Hi,
On 08-03-19 08:33, Heikki Krogerus wrote:
On Thu, Mar 07, 2019 at 10:12:59AM -0800, Guenter Roeck wrote:
On Thu, Mar 07, 2019 at 05:36:01PM +0100, Hans de Goede wrote:
After commit ea3b4d5523bc ("usb: typec: fusb302: Resolve fixed power role
contract setup"), tcpm_set_cc always calls fusb302_set_toggling.
Before this refactor tcpm_set_cc does the following:
1) fusb302_set_toggling(TOGGLING_MODE_OFF),
this sets both FUSB_REG_MASK_BC_LVL and FUSB_REG_MASK_COMP_CHNG.
2) fusb302_set_cc_pull(...).
3) "reset cc status".
4) if pull-up fusb302_set_src_current(...).
5) if pull-up or pull-down enable bc-lvl resp comp-chng irq.
6) fusb302_set_toggling(new-toggling-mode), which again
sets both FUSB_REG_MASK_BC_LVL and FUSB_REG_MASK_COMP_CHNG disabling
the just enabled irq. fusb302_set_toggling is skipped when the new
toggling mode is TOGGLING_MODE_OFF because this is already done in 1,
note in this case 5) is a no-op.
When we are toggling the bits set by fusb302_set_cc_pull will be ignored
until we turn toggling off, so we can safely move the fusb302_set_cc_pull
call to before setting TOGGLING_MODE_OFF.
Either we are not toggling yet, or the src-current has already been set,
so we can also safely set the src-current earlier, allowing us to do the
fusb302_set_toggling(TOGGLING_MODE_OFF) call at the same time as we
set the other toggling modes. Also setting the src-current is a no-op
when not enabling pull-ups, so we can drop the if.
And since the second fusb302_set_toggling undoes the effects of step 5,
we can skip step 5, the bc-lvl resp comp-chng irq wil be enabled by
fusb302_handle_togdone_snk resp. fusb302_handle_togdone_src when toggling
is done.
Together this allows us to simplify things to:
1) fusb302_set_cc_pull(...)
2) "reset cc status"
3) fusb302_set_src_current(...)
4) fusb302_set_toggling(new-toggling-mode)
This commit does this, leading to a nice cleanup.
Signed-off-by: Hans de Goede <hdegoede@xxxxxxxxxx>
I think I understand the logic, but I think it would be really beneficial
if this (and the rest of the series) can be tested independently (ie
by someone with hardware).
Reviewed-by: Guenter Roeck <linux@xxxxxxxxxxxx>
I'll test this series with my GPD win board today.
Note I've already tested this series on both a GPD win and a GPD pocket.
Also Guenter's reviewed has shown that a small error has crept up
in patch 5, patch 5 adds 2:
usleep_range(5000, 10000);
lines which should be:
usleep_range(50, 100);
So if you test please make that change, or wait for v3 of the series.
Regards,
Hans
