Messages in this thread | | | Date | Wed, 25 Apr 2012 22:01:49 +0530 | Subject | Re: [PATCH] regulator: core: Keep boot_on regulators powered during init | From | Jassi Brar <> |
| |
On 25 April 2012 21:14, Ulf Hansson <ulf.hansson@stericsson.com> wrote: > On 04/25/2012 05:34 PM, Jassi Brar wrote: >>> >> Shouldn't the bootloader be responsible for ensuring MMC is >> put down after using it and before passing control to the kernel. > > This is kind of complicated. eMMC is powered by two regulators. VCC and > VCCQ. Cutting VCC uncontrolled is not allowed. If doing so anyway there is > two options to recover. > > 1. Cut VCCQ as well and do full reinit of the eMMC > 2. Pull a hw-reset pin to the eMMC and do a full reinit of the eMMC. > > In some cases neither is possible due to HW constraints. Thus we have to > make sure VCC is not cut. > I meant power-down to whatever extent your h/w permits. The rest you'll have to take care by appropriately flagging supplies in the kernel.
If not full power-off, couldn't your bootloader atleast send eMMC the necessary "power-down sequence of commands", so that it responds well to next power-up sequence of commands ?
The point being, the board files, in both your bootloader and kernel, ought to know the h/w constraints and program the regulator api accordingly.
Generalizing the case, IMHO it's not feasible for every stack in the kernel to assume it has to first cleanly shutdown the h/w before probing it.
>> Even if the bootloader didn't power off properly, unless there is >> some outstanding data to be written when the kernel is passed >> control, perhaps it's not that serious? > > The kernel will not be to re-init the eMMC... quite serious. :-) > Not unless your board files flag the supplies accordingly, always_on, and the bootloader sent eMMC the "deadly command seq" ;) -- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/
| |