* Peter Ujfalusi peter.ujfalusi@ti.com [111222 09:36]:
Hi Tony,
On 12/22/2011 07:32 PM, Tony Lindgren wrote:
Hi,
@@ -548,6 +588,19 @@ void omap4_panda_display_init(void) omap_display_init(&omap4_panda_dss_data); }
+static void omap4_panda_audio_init(void) +{
- if (cpu_is_omap4430()) {
/* PandaBoard 4430 */panda_abe_audio_data.card_name = "PandaBoard";/* Audio in is connected to HSMIC */panda_abe_audio_data.has_hsmic = 1;- } else {
/* PandaBoard ES */panda_abe_audio_data.card_name = "PandaBoardES";- }
+}
Please move the cpu_is_omap4430() check from audio_init to..
static void __init omap4_panda_init(void) { int package = OMAP_PACKAGE_CBS; @@ -560,6 +613,7 @@ static void __init omap4_panda_init(void) pr_err("error setting wl12xx data\n");
omap4_panda_i2c_init();
- omap4_panda_audio_init(); platform_add_devices(panda_devices, ARRAY_SIZE(panda_devices)); platform_device_register(&omap_vwlan_device); board_serial_init();
..omap4_panda_init() instead and set some flag that other devices can use too so we don't have cpu_is_omap tests for each init function. I'd assume that will be needed for other devices too, but that's just a guess of course :)
I was planning to have omap4_panda_init_rev(), which would be called just before the i2c_init(). If other drivers need to differentiate between Panda and PandaES they can add their code there.
OK, sounds good to me. Maybe just make omap4_panda_init_rev() do the cpu_is_omap check for now and then you can patch in the dynamic detection later on?
We can use GPIOs to get the board ID, I'm looking at those for a followup patch. The only issue is that the ID1 GPIO number is different between 4430, and 4460 based Panda. We can have similar revision handling as we have for Beagle at the end.
OK great.
Tony