Add and test a simple kernel matcher

This matcher is for up to four fingers, and has these properties:

* Approximately 1.4 times faster at 4 fingers
* Approximately 4.0 times faster at 2 fingers
* Roughly 100 lines of code

Signed-off-by: Henrik Rydberg <rydberg@euromail.se>

1 files changed, 129 insertions, 0 deletions

diff --git a/src/match_four.c b/src/match_four.c
new file mode 100644
index 0000000..c7d2cd3
--- /dev/null
+++ b/src/match_four.c
@@ -0,0 +1,129 @@
+#include "match.h"
+#include <limits.h>
+
+typedef unsigned char u8;
+typedef unsigned int u32;
+
+/* generated by mtdev-kernel - do not edit */
+static const u8 match_data[] = {
+        0, 0, 1, 0, 1, 2, 0, 1, 2, 3, 0, 0, 0, 0, 1, 1,
+        1, 0, 0, 0, 1, 2, 1, 1, 0, 2, 2, 1, 2, 0, 0, 0,
+        1, 2, 3, 1, 1, 0, 2, 3, 2, 1, 2, 0, 3, 3, 1, 2,
+        3, 0, 0, 0, 1, 1, 1, 2, 1, 0, 0, 3, 0, 1, 1, 3,
+        1, 0, 2, 2, 3, 1, 2, 0, 0, 4, 0, 1, 2, 2, 4, 2,
+        1, 0, 0, 5, 0, 2, 1, 1, 5, 2, 0, 1, 1, 4, 1, 0,
+        2, 3, 2, 4, 1, 2, 0, 3, 3, 4, 1, 2, 3, 0, 0, 5,
+        0, 1, 2, 3, 2, 5, 2, 1, 0, 3, 3, 5, 2, 1, 3, 0,
+        0, 6, 0, 2, 1, 3, 1, 6, 2, 0, 1, 3, 3, 6, 2, 3,
+        1, 0, 0, 7, 0, 2, 3, 1, 1, 7, 2, 0, 3, 1, 2, 7,
+        2, 3, 0, 1, 0, 0, 1, 1, 2, 2, 1, 2, 1, 0, 0, 3,
+        0, 1, 1, 4, 2, 0, 3, 4, 2, 1, 0, 5, 0, 2, 2, 5,
+        1, 2, 2, 4, 6, 2, 1, 0, 1, 5, 6, 2, 0, 1, 2, 3,
+        7, 1, 2, 0, 0, 5, 7, 0, 2, 1, 1, 3, 8, 1, 0, 2,
+        0, 4, 8, 0, 1, 2, 2, 5, 8, 2, 1, 0, 3, 3, 5, 8,
+        2, 1, 3, 0, 1, 6, 8, 2, 0, 1, 3, 3, 6, 8, 2, 3,
+        1, 0, 1, 7, 8, 2, 0, 3, 1, 2, 7, 8, 2, 3, 0, 1,
+        2, 4, 9, 1, 2, 0, 3, 3, 4, 9, 1, 2, 3, 0, 0, 6,
+        9, 0, 2, 1, 3, 3, 6, 9, 3, 2, 1, 0, 0, 7, 9, 0,
+        2, 3, 1, 2, 7, 9, 3, 2, 0, 1, 1, 4, 10, 1, 0, 2,
+        3, 3, 4, 10, 1, 3, 2, 0, 0, 5, 10, 0, 1, 2, 3, 3,
+        5, 10, 3, 1, 2, 0, 0, 7, 10, 0, 3, 2, 1, 1, 7, 10,
+        3, 0, 2, 1, 1, 4, 11, 1, 0, 3, 2, 2, 4, 11, 1, 3,
+        0, 2, 0, 5, 11, 0, 1, 3, 2, 2, 5, 11, 3, 1, 0, 2,
+        0, 6, 11, 0, 3, 1, 2, 1, 6, 11, 3, 0, 1, 2, 0, 0,
+        1, 1, 2, 2, 3, 3, 1, 2, 1, 0, 0, 3, 0, 1, 1, 4,
+        2, 0, 3, 4, 2, 1, 0, 5, 0, 2, 2, 5, 1, 2, 1, 6,
+        3, 0, 3, 6, 3, 1, 5, 6, 3, 2, 0, 7, 0, 3, 2, 7,
+        1, 3, 4, 7, 2, 3, 2, 4, 6, 2, 1, 0, 1, 5, 6, 2,
+        0, 1, 2, 3, 7, 1, 2, 0, 0, 5, 7, 0, 2, 1, 1, 3,
+        8, 1, 0, 2, 0, 4, 8, 0, 1, 2, 2, 4, 9, 3, 1, 0,
+        1, 5, 9, 3, 0, 1, 2, 7, 9, 3, 2, 0, 5, 7, 9, 3,
+        2, 1, 1, 8, 9, 3, 0, 2, 4, 8, 9, 3, 1, 2, 2, 3,
+        10, 1, 3, 0, 0, 5, 10, 0, 3, 1, 2, 6, 10, 2, 3, 0,
+        5, 6, 10, 2, 3, 1, 0, 8, 10, 0, 3, 2, 3, 8, 10, 1,
+        3, 2, 1, 3, 11, 1, 0, 3, 0, 4, 11, 0, 1, 3, 1, 6,
+        11, 2, 0, 3, 4, 6, 11, 2, 1, 3, 0, 7, 11, 0, 2, 3,
+        3, 7, 11, 1, 2, 3, 3, 6, 9, 12, 3, 2, 1, 0, 2, 7,
+        9, 12, 3, 2, 0, 1, 3, 5, 10, 12, 3, 1, 2, 0, 1, 7,
+        10, 12, 3, 0, 2, 1, 2, 5, 11, 12, 3, 1, 0, 2, 1, 6,
+        11, 12, 3, 0, 1, 2, 3, 6, 8, 13, 2, 3, 1, 0, 2, 7,
+        8, 13, 2, 3, 0, 1, 3, 4, 10, 13, 1, 3, 2, 0, 0, 7,
+        10, 13, 0, 3, 2, 1, 2, 4, 11, 13, 1, 3, 0, 2, 0, 6,
+        11, 13, 0, 3, 1, 2, 3, 5, 8, 14, 2, 1, 3, 0, 1, 7,
+        8, 14, 2, 0, 3, 1, 3, 4, 9, 14, 1, 2, 3, 0, 0, 7,
+        9, 14, 0, 2, 3, 1, 1, 4, 11, 14, 1, 0, 3, 2, 0, 5,
+        11, 14, 0, 1, 3, 2, 2, 5, 8, 15, 2, 1, 0, 3, 1, 6,
+        8, 15, 2, 0, 1, 3, 2, 4, 9, 15, 1, 2, 0, 3, 0, 6,
+        9, 15, 0, 2, 1, 3, 1, 4, 10, 15, 1, 0, 2, 3, 0, 5,
+        10, 15, 0, 1, 2, 3,
+};
+
+/* generated by mtdev-kernel - do not edit */
+static const int match_index[][5] = {
+        { 0, 0, 1, 3, 6 },
+        { 10, 10, 12, 18, 30 },
+        { 50, 50, 54, 62, 92 },
+        { 164, 164, 170, 194, 230 },
+        { 398, 398, 406, 454, 598 },
+        { 790 }
+};
+
+static void set_dist(u32 *dist,
+                     const struct trk_coord *b1, const struct trk_coord *e1,
+                     const struct trk_coord *b2, const struct trk_coord *e2)
+{
+        const struct trk_coord *p, *q;
+
+        for (p = b1; p != e1; p++)
+                for (q = b2; q != e2; q++)
+                        *dist++ = abs(q->x - p->x) + abs(q->y - p->y);
+}
+
+const u8 *mtdev_match_four(const struct trk_coord *old, int nslot,
+                           const struct trk_coord *pos, int npos)
+{
+        u32 d[16], obj, t;
+        const u8 *p, *b, *e;
+        const int *at;
+
+        set_dist(d, old, old + nslot, pos, pos + npos);
+
+        at = &match_index[nslot][npos];
+        b = &match_data[at[0]];
+        e = &match_data[at[1]];
+
+        obj = UINT_MAX, p = b;
+
+        switch (minval(nslot, npos)) {
+        case 1:
+                for (; b != e; b += npos) {
+                        t = d[*b++];
+                        if (t < obj)
+                                obj = t, p = b;
+                }
+                break;
+        case 2:
+                for (; b != e; b += npos) {
+                        t = d[*b++], t += d[*b++];
+                        if (t < obj)
+                                obj = t, p = b;
+                }
+                break;
+        case 3:
+                for (; b != e; b += npos) {
+                        t = d[*b++], t += d[*b++], t += d[*b++];
+                        if (t < obj)
+                                obj = t, p = b;
+                }
+                break;
+        case 4:
+                for (; b != e; b += npos) {
+                        t = d[*b++], t += d[*b++], t += d[*b++], t += d[*b++];
+                        if (t < obj)
+                                obj = t, p = b;
+                }
+                break;
+        }
+
+        return p;
+}