On 02/24/2010 06:02 PM, Anthony Liguori wrote:
Hi Chris,
On 02/22/2010 04:52 PM, Chris Wright wrote:
Please send in any agenda items you are interested in covering.
I thought it might be easier to use the wiki for collecting agenda
items instead of polling the list. It also us do a little b
On Wed, Feb 24, 2010 at 09:16:02PM +, Natalia Portillo wrote:
> When the guest is in text mode only the first line is shown.
>
> Tested under KDE4 + NVidia binary drivers + Linux 2.6.31-gentoo-r1
> (x86-64) using QEMU 0.12.2.
>
Is this with kvm or TCG? If this is with kvm can you run "info cp
On Thu, Feb 25, 2010 at 12:58:26AM +0100, Jan Kiszka wrote:
> Marcelo Tosatti wrote:
> > On Thu, Feb 25, 2010 at 12:45:55AM +0100, Jan Kiszka wrote:
> >> Marcelo Tosatti wrote:
> >>> On Wed, Feb 24, 2010 at 03:17:55PM +0100, Jan Kiszka wrote:
> Drop kvm_load_tsc in favor of level-dependent wri
Marcelo Tosatti wrote:
> On Wed, Feb 24, 2010 at 03:17:52PM +0100, Jan Kiszka wrote:
>> Push mpstate reading/writing into kvm_arch_load/save_regs and, on x86,
>> properly synchronize with halted in the accessor functions.
>>
>> Signed-off-by: Jan Kiszka
>
>> @@ -1290,6 +1318,7 @@ int kvm_arch_ini
Marcelo Tosatti wrote:
> On Thu, Feb 25, 2010 at 12:45:55AM +0100, Jan Kiszka wrote:
>> Marcelo Tosatti wrote:
>>> On Wed, Feb 24, 2010 at 03:17:55PM +0100, Jan Kiszka wrote:
Drop kvm_load_tsc in favor of level-dependent writeback in
kvm_arch_load_regs. KVM's PV clock MSRs fall in the sam
Hello,
I just want to note the fact that although Linux got it badly wrong,
according to POSIX, MADV_DONTNEED is _not_ supposed to drop the content
of the memory, but just to tune the write-back heuristics and such.
(see glibc's ./sysdeps/unix/sysv/linux/posix_madvise.c if you're not
convinced)
q
Marcelo Tosatti wrote:
> On Wed, Feb 24, 2010 at 03:17:48PM +0100, Jan Kiszka wrote:
>> Let's proceed with cleaning up the VCPU state writeback. The differences
>> to v2 are:
>> - included guest debugging update patch and rebased on top of it
>> - renamed KVM_PUT_ASYNC_STATE->KVM_PUT_RUNTIME_STAT
Marcelo Tosatti wrote:
> On Wed, Feb 24, 2010 at 03:17:53PM +0100, Jan Kiszka wrote:
>> Do not write nmi_pending, sipi_vector, and mpstate unless we at least go
>> through a reset. And TSC as well as KVM wallclocks should only be
>> written on full sync, otherwise we risk to drop some time on durin
On Thu, Feb 25, 2010 at 12:45:55AM +0100, Jan Kiszka wrote:
> Marcelo Tosatti wrote:
> > On Wed, Feb 24, 2010 at 03:17:55PM +0100, Jan Kiszka wrote:
> >> Drop kvm_load_tsc in favor of level-dependent writeback in
> >> kvm_arch_load_regs. KVM's PV clock MSRs fall in the same category and
> >> should
Marcelo Tosatti wrote:
> On Wed, Feb 24, 2010 at 03:17:55PM +0100, Jan Kiszka wrote:
>> Drop kvm_load_tsc in favor of level-dependent writeback in
>> kvm_arch_load_regs. KVM's PV clock MSRs fall in the same category and
>> should therefore only be written back on full sync.
>>
>> Signed-off-by: Jan
On Wed, Feb 24, 2010 at 03:17:55PM +0100, Jan Kiszka wrote:
> Drop kvm_load_tsc in favor of level-dependent writeback in
> kvm_arch_load_regs. KVM's PV clock MSRs fall in the same category and
> should therefore only be written back on full sync.
>
> Signed-off-by: Jan Kiszka
> ---
> qemu-kvm-x8
On Wed, Feb 24, 2010 at 03:17:53PM +0100, Jan Kiszka wrote:
> Do not write nmi_pending, sipi_vector, and mpstate unless we at least go
> through a reset. And TSC as well as KVM wallclocks should only be
> written on full sync, otherwise we risk to drop some time on during
> state read-modify-write.
On Wed, Feb 24, 2010 at 03:17:52PM +0100, Jan Kiszka wrote:
> Push mpstate reading/writing into kvm_arch_load/save_regs and, on x86,
> properly synchronize with halted in the accessor functions.
>
> Signed-off-by: Jan Kiszka
> @@ -1290,6 +1318,7 @@ int kvm_arch_init_vcpu(CPUState *cenv)
> #ifde
On Wed, Feb 24, 2010 at 03:17:48PM +0100, Jan Kiszka wrote:
> Let's proceed with cleaning up the VCPU state writeback. The differences
> to v2 are:
> - included guest debugging update patch and rebased on top of it
> - renamed KVM_PUT_ASYNC_STATE->KVM_PUT_RUNTIME_STATE and added comments
> - fix
On 02/24/2010 02:59 PM, Blue Swirl wrote:
On 2/24/10, Anthony Liguori wrote:
If you compare the RSS of a freshly booted guest and the same guest after a
reboot, it's very likely the freshly booted guest will have an RSS that is
much lower the the rebooted guest because the previous run
What I've found is that the PPMs are still opened while QEMU is not,
however the PPM corruption only happens on network shares, while the
QCOW2 happens also in local disks.
zeus:feo claunia$ ps aux | grep -i qemu
claunia 80680 0,7 0,0 2435032472 s002 R+9:16PM 0:00.00
grep -i
From: Avi Kivity
Instead of allocating a separate chunk for the first 640KB and another
for 1MB+, allocate one large chunk. This plays well in terms of alignment
and size with large pages.
Signed-off-by: Avi Kivity
(cherry picked from commit cfe0cef63988a7876a9bbcb098500a3983e63814)
Index: qe
Port qemu-kvm's -mem-path and -mem-prealloc options. These are useful
for backing guest memory with huge pages via hugetlbfs.
Signed-off-by: Marcelo Tosatti
CC: john cooper
Index: qemu-kvm/cpu-all.h
===
--- qemu-kvm.orig/cpu-all.h
-mem-path option allows file backed guest memory.
When the guest is in text mode only the first line is shown.
Tested under KDE4 + NVidia binary drivers + Linux 2.6.31-gentoo-r1
(x86-64) using QEMU 0.12.2.
Attached example:
<>
It should be like (took in Mac OS X 10.6.2 using QEMU 0.12.2):
<>
On 2/24/10, Anthony Liguori wrote:
> If you compare the RSS of a freshly booted guest and the same guest after a
> reboot, it's very likely the freshly booted guest will have an RSS that is
> much lower the the rebooted guest because the previous run of the guest
> faulted
> in all available m
If you compare the RSS of a freshly booted guest and the same guest after a
reboot, it's very likely the freshly booted guest will have an RSS that is
much lower the the rebooted guest because the previous run of the guest faulted
in all available memory.
This patch addresses this issue by using m
A monitor may not even exist.
Change load_vmstate() to use qemu_error() instead of monitor_printf().
Parameter mon is now unused, remove it.
Signed-off-by: Markus Armbruster
---
monitor.c |2 +-
savevm.c | 24 +++-
sysemu.h |2 +-
vl.c |2 +-
4 files cha
Signed-off-by: Markus Armbruster
---
qerror.c |4
qerror.h |3 +++
2 files changed, 7 insertions(+), 0 deletions(-)
diff --git a/qerror.c b/qerror.c
index 38db6b2..8a38e5d 100644
--- a/qerror.c
+++ b/qerror.c
@@ -61,6 +61,10 @@ static const QErrorStringTable qerror_table[] = {
Paul Brook schrieb:
>> I suggest these steps:
>>
>> 1. Debug output to stdout is no longer accepted for new / modified code.
>>
>> 2. New or modified debug messages should go to stderr.
>
> I don't see this as a real improvement. Arguably these aren't errors, so
> stdout is where they should be goi
qemu_error_sink can either point to a monitor or a file. In practice,
it always points to the current monitor if we have one, else to
stderr. Simply route errors to the current monitor or else to stderr,
and remove qemu_error_sink along with the functions to control it.
Actually, the old code sw
The monitor_printf() reports failure. Printing is wrong, because the
caller tries various arguments, and expects the function to fail for
some or all.
Disabled since commit 26a9e82a. Remove it.
Signed-off-by: Markus Armbruster
---
usb-linux.c |8
1 files changed, 0 insertions(+),
Signed-off-by: Markus Armbruster
---
hw/qdev-properties.c |6 ++
hw/qdev.c|2 --
2 files changed, 2 insertions(+), 6 deletions(-)
diff --git a/hw/qdev-properties.c b/hw/qdev-properties.c
index 8e96805..a3572b0 100644
--- a/hw/qdev-properties.c
+++ b/hw/qdev-properties.c
Signed-off-by: Markus Armbruster
---
qerror.c |4
qerror.h |3 +++
2 files changed, 7 insertions(+), 0 deletions(-)
diff --git a/qerror.c b/qerror.c
index 45df41e..e555c33 100644
--- a/qerror.c
+++ b/qerror.c
@@ -105,6 +105,10 @@ static const QErrorStringTable qerror_table[] = {
Signed-off-by: Markus Armbruster
---
qerror.c |4
qerror.h |3 +++
2 files changed, 7 insertions(+), 0 deletions(-)
diff --git a/qerror.c b/qerror.c
index ea3971e..51b68bc 100644
--- a/qerror.c
+++ b/qerror.c
@@ -41,6 +41,10 @@ static const QType qerror_type = {
*/
static const
New LOC_FILE. Use it for tracking file name and line number in
qemu_config_parse(). We now report errors like
qemu:foo.conf:42: Did not find I2C bus for smbus-eeprom
In particular, gems like this message:
-device: no driver specified
become almost nice now:
qemu:foo.conf:44: -dev
While there, improve the params help text.
Signed-off-by: Markus Armbruster
---
hw/qdev.c |3 +--
qemu-monitor.hx |4 ++--
2 files changed, 3 insertions(+), 4 deletions(-)
diff --git a/hw/qdev.c b/hw/qdev.c
index 80a027a..bcd989c 100644
--- a/hw/qdev.c
+++ b/hw/qdev.c
@@ -766,8 +7
Signed-off-by: Markus Armbruster
---
qemu-error.c | 11 +++
qemu-error.h |2 ++
2 files changed, 13 insertions(+), 0 deletions(-)
diff --git a/qemu-error.c b/qemu-error.c
index a726d32..95e87c8 100644
--- a/qemu-error.c
+++ b/qemu-error.c
@@ -41,6 +41,17 @@ void error_printf(const
To make it obvious that -device and device_add can't die in
hw_error().
Signed-off-by: Markus Armbruster
---
hw/qdev.c | 37 ++---
1 files changed, 22 insertions(+), 15 deletions(-)
diff --git a/hw/qdev.c b/hw/qdev.c
index 7def97d..e5f6820 100644
--- a/hw/qdev.
Signed-off-by: Markus Armbruster
---
qerror.c |4
qerror.h |3 +++
2 files changed, 7 insertions(+), 0 deletions(-)
diff --git a/qerror.c b/qerror.c
index 7ebc2bb..4761ae5 100644
--- a/qerror.c
+++ b/qerror.c
@@ -41,6 +41,10 @@ static const QType qerror_type = {
*/
static const
Next commit wants to use it.
Signed-off-by: Markus Armbruster
---
qemu-config.c |8
qemu-config.h |1 +
2 files changed, 5 insertions(+), 4 deletions(-)
diff --git a/qemu-config.c b/qemu-config.c
index 72b6279..7346aa0 100644
--- a/qemu-config.c
+++ b/qemu-config.c
@@ -309,7 +3
Signed-off-by: Markus Armbruster
---
qerror.c |4
qerror.h |3 +++
2 files changed, 7 insertions(+), 0 deletions(-)
diff --git a/qerror.c b/qerror.c
index 8a38e5d..2f3b89c 100644
--- a/qerror.c
+++ b/qerror.c
@@ -125,6 +125,10 @@ static const QErrorStringTable qerror_table[] = {
Signed-off-by: Markus Armbruster
---
qemu-config.c |4
1 files changed, 4 insertions(+), 0 deletions(-)
diff --git a/qemu-config.c b/qemu-config.c
index 78e5c67..acd4db7 100644
--- a/qemu-config.c
+++ b/qemu-config.c
@@ -475,6 +475,10 @@ int qemu_config_parse(FILE *fp, const char *fnam
This reverts commit 3ced9f7a36189aed94d8bf86f3f5087a53012455.
The next commit will convert all of qdev_device_add() to QError, and
it'll be clearer with this partial conversion reverted.
Signed-off-by: Markus Armbruster
---
hw/qdev.c |4 ++--
1 files changed, 2 insertions(+), 2 deletions(-)
FIXME They should return int, so callers can calculate width.
Signed-off-by: Markus Armbruster
---
qemu-error.c | 49 ++---
qemu-error.h | 14 ++
2 files changed, 56 insertions(+), 7 deletions(-)
diff --git a/qemu-error.c b/qemu-error.
Commit 30d335d6 converted an informational message from
monitor_printf() to qemu_error(), probably because the latter doesn't
need a mon argument. A later commit will make qemu_error() print
additional stuff that is only appropriate for proper errors, and then
this will break. Clean it up.
Signe
Signed-off-by: Markus Armbruster
---
qerror.c |4
qerror.h |3 +++
2 files changed, 7 insertions(+), 0 deletions(-)
diff --git a/qerror.c b/qerror.c
index 56407be..7ebc2bb 100644
--- a/qerror.c
+++ b/qerror.c
@@ -73,6 +73,10 @@ static const QErrorStringTable qerror_table[] = {
Why this is such a big job? There are two issues with a naive
conversion:
* Error message degradation
The error messages are worded for -device. They aren't so hot to
begin with: we typically have many -device options, and to which one
a message applies is often not obvious.
Now, QMP w
Signed-off-by: Markus Armbruster
---
hw/qdev.c | 25 ++---
1 files changed, 14 insertions(+), 11 deletions(-)
diff --git a/hw/qdev.c b/hw/qdev.c
index 6495894..1887dde 100644
--- a/hw/qdev.c
+++ b/hw/qdev.c
@@ -216,18 +216,21 @@ DeviceState *qdev_device_add(QemuOpts *opts)
Signed-off-by: Markus Armbruster
---
qerror.c |4
qerror.h |3 +++
2 files changed, 7 insertions(+), 0 deletions(-)
diff --git a/qerror.c b/qerror.c
index 51b68bc..56407be 100644
--- a/qerror.c
+++ b/qerror.c
@@ -57,6 +57,10 @@ static const QErrorStringTable qerror_table[] = {
Commit 0ecdffbb created pc_boot_set() for use from monitor command
"boot_set", via qemu_boot_set(). pc_boot_set() reports errors to
cur_mon, which works fine for monitor code.
Commit e0f084bf reused the function int reset handler
restore_boot_devices(). Use of cur_mon is problematic in that
cont
qdev_device_help() prints device information with qemu_error(). A
later commit will make qemu_error() print additional stuff that is
only appropriate for proper errors, and then this will break. Use
error_printf() instead.
While there, simplify: instead of printing a buffer filled by
qdev_print_
Signed-off-by: Markus Armbruster
---
qemu-tool.c |2 --
1 files changed, 0 insertions(+), 2 deletions(-)
diff --git a/qemu-tool.c b/qemu-tool.c
index 18b48af..bbce99f 100644
--- a/qemu-tool.c
+++ b/qemu-tool.c
@@ -33,8 +33,6 @@ void qemu_service_io(void)
{
}
-Monitor *cur_mon;
-
void m
Commits 376253ec..731b0364 introduced global variable cur_mon, which
points to the "default monitor" (if any), except during execution of
monitor_read() or monitor_control_read() it points to the monitor from
which we're reading instead (the "current monitor"). Monitor command
handlers run within
Code duplicated in commit 0ecdffbb. The two versions are similar, but
not identical:
* cmos_init() reports errors to stderr, pc_boot_set() via
qemu_error(). The latter is fine for both, so pick that for the
common code.
* cmos_init() obeys fd_bootchk, pc_boot_set() ignores it. Make it a
The old test assumes that "hotplugged" implies "we have a current
monitor for reading the key". This is in fact true, but it's not
obviously true.
Aside: if it were false, we could pass a null pointer to
monitor_read_bdrv_key_start(), which would then crash.
The previous commit permits us to che
Signed-off-by: Markus Armbruster
---
hw/scsi-bus.c |2 +-
hw/scsi-disk.c|3 +--
hw/scsi-generic.c |1 +
hw/usb-serial.c |1 +
hw/virtio-net.c |1 +
hw/virtio-pci.c |2 +-
monitor.h |1 +
net/dump.c|1 +
net/socket.c |2 +-
ne
Signed-off-by: Markus Armbruster
---
qerror.c |4
qerror.h |3 +++
2 files changed, 7 insertions(+), 0 deletions(-)
diff --git a/qerror.c b/qerror.c
index 4761ae5..38db6b2 100644
--- a/qerror.c
+++ b/qerror.c
@@ -49,6 +49,10 @@ static const QErrorStringTable qerror_table[] = {
Signed-off-by: Markus Armbruster
---
monitor.c |5 +
monitor.h |2 ++
2 files changed, 7 insertions(+), 0 deletions(-)
diff --git a/monitor.c b/monitor.c
index a4263af..5c87a98 100644
--- a/monitor.c
+++ b/monitor.c
@@ -194,6 +194,11 @@ static inline int monitor_ctrl_mode(const Moni
Signed-off-by: Markus Armbruster
---
qemu-error.c | 20 +++-
qemu-error.h |1 +
vl.c |2 ++
3 files changed, 22 insertions(+), 1 deletions(-)
diff --git a/qemu-error.c b/qemu-error.c
index 0778001..b2c0a80 100644
--- a/qemu-error.c
+++ b/qemu-error.c
@@ -113,13
Signed-off-by: Markus Armbruster
---
hw/qdev.c | 29 +
hw/qdev.h |2 +-
qemu-monitor.hx |3 ++-
3 files changed, 24 insertions(+), 10 deletions(-)
diff --git a/hw/qdev.c b/hw/qdev.c
index bcd989c..a631492 100644
--- a/hw/qdev.c
+++ b/hw/qdev.c
@
Conversion to QObject is still missing.
Signed-off-by: Markus Armbruster
---
hw/qdev.c | 19 +--
1 files changed, 9 insertions(+), 10 deletions(-)
diff --git a/hw/qdev.c b/hw/qdev.c
index 4e348fd..a028e9c 100644
--- a/hw/qdev.c
+++ b/hw/qdev.c
@@ -200,15 +200,15 @@ DeviceState
Also put error definitions in alphabetical order
Signed-off-by: Markus Armbruster
---
qerror.c | 28 ++--
qerror.h | 14 +++---
2 files changed, 21 insertions(+), 21 deletions(-)
diff --git a/qerror.c b/qerror.c
index 92b05eb..45df41e 100644
--- a/qerror.c
++
This fixes a few error messages lacking a newline:
net_handle_fd_param()'s "No file descriptor named %s found", and
tap_open()'s "vnet_hdr=1 requested, but no kernel support for
IFF_VNET_HDR available" (all three versions).
There's one place that passes arguments without newlines
intentionally: lo
Treat multiple successive slashes as a one slash. Ignore trailing
slashes. This is how POSIX pathnames work.
Signed-off-by: Markus Armbruster
---
hw/qdev.c | 26 +-
1 files changed, 17 insertions(+), 9 deletions(-)
diff --git a/hw/qdev.c b/hw/qdev.c
index b170081..64
FIXME Only string options are implemented.
Signed-off-by: Markus Armbruster
---
qemu-option.c | 54 ++
qemu-option.h |3 +++
2 files changed, 57 insertions(+), 0 deletions(-)
diff --git a/qemu-option.c b/qemu-option.c
index ab488e4..f86a
In the human monitor, it declares a single optional argument to be
parsed according to the QemuOptsList given by its name.
In QMP, it declares an optional argument for each member of the
QemuOptsList.
FIXME Only string options are implemented.
Signed-off-by: Markus Armbruster
---
monitor.c |
Users can't set them, so qdev_device_help() shouldn't list them. Fix
that. Also make qdev_prop_parse() hide them instead of printing a
meaningless "has no parser" error message.
Their value means nothing to users, so qdev_print_props() shouldn't
print it. Fix by removing their print method.
Th
We sometimes permit omitting the first option name, for example
-device foo is short for -device driver=foo. The name to use
("driver" in the example) is passed as argument to qemu_opts_parse().
For each QemuOptsList, we use at most one such name.
Move the name into QemuOptsList, and pass whether
"device_add isa-serial,bus=pci.0" kills QEMU. Not good.
Signed-off-by: Markus Armbruster
---
hw/qdev.c |5 +
1 files changed, 5 insertions(+), 0 deletions(-)
diff --git a/hw/qdev.c b/hw/qdev.c
index 38fc4e7..7def97d 100644
--- a/hw/qdev.c
+++ b/hw/qdev.c
@@ -204,6 +204,11 @@ DeviceStat
Users can't create them, so qdev_device_help() shouldn't list them.
Fix that.
Also make qdev_device_add() pretend they don't exist. Before, it
rejected them with a "can't be added via command line" message, which
wasn't quite right for monitor command device_add.
Signed-off-by: Markus Armbruster
This separates the monitor part from the QError part.
Signed-off-by: Markus Armbruster
---
monitor.c | 21 +
monitor.h |3 +++
2 files changed, 16 insertions(+), 8 deletions(-)
diff --git a/monitor.c b/monitor.c
index b1a6edc..fcd8718 100644
--- a/monitor.c
+++ b/monit
Signed-off-by: Markus Armbruster
---
qerror.c |4
qerror.h |3 +++
2 files changed, 7 insertions(+), 0 deletions(-)
diff --git a/qerror.c b/qerror.c
index e555c33..ea3971e 100644
--- a/qerror.c
+++ b/qerror.c
@@ -109,6 +109,10 @@ static const QErrorStringTable qerror_table[] = {
New LOC_CMDLINE. Use it for tracking option with argument in
lookup_opt(). We now report errors like this
qemu: -device smbus-eeprom: Did not find I2C bus for smbus-eeprom
Signed-off-by: Markus Armbruster
---
qemu-error.c | 20
qemu-error.h |3 ++-
vl.c
qbus_find() adds an informational line to error messages, and prints
both lines with one qemu_error(). Use error_printf() for the
informational line instead.
While there, simplify: instead of printing buffers filled by
qbus_list_bus() and qbus_list_dev() in one go, make them print it.
Signed-off
New struct Location holds a location. So far, the only location is
LOC_NONE, so this doesn't do anything useful yet.
Passing the current location all over the place would be too
cumbersome. Hide it away in static cur_loc instead, and provide
accessors. Print it in qemu_error().
Store it in QEr
They're about reporting errors, not about the monitor.
Signed-off-by: Markus Armbruster
---
Makefile.target |1 +
monitor.c | 89 -
qemu-error.c| 92 +++
3 files changed, 93
I am trying to emulate ARMv6 on QEMU and cannot get past the early
kernel boot process. I wish to emulate the arm1136 cpu on the
realview-eb platform.
The config file for the kernel and my binaries can be found here:
http://www.cs.columbia.edu/~cdall/qemu-realview.tar.gz
To configure and compile
Hi Chris,
On 02/22/2010 04:52 PM, Chris Wright wrote:
Please send in any agenda items you are interested in covering.
I thought it might be easier to use the wiki for collecting agenda items
instead of polling the list. It also us do a little bit more long term
planning.
http://www.li
On 02/24/2010 09:13 AM, Anthony Liguori wrote:
to work on that if we can agree on what the underlying device looks
like whether it should be VirtIO or just PCI.
Agreed, you had suggested uio_pci for my PCI driver and I'd be happy
I feel pretty strongly that we shouldn't add shared memory to vir
If the parser fails to parse the key in parse_pair, it will access a NULL
pointer. A simple way to trigger this is sending {foo} via QMP. This patch
turns the segfault into a syntax error reply.
Signed-off-by: Kevin Wolf
---
json-parser.c |2 +-
1 files changed, 1 insertions(+), 1 deletions(
On 02/24/2010 05:46 AM, Michael S. Tsirkin wrote:
Ah, you mean telling the guest to switch features on and off: natureally
this would require guest driver changes. This might be also non-trivial
to implement. Consider a request that has been posted without checksum,
suddenly we disable checksum
On 02/24/2010 12:20 AM, Cam Macdonell wrote:
On Tue, Feb 23, 2010 at 2:05 PM, Anthony Liguori wrote:
Hi Cam,
On 02/23/2010 02:52 PM, Cam Macdonell wrote:
Support for passing memory regions via VirtIO to remove need for PCI
support in the guest.
Adds new vectors to VirtIO config spa
Drop kvm_load_tsc in favor of level-dependent writeback in
kvm_arch_load_regs. KVM's PV clock MSRs fall in the same category and
should therefore only be written back on full sync.
Signed-off-by: Jan Kiszka
---
qemu-kvm-x86.c| 19 +--
qemu-kvm.h|4
On 02/24/2010 05:14 AM, Avi Kivity wrote:
On 02/24/2010 08:20 AM, Cam Macdonell wrote:
Virtio is really a DMA engine. One of the nice things about it's
design is
that you can do things like transparent bounce buffering if needed.
Adding
a mechanism like this breaks this abstract and turns v
On 02/23/2010 08:58 PM, Paul Brook wrote:
Bottom halves are run at the very end of the event loop which means that
they're guaranteed to be the last thing run. idle bottom halves can be
rescheduled without causing an infinite loop and do not affect the
select timeout (which normal bottom halves
On 02/23/2010 09:14 PM, Paul Brook wrote:
vnet_hdr is IMHO a really bad example to copy from.
vnet_hdr leaks into the migration state via n->has_vnet_hdr. What this
means is that if you want to migrate from -net tap -net nic,model=virtio
to -net user -net nic,model=virtio, it will fail.
This i
Push mpstate reading/writing into kvm_arch_load/save_regs and, on x86,
properly synchronize with halted in the accessor functions.
Signed-off-by: Jan Kiszka
---
hw/apic.c |7 ---
qemu-kvm-ia64.c |4 +-
qemu-kvm-x86.c| 102 ++-
Do not write nmi_pending, sipi_vector, and mpstate unless we at least go
through a reset. And TSC as well as KVM wallclocks should only be
written on full sync, otherwise we risk to drop some time on during
state read-modify-write.
Signed-off-by: Jan Kiszka
---
kvm.h |2 +-
q
Push reading/writing of vcpu_events into kvm_arch_load/save_regs to
avoid KVM-specific hooks in generic code.
Signed-off-by: Jan Kiszka
---
kvm.h |2 --
qemu-kvm-x86.c|6 --
target-i386/kvm.c |4 ++--
target-i386/machine.c |6 --
4 files change
The APIC is part of the VCPU state, so trigger its readout and writeback
from kvm_arch_save/load_regs. Thanks to the transparent sync on reset
and vmsave/load, we can also drop explicit sync code, reducing the diff
to upstream.
Signed-off-by: Jan Kiszka
---
hw/apic.c | 37 +---
There is no need for the this hack anymore, initialization is now robust
against reordering as it doesn't try to write the VCPU state on its own.
Signed-off-by: Jan Kiszka
---
hw/pc.c |5 -
target-i386/helper.c |2 ++
2 files changed, 2 insertions(+), 5 deletions(-)
dif
Setting the boot CPU ID is arch-specific KVM stuff. So push it where it
belongs to.
Signed-off-by: Jan Kiszka
---
hw/pc.c|3 ---
qemu-kvm-x86.c |3 ++-
2 files changed, 2 insertions(+), 4 deletions(-)
diff --git a/hw/pc.c b/hw/pc.c
index 8b5af35..971ae70 100644
--- a/hw/pc.c
+++
Let's proceed with cleaning up the VCPU state writeback. The differences
to v2 are:
- included guest debugging update patch and rebased on top of it
- renamed KVM_PUT_ASYNC_STATE->KVM_PUT_RUNTIME_STATE and added comments
- fixed mp_state corruption due to early use of cpu_is_bsp
Some patches ta
This grand cleanup drops all reset and vmsave/load related
synchronization points in favor of four(!) generic hooks:
- cpu_synchronize_all_states in qemu_savevm_state_complete
(initial sync from kernel before vmsave)
- cpu_synchronize_all_post_init in qemu_loadvm_state
(writeback after vmload)
As we hard-wire the BSP to CPU 0 anyway and cpuid_apic_id equals
cpu_index, cpu_is_bsp can also be based on the latter directly. This
will help an early user of it: KVM while initializing mp_state.
Signed-off-by: Jan Kiszka
---
hw/pc.c |3 ++-
1 files changed, 2 insertions(+), 1 deletions(-)
This add-on patch to recent guest debugging refactorings adds the
requested awareness for KVM_CAP_X86_ROBUST_SINGLESTEP to both the
upstream as well as qemu-kvm's own code. Fortunately, code sharing
increased once again.
Signed-off-by: Jan Kiszka
---
kvm-all.c | 12 ++
kvm.h
On Wed, Feb 24, 2010 at 12:26:53PM +, Paul Brook wrote:
> > > If we do have a software
> > > fallback then the feature bit is just for backwards compatibility, and
> > > should be enabled unconditionally (on current machine types).
> >
> > Software fallback might turn out to be slower than disa
> > If we do have a software
> > fallback then the feature bit is just for backwards compatibility, and
> > should be enabled unconditionally (on current machine types).
>
> Software fallback might turn out to be slower than disabling the feature
> in the guest. For example, and extra pass over pac
On Wed, Feb 24, 2010 at 11:30:18AM +, Paul Brook wrote:
> > On Wed, Feb 24, 2010 at 03:14:25AM +, Paul Brook wrote:
> > > > vnet_hdr is IMHO a really bad example to copy from.
> > > >
> > > > vnet_hdr leaks into the migration state via n->has_vnet_hdr. What this
> > > > means is that if yo
> On Wed, Feb 24, 2010 at 03:14:25AM +, Paul Brook wrote:
> > > vnet_hdr is IMHO a really bad example to copy from.
> > >
> > > vnet_hdr leaks into the migration state via n->has_vnet_hdr. What this
> > > means is that if you want to migrate from -net tap -net
> > > nic,model=virtio to -net us
On 02/23/2010 02:58 PM, Aurelien Jarno wrote:
I have totally rewritten the patch to be more along the line
that Laurent was suggesting, in that the rounding mode and other
qualifiers are totally parsed within the translator. I no longer
pass the FN11 field to the helper functions.
What's the b
On 02/24/2010 08:20 AM, Cam Macdonell wrote:
Virtio is really a DMA engine. One of the nice things about it's design is
that you can do things like transparent bounce buffering if needed. Adding
a mechanism like this breaks this abstract and turns virtio into something
that's more than I thin
On 02/24/2010 02:36 AM, Natalia Portillo wrote:
10% of saved PPMs are corrupted or empty,
This seems easier to bisect, can you do that?
Thanks!
Paolo
99 matches
Mail list logo
