Re: [PATCH v2] ALSA: line6: add hw monitor volume control for POD HD500

From: Vasily Khoruzhick
Date: Sun May 24 2020 - 13:21:31 EST


On Sat, May 23, 2020 at 11:59 PM Takashi Iwai <tiwai@xxxxxxx> wrote:

> Thanks for the patch.
> The basic implementation looks good, but I see a few issues.

> > +EXPORT_SYMBOL(line6_send_raw_message);
>
> Let's use EXPORT_SYMBOL_GPL() instead.

Sure, will do.

> > +static const unsigned int float_zero_to_one_lookup[] = {
> > +0x00000000, 0x3C23D70A, 0x3CA3D70A, 0x3CF5C28F, 0x3D23D70A, 0x3D4CCCCD,
> > +0x3D75C28F, 0x3D8F5C29, 0x3DA3D70A, 0x3DB851EC, 0x3DCCCCCD, 0x3DE147AE,
> > +0x3DF5C28F, 0x3E051EB8, 0x3E0F5C29, 0x3E19999A, 0x3E23D70A, 0x3E2E147B,
> > +0x3E3851EC, 0x3E428F5C, 0x3E4CCCCD, 0x3E570A3D, 0x3E6147AE, 0x3E6B851F,
> > +0x3E75C28F, 0x3E800000, 0x3E851EB8, 0x3E8A3D71, 0x3E8F5C29, 0x3E947AE1,
> > +0x3E99999A, 0x3E9EB852, 0x3EA3D70A, 0x3EA8F5C3, 0x3EAE147B, 0x3EB33333,
> > +0x3EB851EC, 0x3EBD70A4, 0x3EC28F5C, 0x3EC7AE14, 0x3ECCCCCD, 0x3ED1EB85,
> > +0x3ED70A3D, 0x3EDC28F6, 0x3EE147AE, 0x3EE66666, 0x3EEB851F, 0x3EF0A3D7,
> > +0x3EF5C28F, 0x3EFAE148, 0x3F000000, 0x3F028F5C, 0x3F051EB8, 0x3F07AE14,
> > +0x3F0A3D71, 0x3F0CCCCD, 0x3F0F5C29, 0x3F11EB85, 0x3F147AE1, 0x3F170A3D,
> > +0x3F19999A, 0x3F1C28F6, 0x3F1EB852, 0x3F2147AE, 0x3F23D70A, 0x3F266666,
> > +0x3F28F5C3, 0x3F2B851F, 0x3F2E147B, 0x3F30A3D7, 0x3F333333, 0x3F35C28F,
> > +0x3F3851EC, 0x3F3AE148, 0x3F3D70A4, 0x3F400000, 0x3F428F5C, 0x3F451EB8,
> > +0x3F47AE14, 0x3F4A3D71, 0x3F4CCCCD, 0x3F4F5C29, 0x3F51EB85, 0x3F547AE1,
> > +0x3F570A3D, 0x3F59999A, 0x3F5C28F6, 0x3F5EB852, 0x3F6147AE, 0x3F63D70A,
> > +0x3F666666, 0x3F68F5C3, 0x3F6B851F, 0x3F6E147B, 0x3F70A3D7, 0x3F733333,
> > +0x3F75C28F, 0x3F7851EC, 0x3F7AE148, 0x3F7D70A4, 0x3F800000
>
> Just nitpick: better to align with lower hex letters (a-f).

Sorry, I don't really understand what you meant here. Could you elaborate?

> > + buf = kmalloc(sizeof(msg), GFP_ATOMIC);
>
> Is this function called from the irq context like the urb callback?
> I saw that it's called from the mixer put callback, and it's in the
> normal sleepable context, hence GFP_KERNEL can be used safely.

It's also called from podhd_init() to set initial value but I assume
it can sleep there. I'll change it to GFP_KERNEL.

Thanks for review!

Regards,
Vasily