Colin Cross wrote:
> Sent: Wednesday, November 24, 2010 3:18 PM
>
> On Fri, Nov 19, 2010 at 9:07 AM, Stephen Warren <swarren@xxxxxxxxxx> wrote:
> > ...
> > #define APB_DMA_GEN 0x000
> > #define GEN_ENABLE (1<<31)
> > @@ -773,6 +775,16 @@ int __init tegra_dma_init(void)
> > int i;
> > unsigned int irq;
> > void __iomem *addr;
> > + struct clk *c;
> > +
> > + c = tegra_get_clock_by_name("apbdma");
> tegra_get_clock_by_name is private to the clock implementation, it
> should not be used outside clock.c and tegra2_clocks.c. Use
> clk_get_sys. Also, after a recent change in the Tegra tree, DMA is
> initialized before clocks, and this introduces a dependency loop:
> Reading fuses requires dma (hw bug), dma requires clocks, and clocks
> has to read the fuses to determine how fast the cpu can go. I'll have
> to fix that in the fuse API.
Colin, it looks like the initialization rework you mentioned has been
checked in now:
commit 6236e32391e854d8b247b96387bfcc8043518879
Author: Colin Cross <ccross@xxxxxxxxxxx>
Date: Wed Nov 24 14:52:02 2010 -0800
ARM: tegra: Rearrange subsystem initialization order
Can the updated version of the DMA clock enable patch that I sent out
be committed now too?
Thanks.
--
nvpublic
--
To unsubscribe from this list: send the line "unsubscribe linux-tegra" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
