On Wed, Oct 31, 2012 at 04:56:40, Paul Walmsley wrote:
+ Vaibhav Hiremath
On Tue, 30 Oct 2012, Tony Lindgren wrote:
quoted
* Pantelis Antoniou [off-list ref] [121030 11:04]:
quoted
Looks like the lcdc clock definition got dropped.
It is required for the LCD controller to work. Reintroduce.
This looks like a regression, can you also add the commit
causing it?
Looks like probably a new "feature," in that this clock didn't exist in
the original check-in. Would be good to get Vaibhav's opinion on this;
also the common clock patches will need to be updated.
Thanks Paul for looping me in, something went wrong with my l-o subscription,
so I didn't receive these Patches.
As far as lck clock node is concerned, we had deliberately dropped all leaf-
node clocks from the clock tree, please refer to the description mentioned
in -
http://lists.infradead.org/pipermail/linux-arm-kernel/2012-May/101987.html
From LCDC driver perspective, driver is using,
fb_clk = clk_get(&device->dev, NULL);
This I feel needs to be corrected for valid name as per Spec (mostly I would
vote for "fck") and then every platform should make sure that it returns
valid clock-node for it.
Change in Driver would be,
fb_clk = clk_get(&device->dev, "fck");
Thanks,
Vaibhav
- Paul
On Wed, 31 Oct 2012, Hiremath, Vaibhav wrote:
As far as lck clock node is concerned, we had deliberately dropped all leaf-
node clocks from the clock tree, please refer to the description mentioned
in -
http://lists.infradead.org/pipermail/linux-arm-kernel/2012-May/101987.html
Ach, should have remembered that :-( Indeed there is an LCDC hwmod:
static struct omap_hwmod am33xx_lcdc_hwmod = {
.name = "lcdc",
.class = &am33xx_lcdc_hwmod_class,
.clkdm_name = "lcdc_clkdm",
.mpu_irqs = am33xx_lcdc_irqs,
.flags = HWMOD_SWSUP_SIDLE | HWMOD_SWSUP_MSTANDBY,
.main_clk = "lcd_gclk",
.prcm = {
.omap4 = {
.clkctrl_offs = AM33XX_CM_PER_LCDC_CLKCTRL_OFFSET,
.modulemode = MODULEMODE_SWCTRL,
},
},
};
quoted
From LCDC driver perspective, driver is using,
fb_clk = clk_get(&device->dev, NULL);
This I feel needs to be corrected for valid name as per Spec (mostly I would
vote for "fck") and then every platform should make sure that it returns
valid clock-node for it.
Change in Driver would be,
fb_clk = clk_get(&device->dev, "fck");
Indeed.
- Paul
On Wed, Oct 31, 2012 at 11:19:44, Paul Walmsley wrote:
On Wed, 31 Oct 2012, Hiremath, Vaibhav wrote:
quoted
As far as lck clock node is concerned, we had deliberately dropped all leaf-
node clocks from the clock tree, please refer to the description mentioned
in -
http://lists.infradead.org/pipermail/linux-arm-kernel/2012-May/101987.html
Ach, should have remembered that :-( Indeed there is an LCDC hwmod:
static struct omap_hwmod am33xx_lcdc_hwmod = {
.name = "lcdc",
.class = &am33xx_lcdc_hwmod_class,
.clkdm_name = "lcdc_clkdm",
.mpu_irqs = am33xx_lcdc_irqs,
.flags = HWMOD_SWSUP_SIDLE | HWMOD_SWSUP_MSTANDBY,
.main_clk = "lcd_gclk",
.prcm = {
.omap4 = {
.clkctrl_offs = AM33XX_CM_PER_LCDC_CLKCTRL_OFFSET,
.modulemode = MODULEMODE_SWCTRL,
},
},
};
quoted
quoted
From LCDC driver perspective, driver is using,
fb_clk = clk_get(&device->dev, NULL);
This I feel needs to be corrected for valid name as per Spec (mostly I would
vote for "fck") and then every platform should make sure that it returns
valid clock-node for it.
Change in Driver would be,
fb_clk = clk_get(&device->dev, "fck");
Ok, thanks. Let me submit patch for this.
Thanks,
Vaibhav
Indeed.
- Paul