Hi,

On Tue, Feb 19, 2013 at 12:36 AM, li guang [off-list ref] wrote:

在 2013-02-15五的 20:16 -0800，Simon Glass写道：

quoted

Use the key-matrix layer to interpret key scan information from the EC
and inject input based on the FDT-supplied key map. This driver registers
itself with the ChromeOS EC driver to perform communications.

[snip ...]

quoted

+/*
+ * Returns true when there is at least one combination of pressed keys that
+ * results in ghosting.
+ */
+static bool cros_ec_keyb_has_ghosting(struct cros_ec_keyb *ckdev, uint8_t *buf)
+{
+     int row;
+
+     /*
+      * Ghosting happens if for any pressed key X there are other keys
+      * pressed both in the same row and column of X as, for instance,
+      * in the following diagram:
+      *
+      * . . Y . g .
+      * . . . . . .
+      * . . . . . .
+      * . . X . Z .
+      *
+      * In this case only X, Y, and Z are pressed, but g appears to be
+      * pressed too (see Wikipedia).
+      *
+      * We can detect ghosting in a single pass (*) over the keyboard state
+      * by maintaining two arrays.  pressed_in_row counts how many pressed
+      * keys we have found in a row.  row_has_teeth is true if any of the
+      * pressed keys for this row has other pressed keys in its column.  If
+      * at any point of the scan we find that a row has multiple pressed
+      * keys, and at least one of them is at the intersection with a column
+      * with multiple pressed keys, we're sure there is ghosting.
+      * Conversely, if there is ghosting, we will detect such situation for
+      * at least one key during the pass.
+      *
+      * (*) This looks linear in the number of keys, but it's not.  We can
+      * cheat because the number of rows is small.
+      */
+     for (row = 0; row < ckdev->rows; row++) {
+             if (cros_ec_keyb_row_has_ghosting(ckdev, buf, row))
+                     return true;
+     }
+
+     return false;
+}

are you sure your EC's firmware did not do ghost-key detection?
or, did you test ghost-key with/without your own ghost-key detection?
as far as I know, ghost-key should be take care either by keyboard
designer or firmware.

Yes, the matrix scans are sent from the EC in a raw form - in fact the
EC on snow does not even know the keycode map. The EC does handle
debouncing though. The idea is to reduce code/complexity in the EC
where we are space-constrained.

[snip]

Regards,
Simon