDynamics Lite!
It's a free troubleshooting tool designed for production.
Get down to code-level detail for bottlenecks, with <2% overhead. 
Download for free and get started troubleshooting in minutes. 
http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
_______________________________________________
User-mode-linux-devel mailing list
User-mode-linux-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/user-mode-linux-devel


^ permalink raw reply	[flat|nested] 17+ messages in thread
* Re: [uml-devel] Performance tests (was: Using UML in libguestfs)
  2013-08-15 17:59     ` Richard W.M. Jones
@ 2013-08-15 18:41       ` richard -rw- weinberger
  2013-08-15 18:59         ` Richard W.M. Jones
  0 siblings, 1 reply; 17+ messages in thread
From: richard -rw- weinberger @ 2013-08-15 18:41 UTC (permalink / raw)
  To: Richard W.M. Jones; +Cc: user-mode-linux-devel

On Thu, Aug 15, 2013 at 7:59 PM, Richard W.M. Jones <rjones@redhat.com> wrote:
> On Thu, Aug 15, 2013 at 07:37:06PM +0200, richard -rw- weinberger wrote:
>> On Wed, Aug 14, 2013 at 7:59 PM, Richard W.M. Jones <rjones@redhat.com> wrote:
>> >
>> > I performed some benchmarks against KVM:
>> >
>> > http://rwmj.wordpress.com/2013/08/14/performance-of-user-mode-linux-as-a-libguestfs-backend/#content
>> >
>> > Notes for those that don't want to read all that, there are at least
>> > two bugs I found:
>> >
>> > (1) fsync /dev/ubda (in the guest) is not honoured by the user-mode
>> > linux process.  It should do an fsync on the corresponding backing file.
>>
>> What exactly does libguestfs do?
>
> http://libguestfs.org/ ...

Meant regarding fsync()...

>> > (2) The ssl<N>=... serial port is really slow, at least 10 times
>> > slower than virtio-serial which is the equivalent KVM feature.
>> > libguestfs is chatty over the serial port, and so this has a
>> > significant knock-on impact all over the place.
>>
>> So, libguestfs transfers all data over a serial line?!
>
> Well, it requires an 8-bit clean, reliable, bi-directional connection
> into the guest.  We used to use TCP/IP but it's pretty slow compared
> to virtio-serial and has a bunch of other problems.
>
> For the UML backend I used the serial port.  There's no problem with
> it -- it works and it's reliable, it's just slow.
>
>> We could hack a faster transport into UML.
>> As UML is only a process we can very easily access
>> the host.
>> Have looked at hostfs?
>
> Not sure how hostfs would help?

hostfs show how you can directly access the host...

-- 
Thanks,
//richard

------------------------------------------------------------------------------
Get 100% visibility into Java/.NET code with AppDynamics Lite!
It's a free troubleshooting tool designed for production.
Get down to code-level detail for bottlenecks, with <2% overhead. 
Download for free and get started troubleshooting in minutes. 
http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
_______________________________________________
User-mode-linux-devel mailing list
User-mode-linux-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/user-mode-linux-devel


^ permalink raw reply	[flat|nested] 17+ messages in thread
* Re: [uml-devel] Performance tests (was: Using UML in libguestfs)
  2013-08-15 18:41       ` richard -rw- weinberger
@ 2013-08-15 18:59         ` Richard W.M. Jones
  2013-08-15 20:31           ` richard -rw- weinberger
  0 siblings, 1 reply; 17+ messages in thread
From: Richard W.M. Jones @ 2013-08-15 18:59 UTC (permalink / raw)
  To: richard -rw- weinberger; +Cc: user-mode-linux-devel

On Thu, Aug 15, 2013 at 08:41:53PM +0200, richard -rw- weinberger wrote:
> Meant regarding fsync()...

Ah right, see what you mean :-)

It calls sync(2).  Then it opens each /dev/ubdX device and calls fsync
on the file descriptor:

https://github.com/libguestfs/libguestfs/blob/master/daemon/sync.c#L54

The reason for this is a bit complicated, but has to do with write
barriers.  I believe this has been fixed since then.

https://github.com/libguestfs/libguestfs/commit/c0a3c9ce70b98171e737e49e6dccc4457963f2ec

In any case, we're calling sync & fsync in the guest, and that ain't
causing the host cache to be flushed.

It's pretty easy to show this with libguestfs:

LIBGUESTFS_BACKEND=uml LIBGUESTFS_HV=~/d/linux-um/linux \
time ./run ./fish/guestfish -N fs:ext2:1G -m /dev/sda1 fallocate64 /data 800M
time sync

The first command finishes in 8 seconds, with barely any disk
activity.  The sync afterwards takes 9 seconds with the disk light on
the whole time, while it actually writes the whole 800 MB of data.

Rich.

-- 
Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones
virt-top is 'top' for virtual machines.  Tiny program with many
powerful monitoring features, net stats, disk stats, logging, etc.
http://people.redhat.com/~rjones/virt-top

------------------------------------------------------------------------------
Get 100% visibility into Java/.NET code with AppDynamics Lite!
It's a free troubleshooting tool designed for production.
Get down to code-level detail for bottlenecks, with <2% overhead. 
Download for free and get started troubleshooting in minutes. 
http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
_______________________________________________
User-mode-linux-devel mailing list
User-mode-linux-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/user-mode-linux-devel


^ permalink raw reply	[flat|nested] 17+ messages in thread
* Re: [uml-devel] Performance tests (was: Using UML in libguestfs)
  2013-08-15 18:59         ` Richard W.M. Jones
@ 2013-08-15 20:31           ` richard -rw- weinberger
  2013-08-16  7:54             ` Richard W.M. Jones
  0 siblings, 1 reply; 17+ messages in thread
From: richard -rw- weinberger @ 2013-08-15 20:31 UTC (permalink / raw)
  To: Richard W.M. Jones; +Cc: user-mode-linux-devel

On Thu, Aug 15, 2013 at 8:59 PM, Richard W.M. Jones <rjones@redhat.com> wrote:
> On Thu, Aug 15, 2013 at 08:41:53PM +0200, richard -rw- weinberger wrote:
>> Meant regarding fsync()...
>
> Ah right, see what you mean :-)
>
> It calls sync(2).  Then it opens each /dev/ubdX device and calls fsync
> on the file descriptor:
>
> https://github.com/libguestfs/libguestfs/blob/master/daemon/sync.c#L54
>
> The reason for this is a bit complicated, but has to do with write
> barriers.  I believe this has been fixed since then.
>
> https://github.com/libguestfs/libguestfs/commit/c0a3c9ce70b98171e737e49e6dccc4457963f2ec
>
> In any case, we're calling sync & fsync in the guest, and that ain't
> causing the host cache to be flushed.

Hmm, AFAIK a fsync() after sync() is pointless.

> It's pretty easy to show this with libguestfs:
>
> LIBGUESTFS_BACKEND=uml LIBGUESTFS_HV=~/d/linux-um/linux \
> time ./run ./fish/guestfish -N fs:ext2:1G -m /dev/sda1 fallocate64 /data 800M
> time sync
>
> The first command finishes in 8 seconds, with barely any disk
> activity.  The sync afterwards takes 9 seconds with the disk light on
> the whole time, while it actually writes the whole 800 MB of data.

I think you don't see the desired effect because UML does not open
the backing file with O_SYNC.
But you can tell UML to do so.
e.g. write ubdas=/foo instead of ubda=/foo.

See:
ubd<n><flags>=<filename>[(:|,)<filename2>]
    This is used to associate a device with a file in the underlying
    filesystem. When specifying two filenames, the first one is the
    COW name and the second is the backing file name. As separator you can
    use either a ':' or a ',': the first one allows writing things like;
        ubd0=~/Uml/root_cow:~/Uml/root_backing_file
    while with a ',' the shell would not expand the 2nd '~'.
    When using only one filename, UML will detect whether to treat it like
    a COW file or a backing file. To override this detection, add the 'd'
    flag:
        ubd0d=BackingFile
    Usually, there is a filesystem in the file, but
    that's not required. Swap devices containing swap files can be
    specified like this. Also, a file which doesn't contain a
    filesystem can have its contents read in the virtual
    machine by running 'dd' on the device. <n> must be in the range
    0 to 7. Appending an 'r' to the number will cause that device
    to be mounted read-only. For example ubd1r=./ext_fs. Appending
    an 's' will cause data to be written to disk on the host immediately.
    'c' will cause the device to be treated as being shared between multiple
    UMLs and file locking will be turned off - this is appropriate for a
    cluster filesystem and inappropriate at almost all other times.

Please let me know if that helps. :-)

-- 
Thanks,
//richard

------------------------------------------------------------------------------
Get 100% visibility into Java/.NET code with AppDynamics Lite!
It's a free troubleshooting tool designed for production.
Get down to code-level detail for bottlenecks, with <2% overhead. 
Download for free and get started troubleshooting in minutes. 
http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
_______________________________________________
User-mode-linux-devel mailing list
User-mode-linux-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/user-mode-linux-devel


^ permalink raw reply	[flat|nested] 17+ messages in thread
* Re: [uml-devel] Performance tests (was: Using UML in libguestfs)
  2013-08-15 20:31           ` richard -rw- weinberger
@ 2013-08-16  7:54             ` Richard W.M. Jones
  0 siblings, 0 replies; 17+ messages in thread
From: Richard W.M. Jones @ 2013-08-16  7:54 UTC (permalink / raw)
  To: richard -rw- weinberger; +Cc: user-mode-linux-devel

On Thu, Aug 15, 2013 at 10:31:54PM +0200, richard -rw- weinberger wrote:
> On Thu, Aug 15, 2013 at 8:59 PM, Richard W.M. Jones <rjones@redhat.com> wrote:
> > On Thu, Aug 15, 2013 at 08:41:53PM +0200, richard -rw- weinberger wrote:
> >> Meant regarding fsync()...
> >
> > Ah right, see what you mean :-)
> >
> > It calls sync(2).  Then it opens each /dev/ubdX device and calls fsync
> > on the file descriptor:
> >
> > https://github.com/libguestfs/libguestfs/blob/master/daemon/sync.c#L54
> >
> > The reason for this is a bit complicated, but has to do with write
> > barriers.  I believe this has been fixed since then.
> >
> > https://github.com/libguestfs/libguestfs/commit/c0a3c9ce70b98171e737e49e6dccc4457963f2ec
> >
> > In any case, we're calling sync & fsync in the guest, and that ain't
> > causing the host cache to be flushed.
> 
> Hmm, AFAIK a fsync() after sync() is pointless.

That's definitely what you'd expect, but Linux was buggy until last
year: it only synched mounted filesystems, not devices in general.

This was fixed:

  commit a8c7176b6ded413d5044a00f1d05477b95a6d7ad
  Author: Jan Kara <jack@suse.cz>
  Date:   Tue Jul 3 16:45:32 2012 +0200

    vfs: Make sys_sync writeout also block device inodes

but in libguestfs we can't rely on people using a new enough kernel so
we do the safe thing.

> > It's pretty easy to show this with libguestfs:
> >
> > LIBGUESTFS_BACKEND=uml LIBGUESTFS_HV=~/d/linux-um/linux \
> > time ./run ./fish/guestfish -N fs:ext2:1G -m /dev/sda1 fallocate64 /data 800M
> > time sync
> >
> > The first command finishes in 8 seconds, with barely any disk
> > activity.  The sync afterwards takes 9 seconds with the disk light on
> > the whole time, while it actually writes the whole 800 MB of data.
> 
> I think you don't see the desired effect because UML does not open
> the backing file with O_SYNC.
> But you can tell UML to do so.
> e.g. write ubdas=/foo instead of ubda=/foo.

I didn't know about these flags.

I modified libguestfs to use the 's' flag, and reran the test above:

  UML (with O_SYNC):           598 seconds
  UML (unsafe):                  9 seconds
  UML + explicit host sync:     20 seconds
  KVM:                          20 seconds

However using O_SYNC is not the same as properly handling write
barriers.  UML is buggy in this respect because a user in the guest
who does 'sync' (and the sync returns to the prompt) wouldn't expect
that if the power failed immediately afterwards they would lose all
their data.  If they are using a journaled filesystem they'd also risk
unrecoverable data corruption during general usage.

KVM and LVM have both been here a few years ago and have fixed these
issues (mind you, KVM only does the right thing if you use the right
cache=... option).

Rich.

-- 
Richard Jones, Virtualization Group, Red Hat http://people.redhat.com/~rjones
Read my programming blog: http://rwmj.wordpress.com
Fedora now supports 80 OCaml packages (the OPEN alternative to F#)

------------------------------------------------------------------------------
Get 100% visibility into Java/.NET code with AppDynamics Lite!
It's a free troubleshooting tool designed for production.
Get down to code-level detail for bottlenecks, with <2% overhead. 
Download for free and get started troubleshooting in minutes. 
http://pubads.g.doubleclick.net/gampad/clk?id=48897031&iu=/4140/ostg.clktrk
_______________________________________________
User-mode-linux-devel mailing list
User-mode-linux-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/user-mode-linux-devel


^ permalink raw reply	[flat|nested] 17+ messages in thread
end of thread, other threads:[~2013-08-16  7:55 UTC | newest]

Thread overview: 17+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2013-08-09 20:43 [uml-devel] Using UML in libguestfs Richard W.M. Jones
2013-08-09 20:57 ` richard -rw- weinberger
2013-08-09 21:31   ` Richard W.M. Jones
2013-08-10  4:16   ` Mattia Dongili
2013-08-09 21:15 ` Richard W.M. Jones
2013-08-09 21:18   ` Richard Weinberger
2013-08-09 21:24     ` Richard W.M. Jones
2013-08-11 19:05 ` Geert Uytterhoeven
2013-08-11 19:27   ` Richard W.M. Jones
2013-08-13 11:16     ` Geert Uytterhoeven
2013-08-14 17:59 ` [uml-devel] Performance tests (was: Using UML in libguestfs) Richard W.M. Jones
2013-08-15 17:37   ` richard -rw- weinberger
2013-08-15 17:59     ` Richard W.M. Jones
2013-08-15 18:41       ` richard -rw- weinberger
2013-08-15 18:59         ` Richard W.M. Jones
2013-08-15 20:31           ` richard -rw- weinberger
2013-08-16  7:54             ` Richard W.M. Jones

This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.