[Activity] Power Management WG Weekly Status report for week ending 2012-01-27

2012-01-27 Thread Mounir Bsaibes 
Enclosed  please find the link to the  Weekly Status report & meeting
minutes
for the Power Management working group for the week ending 2012-01-27


== Weekly Status Report ==
https://wiki.linaro.org/WorkingGroups/PowerManagement/Status/2012-01-26

== Meeting Minutes ==
https://wiki.linaro.org/WorkingGroups/PowerManagement/Meetings/2012-01-25


== Summary ==
(For details, see the Weekly Status Report and Meeting Minutes )

 * Linaro Connect sessions  planning & setup
* http://summit.ubuntu.com/lcq1-12/track/linaro-power/

 * Common Clock
   * Working on changing the code based on feedback from v2-v4 submission,
quite a bit of code change for v5.
   * About 50% done on converting OMAP4 to use the new code.

 * Thermal Management
   * Working on fine tuning of cpufreq/cpuhotplug based cooling devices to
be now  instance based (needed for multi core systems)
   * Submitted V4 of imx thermal driver

 * hrtimer
* Continue working on  implementing core high-res timers feature to
return time spent in usleep_range()

 * cpuidle
* Submitted at91 standby functions
* Implemented cpuidle WFI state for ux500 based on common framework
* Working on dynamic init function for the common cpuidle code will add
to the next version

 * Sched_mc
   * Preparing for the scheduler mini-summit at Connect
   * Taking measurement on the new sched_mc patch before submitting it,
will try to summit the
  patch before connect and cc the attendees of the mini-summit

* Issue
   * Team is not getting fast enough feedback on their patches.


Best regards,
Mounir

-- 
Mounir Bsaibes
Project Manager

Follow Linaro.org:
facebook.com/pages/Linaro/155974581091106
http://twitter.com/#!/linaroorg
http://www.linaro.org/linaro-blog 
___
linaro-dev mailing list
linaro-dev@lists.linaro.org
http://lists.linaro.org/mailman/listinfo/linaro-dev

Re: [PATCH] module: avoid call vmalloc if init size is zero

2012-01-27 Thread Rusty Russell 
On Fri, 27 Jan 2012 11:19:16 +0400, Dmitry Antipov  
wrote:
> For the architectures with it's own module_alloc(), if module init
> size is zero, avoiding module_alloc_update_bounds() and memset()
> no-op calls also eliminates warn_alloc_failed() zero-size warning
> in __vmalloc_node_range().
> 
> Signed-off-by: Dmitry Antipov 

Please fix __vmalloc_node_range() instead to return ZERO_SIZE_PTR.
And of course, vfree() to use ZERO_OR_NULL_PTR().

Thanks,
Rusty.

___
linaro-dev mailing list
linaro-dev@lists.linaro.org
http://lists.linaro.org/mailman/listinfo/linaro-dev

[Activity] Kernel WG Weekly Status report for week ending 2012-01-27

2012-01-27 Thread Mounir Bsaibes 
Enclosed  please find the link to the Weekly Status & Individual Activity
reports
for the kernel  working group for the week ending 2012-01-27

== Weekly Status Report ==
https://wiki.linaro.org/WorkingGroups/Kernel/Status/2012-01-26
 
== Individual Activity Report ==
 https://wiki.linaro.org/WorkingGroups/Kernel/ActivityReports/2012-01-20
 
== Summary ==
(For more details, please see the Status and the Individual Activity
reports above)

* Released linux-linaro-3.2-2021.01-0
*
http://launchpad.net/linux-linaro/3.2/3.2-2012.01/+download/linux-linaro-3.2-2012.01-0.tar.bz2

* Connect sessions planning
 * http://summit.ubuntu.com/lcq1-12/track/linaro-kernel

 * Pinctrl
* Created a new pin control patch (v2), will make a third (v3) for
per-device named states.
* Tony Lindgren, OMAP maintainer, attended the pinctrl meeting
* See the various  topics / issues discussed at:
https://wiki.linaro.org/WorkingGroups/Kernel/PinctrlMeetings/2012-01-24

* Android upstreaming
  * evdev timestamp patch was merged in maintainer's tree and queued for 3.4
  * Worked range-tree implementation for the fadvise volatile code. Got
insert/delete/lookup working & also added splay-tree style self-balancing.
  * Started looking into interactive cpufreq governor, forward-ported it
onto latest kernels.

* kconfig
  * merge_config.sh script that handles config fragments was merged into 3.3

* Storage
   * Prepared a patch for Context ID for Samsung eMMC.

* big.LITTLE switcher
* Identifying the needed code to support the b.L switcher  from the
Linux kernel environment.

* Single  Kernel
 * Rebased patch series removing arch_idle() and most instances of
mach/system.h from the source tree onto v3.3-rc1.

Best regards,

Mounir Bsaibes
Project Manager

Follow Linaro.org:
facebook.com/pages/Linaro/155974581091106
http://twitter.com/#!/linaroorg
http://www.linaro.org/linaro-blog 
___
linaro-dev mailing list
linaro-dev@lists.linaro.org
http://lists.linaro.org/mailman/listinfo/linaro-dev

Re: [Linaro-validation] Linaro kernel tree for A15 for daily builds

2012-01-27 Thread Zygmunt Krynicki 
On Fri, Jan 27, 2012 at 12:47 PM, Alexander Sack  wrote:
> On Wed, Jan 25, 2012 at 9:33 PM, Zygmunt Krynicki
>  wrote:
>> HI.
>>
>> I'm building a LAVA service for running fast models. Quite soon (*)
>> we'll be ready to open an alpha access. Right now you will need to
>> bring your own root filesystem and kernel image to use it. With that
>> in mind I wanted to start a discussion about the state of A15 support
>> in Linaro kernel(s). I need to understand two things:
>>
>> 1) Are we ready to do automatic builds for A15 kernels?
>
> We can do builds on ci.linaro.org - as soon as we have a branch/tag of
> a public git tree and a defconfig we would be ready to go. Might
> require a few minor tweaks if the LAVA job for A15 needs different
> input parameters, so if that's the case let Deepti and Danilo know
> about the details.

It definitely will. There will be a different job specification for
simulation targets. I'll send an update when this part becomes ready

ZK

___
linaro-dev mailing list
linaro-dev@lists.linaro.org
http://lists.linaro.org/mailman/listinfo/linaro-dev

Re: [PATCH] hrtimers: move default_timer_slack_ns to system-wide sysctl

2012-01-27 Thread Amit Kucheria 
Dmitry,

On Fri, Jan 27, 2012 at 12:30 PM, Dmitry Antipov
 wrote:
>  - moving default_timer_slack_ns from per-task field to global variable
>   controlled via sysctl() in [0..100] range (usecs) provides some
>   more system-wide control beyond per-task prctl();

What benefits do I get from this change? In other words, don't just
describe the change, please describe why it is desirable.

Have you done any tests where this improves things - reduction in
wakeups, increase in throughput? Those results will also help in
getting more attention on the patch.

>  - task_time_slack() should be used to get default timer slack if no
>   special requirements exists.
>
> Signed-off-by: Dmitry Antipov 
> ---
>  include/linux/hrtimer.h |    1 +
>  include/linux/sched.h   |   11 ---
>  kernel/fork.c           |    3 ---
>  kernel/futex.c          |    4 ++--
>  kernel/hrtimer.c        |   10 +++---
>  kernel/sys.c            |    3 ++-
>  kernel/sysctl.c         |   10 ++
>  7 files changed, 30 insertions(+), 12 deletions(-)
>
> diff --git a/include/linux/hrtimer.h b/include/linux/hrtimer.h
> index fd0dc30..855d45e 100644
> --- a/include/linux/hrtimer.h
> +++ b/include/linux/hrtimer.h
> @@ -323,6 +323,7 @@ extern ktime_t ktime_get_monotonic_offset(void);
>
>  DECLARE_PER_CPU(struct tick_device, tick_cpu_device);
>
> +extern int default_timer_slack_ns;
>
>  /* Exported timer functions: */
>
> diff --git a/include/linux/sched.h b/include/linux/sched.h
> index 2234985..4f475aa 100644
> --- a/include/linux/sched.h
> +++ b/include/linux/sched.h
> @@ -1551,11 +1551,11 @@ struct task_struct {
>        struct latency_record latency_record[LT_SAVECOUNT];
>  #endif
>        /*
> -        * time slack values; these are used to round up poll() and
> -        * select() etc timeout values. These are in nanoseconds.
> +        * High-resolution timer slack value, in nanoseconds.
> +        * Used to round up poll()/select(), nanosleep, futex
> +        * waiting, etc. timeout values for non-realtime tasks.
>         */
>        unsigned long timer_slack_ns;
> -       unsigned long default_timer_slack_ns;
>
>        struct list_head        *scm_work_list;
>  #ifdef CONFIG_FUNCTION_GRAPH_TRACER
> @@ -2622,6 +2622,11 @@ static inline int spin_needbreak(spinlock_t *lock)
>  #endif
>  }
>
> +static inline unsigned long task_timer_slack(struct task_struct *tsk)
> +{
> +       return rt_task(tsk) ? 0 : tsk->timer_slack_ns;
> +}
> +
>  /*
>  * Thread group CPU time accounting.
>  */
> diff --git a/kernel/fork.c b/kernel/fork.c
> index 051f090..b0e8f63 100644
> --- a/kernel/fork.c
> +++ b/kernel/fork.c
> @@ -1144,9 +1144,6 @@ static struct task_struct *copy_process(unsigned long 
> clone_flags,
>  #if defined(SPLIT_RSS_COUNTING)
>        memset(&p->rss_stat, 0, sizeof(p->rss_stat));
>  #endif
> -
> -       p->default_timer_slack_ns = current->timer_slack_ns;
> -
>        task_io_accounting_init(&p->ioac);
>        acct_clear_integrals(p);
>
> diff --git a/kernel/futex.c b/kernel/futex.c
> index 1614be2..a0d302d 100644
> --- a/kernel/futex.c
> +++ b/kernel/futex.c
> @@ -1887,7 +1887,7 @@ static int futex_wait(u32 __user *uaddr, unsigned int 
> flags, u32 val,
>                                      HRTIMER_MODE_ABS);
>                hrtimer_init_sleeper(to, current);
>                hrtimer_set_expires_range_ns(&to->timer, *abs_time,
> -                                            current->timer_slack_ns);
> +                                            task_timer_slack(current));
>        }
>
>  retry:
> @@ -2281,7 +2281,7 @@ static int futex_wait_requeue_pi(u32 __user *uaddr, 
> unsigned int flags,
>                                      HRTIMER_MODE_ABS);
>                hrtimer_init_sleeper(to, current);
>                hrtimer_set_expires_range_ns(&to->timer, *abs_time,
> -                                            current->timer_slack_ns);
> +                                            task_timer_slack(current));
>        }
>
>        /*
> diff --git a/kernel/hrtimer.c b/kernel/hrtimer.c
> index ae34bf5..0704559 100644
> --- a/kernel/hrtimer.c
> +++ b/kernel/hrtimer.c
> @@ -51,6 +51,12 @@
>  #include 
>
>  /*
> + * Default hrtimer slack value, in nanoseconds. May be
> + * changed via sysctl within [0..100] range.
> + */
> +int default_timer_slack_ns = 5;
> +
> +/*
>  * The timer bases:
>  *
>  * There are more clockids then hrtimer bases. Thus, we index
> @@ -1564,9 +1570,7 @@ long hrtimer_nanosleep(struct timespec *rqtp, struct 
> timespec __user *rmtp,
>        int ret = 0;
>        unsigned long slack;
>
> -       slack = current->timer_slack_ns;
> -       if (rt_task(current))
> -               slack = 0;
> +       slack = task_timer_slack(current);
>
>        hrtimer_init_on_stack(&t.timer, clockid, mode);
>        hrtimer_set_expires_range_ns(&t.timer, timespec_to_ktime(*rqtp), 
> slack);
> diff --git a/kernel/sys.c b/kernel/sys.c
> index 4070153..db36543 100644
> --- a/kernel/sys.c
> +++ b/kernel/sys

Re: [Linaro-validation] Linaro kernel tree for A15 for daily builds

2012-01-27 Thread Alexander Sack 
On Wed, Jan 25, 2012 at 9:33 PM, Zygmunt Krynicki
 wrote:
> HI.
>
> I'm building a LAVA service for running fast models. Quite soon (*)
> we'll be ready to open an alpha access. Right now you will need to
> bring your own root filesystem and kernel image to use it. With that
> in mind I wanted to start a discussion about the state of A15 support
> in Linaro kernel(s). I need to understand two things:
>
> 1) Are we ready to do automatic builds for A15 kernels?

We can do builds on ci.linaro.org - as soon as we have a branch/tag of
a public git tree and a defconfig we would be ready to go. Might
require a few minor tweaks if the LAVA job for A15 needs different
input parameters, so if that's the case let Deepti and Danilo know
about the details.


-- 
Alexander Sack
Technical Director, Linaro Platform Teams
http://www.linaro.org | Open source software for ARM SoCs
http://twitter.com/#!/linaroorg - http://www.linaro.org/linaro-blog

___
linaro-dev mailing list
linaro-dev@lists.linaro.org
http://lists.linaro.org/mailman/listinfo/linaro-dev

Re: Sources for 11.11 kernel release

2012-01-27 Thread Alexander Sack 
On Fri, Jan 27, 2012 at 11:40 AM, Loïc Minier  wrote:
> On Fri, Jan 27, 2012, Alexander Sack wrote:
>>  * hide people/ trees and show them on people.git.linaro.org instead
>> of main git.linaro.org. At best git:// urls would be valid still
>> aftterwards
>>  * hide the android/ trees and don't show them anywhere else; keep
>> git:// urls and/or http:// clone urls wprking too
>>
>> This would definitely air to breath and we can continue to improve
>> things from there.
>
>  The arago-project layout seems nicer, effectively just hiding subtrees

Yeah. That makes sense. Happy if we get that.

>
>>                             As one example, ubuntu packages having a
>> vcs-bzr field and a reasonable tagging/versioning scheme could
>> probably work, but we have to look case by case.
>
>  Problem is that a Vcs-* field is not consistently applied   :-/

Right. But I don't think it matters much, no?. Release Team will work
with folks to sort out the details. And if they decide that release
team will track Vcs- fields for mining the source info for ubuntu
packages, we probably would ensure that it's consistently applied for
the packages we care about :).

>  Sometimes it's not set, or doesn't point at the right repo, or points
>  at a repo with a special tree where people have to do magic to get
>  things done.

I see three main problems to solve:

 1. How do I find the vcs component for this source component?
 2. How can I hack on this tree and how does this tree work?
 3. What trees/tags/components are in the LEB?

Point 1. can be solved through the release team working with component
owners and sourcing the location of the tree used to produce a
component/package

Point 2. can be solved by the tree owner through maintaining good
documentation on the git repo main gitweb page that explains in simple
instructions how things work. For example:

 a. how the trees are produced
 b. what tag/branch syntax rules apply
 c. how to hack and contribute to this tree
 d. how to participate in meetings where this baseline is on topic to
discuss

Point 3. could be addressed by providing an explicit list of
integrated/improved components alongside the LEB (on top of the usual
android repo info etc.) to make this easier.

-- 
Alexander Sack
Technical Director, Linaro Platform Teams
http://www.linaro.org | Open source software for ARM SoCs
http://twitter.com/#!/linaroorg - http://www.linaro.org/linaro-blog

___
linaro-dev mailing list
linaro-dev@lists.linaro.org
http://lists.linaro.org/mailman/listinfo/linaro-dev

Re: [Linaro-validation] Linaro kernel tree for A15 for daily builds

2012-01-27 Thread Dave Pigott 
On 27 Jan 2012, at 10:33, Pawel Moll wrote:

> On Thu, 2012-01-26 at 19:20 +0100, Zygmunt Krynicki wrote:
>> It also requires a license from arm for each machine
>> that runs it (it binds to mac address) 
> 
> This could be mitigated by setting up a Flex licence server for machines
> in the farm, with floating licences instead of noded-locked ones. After
> all that's what we have internally at ARM - we wouldn't cope with
> generating licenses for every single machine here ;-)

I have some floating licences, and am intending to put flex on the server farm 
once I've re-installed and got OpenStack running.

Dave
___
linaro-dev mailing list
linaro-dev@lists.linaro.org
http://lists.linaro.org/mailman/listinfo/linaro-dev

Re: Sources for 11.11 kernel release

2012-01-27 Thread Loïc Minier 
On Fri, Jan 27, 2012, Alexander Sack wrote:
>  * hide people/ trees and show them on people.git.linaro.org instead
> of main git.linaro.org. At best git:// urls would be valid still
> aftterwards
>  * hide the android/ trees and don't show them anywhere else; keep
> git:// urls and/or http:// clone urls wprking too
> 
> This would definitely air to breath and we can continue to improve
> things from there.

 The arago-project layout seems nicer, effectively just hiding subtrees
 rather than using separate DNSes; also, they are linked from the main
 repos to the other ones, whereas if we create a people.git.linaro.org,
 how will people discover it?  It's best to avoid having 3
 places/hostnames for people to host stuff.

> As one example, ubuntu packages having a
> vcs-bzr field and a reasonable tagging/versioning scheme could
> probably work, but we have to look case by case.

 Problem is that a Vcs-* field is not consistently applied   :-/
 Sometimes it's not set, or doesn't point at the right repo, or points
 at a repo with a special tree where people have to do magic to get
 things done.

-- 
Loïc Minier

___
linaro-dev mailing list
linaro-dev@lists.linaro.org
http://lists.linaro.org/mailman/listinfo/linaro-dev

Re: [Linaro-validation] Linaro kernel tree for A15 for daily builds

2012-01-27 Thread Peter Maydell 
On 27 January 2012 10:30, Pawel Moll  wrote:
> I'm planning to modify the boot.S to be able to create an AXF file from
> separate kernel and DTB (will happen next week, if anyone wants to try
> it let me know and I'll notify you when it's ready)

I would suggest you start out with the patches I did to
let the model load the kernel and initrd via semihosting:

https://lists.cs.columbia.edu/pipermail/android-virt/2012-January/000570.html

extending those to also support --dtb would be much more useful
and flexible than the make-an-AXF-file approach.

-- PMM

___
linaro-dev mailing list
linaro-dev@lists.linaro.org
http://lists.linaro.org/mailman/listinfo/linaro-dev

Re: Linaro kernel tree for A15 for daily builds

2012-01-27 Thread Pawel Moll 
On Thu, 2012-01-26 at 19:20 +0100, Zygmunt Krynicki wrote:
> It also requires a license from arm for each machine
> that runs it (it binds to mac address) 

This could be mitigated by setting up a Flex licence server for machines
in the farm, with floating licences instead of noded-locked ones. After
all that's what we have internally at ARM - we wouldn't cope with
generating licenses for every single machine here ;-)

I'm far from being expert in the subject, but let me know if you wanted
to implement it and I'll get the right people look into it.

Cheers!

Paweł



___
linaro-dev mailing list
linaro-dev@lists.linaro.org
http://lists.linaro.org/mailman/listinfo/linaro-dev

[PATCH] hrtimers: move default_timer_slack_ns to system-wide sysctl

2012-01-27 Thread Dmitry Antipov 
 - moving default_timer_slack_ns from per-task field to global variable
   controlled via sysctl() in [0..100] range (usecs) provides some
   more system-wide control beyond per-task prctl();

 - task_time_slack() should be used to get default timer slack if no
   special requirements exists.

Signed-off-by: Dmitry Antipov 
---
 include/linux/hrtimer.h |1 +
 include/linux/sched.h   |   11 ---
 kernel/fork.c   |3 ---
 kernel/futex.c  |4 ++--
 kernel/hrtimer.c|   10 +++---
 kernel/sys.c|3 ++-
 kernel/sysctl.c |   10 ++
 7 files changed, 30 insertions(+), 12 deletions(-)

diff --git a/include/linux/hrtimer.h b/include/linux/hrtimer.h
index fd0dc30..855d45e 100644
--- a/include/linux/hrtimer.h
+++ b/include/linux/hrtimer.h
@@ -323,6 +323,7 @@ extern ktime_t ktime_get_monotonic_offset(void);
 
 DECLARE_PER_CPU(struct tick_device, tick_cpu_device);
 
+extern int default_timer_slack_ns;
 
 /* Exported timer functions: */
 
diff --git a/include/linux/sched.h b/include/linux/sched.h
index 2234985..4f475aa 100644
--- a/include/linux/sched.h
+++ b/include/linux/sched.h
@@ -1551,11 +1551,11 @@ struct task_struct {
struct latency_record latency_record[LT_SAVECOUNT];
 #endif
/*
-* time slack values; these are used to round up poll() and
-* select() etc timeout values. These are in nanoseconds.
+* High-resolution timer slack value, in nanoseconds.
+* Used to round up poll()/select(), nanosleep, futex
+* waiting, etc. timeout values for non-realtime tasks.
 */
unsigned long timer_slack_ns;
-   unsigned long default_timer_slack_ns;
 
struct list_head*scm_work_list;
 #ifdef CONFIG_FUNCTION_GRAPH_TRACER
@@ -2622,6 +2622,11 @@ static inline int spin_needbreak(spinlock_t *lock)
 #endif
 }
 
+static inline unsigned long task_timer_slack(struct task_struct *tsk)
+{
+   return rt_task(tsk) ? 0 : tsk->timer_slack_ns;
+}
+
 /*
  * Thread group CPU time accounting.
  */
diff --git a/kernel/fork.c b/kernel/fork.c
index 051f090..b0e8f63 100644
--- a/kernel/fork.c
+++ b/kernel/fork.c
@@ -1144,9 +1144,6 @@ static struct task_struct *copy_process(unsigned long 
clone_flags,
 #if defined(SPLIT_RSS_COUNTING)
memset(&p->rss_stat, 0, sizeof(p->rss_stat));
 #endif
-
-   p->default_timer_slack_ns = current->timer_slack_ns;
-
task_io_accounting_init(&p->ioac);
acct_clear_integrals(p);
 
diff --git a/kernel/futex.c b/kernel/futex.c
index 1614be2..a0d302d 100644
--- a/kernel/futex.c
+++ b/kernel/futex.c
@@ -1887,7 +1887,7 @@ static int futex_wait(u32 __user *uaddr, unsigned int 
flags, u32 val,
  HRTIMER_MODE_ABS);
hrtimer_init_sleeper(to, current);
hrtimer_set_expires_range_ns(&to->timer, *abs_time,
-current->timer_slack_ns);
+task_timer_slack(current));
}
 
 retry:
@@ -2281,7 +2281,7 @@ static int futex_wait_requeue_pi(u32 __user *uaddr, 
unsigned int flags,
  HRTIMER_MODE_ABS);
hrtimer_init_sleeper(to, current);
hrtimer_set_expires_range_ns(&to->timer, *abs_time,
-current->timer_slack_ns);
+task_timer_slack(current));
}
 
/*
diff --git a/kernel/hrtimer.c b/kernel/hrtimer.c
index ae34bf5..0704559 100644
--- a/kernel/hrtimer.c
+++ b/kernel/hrtimer.c
@@ -51,6 +51,12 @@
 #include 
 
 /*
+ * Default hrtimer slack value, in nanoseconds. May be
+ * changed via sysctl within [0..100] range.
+ */
+int default_timer_slack_ns = 5;
+
+/*
  * The timer bases:
  *
  * There are more clockids then hrtimer bases. Thus, we index
@@ -1564,9 +1570,7 @@ long hrtimer_nanosleep(struct timespec *rqtp, struct 
timespec __user *rmtp,
int ret = 0;
unsigned long slack;
 
-   slack = current->timer_slack_ns;
-   if (rt_task(current))
-   slack = 0;
+   slack = task_timer_slack(current);
 
hrtimer_init_on_stack(&t.timer, clockid, mode);
hrtimer_set_expires_range_ns(&t.timer, timespec_to_ktime(*rqtp), slack);
diff --git a/kernel/sys.c b/kernel/sys.c
index 4070153..db36543 100644
--- a/kernel/sys.c
+++ b/kernel/sys.c
@@ -22,6 +22,7 @@
 #include 
 #include 
 #include 
+#include 
 #include 
 #include 
 #include 
@@ -1919,7 +1920,7 @@ SYSCALL_DEFINE5(prctl, int, option, unsigned long, arg2, 
unsigned long, arg3,
case PR_SET_TIMERSLACK:
if (arg2 <= 0)
current->timer_slack_ns =
-   current->default_timer_slack_ns;
+   default_timer_slack_ns;
else
current->timer_slack_ns = arg2;

Re: [Linaro-validation] Linaro kernel tree for A15 for daily builds

2012-01-27 Thread Pawel Moll 
On Thu, 2012-01-26 at 21:19 +0100, Zygmunt Krynicki wrote:
> >>> With Pawel Moll's VExpress device tree support patches on top of
> >>> mainline, we can produce a single kernel config and a set of device
> >>> trees which allow booting on all the A15-based model variants.
> 
> About device trees, does the simulator need an explicitly provided
> device tree (in a way we currently provide the kernel image and
> ramdisk) or is the dt table built into the image?

Currently the simplest way of booting a DT kernel on a model is to use
the CONFIG_ARM_APPENDED_DTB=y and CONFIG_ARM_ATAG_DTB_COMPAT=y and
simply feed the model with concatenated zImage+DTB. Tested last night.
One hint: do _not_ use earlyprintk on models - current version of
DEBUG_LL in my DT4VE patches can fail in such case.

I'm planning to modify the boot.S to be able to create an AXF file from
separate kernel and DTB (will happen next week, if anyone wants to try
it let me know and I'll notify you when it's ready), but the long term
(hopefully next FM release, mid-this-year) plan is to teach the model
environment about Linux and DT so no wrapper would be necessary at all.

Cheers!

Paweł



___
linaro-dev mailing list
linaro-dev@lists.linaro.org
http://lists.linaro.org/mailman/listinfo/linaro-dev

Re: Sources for 11.11 kernel release

2012-01-27 Thread Fathi Boudra 
2012/1/27 Alexander Sack :
> On Wed, Jan 25, 2012 at 7:14 PM, Christian Robottom Reis
>  wrote:
>> On Tue, Jan 24, 2012 at 10:50:51AM -0500, Chris Lalancette wrote:
>
>>> 1)  Attempt to reduce the number of trees on git.linaro.org.  I
>>> understand that there is probably a lot going on, but the sheer
>>> number of trees makes it confusing.  It might be a good idea to
>>> remove some of the very stale or no longer active trees.
>>
>> What do Deepak and Loďc think of this in general?
>
> I would propose to:
>  * hide people/ trees and show them on people.git.linaro.org instead
> of main git.linaro.org. At best git:// urls would be valid still
> aftterwards
>  * hide the android/ trees and don't show them anywhere else; keep
> git:// urls and/or http:// clone urls wprking too
>
> This would definitely air to breath and we can continue to improve
> things from there.
>
>>
>>> 2)  Document on the wiki where the releases are built from, so there
>>> is a running record per release
>>
>> Where would we record this, Deepak, Alexander, Loďc?
>
> I am thinking ...
>
> If it's about documenting released bits it's something that should
> come out of our release process. The release team can bring that
> together. For that the information needs to be shipped along- or
> inside the source artifact. As one example, ubuntu packages having a
> vcs-bzr field and a reasonable tagging/versioning scheme could
> probably work, but we have to look case by case.
>
> Where are we putting it?
> One place that comes to my mind is obviously the release wiki page:
> https://wiki.linaro.org/Cycles/1201/Release/?

Another one is https://wiki.linaro.org/Cycles/ProjectsSummary
I'd like to generate this table automatically to keep it up-to-date,
based on inputs from packages meta data (using additional user-defined
fields to debian/control file).

___
linaro-dev mailing list
linaro-dev@lists.linaro.org
http://lists.linaro.org/mailman/listinfo/linaro-dev

Re: Sources for 11.11 kernel release

2012-01-27 Thread Alexander Sack 
On Wed, Jan 25, 2012 at 7:14 PM, Christian Robottom Reis
 wrote:
> On Tue, Jan 24, 2012 at 10:50:51AM -0500, Chris Lalancette wrote:

>> 1)  Attempt to reduce the number of trees on git.linaro.org.  I
>> understand that there is probably a lot going on, but the sheer
>> number of trees makes it confusing.  It might be a good idea to
>> remove some of the very stale or no longer active trees.
>
> What do Deepak and Loďc think of this in general?

I would propose to:
 * hide people/ trees and show them on people.git.linaro.org instead
of main git.linaro.org. At best git:// urls would be valid still
aftterwards
 * hide the android/ trees and don't show them anywhere else; keep
git:// urls and/or http:// clone urls wprking too

This would definitely air to breath and we can continue to improve
things from there.

>
>> 2)  Document on the wiki where the releases are built from, so there
>> is a running record per release
>
> Where would we record this, Deepak, Alexander, Loďc?

I am thinking ...

If it's about documenting released bits it's something that should
come out of our release process. The release team can bring that
together. For that the information needs to be shipped along- or
inside the source artifact. As one example, ubuntu packages having a
vcs-bzr field and a reasonable tagging/versioning scheme could
probably work, but we have to look case by case.

Where are we putting it?
One place that comes to my mind is obviously the release wiki page:
https://wiki.linaro.org/Cycles/1201/Release/?

-- 
Alexander Sack
Technical Director, Linaro Platform Teams
http://www.linaro.org | Open source software for ARM SoCs
http://twitter.com/#!/linaroorg - http://www.linaro.org/linaro-blog

___
linaro-dev mailing list
linaro-dev@lists.linaro.org
http://lists.linaro.org/mailman/listinfo/linaro-dev