I don't suppose there are any guidelines from ARM about compatibility
between 32bit userspace and 64bit kernel on some hypothetical future
version of the ARM architecture? Some hints about what to do or not
do could perhaps save some ioctl compat glue later on down the line..
BR,
-R
---------- Forwarded message ----------
From: Rob Clark <rob.clark(a)linaro.org>
Date: Sun, Sep 18, 2011 at 3:15 PM
Subject: Re: [PATCH] RFCv2: omapdrm DRM/KMS driver for TI OMAP platforms
To: Thomas Hellstrom <thomas(a)shipmail.org>
Cc: Inki Dae <inki.dae(a)samsung.com>, dri-devel(a)lists.freedesktop.org,
patches(a)linaro.org
On Sun, Sep 18, 2011 at 3:00 PM, Thomas Hellstrom <thomas(a)shipmail.org> wrote:
> On 09/18/2011 09:50 PM, Rob Clark wrote:
>>
>> On Sun, Sep 18, 2011 at 2:36 PM, Thomas Hellstrom<thomas(a)shipmail.org>
>> wrote:
>>
>>>
>>> On 09/17/2011 11:32 PM, Rob Clark wrote:
>>>
>>>>
>>>> From: Rob Clark<rob(a)ti.com>
>>>>
[snip]
>>>>
>>>> diff --git a/include/drm/omap_drm.h b/include/drm/omap_drm.h
>>>> new file mode 100644
>>>> index 0000000..ea0ae8e
>>>> --- /dev/null
>>>> +++ b/include/drm/omap_drm.h
>>>> @@ -0,0 +1,111 @@
>>>> +/*
>>>> + * linux/include/drm/omap_drm.h
>>>> + *
>>>> + * Copyright (C) 2011 Texas Instruments
>>>> + * Author: Rob Clark<rob(a)ti.com>
>>>> + *
>>>> + * This program is free software; you can redistribute it and/or modify
>>>> it
>>>> + * under the terms of the GNU General Public License version 2 as
>>>> published by
>>>> + * the Free Software Foundation.
>>>> + *
>>>> + * This program is distributed in the hope that it will be useful, but
>>>> WITHOUT
>>>> + * ANY WARRANTY; without even the implied warranty of MERCHANTABILITY
>>>> or
>>>> + * FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public
>>>> License
>>>> for
>>>> + * more details.
>>>> + *
>>>> + * You should have received a copy of the GNU General Public License
>>>> along with
>>>> + * this program. If not, see<http://www.gnu.org/licenses/>.
>>>> + */
>>>> +
>>>> +#ifndef __OMAP_DRM_H__
>>>> +#define __OMAP_DRM_H__
>>>> +
>>>> +#include "drm.h"
>>>> +
>>>> +/* Please note that modifications to all structs defined here are
>>>> + * subject to backwards-compatibility constraints.
>>>> + */
>>>> +
>>>> +#define OMAP_PARAM_CHIPSET_ID 1 /* ie. 0x3430, 0x4430, etc */
>>>> +
>>>> +struct drm_omap_param {
>>>> + uint64_t param; /* in */
>>>> + uint64_t value; /* in (set_param), out
>>>> (get_param)
>>>> */
>>>> +};
>>>> +
>>>> +struct drm_omap_get_base {
>>>> + char plugin_name[64]; /* in */
>>>> + uint32_t ioctl_base; /* out */
>>>> +};
>>>>
>>>>
>>>
>>> What about future ARM 64-bit vs 32-bit structure sizes? On x86 we always
>>> take care to make structures appearing in the drm user-space interfaces
>>> having sizes that are a multiple of 64-bits, to avoid having to maintain
>>> compat code for 32-bit apps running on 64 bit kernels. For the same
>>> reasons, structure members with 64 bit alignment requirements on 64-bit
>>> systems need to be carefully places.
>>>
>>> I don't know whether there is or will be a 64-bit ARM, but it might be
>>> worth
>>> taking into consideration.
>>>
>>
>> There isn't currently any 64-bit ARM, but it is a safe assumption that
>> there will be some day.. I guess we'll have enough fun w/ various
>> random 32b devices when LPAE arrives w/ the cortex-a15..
>>
>> I did try to make sure any uint64_t's were aligned to a 64bit offset,
>> but beyond that I confess to not being an expert on how 64 vs 32b
>> ioctl compat stuff is handled or what the issues going from 32->64b
>> are. If there are some additional considerations that should be taken
>> care of, then now is the time. So far I don't have any pointer fields
>> in any of the ioctl structs. Beyond that, I'm not entirely sure what
>> else needs to be done, but would appreciate any pointers to docs about
>> how the compat stuff works.
>>
>> BR,
>> -R
>>
>
> I've actually avoided writing compat ioctl code myself, by trying to make
> sure that structures look identical in the 64-bit and 32-bit x86 ABIs, but
> the compat code is there to translate pointers and to overcome alignment
> incompatibilities.
>
> A good way to at least try to avoid having to maintain compat code once the
> 64-bit ABI shows up is to make sure that 64-bit scalars and embedded
> structures are placed on 64-bit boundaries, padding if necessary, and to
> make sure (using padding) that struct sizes are always multiples of 64 bits.
So far this is true for 64bit scalars.. I'm using stdint types
everywhere so there is no chance for fields having different sizes on
64b vs 32b. And the only structs contained within ioctl structs so
far are starting at offset==0.
Is it necessary to ensure that the ioctl structs themselves (as
opposed to structs within those structs) have sizes that are multiple
of 64b? The ioctl structs are copied
(copy_from_user()/copy_to_user()), which I would have assumed would be
sufficient?
> But since there is no 64-bit ARM yet, it might be better to rely on using
> compat code in the future than on making guesses about alignment
> restrictions of the ABI...
hmm, it might be nice to get some guidelines from ARM on this, since I
really have no idea what a 64b ARM architecture would look like..
BR,
-R
> /Thomas
>
>
>
>
> _______________________________________________
> dri-devel mailing list
> dri-devel(a)lists.freedesktop.org
> http://lists.freedesktop.org/mailman/listinfo/dri-devel
>
I'm writing up some notes on building Android from scratch and
replacing the kernel in an Android build from one built locally, and I
realized that's it a bit of a chore to extract the kernel config that
got used. I thought, it may make sense to provide an way in
android-build to control what gets used - which would make finding
this information easier since if would one of the configs that gets
passed to the build like TARGET_PRODUCT. Thoughts?
--
Zach Pfeffer
Android Platform Team Lead, Linaro Platform Teams
Linaro.org | Open source software for ARM SoCs
Follow Linaro: http://www.facebook.com/pages/Linarohttp://twitter.com/#!/linaroorg - http://www.linaro.org/linaro-blog
> -----Original Message-----
> From: Mark Brown [mailto:broonie@opensource.wolfsonmicro.com]
> Sent: Monday, September 19, 2011 3:56 PM
> To: Ashish Jangam
> Subject: Re: LKML - Battery charge current issue
>
> On Mon, Sep 19, 2011 at 06:26:10AM +0000, Ashish Jangam wrote:
>
> Questions like this should always be CCed to the list.
>
> > Some Power Management Controllers has a feature of configuring battery
> > charge current for faster battery charging.
>
> > In the Linux power supply core there is no provision that can help the user
> > for configuring battery charge current at "runtime". In current Linux
> implementation
> > this is done in battery "probe" probably using platform data.
>
> > Should an additional entry in sysfs be added for configuring battery charge
> current
> > since most of our clients desire such feature?
>
> This depends what you're looking to do. The specific currents used
> should be configured using platform data, if the user gets these wrong
> they can result in physical damage to the system so it's generally
> undesirable to have userspace poking at them. The switch from trickle
> to fast charge is normally managed automatically, often by hardware.
> If there's absolutely no way to figure this out from kernel I guess a
> userspace control would be OK but it'd be pretty surprising.
Hardware that don't support/have automatic upgrade then, quick battery charge feature
will be of no use to end user and especially in smart phone this could be problematic.
If sysfs is not a good way of doing this then some alternate mechanism will be very much
appreciated.
Hi,
For the last few days I've been fighting what I thought was a
touchscreen software problem with the Origen.
I finally went back to a known good version and it didn't work either.
I've figured out what happened to my board I'm just not sure why as
there has been no mechanical stress on that connection. One of the
connectors has become de-laminated so the signals aren't getting
through.
I've attached a picture with the area of the fault indicated.
Applying pressure to the indicted area re-establishes the connection.
Angus
--
Angus Ainslie <angus.ainslie(a)linaro.org>
Team Lead, Samsung Landing Team
Hi Linaro'ers
I need to boot Linaro in very short time as my project has a constraint on
energy.
The normal linaro boot up time is 2 minutes and 15 seconds on my overo fire.
I did a little startup tweaks and achieved 2:00 minutes.
Is there a way to boot up Linaro in under 40 sec. ?? This could include
increasing CPU speeds or more OS tweaks....
Has anybody worked on this(Linaro boot up times) so far..??
Thanks & Regards.
---------------------------------
Sudhangathan BS
Ph:(+91) 9731-905-205
---------------------------------
Prevent a syntax error when running the test which leads to an
exception in the script and exiting wildely the loop.
The cpuburn stays running in background.
Signed-off-by: Daniel Lezcano <daniel.lezcano(a)linaro.org>
---
include/functions.sh | 14 ++++++++++++--
1 files changed, 12 insertions(+), 2 deletions(-)
diff --git a/include/functions.sh b/include/functions.sh
index 2cc44b3..f003414 100644
--- a/include/functions.sh
+++ b/include/functions.sh
@@ -109,8 +109,18 @@ get_governor() {
wait_latency() {
local cpu=$1
local dirpath=$CPU_PATH/$cpu/cpufreq
- local latency=$(cat $dirpath/cpuinfo_transition_latency)
- local nrfreq=$(cat $dirpath/scaling_available_frequencies | wc -w)
+ local latency=
+ local nrfreq=
+
+ latency=$(cat $dirpath/cpuinfo_transition_latency)
+ if [ $? != 0 ]; then
+ return $?
+ fi
+
+ nrfreq=$(cat $dirpath/scaling_available_frequencies | wc -w)
+ if [ $? != 0 ]; then
+ return $?
+ fi
nrfreq=$((nrfreq + 1))
../utils/nanosleep $(($nrfreq * $latency))
--
1.7.4.1
Enclosed please find the link to the Weekly Status report
for the Power Management working group for the week ending 2011-09-16.
== Weekly Status Report ==
https://wiki.linaro.org/WorkingGroups/PowerManagement/Status/2011-09-15
== Summary ==
* The PMWG will handle the common struck clk work, going forward
* Steve Jahnke gave thermal talk at Linux Plumbers Conference which was
pretty well received. Got a lot of comments and feedback from not just
upstream, but also other Si vendors (Intel, Nvidia, Qualcomm).
* Basic feedback was that existing framework SHOULD be able to handle
what we need (governor structure, etc.) so we should submit patches to that
first.
* The more complex items (tying multiple sensors into a single thermal
zone) would be welcomed additions so we should propose and take the lead
here
* Samsung: Hosted the thermal sensor driver, thermal governor , cooling
devices and the DVFS fixes that are not in mainline
* Started testing sched_mc features in samsung exynos board.
* Continuing the work on imx cpuidle driver, modifying code to be more
community friendly and testing and fixing issues for compatibility with
i.MX6Q.
* Initial tests on the cpuidle menu governor showed some significant
power inefficiency for some contexts, will discuss the findings with Linaro
PMWG.
* Studying the impact of flags and fields of struct sched_domain in
scheduling policy
Best regards,
Mounir
--
Mounir Bsaibes
Project Manager
Follow Linaro.org:
facebook.com/pages/Linaro/155974581091106http://twitter.com/#!/linaroorghttp://www.linaro.org/linaro-blog <http://www.linaro.org/linaro-blog>
