Re: [PATCH v6 03/15] soc/tegra: Add ram code reader helper

From: Mikko Perttunen
Date: Mon Mar 02 2015 - 04:52:20 EST


On 03/02/2015 10:46 AM, Alexandre Courbot wrote:
On Thu, Feb 12, 2015 at 11:06 PM, Tomeu Vizoso
<tomeu.vizoso@xxxxxxxxxxxxx> wrote:
From: Mikko Perttunen <mperttunen@xxxxxxxxxx>

Needed for the EMC and MC drivers to know what timings from the DT to use.

Signed-off-by: Mikko Perttunen <mperttunen@xxxxxxxxxx>
Signed-off-by: Tomeu Vizoso <tomeu.vizoso@xxxxxxxxxxxxx>

---

v4: Replace magic number with PMC_STRAPPING_OPT_A_RAM_CODE_SHIFT
---
drivers/soc/tegra/fuse/tegra-apbmisc.c | 19 +++++++++++++++++++
include/soc/tegra/fuse.h | 1 +
2 files changed, 20 insertions(+)

diff --git a/drivers/soc/tegra/fuse/tegra-apbmisc.c b/drivers/soc/tegra/fuse/tegra-apbmisc.c
index 3bf5aba..dc96a62 100644
--- a/drivers/soc/tegra/fuse/tegra-apbmisc.c
+++ b/drivers/soc/tegra/fuse/tegra-apbmisc.c
@@ -28,8 +28,13 @@
#define APBMISC_SIZE 0x64
#define FUSE_SKU_INFO 0x10

+#define PMC_STRAPPING_OPT_A_RAM_CODE_SHIFT 4
+#define PMC_STRAPPING_OPT_A_RAM_CODE_MASK_LONG (0xf << PMC_STRAPPING_OPT_A_RAM_CODE_SHIFT)
+#define PMC_STRAPPING_OPT_A_RAM_CODE_MASK_SHORT (0x3 << PMC_STRAPPING_OPT_A_RAM_CODE_SHIFT)

I think you are slightly over 80 chars here.

+
static void __iomem *apbmisc_base;
static void __iomem *strapping_base;
+static bool long_ram_code;

u32 tegra_read_chipid(void)
{
@@ -54,6 +59,18 @@ u32 tegra_read_straps(void)
return 0;
}

+u32 tegra_read_ram_code(void)
+{
+ u32 straps = tegra_read_straps();
+
+ if (long_ram_code)
+ straps &= PMC_STRAPPING_OPT_A_RAM_CODE_MASK_LONG;
+ else
+ straps &= PMC_STRAPPING_OPT_A_RAM_CODE_MASK_SHORT;
+
+ return straps >> PMC_STRAPPING_OPT_A_RAM_CODE_SHIFT;
+}
+
static const struct of_device_id apbmisc_match[] __initconst = {
{ .compatible = "nvidia,tegra20-apbmisc", },
{},
@@ -112,4 +129,6 @@ void __init tegra_init_apbmisc(void)
strapping_base = of_iomap(np, 1);
if (!strapping_base)
pr_err("ioremap tegra strapping_base failed\n");
+
+ long_ram_code = of_property_read_bool(np, "nvidia,long-ram-code");

Is this long-ram property chip-dependent? If so, couldn't we just use
the chip ID or matched compatible string to decide whether the RAM
code length, instead of having a dedicated property for it?


I am under the impression that it is board-specific.

Mikko

--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/