Re: [PATCH -next] scsi: hisi_sas: drop free_irq of devm_request_irq allocated irq

[John Garry]

On 19/05/2021 04:36, Yang Yingliang wrote:
> On 2021/5/18 23:34, John Garry wrote:
> > On 18/05/2021 14:09, Yang Yingliang wrote:
> > > irq allocated with devm_request_irq should not be freed using
> > > free_irq, because doing so causes a dangling pointer, and a
> > > subsequent double free.
> > >
> > > Reported-by: Hulk Robot <hulkci@xxxxxxxxxx>
> > > Signed-off-by: Yang Yingliang <yangyingliang@xxxxxxxxxx>
> > > ---
> > >   drivers/scsi/hisi_sas/hisi_sas_v3_hw.c | 6 +++---
> > >   1 file changed, 3 insertions(+), 3 deletions(-)
> > >
> > > diff --git a/drivers/scsi/hisi_sas/hisi_sas_v3_hw.c 
> > > b/drivers/scsi/hisi_sas/hisi_sas_v3_hw.c
> > > index 499c770d405c..684f762bcfb3 100644
> > > --- a/drivers/scsi/hisi_sas/hisi_sas_v3_hw.c
> > > +++ b/drivers/scsi/hisi_sas/hisi_sas_v3_hw.c
> > > @@ -4811,9 +4811,9 @@ hisi_sas_v3_destroy_irqs(struct pci_dev *pdev, 
> > > struct hisi_hba *hisi_hba)
> > >   {
> > >       int i;
> > >   -    free_irq(pci_irq_vector(pdev, 1), hisi_hba);
> > > -    free_irq(pci_irq_vector(pdev, 2), hisi_hba);
> > > -    free_irq(pci_irq_vector(pdev, 11), hisi_hba);
> > > +    devm_free_irq(&pdev->dev, pci_irq_vector(pdev, 1), hisi_hba);
> > > +    devm_free_irq(&pdev->dev, pci_irq_vector(pdev, 2), hisi_hba);
> > > +    devm_free_irq(&pdev->dev, pci_irq_vector(pdev, 11), hisi_hba);
> > >       for (i = 0; i < hisi_hba->cq_nvecs; i++) {
> > >           struct hisi_sas_cq *cq = &hisi_hba->cq[i];
> > >           int nr = hisi_sas_intr_conv ? 16 : 16 + i;
> >
> > Does the free_irq(pci_irq_vector(pdev, nr, cq)) call also need to 
> > change (not shown)?
> Yes, I missed that, it should be changed too.

So I think that we need this addition:
 devm_free_irq(&pdev->dev, pci_irq_vector(pdev, nr), cq);

> > Having said that, why have these at all if we use devm_request_irq()? 
> > devm_irq_release() calls free_irq().
> I keep the original logic here, only avoid double free.

Kasan doesn't complain. Anyway, I think we can't rely on device-managed 
method (for calling free_irq()) as it conflicts with pci free vectors 
call. I thought that someone was developed a device-managed version of 
that (pci_alloc_irq_vectors()).

Anyway, please proceed with your change, but with the suggested addition.

Thanks