A script to create ACPI based GPIO mapping.

A script to create ACPI based GPIO mapping.

See the doc string for theory of operation.

Testing
=======

On an ST8 system with the Insyde BIOS implementing a mock ACPI device:

localhost tmp # ./gpio_setup.py
gpio_setup.py: could not find GPIO PCI device 0000:00:1f.0
localhost tmp # modprobe nm10_gpio.ko
localhost tmp # ./gpio_setup.py
gpio_setup.py: no gpio mapping found. Is ACPI driver installed?
localhost tmp # modprobe chromeos_acpi
localhost tmp # ./gpio_setup.py
gpio_setup.py: /home/gpio does not exist
localhost tmp # mkdir /home/gpio
localhost tmp # ./gpio_setup.py
localhost tmp # ls /home/gpio
debug_header_0  debug_header_1  debug_header_2  debug_header_3  developer_switch  recovery_button  write_protect
localhost tmp # cat /home/gpio/developer_switch
1
 # At this point press and hold the developer button
localhost tmp # cat /home/gpio/developer_switch
0
 # At this point release the developer button
localhost tmp # cat /home/gpio/developer_switch
1
localhost tmp # ./gpio_setup.py --help
Usage: gpio_setup.py [options]

Options:
  -h, --help            show this help message and exit
  --symlink_root=SYMLINK_ROOT
  --pci_address=PCI_ADDRESS
  --acpi_root=ACPI_ROOT
localhost tmp # ./gpio_setup.py --acpi_root=/tmp/xuz
gpio_setup.py: no gpio mapping found. Is ACPI driver installed?
localhost tmp # ./gpio_setup.py --pci_address=0.1.2.3
gpio_setup.py: could not find GPIO PCI device 0.1.2.3
localhost tmp # ./gpio_setup.py --symlink_root=/tmp/xyz
gpio_setup.py: /tmp/xyz does not exist

[vb, 2010-06-02 17:30:48]

Hung Te, please have a look as we discussed.

[Hung-Te, 2010-06-02 19:15:51]

http://codereview.chromium.org/2421006/diff/4001/5001
File src/platform/utils/gpio_setup.py (right):

http://codereview.chromium.org/2421006/diff/4001/5001#newcode116
src/platform/utils/gpio_setup.py:116: dir = os.path.dirname(f)
changing built-in names like 'dir' may cause potential bugs in future - maybe
using some other variable names here, for example folder or dirname ?

http://codereview.chromium.org/2421006/diff/4001/5001#newcode128
src/platform/utils/gpio_setup.py:128: /sys/class/gpio/export, if this pin has
not been enabled yet.
Just to make sure - the write operation will trigger pin to be enabled
immediately, no need to wait for any signal / synchronization, right?

[vb, 2010-06-02 20:51:49]

Hung-Te, I addressed your comments, please look again.

http://codereview.chromium.org/2421006/diff/4001/5001
File src/platform/utils/gpio_setup.py (right):

http://codereview.chromium.org/2421006/diff/4001/5001#newcode116
src/platform/utils/gpio_setup.py:116: dir = os.path.dirname(f)
On 2010/06/02 19:15:51, Hung-Te wrote:
> changing built-in names like 'dir' may cause potential bugs in future - maybe
> using some other variable names here, for example folder or dirname ?

Good point, changed 'dir' to 'directory'

http://codereview.chromium.org/2421006/diff/4001/5001#newcode128
src/platform/utils/gpio_setup.py:128: /sys/class/gpio/export, if this pin has
not been enabled yet.
On 2010/06/02 19:15:51, Hung-Te wrote:
> Just to make sure - the write operation will trigger pin to be enabled
> immediately, no need to wait for any signal / synchronization, right?

I believe so, this is an action taken by the kernel on a write to this file.

Note that the file gets implicitly closed when this function exits, so I don't
think there is a need to worry about synchronization here.

[Hung-Te, 2010-06-02 21:12:40]

LGTM.