sm5803: avoid requesting 0 fast-charge current

Silicon Mitus claim that setting the fast charge current to zero can
cause charger instability, so force requests for zero current to set
a nominal value instead of zero. This should be safe because this
function only sets the fast charge current limit, and situations where
the charge manager sets zero current should be outside the domain where
the charger will use fast-charge mode.

BUG=b:230712704
TEST=Fast charge current limit on Nereid is set to 100 mA instead of
     zero when the battery is fully charged; no errors are reported.
BRANCH=none

Signed-off-by: Peter Marheine <pmarheine@chromium.org>
Change-Id: I3f625552913d40a7a8f8b3d0ac9874e3af7f0a56
Reviewed-on: https://chromium-review.googlesource.com/c/chromiumos/platform/ec/+/3826712
Reviewed-by: Aseda Aboagye <aaboagye@chromium.org>

1 files changed, 12 insertions, 0 deletions

diff --git a/driver/charger/sm5803.c b/driver/charger/sm5803.c
index 363e320f89..a1fbb4a884 100644
--- a/driver/charger/sm5803.c
+++ b/driver/charger/sm5803.c
@@ -1434,6 +1434,18 @@ static enum ec_error_list sm5803_set_current(int chgnum, int current)
 	enum ec_error_list rv;
 	int reg;
 
+	if (current == 0) {
+		/*
+		 * Per Silicon Mitus, setting the fast charge current limit to 0
+		 * causes "much unstable". This normally happens when the
+		 * battery is fully charged (so we don't expect fast charge to
+		 * be in use): turn 0 into the minimum nonzero value so we
+		 * avoid setting this register to 0 but still make the requested
+		 * current as small as possible.
+		 */
+		current = SM5803_REG_TO_CURRENT(1);
+	}
+
 	rv = chg_read8(chgnum, SM5803_REG_FAST_CONF4, &reg);
 	if (rv)
 		return rv;