[ltt-dev] [RELEASE] Userspace RCU 0.5.0

[ltt-dev] [RELEASE] Userspace RCU 0.5.0

[Mathieu Desnoyers]

liburcu is a LGPLv2.1 userspace RCU (read-copy-update) library. This data
synchronization library provides read-side access which scales linearly with the
number of cores. It does so by allowing multiples copies of a given data
structure to live at the same time, and by monitoring the data structure
accesses to detect grace periods after which memory reclamation is possible.

Changelog:

* Version 0.5.0 changes the API presented by memory model, architecture
  abstraction and data structure primitives in headers. The prefixes are,
  respectively:
  - cmm_ : Concurrent Memory Model
  - caa_ : Concurrent Architecture Abstraction
  - cds_ : Concurrent Data Structures

This will make UST's life easier for instrumentation of large projects like
MariaDB and qemu.

Project website: http://lttng.org/urcu
Download link: http://www.lttng.org/files/urcu/


-- 
Mathieu Desnoyers
Operating System Efficiency R&D Consultant
EfficiOS Inc.
http://www.efficios.com

[ltt-dev] [RELEASE] Userspace RCU 0.5.1

[Mathieu Desnoyers]

Well well, a renaming error seems to have crept into 0.5.0, which caused
problems with atomic operations on 64-bit architectures.

I made sure no other similar problem exist. Anyone using 0.5.0 (released a few
hours ago) should upgrade to 0.5.1.

Thank you,

Mathieu

* Mathieu Desnoyers (mathieu.desnoyers at efficios.com) wrote:
> liburcu is a LGPLv2.1 userspace RCU (read-copy-update) library. This data
> synchronization library provides read-side access which scales linearly with the
> number of cores. It does so by allowing multiples copies of a given data
> structure to live at the same time, and by monitoring the data structure
> accesses to detect grace periods after which memory reclamation is possible.
> 
> Changelog:
> 
> * Version 0.5.0 changes the API presented by memory model, architecture
>   abstraction and data structure primitives in headers. The prefixes are,
>   respectively:
>   - cmm_ : Concurrent Memory Model
>   - caa_ : Concurrent Architecture Abstraction
>   - cds_ : Concurrent Data Structures
> 
> This will make UST's life easier for instrumentation of large projects like
> MariaDB and qemu.
> 
> Project website: http://lttng.org/urcu
> Download link: http://www.lttng.org/files/urcu/
> 
> 
> -- 
> Mathieu Desnoyers
> Operating System Efficiency R&D Consultant
> EfficiOS Inc.
> http://www.efficios.com

-- 
Mathieu Desnoyers
Operating System Efficiency R&D Consultant
EfficiOS Inc.
http://www.efficios.com

[ltt-dev] [RELEASE] Userspace RCU 0.5.2

[Mathieu Desnoyers]

s/0.5.1/0.5.2/g

Now works fine on non-x86 architectures. I'll have to be more cautious in the
future for such large changes.

Mathieu

* Mathieu Desnoyers (mathieu.desnoyers at efficios.com) wrote:
> Well well, a renaming error seems to have crept into 0.5.0, which caused
> problems with atomic operations on 64-bit architectures.
> 
> I made sure no other similar problem exist. Anyone using 0.5.0 (released a few
> hours ago) should upgrade to 0.5.1.
> 
> Thank you,
> 
> Mathieu
> 
> * Mathieu Desnoyers (mathieu.desnoyers at efficios.com) wrote:
> > liburcu is a LGPLv2.1 userspace RCU (read-copy-update) library. This data
> > synchronization library provides read-side access which scales linearly with the
> > number of cores. It does so by allowing multiples copies of a given data
> > structure to live at the same time, and by monitoring the data structure
> > accesses to detect grace periods after which memory reclamation is possible.
> > 
> > Changelog:
> > 
> > * Version 0.5.0 changes the API presented by memory model, architecture
> >   abstraction and data structure primitives in headers. The prefixes are,
> >   respectively:
> >   - cmm_ : Concurrent Memory Model
> >   - caa_ : Concurrent Architecture Abstraction
> >   - cds_ : Concurrent Data Structures
> > 
> > This will make UST's life easier for instrumentation of large projects like
> > MariaDB and qemu.
> > 
> > Project website: http://lttng.org/urcu
> > Download link: http://www.lttng.org/files/urcu/
> > 
> > 
> > -- 
> > Mathieu Desnoyers
> > Operating System Efficiency R&D Consultant
> > EfficiOS Inc.
> > http://www.efficios.com
> 
> -- 
> Mathieu Desnoyers
> Operating System Efficiency R&D Consultant
> EfficiOS Inc.
> http://www.efficios.com

-- 
Mathieu Desnoyers
Operating System Efficiency R&D Consultant
EfficiOS Inc.
http://www.efficios.com

[ltt-dev] [RELEASE] Userspace RCU 0.5.2

[Alexandre Montplaisir]

On 10-11-18 10:22 PM, Mathieu Desnoyers wrote:
> s/0.5.1/0.5.2/g
>
> Now works fine on non-x86 architectures. I'll have to be more cautious in the
> future for such large changes.
>
> Mathieu

Maybe we should have 1- or 2-day "beta testing" periods before pushing 
releases. ;)

That or unit testing... Yannick?

[ltt-dev] [RELEASE] Userspace RCU 0.5.2

[Brosseau, Yannick]

>>
>> Now works fine on non-x86 architectures. I'll have to be more cautious in
>> the
>> future for such large changes.
>>
>> Mathieu
>
> Maybe we should have 1- or 2-day "beta testing" periods before pushing
> releases. ;)
>
> That or unit testing... Yannick?

Better than that, an automated, multiplatform test infrastructure. But
yes, that will need unit tests.
I think it will be my next project, when I finish working my current
lttv improvement.

-- 
Yannick Brosseau
www.projetmontreal.org
Consultant en logiciel libre
yannickbrosseau.com

[ltt-dev] [RELEASE] Userspace RCU 0.5.2

[Mathieu Desnoyers]

* Brosseau, Yannick (yannick.brosseau at gmail.com) wrote:
> >>
> >> Now works fine on non-x86 architectures. I'll have to be more cautious in
> >> the
> >> future for such large changes.
> >>
> >> Mathieu
> >
> > Maybe we should have 1- or 2-day "beta testing" periods before pushing
> > releases. ;)
> >
> > That or unit testing... Yannick?
> 
> Better than that, an automated, multiplatform test infrastructure. But
> yes, that will need unit tests.
> I think it will be my next project, when I finish working my current
> lttv improvement.

Please coordinate your effort with Lo?c Minier, from Linaro, who showed great
interest in helping us to deploy an automated test setup.

Thanks,

Mathieu

-- 
Mathieu Desnoyers
Operating System Efficiency R&D Consultant
EfficiOS Inc.
http://www.efficios.com

[ltt-dev] [RELEASE] Userspace RCU 0.5.2

[Loïc Minier]

        Hey folks!

On Thu, Nov 18, 2010, Mathieu Desnoyers wrote:
> > Better than that, an automated, multiplatform test infrastructure. But
> > yes, that will need unit tests.
> > I think it will be my next project, when I finish working my current
> > lttv improvement.
> 
> Please coordinate your effort with Lo?c Minier, from Linaro, who showed great
> interest in helping us to deploy an automated test setup.

 Yes!  So Linaro is ramping up some automated testing infrastructure
 this cycle; right now, it's x86 only, so not terribly exciting, but we
 intend to run various workloads, testsuites, benchmarks etc. on a set
 of ARM boards from various SoCs as well.  I can only offer x86 testing
 right now though

   Cheers,
-- 
Lo?c Minier

[ltt-dev] [RELEASE] Userspace RCU 0.5.2

[Mathieu Desnoyers]

* Lo?c Minier (lool at dooz.org) wrote:
>         Hey folks!
> 
> On Thu, Nov 18, 2010, Mathieu Desnoyers wrote:
> > > Better than that, an automated, multiplatform test infrastructure. But
> > > yes, that will need unit tests.
> > > I think it will be my next project, when I finish working my current
> > > lttv improvement.
> > 
> > Please coordinate your effort with Lo?c Minier, from Linaro, who showed great
> > interest in helping us to deploy an automated test setup.
> 
>  Yes!  So Linaro is ramping up some automated testing infrastructure
>  this cycle; right now, it's x86 only, so not terribly exciting, but we
>  intend to run various workloads, testsuites, benchmarks etc. on a set
>  of ARM boards from various SoCs as well.  I can only offer x86 testing
>  right now though

I remember you did a presentation on a tool to help deploying this kind of
testing infrastructure. Do you have pointers to it ?

The research lab I collaborate with at Polytechnique will very probably be
interested to deploy their own test setup, and I think the tool you use might be
of great help to them.

Thanks,

Mathieu


-- 
Mathieu Desnoyers
Operating System Efficiency R&D Consultant
EfficiOS Inc.
http://www.efficios.com

[ltt-dev] [RELEASE] Userspace RCU 0.5.2

[Loïc Minier]

On Fri, Nov 19, 2010, Mathieu Desnoyers wrote:
> I remember you did a presentation on a tool to help deploying this kind of
> testing infrastructure. Do you have pointers to it ?

 in terms of tools, the current tool is hudson:
   http://hudson-ci.org/
 my own instance is at:
   http://hudson.dooz.org/

 but it might be that we need to scale this up a bit, perhaps using
   http://cloudbees.com/
 or simply changing software

> The research lab I collaborate with at Polytechnique will very
> probably be interested to deploy their own test setup, and I think the
> tool you use might be of great help to them.

 Sounds good; after evaluating both buildbot and hudson, I really really
 recommend hudson, and it's to try out and back out from (less than a
 couple of hours to get something useful working).  Happy to help in
 this project, if it helps testing LTTng

   Cheers
-- 
Lo?c Minier

[ltt-dev] [RELEASE] Userspace RCU 0.5.2

[Yannick Brosseau]

On 11/19/2010 09:07 AM, Lo?c Minier wrote:
> Sounds good; after evaluating both buildbot and hudson, I really really
>   recommend hudson, and it's to try out and back out from (less than a
>   couple of hours to get something useful working).  Happy to help in
>   this project, if it helps testing LTTng
>
>     Cheers
>    

It looks better than CruiseControl which I've used in the past. I will 
definitely look into it.
I just sad that every good Continuous Integration tools seems to be 
written in java! ;-)

Yannick

[ltt-dev] [RELEASE] Userspace RCU 0.5.2

[Jon Bernard]

* Yannick Brosseau <yannick.brosseau at gmail.com> wrote:
> On 11/19/2010 09:07 AM, Lo?c Minier wrote:
> >Sounds good; after evaluating both buildbot and hudson, I really really
> >  recommend hudson, and it's to try out and back out from (less than a
> >  couple of hours to get something useful working).  Happy to help in
> >  this project, if it helps testing LTTng
> >
> >    Cheers
> 
> It looks better than CruiseControl which I've used in the past. I
> will definitely look into it.
> I just sad that every good Continuous Integration tools seems to be
> written in java! ;-)

You may also want to look at "CI Joe" [1,2], I've heard good things about it
from a few friends. May be worth a look.

[1]: https://github.com/defunkt/cijoe
[2]: https://github.com/blog/471-continuous-integration-spring-cleaning

-- 
Jon