From mboxrd@z Thu Jan 1 00:00:00 1970 From: anarsoul@gmail.com (Vasily Khoruzhick) Date: Fri, 9 Jul 2010 17:07:43 +0300 Subject: [PATCH 2/3] Add s3c-adc-battery driver In-Reply-To: <20100709135345.GA4351@oksana.dev.rtsoft.ru> References: <1278674842-17583-1-git-send-email-anarsoul@gmail.com> <201007091619.20454.anarsoul@gmail.com> <20100709135345.GA4351@oksana.dev.rtsoft.ru> Message-ID: <201007091707.47807.anarsoul@gmail.com> To: linux-arm-kernel@lists.infradead.org List-Id: linux-arm-kernel.lists.infradead.org ? ????????? ?? 9 ???? 2010 16:53:45 ????? Anton Vorontsov ???????: > > I guess you can remove all the cable_plugged handling from s3c > battery driver, and then use power_supply.external_power_changed > callback to get cable_plugged notification (you should fill > pda_power.supplied_to properly to make it work). > > See drivers/power/ds2760_battery.c and arch/arm/mach-pxa/hx4700.c > as an example. > > You probably will need to change .external_power_changed() hook > to accept 'pst' argument (the supply that caused the change), > so that you could write something like this: > > s3c_external_power_changed(struct power_supply *psy, > struct power_supply *ext) > { > ... > ext->get_property(ext, ONLINE, &val); > s3c->is_plugged = val.intval; > > s3c_kick_cable_plugged_handler(s3c); > } > > Or, instead of ext->get_property() you could just use > power_supply_is_system_supplied(). Not very elegant, but > should work. You didn't get the point. Here's workflow: cable plugged -> GPF2 goes to 0, irq is generated cable plugged irq handler -> cable_plugged ? enable_charger : disable_charger charging.... battery is charged, GPF3 goes to 0, irq is generated battery charged irq handler -> disable_charger cable unplugged -> GPF2 goest to 1, irq is generated pda_power driver does not support 'battery_charged' pin handling, but I need to handle this pin (and IRQ from it) to prevent battery overcharge > Timer is a softirq, atomic context. It adds latency to your > embedded, underpowered device. Plus with timer you can't use > sleepable GPIOs (we should change it for pda_power too, btw). > > Jitter filtering isn't urgent task for the kernel (or user) > needs, right? So it's fine to postpone it for non-atomic > context. Ok, will replace timer with delayed works Regards Vasily -------------- next part -------------- A non-text attachment was scrubbed... Name: not available Type: application/pgp-signature Size: 198 bytes Desc: This is a digitally signed message part. URL: