Re: [PATCH] ALSA: emu10k1: Stop using iommu_present()

[Robin Murphy]

On 2022-04-05 13:29, Takashi Iwai wrote:
> On Tue, 05 Apr 2022 14:13:33 +0200,
> Robin Murphy wrote:
> > iommu_get_domain_for_dev() is already perfectly happy to return NULL
> > if the given device has no IOMMU. Drop the unnecessary check.
> >
> > Signed-off-by: Robin Murphy <robin.murphy@xxxxxxx>
>
> This will change the code behavior.  The current code does nothing if
> no IOMMU is found, but after your removal of the check, the code
> reaches to emu->iommu_workaround = true incorrectly.

Ah, right you are - I should have looked more closely at the second 
check, where of course "domain &&" wants tweaking to "!domain ||" as 
well. I'll send a v2 in a bit...

Thanks,
Robin.

> thanks,
>
> Takashi
>
> > ---
> >   sound/pci/emu10k1/emu10k1_main.c | 3 ---
> >   1 file changed, 3 deletions(-)
> >
> > diff --git a/sound/pci/emu10k1/emu10k1_main.c b/sound/pci/emu10k1/emu10k1_main.c
> > index 86cc1ca025e4..5ffab343b89c 100644
> > --- a/sound/pci/emu10k1/emu10k1_main.c
> > +++ b/sound/pci/emu10k1/emu10k1_main.c
> > @@ -1751,9 +1751,6 @@ static void snd_emu10k1_detect_iommu(struct snd_emu10k1 *emu)
> >   
> >   	emu->iommu_workaround = false;
> >   
> > -	if (!iommu_present(emu->card->dev->bus))
> > -		return;
> > -
> >   	domain = iommu_get_domain_for_dev(emu->card->dev);
> >   	if (domain && domain->type == IOMMU_DOMAIN_IDENTITY)
> >   		return;
> > --
> > 2.28.0.dirty