Ben Nagy reported a scalability problem with KVM/QEMU that hit very hard
a single spinlock (idr_lock) in posix-timers code, on its 48 core
machine.
Even on a 16 cpu machine (2x4x2), a single test can show 98% of cpu time
used in ticket_spin_lock, from lock_timer
Ref: http://www.spinics.net/lists/
On Tue, Mar 22, 2011 at 12:54 PM, Eric Dumazet wrote:
> Ben Nagy reported a scalability problem with KVM/QEMU that hit very hard
> a single spinlock (idr_lock) in posix-timers code, on its 48 core
> machine.
Hi all,
Thanks a lot for all the help so far. We've tested with Eric's patch.
First up,
On 03/22/2011 01:37 AM, Rafael J. Wysocki wrote:
From: Rafael J. Wysocki
KVM uses a sysdev class and a sysdev for executing kvm_suspend()
after interrupts have been turned off on the boot CPU (during system
suspend) and for executing kvm_resume() before turning on interrupts
on the boot CPU (dur
On 03/21/2011 06:51 PM, Avi Kivity wrote:
> On 03/21/2011 06:26 PM, Avi Kivity wrote:
> > On 03/21/2011 02:35 PM, Avi Kivity wrote:
> > > On 03/21/2011 04:47 AM, Ren, Yongjie wrote:
> > > > Hi all,
> > > > This is KVM test result against kvm.git
> > > > a72e315c509376bbd1e121219c3ad9f23973923f bas
>> Part 1 of the patch implements the following QEMU command line option.
>>
>> -hpet [device=none|present][,driftfix=none|slew]
>
> Just define driftfix as property of the hpet device. That way it can be
> controlled both globally (-global hpet.driftfix=...) and per hpet block
> (once we support
>> Part 3 of the patch implements the following options for the
>> 'configure' script.
>>
>> --disable-hpet-driftfix
>> --enable-hpet-driftfix
>
> I see no benefit in this configurability. Just make the driftfix
> unconditionally available, runtime-disabled by default for now until it
> matured a
Hello Vadim,
Can you check this issues?
Thanks,
Yan.
> -Original Message-
> From: ya su [mailto:suya94...@gmail.com]
> Sent: Saturday, March 19, 2011 5:28 AM
> To: Yan Vugenfirer
> Cc: kvm@vger.kernel.org
> Subject: Re: [COMMIT] [WIN-GUEST-DRIVERS] Balloon - remove WMI usage.
> Remove wm
On 03/18/2011 05:54 PM, Ulrich Obergfell wrote:
Please review and please comment.
Some procedural remarks:
- put all patches in the same thread. 'git send-email' can do that
automatically for you. This really helps reviewers, at least with some
email readers
- give individual patches meani
On 2011-03-22 11:03, Ulrich Obergfell wrote:
>
>>> Part 3 of the patch implements the following options for the
>>> 'configure' script.
>>>
>>> --disable-hpet-driftfix
>>> --enable-hpet-driftfix
>>
>> I see no benefit in this configurability. Just make the driftfix
>> unconditionally available, ru
* Rafael J. Wysocki wrote:
> > If there are no objectsions, I'd like to push these patches through the
> > suspend
> > tree.
>
> [1/8] has been merged in the meantime and [3/8] has been included into the
> ACPI tree. if there are no objections, I'm going to push the following
> patches to Lin
On 03/22/2011 10:59 AM, Ben Nagy wrote:
On Tue, Mar 22, 2011 at 12:54 PM, Eric Dumazet wrote:
> Ben Nagy reported a scalability problem with KVM/QEMU that hit very hard
> a single spinlock (idr_lock) in posix-timers code, on its 48 core
> machine.
Hi all,
Thanks a lot for all the help so fa
On Mon, 21 Mar 2011, Wei Xu wrote:
> Avi and Jiri:
>
> I implemented emulation of movq(64bit) and movdqa(128 bit). If you guys
> still need it let me know and I can post somewhere...
I found a way around it, so I don't need it any more in the setup that has
been affected.
But it is definitely
On Mon, Mar 21, 2011 at 07:36:17PM -0400, Rafael J. Wysocki wrote:
> drivers/pci/intel-iommu.c | 38 +-
> 1 file changed, 9 insertions(+), 29 deletions(-)
Looks good. I prepare a patch to convert AMD IOMMU to syscore_ops too.
Joerg
--
AMD Operating
I have some troubles with passing usb into kvm virtual machine.
I trying to provide the usb device 2022:0008 (Amikon vpn key) or 19d2:2000
(ZTE MF112 hsdpa modem) into virtual machine with windows xp iax32.
kvm virtual machine was hung completely, and stop responding to anything.
Maybe I need ther
On 03/16/2011 12:26 PM, Dmitry Konishchev wrote:
Hi! I use QEMU via QMP and discovered that for some tasks there is no
proper way to do them via QMP. I've written few patches:
* One of them modifies "pci_add" command to return pci address of the
added device, when user hasn't specified it (to be
On Mon, Mar 21, 2011 at 11:03:07AM -0700, Shirley Ma wrote:
> On Fri, 2011-03-18 at 18:41 -0700, Shirley Ma wrote:
> > > > + /* Drop packet instead of stop queue for better
> > performance
> > > */
> > >
> > > I would like to see some justification as to why this is the right
> > > way to go
On 03/17/2011 10:53 AM, Michael S. Tsirkin wrote:
irqfd in kvm used flush_work incorrectly:
it assumed that work scheduled previously can't run
after flush_work, but since kvm uses a non-reentrant
workqueue (by means of schedule_work)
we need flush_work_sync to get that guarantee.
Signed-off-by:
On 03/16/2011 06:37 PM, Bharat Bhushan wrote:
Following dump is observed on host when clearing the exit timing counters
[root@p1021mds kvm]# echo -n 'c'> vm1200_vcpu0_timing
INFO: task echo:1276 blocked for more than 120 seconds.
"echo 0> /proc/sys/kernel/hung_task_timeout_secs" disables this
On Tue, Mar 22, 2011 at 02:37:27PM +0200, Avi Kivity wrote:
> On 03/17/2011 10:53 AM, Michael S. Tsirkin wrote:
> >irqfd in kvm used flush_work incorrectly:
> >it assumed that work scheduled previously can't run
> >after flush_work, but since kvm uses a non-reentrant
> >workqueue (by means of sched
On 03/15/2011 04:06 PM, Takuya Yoshikawa wrote:
> > So I just decided to treat CMPS and SCAS in another patch.
> > I mean I may introduce em_cmps or em_scas later if needed.
> >
> scas will likely just call em_cmp.
>
> > You prefer to treat these in this patch?
> >
> If there will be ot
On Tue, 2011-03-22 at 06:06 -0400, Yan Vugenfirer wrote:
> Hello Vadim,
>
> Can you check this issues?
>
http://www.mail-archive.com/kvm@vger.kernel.org/msg51061.html
> Thanks,
> Yan.
>
> > -Original Message-
> > From: ya su [mailto:suya94...@gmail.com]
> > Sent: Saturday, March 19, 201
On 03/13/2011 05:17 PM, Takuya Yoshikawa wrote:
From: Takuya Yoshikawa
ADD, OR, ADC, SBB, AND, SUB, XOR, CMP are converted using a new macro
I6ALU(_f, _e).
CMPS, SCAS will be converted later.
@@ -2337,10 +2401,20 @@ static int em_mov(struct x86_emulate_ctxt *ctxt)
#define D6ALU(_f) D2bv((_
On Mon, 21 Mar 2011 13:58:35 +0100
Juan Quintela wrote:
>
> Please, send in any agenda items you are interested in covening.
>
> - Merge patches speed. I just "feel", that patches are not being
> handled fast enough, so ... I looked how much patches have been
> integrated since Mars 1st:
On 03/13/2011 05:19 PM, Takuya Yoshikawa wrote:
From: Takuya Yoshikawa
PUSH ES/CS/SS/DS/FS/GS and POP ES/SS/DS/FS/GS are converted.
+static int em_push_es(struct x86_emulate_ctxt *ctxt)
+{
+ emulate_push_sreg(ctxt, ctxt->ops, VCPU_SREG_ES);
+ return X86EMUL_CONTINUE;
+}
I thought
On 03/13/2011 05:20 PM, Takuya Yoshikawa wrote:
From: Takuya Yoshikawa
POP is converted. RET will be converted later.
There is also POP r/m (8F /0); could be done later.
--
error compiling committee.c: too many arguments to function
--
To unsubscribe from this list: send the line "unsubscri
On 03/13/2011 05:21 PM, Takuya Yoshikawa wrote:
From: Takuya Yoshikawa
PUSHA and POPA are converted.
Signed-off-by: Takuya Yoshikawa
---
arch/x86/kvm/emulate.c | 19 ---
1 files changed, 12 insertions(+), 7 deletions(-)
diff --git a/arch/x86/kvm/emulate.c b/arch/x86/kvm/em
On 03/22/2011 01:43 PM, Avi Kivity wrote:
On 03/16/2011 06:37 PM, Bharat Bhushan wrote:
Following dump is observed on host when clearing the exit timing
counters
[root@p1021mds kvm]# echo -n 'c'> vm1200_vcpu0_timing
INFO: task echo:1276 blocked for more than 120 seconds.
"echo 0> /proc/sys/k
Minutes of today call:
- Patch integration
* Anthony thinks that we need more reviews
* chicken-eng problem to become a maintainer
* more patch reviews
* need for autotest to run faster
* not everybody use autotest
* use patches for infrastructure
* we need to get better integrating
On Tue, 22 Mar 2011 14:53:21 +0200
Avi Kivity wrote:
>
> I prefer to have the patchset fully updated, even if it takes a while.
> Good luck with the recovery!
Things already got back as usual, thanks.
I had expected much longer time.
BTW, is it better to wait until rc1 is released when we se
On Tue, 22 Mar 2011 14:55:57 +0200
Avi Kivity wrote:
> > @@ -2337,10 +2401,20 @@ static int em_mov(struct x86_emulate_ctxt *ctxt)
> > #define D6ALU(_f) D2bv((_f) | DstMem | SrcReg | ModRM),
> > \
> > D2bv(((_f) | DstReg | SrcMem | ModRM)& ~Lock), \
> >
On Tue, 22 Mar 2011 15:03:11 +0200
Avi Kivity wrote:
> > +static int em_push_es(struct x86_emulate_ctxt *ctxt)
> > +{
> > + emulate_push_sreg(ctxt, ctxt->ops, VCPU_SREG_ES);
> > + return X86EMUL_CONTINUE;
> > +}
>
> I thought of adding generic sreg decoding, so we can use
> em_push()/em_pop
On Tue, 22 Mar 2011 15:06:33 +0200
Avi Kivity wrote:
> > POP is converted. RET will be converted later.
>
> There is also POP r/m (8F /0); could be done later.
>
OK, I'll recheck.
I want to put related things into one patch if possible.
Takuya
--
To unsubscribe from this list: send the line
On 03/22/2011 05:35 PM, Takuya Yoshikawa wrote:
On Tue, 22 Mar 2011 14:53:21 +0200
Avi Kivity wrote:
I prefer to have the patchset fully updated, even if it takes a while.
Good luck with the recovery!
Things already got back as usual, thanks.
I had expected much longer time.
Good to hear.
On Tue, 22 Mar 2011 15:07:20 +0200
Avi Kivity wrote:
> > +static int em_pusha(struct x86_emulate_ctxt *ctxt)
> > +{
> > + return emulate_pusha(ctxt, ctxt->ops);
> > +}
> > +
>
> You can simply rename/update emulate_pusha/emulate_popa, since they have
> no other callers.
>
I intentionally le
On 03/22/2011 05:54 PM, Takuya Yoshikawa wrote:
I intentionally left emulate_* in this version because I thought
there might be some reason for introducing new em_* naming.
It's just that their signatures are all the same, and to conserve space
in the decode tables.
--
error compiling commi
On 03/17/2011 10:53 AM, Michael S. Tsirkin wrote:
irqfd in kvm used flush_work incorrectly:
it assumed that work scheduled previously can't run
after flush_work, but since kvm uses a non-reentrant
workqueue (by means of schedule_work)
we need flush_work_sync to get that guarantee.
Applied, and
On 03/22/11 05:03, slavik wrote:
> I have some troubles with passing usb into kvm virtual machine.
> I trying to provide the usb device 2022:0008 (Amikon vpn key) or 19d2:2000
> (ZTE MF112 hsdpa modem) into virtual machine with windows xp iax32.
> kvm virtual machine was hung completely, and sto
slavik wrote:
I have some troubles with passing usb into kvm virtual machine.
I trying to provide the usb device 2022:0008 (Amikon vpn key) or 19d2:2000
(ZTE MF112 hsdpa modem) into virtual machine with windows xp iax32.
kvm virtual machine was hung completely, and stop responding to anything.
M
When enabling the -device usb-host option support for adding automatically
USB devices from the host to the guest, only one device gets detected.
It does not matter if it is added via commandline or via device_add on the
qemu console.
Curious: If a second devices is plugged into the host, nothi
i suggest this is synonyms for single function, doesn't it?
On Tue, Mar 22, 2011 at 11:07 PM, David Ahern wrote:
>
>
> On 03/22/11 11:53, slavik wrote:
>> hi, as usual
>> kvm -boot d -usbdevice tablet -vnc :8 -monitor
>> telnet:0.0.0.0:4008,server,nowait -drive
>> file="/winxp.img",cache=unsaf
On 03/22/11 12:20, slavik wrote:
> i suggest this is synonyms for single function, doesn't it?
>
It's the legacy syntax - you are already using it with the tablet
option. If it cleared your problem then it is also synonymous for bug in
the -device route.
> On Tue, Mar 22, 2011 at 11:07 PM, Da
Hello,
I wanted to give vhost_net a try, but I can't make it work,
when I try to start qemu-kvm with vhost interface, it crashers with
msix_unset_mask_notifier: Assertion `dev->msix_mask_notifier' failed.
according to some threads I've digged up, it is related to missing
eventfd support (I'm using
On Tuesday, March 22, 2011, Ingo Molnar wrote:
>
> * Rafael J. Wysocki wrote:
>
> > > If there are no objectsions, I'd like to push these patches through the
> > > suspend
> > > tree.
> >
> > [1/8] has been merged in the meantime and [3/8] has been included into the
> > ACPI tree. if there ar
On Tuesday, March 22, 2011, Joerg Roedel wrote:
> On Mon, Mar 21, 2011 at 07:36:17PM -0400, Rafael J. Wysocki wrote:
> > drivers/pci/intel-iommu.c | 38 +-
> > 1 file changed, 9 insertions(+), 29 deletions(-)
>
> Looks good.
May I take that as an ACK?
> I p
On Tue, 2011-03-22 at 13:36 +0200, Michael S. Tsirkin wrote:
> diff --git a/drivers/virtio/virtio_ring.c
> b/drivers/virtio/virtio_ring.c
> index cc2f73e..6106017 100644
> --- a/drivers/virtio/virtio_ring.c
> +++ b/drivers/virtio/virtio_ring.c
> @@ -185,11 +185,6 @@ int virtqueue_add_buf_gfp(struct
Hi folks,
Kvm.git: commit 2ee44a580db58f98d85b57bfc468bbc5729ec9b3 Author: Avi Kivity
Date: Mon Mar 21 12:53:58 2011 +0200
My build system: RHEL5u5, Linux kvm-build 2.6.38-rc4+ #1 SMP Sat Feb 19
15:35:09 CST 2011 x86_64 x86_64 x86_64 GNU/Linux
When I make kvm, I get the following error.
d
To utilize ft_trans_file function, savevm needs interfaces to be
exported.
Signed-off-by: Yoshiaki Tamura
---
hw/hw.h |5 ++
savevm.c | 150 ++
2 files changed, 155 insertions(+), 0 deletions(-)
diff --git a/hw/hw.h b/hw/hw.h
ind
When "kemari:" is set in front of URI of migrate command, it will turn
on ft_mode to start FT migration mode (Kemari). On the receiver side,
the option looks like, -incoming kemari:::
Signed-off-by: Yoshiaki Tamura
Acked-by: Paolo Bonzini
---
hmp-commands.hx |4 +++-
migration.c | 12
event-tap function is called only when it is on.
Signed-off-by: Yoshiaki Tamura
---
net.c |9 +
1 files changed, 9 insertions(+), 0 deletions(-)
diff --git a/net.c b/net.c
index ddcca97..a541ede 100644
--- a/net.c
+++ b/net.c
@@ -37,6 +37,7 @@
#include "qemu_socket.h"
#include "hw
Introduce qemu_savevm_trans_{begin,commit} to send the memory and
device info together, while avoiding cancelling memory state tracking.
This patch also abstracts common code between
qemu_savevm_state_{begin,iterate,commit}.
Signed-off-by: Yoshiaki Tamura
---
savevm.c | 157
Hi,
This patch series is a revised version of Kemari for KVM, which
applied comments for the previous post. The current code is based on
qemu.git 4ac8e585c85079f6fd2b2b6da3cb845e3e19459c.
The changes from v0.2.12 -> v0.2.13 are:
- replaced qemu_get_timer() with qemu_get_timer_ns()
- check check
Introduce qemu_loadvm_state_no_header() so that it can be called
iteratively without reading the header, and qemu_loadvm_state()
becomes a wrapper of it.
Signed-off-by: Yoshiaki Tamura
---
savevm.c | 45 +++--
1 files changed, 27 insertions(+), 18 deleti
When ft_mode is set in the header, tcp_accept_incoming_migration()
sets ft_trans_incoming() as a callback, and call
qemu_file_get_notify() to receive FT transaction iteratively. We also
need a hack no to close fd before moving to ft_transaction mode, so
that we can reuse the fd for it. vm_change_
event-tap controls when to start FT transaction, and provides proxy
functions to called from net/block devices. While FT transaction, it
queues up net/block requests, and flush them when the transaction gets
completed.
Signed-off-by: Yoshiaki Tamura
Signed-off-by: OHMURA Kei
---
Makefile.targe
Signed-off-by: Yoshiaki Tamura
---
qemu-char.c |2 +-
qemu_socket.h |1 +
2 files changed, 2 insertions(+), 1 deletions(-)
diff --git a/qemu-char.c b/qemu-char.c
index 31c9e79..fa16d36 100644
--- a/qemu-char.c
+++ b/qemu-char.c
@@ -2111,7 +2111,7 @@ static void tcp_chr_telnet_init(int
This code implements VM transaction protocol. Like buffered_file, it
sits between savevm and migration layer. With this architecture, VM
transaction protocol is implemented mostly independent from other
existing code.
Signed-off-by: Yoshiaki Tamura
Signed-off-by: OHMURA Kei
---
Makefile.objs
Currently buf size is fixed at 32KB. It would be useful if it could
be flexible.
Signed-off-by: Yoshiaki Tamura
---
hw/hw.h |2 ++
savevm.c | 20 +++-
2 files changed, 21 insertions(+), 1 deletions(-)
diff --git a/hw/hw.h b/hw/hw.h
index 1b09039..f90ff15 100644
--- a/hw/
Introduce migrate_ft_trans_put_ready() which kicks the FT transaction
cycle. When ft_mode is on, migrate_fd_put_ready() would open
ft_trans_file and turn on event_tap. To end or cancel FT transaction,
ft_mode and event_tap is turned off. migrate_ft_trans_get_ready() is
called to receive ack from
For regular migration inuse == 0 always as requests are flushed before
save. However, event-tap log when enabled introduces an extra queue
for requests which is not being flushed, thus the last inuse requests
are left in the event-tap queue. Move the last_avail_idx value sent
to the remote back to
event-tap function is called only when it is on, and requests were
sent from device emulators.
Signed-off-by: Yoshiaki Tamura
Acked-by: Kevin Wolf
---
block.c | 15 +++
1 files changed, 15 insertions(+), 0 deletions(-)
diff --git a/block.c b/block.c
index c8e2f97..952543a 100644
Record mmio write event to replay it upon failover.
Signed-off-by: Yoshiaki Tamura
---
exec.c |4
1 files changed, 4 insertions(+), 0 deletions(-)
diff --git a/exec.c b/exec.c
index 964ce31..be71464 100644
--- a/exec.c
+++ b/exec.c
@@ -33,6 +33,7 @@
#include "osdep.h"
#include "kvm.h
Signed-off-by: Yoshiaki Tamura
---
vl.c |3 +++
1 files changed, 3 insertions(+), 0 deletions(-)
diff --git a/vl.c b/vl.c
index 483e2e3..6ed9b20 100644
--- a/vl.c
+++ b/vl.c
@@ -160,6 +160,7 @@ int main(int argc, char **argv)
#include "qemu-queue.h"
#include "cpus.h"
#include "arch_init.h
Record ioport event to replay it upon failover.
Signed-off-by: Yoshiaki Tamura
---
ioport.c |2 ++
1 files changed, 2 insertions(+), 0 deletions(-)
diff --git a/ioport.c b/ioport.c
index 2e971fa..f485bab 100644
--- a/ioport.c
+++ b/ioport.c
@@ -27,6 +27,7 @@
#include "ioport.h"
#include
Make deleting handlers robust against deletion of any elements in a
handler by using a deleted flag like in file descriptors.
Signed-off-by: Yoshiaki Tamura
---
vl.c | 15 ++-
1 files changed, 10 insertions(+), 5 deletions(-)
diff --git a/vl.c b/vl.c
index dbb927d..483e2e3 100644
Currently FdMigrationState doesn't support read(), and this patch
introduces it to get response from the other side. Note that this
won't change the existing migration protocol to be bi-directional.
Signed-off-by: Yoshiaki Tamura
---
migration-tcp.c | 15 +++
migration.c | 1
65 matches
Mail list logo
