On 2018/5/2 20:46, Ulf Hansson wrote:
On 28 April 2018 at 03:11, Shawn Lin <shawn.lin@xxxxxxxxxxxxxx> wrote:
Hi Rob and Ulf,
On 2018/4/28 2:54, Rob Herring wrote:
On Mon, Apr 23, 2018 at 04:10:29PM +0800, Shawn Lin wrote:
...
+- power-delay-ms: Tunable delay waiting for I/O signalling and card
power supply
+ to be stable. Default to 10ms if no available.
How long do you need? Would changing the default to say 20ms work and
still be short enough others don't care?
It varies from board-2-board, but my boards only need 2ms or less. The
hard-coded 10ms was increased from 2ms since 2009, just for fixing one
board. And nobody complained 10ms is too short(including me), so it's
fine for everybody probably. However, nobody complained it's toooo long
as well, expect me, so my best guess is either others don't care it at
all, or haven't noticed it. So the intention for this patch, is to save
the unnecessary waste for the boot-time sensitive environment, by
reducing the delay but don't break anything else.
Can we use the same property as the mmc pwrseq binding defines:
post-power-on-delay-ms
I'm fine with using post-power-on-delay-ms, but it depends on whether
using pwrseq_simple. So I need add parsing it in two places for
different prupose. Is it ok, or better idea?
I don't think the parsing is an issue, but that we need to allow two
different descriptions of the same property name may be a bit
confusing.
I would rather keep them separate, but I have no strong onion.
I also would like to keep the separate. Rob, any suggestion? :)
Kind regards
Uffe
--
To unsubscribe from this list: send the line "unsubscribe devicetree" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
