RE: unstable binaries

RE: unstable binaries

[Ian Pratt]

 
> i am coming from UML, and now i evaluate Xen on my desktop:
> It works, but some desktop applications crash once in a
> while within dom0:

Please can you try 2.0-testing. If this fixes your problem I'll declare
it 2.0.5
Thanks,
Ian
 
>   metacity-2.8.6
>   firefox-1.0.1
>   wnck-applet (from gnome-panel-2.8.1)
> 
> But others (X, xterm, gnome-panel, skype) remain stable, 
> so it's probably not related to my framebuffer X 
> (xorg-x11-Xvfb-6.8.1).
> 
> Before firefox crashes, it sometimes behaves odd:
> the location box does not get updated when i switch tabs; 
> scrollwhell, cursor keys, pageup/down keys stop navigating
> a document, while the scrollbar still does.
> 
> With the fedora kernels everything is fine.
> 
> Any suggestions?
> 
> /nils.
> 
> btw: 2.6.11 is out ...
> 
> -- 
> there is no sig.
> 
> 
> -------------------------------------------------------
> SF email is sponsored by - The IT Product Guide
> Read honest & candid reviews on hundreds of IT Products from 
> real users.
> Discover which products truly live up to the hype. Start reading now.
> http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click
> _______________________________________________
> Xen-devel mailing list
> Xen-devel@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/xen-devel
> 


-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_ide95&alloc_id\x14396&op=click

Re: unstable binaries

[nils toedtmann]

On Wed, Mar 02, 2005 at 05:54:09PM -0000, Ian Pratt wrote:
>  
> > i am coming from UML, and now i evaluate Xen on my desktop:
> > It works, but some desktop applications crash once in a
> > while within dom0:
> 
> Please can you try 2.0-testing. 

Up and running since 5min, no crashes - yet. Navigating
firefox through bloated websites/flash led within 60sec
to a coredump before, and now i works.

Unfortunatly i travel from tommorow to sunday, and i don't 
know if there will be enough "used uptime" until then on 
this desktop to be _sure_ the crashes won't occur any more. 
I'll try to let you know before i leave.


> If this fixes your problem I'll declare it 2.0.5

So xen-2.0-testing is something like RC of the next minor 
release? Will 2.0.5 support 2.6.11?

/nils.


-- 
there is no sig.


-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click

Re: unstable binaries

[nils toedtmann]

On Wed, Mar 02, 2005 at 11:15:10PM +0100, nils toedtmann wrote:
> On Wed, Mar 02, 2005 at 05:54:09PM -0000, Ian Pratt wrote:
> >  
> > > i am coming from UML, and now i evaluate Xen on my desktop:
> > > It works, but some desktop applications crash once in a
> > > while within dom0:
> > 
> > Please can you try 2.0-testing. 
> 
> Up and running since 5min, no crashes - yet. Navigating
> firefox through bloated websites/flash led within 60sec
> to a coredump before, and now i works.
> 
> Unfortunatly i travel from tommorow to sunday, and i don't 
> know if there will be enough "used uptime" until then on 
> this desktop to be _sure_ the crashes won't occur any more. 
> I'll try to let you know before i leave.

wget oopses sometimes, never seen that before:

    wget http://...
    25% [======>                    ] 1,028,840     55.08K/s    ETA 06:31
    wget: retr.c:293: calc_rate: Assertion `msecs >= 0' failed.
    Aborted
    
/nils.

-- 
there is no sig.


-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click

Re: unstable binaries

[nils toedtmann]

On Thu, Mar 03, 2005 at 12:49:24AM +0100, nils toedtmann wrote:
> On Wed, Mar 02, 2005 at 11:15:10PM +0100, nils toedtmann wrote:
> > On Wed, Mar 02, 2005 at 05:54:09PM -0000, Ian Pratt wrote:
> > >  
> > > > i am coming from UML, and now i evaluate Xen on my desktop:
> > > > It works, but some desktop applications crash once in a
> > > > while within dom0:
> > > 
> > > Please can you try 2.0-testing. 
> > 
> > Up and running since 5min, no crashes - yet. Navigating
> > firefox through bloated websites/flash led within 60sec
> > to a coredump before, and now i works.

While firefox is ok, metacity and wnck-applet still like to 
go beserk, consuming 100% cpu until killed.

usb-storage is _very_ slow (~30kByte/s).

I use xen-2.0-testing of yesterday.

/nils.

> wget oopses sometimes, never seen that before:
> 
>     wget http://...
>     25% [======>                    ] 1,028,840     55.08K/s    ETA 06:31
>     wget: retr.c:293: calc_rate: Assertion `msecs >= 0' failed.
>     Aborted



-- 
there is no sig.


-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click

Re: unstable binaries

[Tommi Virtanen]

> wget oopses sometimes, never seen that before:
> 
>     wget http://...
>     25% [======>                    ] 1,028,840     55.08K/s    ETA 06:31
>     wget: retr.c:293: calc_rate: Assertion `msecs >= 0' failed.
>     Aborted

That means time went backwards. I see that sometimes under xen, too.


-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click

Re: unstable binaries

[Robin Green]

On Thu, 3 Mar 2005, nils toedtmann wrote:
> wget oopses sometimes, never seen that before:
>
>    wget http://...
>    25% [======>                    ] 1,028,840     55.08K/s    ETA 06:31
>    wget: retr.c:293: calc_rate: Assertion `msecs >= 0' failed.
>    Aborted

This is a symptom of the famous Xen floating-point corruption bug.
What version (i.e. what date) is your Xen installation?
-- 
Robin



-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click

Re: unstable binaries

[nils toedtmann]

On Fri, Mar 04, 2005 at 07:39:40AM -0500, Robin Green wrote:
> On Thu, 3 Mar 2005, nils toedtmann wrote:
> >wget oopses sometimes, never seen that before:
> >
> >   wget http://...
> >   25% [======>                    ] 1,028,840     55.08K/s    ETA 06:31
> >   wget: retr.c:293: calc_rate: Assertion `msecs >= 0' failed.
> >   Aborted
> 
> This is a symptom of the famous Xen floating-point corruption bug.
> What version (i.e. what date) is your Xen installation?

The domain0 linux kernel was xen-2.0-testing, last change in ChangeLog:

  ChangeSet@1.1751, 2005-03-01 17:40:10+00:00, kaf24@scramble.cl.cam.ac.uk

I am not sure if Xen itself had the same version or if it was
2.0.4 (i first forgot to update the "kernel" line, too). I can
try to reproduce that error again when i'm back on monday to 
verify the versions - shall i?

/nils.
-- 
there is no sig.


-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click

Re: unstable binaries

[Robin Green]

On Fri, 4 Mar 2005, nils toedtmann wrote:

> On Fri, Mar 04, 2005 at 07:39:40AM -0500, Robin Green wrote:
>> On Thu, 3 Mar 2005, nils toedtmann wrote:
>>> wget oopses sometimes, never seen that before:
>>>
>>>   wget http://...
>>>   25% [======>                    ] 1,028,840     55.08K/s    ETA 06:31
>>>   wget: retr.c:293: calc_rate: Assertion `msecs >= 0' failed.
>>>   Aborted
>>
>> This is a symptom of the famous Xen floating-point corruption bug.
>> What version (i.e. what date) is your Xen installation?
>
> The domain0 linux kernel was xen-2.0-testing, last change in ChangeLog:
>
>  ChangeSet@1.1751, 2005-03-01 17:40:10+00:00, kaf24@scramble.cl.cam.ac.uk
>
> I am not sure if Xen itself had the same version or if it was
> 2.0.4 (i first forgot to update the "kernel" line, too). I can
> try to reproduce that error again when i'm back on monday to
> verify the versions - shall i?

Yes, can you please try it with xen-2.0-testing.
-- 
Robin



-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click

RE: unstable binaries

[Ian Pratt]

> Not at all. I never tested unstable. Is it worth a try (i 
> already spent
> LOTs of time in this bug trying different versions)? Then i'll do so.

Unstable has different fpsave/restore code, so worth trying.
 
> > Have you changed your kernel config from the default, 
> > or otherwise installed other kernel modules?
> 
> Yes, heavily. I compile all kernels myself. An example:
> <http://nils.toedtmann.net/stuff/virtualization/config-2.6.10-xen-2.0-
> testing-20050304-dom0-nils-ws.14>

It may well be being caused by some kernel module that we don't test in
our configs. Just because it work on x86 doesn't always mean it will get
away with it on Xen/x86.

> If it will help tracking the bug, i'll test the binary testing &
> unstable xen distribution, too.

Knowing whether you can repeat on unstable would be useful.

Thanks,
Ian


-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_ide95&alloc_id\x14396&op=click

RE: unstable binaries

[Ian Pratt]

 > > Also the strace output of the wget program failing might be handy
to
> > know if there's a link with nanosleep at least.
> 
> See <http://nils.toedtmann.net/stuff/virtualization/wget-
> xen-2.0t.20050314.strace.bz2>

This trace would seem to indicate that gettimeofday is working fine.
It's far more likely to be a floating point issue as the calculation is
done as a double.

What CPU does your system have?
Are you 100% sure you can't repeat these problems on unstable?
Have you changed your kernel config from the default, or otherwise
installed other kernel modules?

Ian


-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_ide95&alloc_id\x14396&op=click

RE: unstable binaries

[Nils Toedtmann]

Am Mittwoch, den 16.03.2005, 10:34 +0000 schrieb Ian Pratt:
> > > Also the strace output of the wget program failing might be handy
> > > to 
> > > know if there's a link with nanosleep at least.
> > 
> > See <http://nils.toedtmann.net/stuff/virtualization/wget-
> > xen-2.0t.20050314.strace.bz2>
> 
> This trace would seem to indicate that gettimeofday is working fine.
> It's far more likely to be a floating point issue as the calculation is
> done as a double.
> 
> What CPU does your system have?

I think i already posted this, but anyway:

  vendor_id       : AuthenticAMD
  cpu family      : 6
  model           : 4
  model name      : AMD Athlon(tm) Processor
  stepping        : 4
  cpu MHz         : 1410.253
  cache size      : 256 KB

I compile kernels with "CONFIG_MK7=y"

> Are you 100% sure you can't repeat these problems on unstable?

Not at all. I never tested unstable. Is it worth a try (i already spent
LOTs of time in this bug trying different versions)? Then i'll do so.

> Have you changed your kernel config from the default, 
> or otherwise installed other kernel modules?

Yes, heavily. I compile all kernels myself. An example:

  <http://nils.toedtmann.net/stuff/virtualization/config-2.6.10-xen-2.0-
testing-20050304-dom0-nils-ws.14>

But i tested the same config with a vanilla 2.6.10 (with ARCH specific
changes:

  <http://nils.toedtmann.net/stuff/virtualization/config-2.6.10-nils-
ws.14>)

and on that kernel wget was fine. So it should not be a .config issue,
right?

If it will help tracking the bug, i'll test the binary testing &
unstable xen distribution, too.

/nils.



-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click

RE: unstable binaries

[Michael Rice]

> > failed on all kernels, so it probably a fedora issue. The tests 
> > "ioperm02", "iopl02" and "nanosleep02" failed on all xen-kernels,
> > and only on them. This is the output for "nanosleep02":
> 
> The iopl02 and ioperm02 failures are expected, but not at all serious.
> We actually have a plan to virtualize iopl just to tidy this up.
> 
> nanosleep2 is a surprise -- I can"t recall seeing this fail before. I
> wander if its an Athlon issue...

I'm seeing a similar problem on different Hardware and OS.  Sorry 
in advance about the verbosity of this.  I am looking for an answer
and a reliably working kernel, so I'm off to try 2.0.5 and more.

hardware:	IBM x335 (2.4Ghz Xeon P4), 1.5G RAM, 2x80G IDE
domain0:	RedHat Enterprise Linux 3ES
domain0 kernel:	2.4.29-xen0
domain1:	(dd copy of dom0) RedHat Enterprise Linux 3ES
domain1 kernel: 2.6.10-xenU
xen:		xen-2.0.4-install.tgz
bridge-utils:   bridge-utils-1.0.4-1
Twisted:	Twisted-1.3.0-1tummy
unixbench:	unixbench-4.1.0
ltp:		ltp-full-20050307
gcc:		gcc version 3.2.3 20030502 (Red Hat Linux 3.2.3-47)

I ran ltp on domain1 and also get the error Nils reported earlier on 
nanosleep02.  Note that I am not getting this error on domain0 or 
native.  

full ltp output: http://www.riceclan.org/~michael/xen/
There are several failed tests in there.   

I was investigating why the unixbench-4.1.0 'speed' series would 
consistently hang during the 'float' tests.  While the tests were 
running (high CPU usage) top died with an error similar to Nils'
on wget (though the exact text escaped me).  tail has exhibited this
as well: 
tail: xnanosleep.c:128: xnanosleep: Assertion `0 <= seconds' failed.

The unixbench script does a 'sleep 1' and a 'sleep 2' inside the
test counter loop.  The first round these are exactly as expected
(see strace output at http://www.riceclan.org/~michael/xen/ and 
excerpt below) but in the second round (sometimes I make it to 
the third) my 'sleep 1' instead makes the third call below:

1367  nanosleep({1, 0}, NULL)           = 0
1369  nanosleep({2, 0}, NULL)           = 0
1375  nanosleep({2147483647, 999999999},

the 'float' test reliably reproduces the problem (./Run -D float).

I don't presume to know why this would happen, but I hope that it
makes sense or helps some of you.

-- 
Michael Rice <michael@riceclan.org>



-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click

RE: unstable binaries

[Ian Pratt]

 

> No. I ran the default testset (ltp-full-20050207) on xen0-2.0.4 and
> repeated the failed tests on different kernels (all 2.6.10): 
> 1.770_FC3 (fedora errata kernel), vanilla, xen0-2.0-testing (2.3.), 
> xen0-2.0-testing (4.3.). Except the FC3 kernel, they all had the same 
> config (except options not present due to ARCH). The test "fcntl23" 
> failed on all kernels, so it probably a fedora issue. The tests 
> "ioperm02", "iopl02" and "nanosleep02" failed on all xen-kernels,
> and only on them. This is the output for "nanosleep02":

The iopl02 and ioperm02 failures are expected, but not at all serious.
We actually have a plan to virtualize iopl just to tidy this up.

nanosleep2 is a surprise -- I can't recall seeing this fail before. I
wander if its an Athlon issue...

Ian


-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_ide95&alloc_id\x14396&op=click

RE: unstable binaries

[Ian Pratt]

> wget oopses sometimes, never seen that before:
> 
>     wget http://...
>     25% [======>                    ] 1,028,840     55.08K/s  
>   ETA 06:31
>     wget: retr.c:293: calc_rate: Assertion `msecs >= 0' failed.
>     Aborted

It sound like time may be screwed on your system, which is rather
surprising.
What kind of CPU do you have?

Does wget still fail if you're in a text mode? (I think we really need
to rule out any interaction with AGP or X). 

Please can you try running LTP on your system. Does it pass all the time
tests?

Thanks,
Ian


-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_ide95&alloc_id\x14396&op=click

Re: unstable binaries

[nils toedtmann]

On Thu, Mar 03, 2005 at 02:38:14PM -0000, Ian Pratt wrote:
> > wget oopses sometimes, never seen that before:
> > 
> >     wget http://...
> >     25% [======>                    ] 1,028,840     55.08K/s  
> >   ETA 06:31
> >     wget: retr.c:293: calc_rate: Assertion `msecs >= 0' failed.
> >     Aborted
> 
> It sound like time may be screwed on your system, which is rather
> surprising.
> What kind of CPU do you have?
> 
> Does wget still fail if you're in a text mode? (I think we really need
> to rule out any interaction with AGP or X). 
> 
> Please can you try running LTP on your system. Does it pass all the time
> tests?

I'll try to check that on monday, is that ok?

/nils.

-- 
there is no sig.


-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click

Re: unstable binaries

[nils toedtmann]

On Fri, Mar 04, 2005 at 02:29:38PM +0100, nils toedtmann wrote:
> On Thu, Mar 03, 2005 at 02:38:14PM -0000, Ian Pratt wrote:
> > > wget oopses sometimes, never seen that before:
> > > 
> > >     wget http://...
> > >     25% [======>                    ] 1,028,840     55.08K/s  
> > >   ETA 06:31
> > >     wget: retr.c:293: calc_rate: Assertion `msecs >= 0' failed.
> > >     Aborted
> > 
> > It sound like time may be screwed on your system, which is rather
> > surprising.

Ok, after lots of compiling, rebooting & testing, i got this picture:

The error "wget: retr.c:293: calc_rate: Assertion `msecs >= 0' failed."
is still present with xen-2.0-testing ("ChangeSet@1.1755, 2005-03-04 
00:57:22"), but now i have to trigger them by higher load. I was not
able to reproduce it on a vanilla kernel with same config (besides 
ARCH specific options).


> > What kind of CPU do you have?

  $ cat /proc/cpuinfo
  processor       : 0
  vendor_id       : AuthenticAMD
  cpu family      : 6
  model           : 4
  model name      : AMD Athlon(tm) Processor
  [...]


> > Does wget still fail if you're in a text mode? (I think we really need
> > to rule out any interaction with AGP or X). 

Yes: could reproduce it on tty1 without any X running. Some kernel config:

  # CONFIG_MODULES is not set
  # CONFIG_AGP is not set
  CONFIG_FB=y


> > Please can you try running LTP on your system. Does it pass all the time
> > tests?

No. I ran the default testset (ltp-full-20050207) on xen0-2.0.4 and
repeated the failed tests on different kernels (all 2.6.10): 
1.770_FC3 (fedora errata kernel), vanilla, xen0-2.0-testing (2.3.), 
xen0-2.0-testing (4.3.). Except the FC3 kernel, they all had the same 
config (except options not present due to ARCH). The test "fcntl23" 
failed on all kernels, so it probably a fedora issue. The tests 
"ioperm02", "iopl02" and "nanosleep02" failed on all xen-kernels,
and only on them. This is the output for "nanosleep02":

  <<<test_output>>>
  nanosleep02    1  FAIL  :  Remaining sleep time 4010000 usec doesn't match with the expected 3999236 usec time
  nanosleep02    1  FAIL  :  child process exited abnormally
  incrementing stop
  <<<execution_status>>>
  duration=1 termination_type=exited termination_id=1 corefile=no
  cutime=0 cstime=0
  <<<test_end>>>

All other default tests passed on all kernels, including 
"gettimeofday02".


> I'll try to check that on monday, is that ok?

So here i am. I hope it was worth it ...

/nils.


-- 
there is no sig.


-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click

Re: unstable binaries

[Vincent Hanquez]

On Mon, Mar 07, 2005 at 09:47:20PM +0100, nils toedtmann wrote:
> Ok, after lots of compiling, rebooting & testing, i got this picture:
> 
> The error "wget: retr.c:293: calc_rate: Assertion `msecs >= 0' failed."
> is still present with xen-2.0-testing ("ChangeSet@1.1755, 2005-03-04 
> 00:57:22"), but now i have to trigger them by higher load. I was not
> able to reproduce it on a vanilla kernel with same config (besides 
> ARCH specific options).

>   <<<test_output>>>
>   nanosleep02    1  FAIL  :  Remaining sleep time 4010000 usec doesn't match with the expected 3999236 usec time
>   nanosleep02    1  FAIL  :  child process exited abnormally
>   incrementing stop
>   <<<execution_status>>>
>   duration=1 termination_type=exited termination_id=1 corefile=no
>   cutime=0 cstime=0
>   <<<test_end>>>

the failure is just a precision range problem. increasing the nanosleep
USEC_PRECISION to 10 times more make the test suceed.

Now running with HZ = 1000 make the precision problem disappear (at the
expense of awakening the test 700ns too soon and still failing ..)

Could you try running with HZ = 1000 to see if there are still problems
with your programs ?

Also the strace output of the wget program failing might be handy to
know if there's a link with nanosleep at least.

-- 
Vincent Hanquez


-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click

Re: unstable binaries

[Nils Toedtmann]

Am Dienstag, den 15.03.2005, 20:16 +0100 schrieb Vincent Hanquez:
> On Mon, Mar 07, 2005 at 09:47:20PM +0100, nils toedtmann wrote:
> > Ok, after lots of compiling, rebooting & testing, i got this picture:
> > 
> > The error "wget: retr.c:293: calc_rate: Assertion `msecs >= 0' failed."
> > is still present with xen-2.0-testing ("ChangeSet@1.1755, 2005-03-04 
> > 00:57:22"), but now i have to trigger them by higher load. I was not
> > able to reproduce it on a vanilla kernel with same config (besides 
> > ARCH specific options).

> >   <<<test_output>>>
> >   nanosleep02    1  FAIL  :  Remaining sleep time 4010000 usec doesn't match with the expected 3999236 usec time
> >   nanosleep02    1  FAIL  :  child process exited abnormally
> >   incrementing stop
> >   <<<execution_status>>>
> >   duration=1 termination_type=exited termination_id=1 corefile=no
> >   cutime=0 cstime=0
> >   <<<test_end>>>

Both problems still present on 2.6.11.3 & xen-2.0-testing(2005/03/14) as
dom0

> the failure is just a precision range problem. increasing the nanosleep
> USEC_PRECISION to 10 times more make the test suceed.

Same with me.

> Now running with HZ = 1000 make the precision problem disappear (at the
> expense of awakening the test 700ns too soon and still failing ..)
> 
> Could you try running with HZ = 1000 to see if there are still problems
> with your programs ?

Did you mean this?

  export HZ=1000; wget -c http://www.kernel.org/pub/linux/...

It did not help. Still "Assertion `msecs >= 0' failed" after some
minutes of downloading at ~50kB/s.

> Also the strace output of the wget program failing might be handy to
> know if there's a link with nanosleep at least.

See <http://nils.toedtmann.net/stuff/virtualization/wget-
xen-2.0t.20050314.strace.bz2>

Is that ok or shall i strace with special options?

/nils.

ps: firefox still behaves odd, too.
-- 
no sig



-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click

Re: unstable binaries

[Vincent Hanquez]

On Wed, Mar 16, 2005 at 11:18:50AM +0100, Nils Toedtmann wrote:
> Did you mean this?
> 
>   export HZ=1000; wget -c http://www.kernel.org/pub/linux/...

no, I meant to modify the built kernel file
include/asm-xen/asm-i386/param.h substituting 100 with 1000

> It did not help. Still "Assertion `msecs >= 0' failed" after some
> minutes of downloading at ~50kB/s.
> 
> > Also the strace output of the wget program failing might be handy to
> > know if there's a link with nanosleep at least.
> 
> See <http://nils.toedtmann.net/stuff/virtualization/wget-
> xen-2.0t.20050314.strace.bz2>

the strace output shows that nanosleep shouldn't not be a part of this
problem ... so float/double issues ?

Maybe adding some debug into the wget program might help. I'll try to
come up with a patch soon.

-- 
Vincent Hanquez


-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click

RE: unstable binaries

[Ian Pratt]

 
> > If this fixes your problem I'll declare it 2.0.5
> 
> So xen-2.0-testing is something like RC of the next minor 
> release? Will 2.0.5 support 2.6.11?

No. I think we'll release 2.0.5 before forward porting to 2.6.11 (which
was released this morning).

Ian 


-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_ide95&alloc_id\x14396&op=click

unstable binaries

[nils toedtmann]

Hi *,

i am coming from UML, and now i evaluate Xen on my desktop:

  Xen-2.0.4
  linux-2.6.10
    "CONFIG_MODULES is not set"
    "CONFIG_AGP is not set"
    "CONFIG_FB_RADEON=y"
  FC3, [/usr]/lib/tls moved away

It works, but some desktop applications crash once in a
while within dom0:

  metacity-2.8.6
  firefox-1.0.1
  wnck-applet (from gnome-panel-2.8.1)

But others (X, xterm, gnome-panel, skype) remain stable, 
so it's probably not related to my framebuffer X 
(xorg-x11-Xvfb-6.8.1).

Before firefox crashes, it sometimes behaves odd:
the location box does not get updated when i switch tabs; 
scrollwhell, cursor keys, pageup/down keys stop navigating
a document, while the scrollbar still does.

With the fedora kernels everything is fine.

Any suggestions?

/nils.

btw: 2.6.11 is out ...

-- 
there is no sig.


-------------------------------------------------------
SF email is sponsored by - The IT Product Guide
Read honest & candid reviews on hundreds of IT Products from real users.
Discover which products truly live up to the hype. Start reading now.
http://ads.osdn.com/?ad_id=6595&alloc_id=14396&op=click