Hi,
On Tue, May 15, 2012 at 12:43:00PM +0800, Yu Xu wrote:
> Hi Balbi,
>
> 2012/5/14 Felipe Balbi <balbi@xxxxxx>:
> > On Mon, May 14, 2012 at 06:11:11PM +0800, Yu Xu wrote:
> >> Hi Balbi,
> >>
> >>
> >> >> + dev_err(&dev->dev, "failed to allocate memory for u3d\n");
> >> >> + retval = -ENOMEM;
> >> >> + goto err_alloc_private;
> >> >> + }
> >> >> +
> >> >> + spin_lock_init(&u3d->lock);
> >> >> +
> >> >> + platform_set_drvdata(dev, u3d);
> >> >> +
> >> >> + u3d->dev = dev;
> >> >> + u3d->pdata = dev->dev.platform_data;
> >> >
> >> > I believe pdata could be freed after init, no ?
> >> >
> >> Actually, pdata is still used after init, such as pdata->phy_init/phy_deinit.
> >
> > k, fair enough.
> >
> >> >> + u3d->clk = clk_get(&dev->dev, pdata->clkname[0]);
> >> >> + if (IS_ERR(u3d->clk)) {
> >> >> + retval = PTR_ERR(u3d->clk);
> >> >> + goto err_get_clk;
> >> >> + }
> >> >> +
> >> >> + r = platform_get_resource_byname(u3d->dev, IORESOURCE_MEM, "capregs");
> >> >> + if (r == NULL) {
> >> >> + dev_err(&dev->dev, "no I/O memory resource defined\n");
> >> >> + retval = -ENODEV;
> >> >> + goto err_get_cap_regs;
> >> >> + }
> >> >> +
> >> >> + u3d->cap_regs = (struct mv_u3d_cap_regs __iomem *)
> >> >> + ioremap(r->start, resource_size(r));
> >> >> + if (u3d->cap_regs == NULL) {
> >> >> + dev_err(&dev->dev, "failed to map I/O memory\n");
> >> >> + retval = -EBUSY;
> >> >> + goto err_map_cap_regs;
> >> >> + } else {
> >> >> + dev_dbg(&dev->dev, "cap_regs address: 0x%x/0x%x\n",
> >> >> + (unsigned int)r->start, (unsigned int)u3d->cap_regs);
> >> >> + }
> >> >> +
> >> >> + u3d->phy_regs = (u32)u3d->cap_regs + USB3_PHY_OFFSET;
> >> >
> >> > PHY shouldn't be handled by this driver. You need to write a PHY driver
> >> > for controlling the PHY, no ?
> >> >
> >> Yes, I'll have another patch for PHY.
> >
> > Then, why even add this here ?
> >
> Here we just get the PHY register address, but no PHY operation.
> Do you mean we need to put it to the PHY driver?
yes.
--
balbi
Attachment:
signature.asc
Description: Digital signature
