aboutsummaryrefslogtreecommitdiffstats
path: root/drivers/media/common/ir-functions.c
diff options
context:
space:
mode:
authorMauro Carvalho Chehab <mchehab@redhat.com>2009-08-29 14:15:55 -0300
committerMauro Carvalho Chehab <mchehab@redhat.com>2009-09-12 12:19:47 -0300
commit715a223323c8c8bcbe7739e20f6c619f7343b595 (patch)
tree0c7eddcdd3c778428fc936009af085940c0394bb /drivers/media/common/ir-functions.c
parentV4L/DVB (12501): gspca - sonixj: Do the ov7660 sensor work again. (diff)
downloadlinux-dev-715a223323c8c8bcbe7739e20f6c619f7343b595.tar.xz
linux-dev-715a223323c8c8bcbe7739e20f6c619f7343b595.zip
V4L/DVB (12595): common/ir: use a struct for keycode tables
Currently, V4L uses a scancode table whose index is the scancode and the value is the keycode. While this works, it has some drawbacks: 1) It requires that the scancode to be at the range 00-7f; 2) keycodes should be masked on 7 bits in order for it to work; 3) due to the 7 bits approach, sometimes it is not possible to replace the default keyboard to another one with a different encoding rule; 4) it is different than what is done with dvb-usb approach; 5) it requires a typedef for it to work. This is not a recommended Linux CodingStyle. This patch is part of a larger series of IR changes. It basically replaces the IR_KEYTAB_TYPE tables by a structured table: struct ir_scancode { u16 scancode; u32 keycode; }; This is very close to what dvb does. So, a further integration with DVB code will be easy. While we've changed the tables, for now, the IR keycode handling is still based on the old approach. The only notable effect is the redution of about 35% of the ir-common module size: text data bss dec hex filename 6721 29208 4 35933 8c5d old/ir-common.ko 5756 18040 4 23800 5cf8 new/ir-common.ko In thesis, we could be using above u8 for scancode, reducing even more the size of the module, but defining it as u16 is more convenient, since, on dvb, each scancode has up to 16 bits, and we currently have a few troubles with rc5, as their scancodes are defined with more than 8 bits. This patch itself shouldn't be doing any functional changes. Signed-off-by: Mauro Carvalho Chehab <mchehab@redhat.com>
Diffstat (limited to 'drivers/media/common/ir-functions.c')
-rw-r--r--drivers/media/common/ir-functions.c15
1 files changed, 13 insertions, 2 deletions
diff --git a/drivers/media/common/ir-functions.c b/drivers/media/common/ir-functions.c
index 16792a68a449..655474b29e21 100644
--- a/drivers/media/common/ir-functions.c
+++ b/drivers/media/common/ir-functions.c
@@ -58,13 +58,24 @@ static void ir_input_key_event(struct input_dev *dev, struct ir_input_state *ir)
/* -------------------------------------------------------------------------- */
void ir_input_init(struct input_dev *dev, struct ir_input_state *ir,
- int ir_type, IR_KEYTAB_TYPE *ir_codes)
+ int ir_type, struct ir_scancode_table *ir_codes)
{
int i;
ir->ir_type = ir_type;
+
+ memset(ir->ir_codes, sizeof(ir->ir_codes), 0);
+
+ /*
+ * FIXME: This is a temporary workaround to use the new IR tables
+ * with the old approach. Later patches will replace this to a
+ * proper method
+ */
+
if (ir_codes)
- memcpy(ir->ir_codes, ir_codes, sizeof(ir->ir_codes));
+ for (i = 0; i < ir_codes->size; i++)
+ if (ir_codes->scan[i].scancode < IR_KEYTAB_SIZE)
+ ir->ir_codes[ir_codes->scan[i].scancode] = ir_codes->scan[i].keycode;
dev->keycode = ir->ir_codes;
dev->keycodesize = sizeof(IR_KEYTAB_TYPE);