Hi On Samstag, 9. August 2008, Vefa Bicakci wrote: > Hello, > > I am sad to say that the patch relayed by Stefan Lippers-Hollman breaks things > for me. This is because my AC adapter has the name "ADP0". (This is a Toshiba > Satellite A100 series laptop running Debian Sid with Linux 2.6.25.15.) Here's > a > demonstration:
You're right, now I could find a system where the previous patch failed as
well, strangely all other notebooks I encountered over the last few days
used /sys/class/power_supply/ACAD, but you're right the test needs to be
more generic.
> === 8< ===
> $ cd /sys/class/power_supply/
> $ ls
> ADP0 BAT0
> $ cd ADP0
> $ ls
> device online power subsystem type uevent
> $ cat type
> Mains
> $ cat online
> 1
> $
> === >8 ===
>
> As you can see, my power supply has the name "ADP0" which
> does not start with "AC".
>
> Now let's take a look at the relevant part of the patch relayed by
> Stefan Lippers-Hollman:
>
> === 8< ===
> + if [ -d /sys/class/power_supply/ ]; then
> + for FN in /sys/class/power_supply/AC*; do
> + local online
> + if [ -r "${FN}/online" ]; then
> + online="$(cat ${FN}/online)"
> + [ "$online" = 1 ] && exit 0
> + [ "$online" = 0 ] && OFF_LINE_P=yes
> === >8 ===
>
> As you can see, the patch assumes that the power supply names are always
> prefixed with the string "AC". However, as I have shown above, this is not the
> case with all laptops.
>
> Hence, here's an improved patch which makes "on_ac_power check" whether each
> power supply is an AC adapter by looking at its type:
>
> === 8< ===
>
> --- a/src/on_ac_power.orig.debian 2008-08-09 14:44:48.000000000 +0300
> +++ b/src/on_ac_power 2008-08-09 14:19:24.000000000 +0300
> @@ -20,19 +20,34 @@
> # that have known states. If any adapter is on-line, we return 0. If
> # no adapters are on-line but one or more are off-line, we return 1.
> #
> -if /sbin/acpi_available && [ -d /proc/acpi/ac_adapter ]; then
> +if /sbin/acpi_available; then
> OFF_LINE_P=no
> - for FN in /proc/acpi/ac_adapter/*; do
> - if [ -d "${FN}" ]; then
> - if [ -r "${FN}/state" ]; then
> - grep --quiet on-line "${FN}/state" && exit 0
> - grep --quiet off-line "${FN}/state" && OFF_LINE_P=yes
> - elif [ -r "${FN}/status" ]; then
> - grep --quiet on-line "${FN}/status" && exit 0
> - grep --quiet off-line "${FN}/status" && OFF_LINE_P=yes
> + if [ -d /sys/class/power_supply/ ]; then
> + for FN in /sys/class/power_supply/*; do
> + if test -d "${FN}" && test -r "${FN}/type"; then
> + type="$(cat ${FN}/type)"
> + if test "x${type}" = "xMains"; then
> + if [ -r "${FN}/online" ]; then
> + online="$(cat ${FN}/online)"
> + [ "$online" = 1 ] && exit 0
> + [ "$online" = 0 ] && OFF_LINE_P=yes
> + fi
> + fi
> fi
> - fi
> - done
> + done
> + elif [ -d /proc/acpi/ac_adapter ]; then
> + for FN in /proc/acpi/ac_adapter/*; do
> + if [ -d "${FN}" ]; then
> + if [ -r "${FN}/state" ]; then
> + grep --quiet on-line "${FN}/state" && exit 0
> + grep --quiet off-line "${FN}/state" && OFF_LINE_P=yes
> + elif [ -r "${FN}/status" ]; then
> + grep --quiet on-line "${FN}/status" && exit 0
> + grep --quiet off-line "${FN}/status" && OFF_LINE_P=yes
> + fi
> + fi
> + done
> + fi
> [ "${OFF_LINE_P}" = "yes" ] && exit 1
> fi
>
>
> === >8 ===
Yes, this is a more robust form of the probing, which seems to be
necessary, confirmed to be working on the systems which worked already and
on the new one where it failed due to a slightly different sysfs naming
(using 2.6.26).
> This patch also fixes an issue in the patch relayed by Stefan Lippers-Hollman
> where the variable named "online" is declared as a "local" variable even
> though
> it is not in a function.
>
> Regards,
>
> M. Vefa Bıçakcı
>
> Note: I am attaching the patch as well. (Just in case my e-mail program messes
> up the patch.)
>
Regards
Stefan Lippers-Hollmann
signature.asc
Description: This is a digitally signed message part.
