This is the mail archive of the
ecos-devel@sourceware.org
mailing list for the eCos project.
Re: Closing devices
On 06/19/2012 09:36 AM, Bernard Fouché wrote:
Le 18/06/2012 16:34, Frank Pagliughi a écrit :
Hello All,
I'm looking for ideas on how to close and re-open devices on eCos.
The needs for this are (1) to support swappable/removable devices,
(2) to have a consistent way to put devices into a low-power state
when they are not being used, and (3) to prevent devices from using
the CPU when they are not needed.
On a current project for a battery-powered device, I have a need for
all of this: a CompactFlash that can be removed; a GPS that sends
data constantly, but only needs to be read once every few hours; and
many of the peripherals need to be put in a consistent state prior to
going into a low power mode.
I've been able to accomplish all of this with ugly application-level
code, but thought that a much better solution would be to propagate
the close() call of a devfs device down to the driver, so you could
do this:
Hello Frank,
I have the exact same needs and I also made my changes in the
application code at the moment, for the same reasons. However if we
look at the low level details:
- The Init() function of a driver is called at boot time, a time you
don't want to initialize much things if you don't know yet if you'll
need them a few moment later. init() isn't visible from anything else
than the startup procedure of eCos.
- lookup() is called when the application 'opens' a channel of a
driver. Usually nothing much is done at low level since the assumption
is that init() made the job before. However it's possible to rework
the drivers to change this and future drivers could be written with
this in mind.
- Since a devtab entry can be looked up many times, even by different
threads, it is probably necessary to have a driver to count the number
of times it is looked up and the number of times it is shutdown. When
the count reaches zero, then the driver knows it can power off things.
- drivers that are shared between different targets do not know about
target specific features, by design they focus on the parts that are
common to all targets they can be used on. Such a driver expects that
the MCU pin setup (and other details) has already been done earlier in
the board init code, it has no way to query something to run again
this procedure. If you need to close an UART, you probably also want
to reconfigure the MCU's pins. You may also want to power off the UART
(from the MCU point of view) if the MCU allows it. So even if
shutdown() is implemented, such a driver wouldn't do much regarding
power savings, at best it could only mask or disable an interrupt, the
most important savings must be handled elsewhere. Even if the board
init code could be accessible, how one could ask this code to perform
a partial initialization? (for instance to avoid reconfiguring all
UARTs while a single one is to be re-initialized).
- power management is very MCU/board/application specific and project
specific code will have different things to do. For instance if you
have an external RS232 device, you save more power by turning off the
level converter between the UART and the RS232 connector. Of course
it's better to be able to turn off both the UART in the MCU and the
level converter. If you don't have a level converter, you may want to
reconfigure the MCU pins, for instance to avoid having power drawn
from the UART TX pin if the connected device is also powered off. You
may also want to setup pull-up/down on the pins to stabilize the
signals: it means changing the pin setup of the MCU to change them
from 'UART' to 'GPIO' and then configure the pull up/down feature.
You may also want to change peripheral clock settings for disabled
peripheral in the MCU, to spare a bit more so you have to re configure
also clocking registers when the peripheral must come back in line.
- There is CYGPKG_POWER. Each driver implementing some kind of power
management can be modified to support this package. But I don't see
how this package can interact with the platform code layer. How can a
target using a shared driver can make use of this package for the
shared driver?
IMHO, beside a shutdown mechanism, one also needs to be able to get
control of what's going on between the hardware drivers and the
packages that use them. A low level application initialization routine
should be able to register callbacks to be triggered when events occur
in the drivers and in the package code managing them, hence the
application could handle the board or MCU specifically when some
expected event occurs. Today only part of this could be done in
platform code, but in such a way that it is very close to application
code, however without any clearly defined API.
Bernard
Thank you, Bernard.
I had not thought about reference counting in the drivers from multiple
lookup() calls, but yes, that would probably be required.
In this context - closable devices - it seems unfortunate that the
lookup() function has the side effect of opening the device. I can
imagine that some code might just be trying to get the devtab entry for
other purposes. Perhaps there needs to be another search function that
returns the entry without triggering the lookup() function?
Does the file I/O code serialize calls to lookup? I examined a few
device drivers the implement lookup() and don't see any explicit
mechanism to prevent race conditions, but that could be due to the
nature of the calls not requiring it. I suppose the drivers should lock
the scheduler when manipulating this new reference count.
I acknowledge that the close of the low-level driver may not be
adequate, and the power management and GPIO pins are problematic.
I saw a recent discussion on the list about layering serial drivers, and
started wondering if the target-level issues might be handled through
layered drivers. Could I, perhaps, make a target-specific driver that
intercepted the shutdown() calls, called the lower-level driver (which
might just turn off the interrupt), then have the upper-layer driver
power-down the device and tri-state the GPIO pins?
Would that be similar to the application callbacks that you mention? If
not, please give more detail about what you're thinking in this regard.
Thanks,
Frank