Re: [PATCH 6/6] thermal: Add Tegra SOCTHERM thermal management driver

On 04/07/14 11:43, Wei Ni wrote:

On 06/27/2014 04:11 PM, Mikko Perttunen wrote:

quoted

This adds support for the Tegra SOCTHERM thermal sensing and management
system found in the Tegra124 system-on-chip. This initial driver supports
the four thermal zones with hardware-tracked trip points.

quoted

diff --git a/drivers/thermal/tegra_soctherm.c b/drivers/thermal/tegra_soctherm.c

quoted

+static int calculate_shared_calibration(struct tsensor_shared_calibration *r)
+{
+       u32 val;
+       u32 shifted_cp, shifted_ft;
+       int err;
+
+       err = tegra_fuse_readl(FUSE_TSENSOR8_CALIB, &val);
+       if (err)
+               return err;
+       r->base_cp = val & 0x3ff;
+       r->base_ft = (val & (0x7ff << 10)) >> 10;
+
+       err = tegra_fuse_readl(FUSE_SPARE_REALIGNMENT_REG_0, &val);
+       if (err)
+               return err;
+       /* Sign extend from 6 bits to 32 bits */
+       shifted_cp = (s32)((val & 0x1f) | ((val & 0x20) ? 0xffffffe0 : 0x0));
+       val = ((val & (0x1f << 21)) >> 21);
+       /* Sign extend from 5 bits to 32 bits */
+       shifted_ft = (s32)((val & 0xf) | ((val & 0x10) ? 0xfffffff0 : 0x0));
+
+       r->actual_temp_cp = 2 * 25 + shifted_cp;

Do we need to define the "25" as constant, just like below line.

I believe downstream just uses "25" which is the reason I used it, but yes, a constant would be nicer.

quoted

+       r->actual_temp_ft = 2 * NOMINAL_CALIB_FT_T124 + shifted_ft;
+
+       return 0;
+}

quoted

+
+static irqreturn_t soctherm_isr(int irq, void *dev_id)
+{
+       struct tegra_thermctl_zone *zone = dev_id;
+       u32 val;
+       u32 intr_mask = 0x03 << t124_thermctl_shifts[zone->sensor];
+
+       val = readl(zone->tegra->regs + THERMCTL_INTR_STATUS);
+
+       if ((val & intr_mask) == 0)
+               return IRQ_NONE;
+
+       writel(val & intr_mask, zone->tegra->regs + THERMCTL_INTR_STATUS);

I think we need to disable the interrupt here, and enable it again after
updating the high/low limited values. If not, there will trigger mass of
interrupts.

Good point. It works now because of-thermal will set up the new trip points during soctherm_thread_isr, and apparently the interrupt is kept disabled until after that.

quoted

+
+static struct platform_driver tegra_soctherm_driver = {
+       .probe = tegra_soctherm_probe,
+       .remove = tegra_soctherm_remove,
+       .driver = {
+               .name = "tegra_soctherm",
+               .owner = THIS_MODULE,
+               .of_match_table = tegra_soctherm_of_match,
+       },
+};

Will you consider to add suspend/resume support?

I guess for this one it should be simple; same driver probe and teardown as normally. Although currently suspend doesn't support LP0 so no-op is enough.

Thanks.
Wei.

Thanks