Re: [PATCH 6/7] PCI: aardvark: Correctly clear and unmask all MSI interrupts

["Pali Rohár" <pali@kernel.org>]

On Friday 06 August 2021 09:32:24 Marc Zyngier wrote:
> On 2021-06-25 10:03, Pali Rohár wrote:
> > Define a new macro PCIE_MSI_ALL_MASK and use it for masking, unmasking
> > and
> > clearing all MSI interrupts.
> > 
> > Signed-off-by: Pali Rohár <pali@kernel.org>
> > Reviewed-by: Marek Behún <kabel@kernel.org>
> > Cc: stable@vger.kernel.org
> > ---
> >  drivers/pci/controller/pci-aardvark.c | 16 ++++++++++------
> >  1 file changed, 10 insertions(+), 6 deletions(-)
> > 
> > diff --git a/drivers/pci/controller/pci-aardvark.c
> > b/drivers/pci/controller/pci-aardvark.c
> > index 0e81d89f307d..7cad6d989f6c 100644
> > --- a/drivers/pci/controller/pci-aardvark.c
> > +++ b/drivers/pci/controller/pci-aardvark.c
> > @@ -117,6 +117,7 @@
> >  #define PCIE_MSI_ADDR_HIGH_REG			(CONTROL_BASE_ADDR + 0x54)
> >  #define PCIE_MSI_STATUS_REG			(CONTROL_BASE_ADDR + 0x58)
> >  #define PCIE_MSI_MASK_REG			(CONTROL_BASE_ADDR + 0x5C)
> > +#define     PCIE_MSI_ALL_MASK			GENMASK(31, 0)
> >  #define PCIE_MSI_PAYLOAD_REG			(CONTROL_BASE_ADDR + 0x9C)
> >  #define     PCIE_MSI_DATA_MASK			GENMASK(15, 0)
> > 
> > @@ -470,19 +471,22 @@ static void advk_pcie_setup_hw(struct advk_pcie
> > *pcie)
> >  	advk_writel(pcie, reg, PCIE_CORE_CTRL2_REG);
> > 
> >  	/* Clear all interrupts */
> > +	advk_writel(pcie, PCIE_MSI_ALL_MASK, PCIE_MSI_STATUS_REG);
> >  	advk_writel(pcie, PCIE_ISR0_ALL_MASK, PCIE_ISR0_REG);
> >  	advk_writel(pcie, PCIE_ISR1_ALL_MASK, PCIE_ISR1_REG);
> >  	advk_writel(pcie, PCIE_IRQ_ALL_MASK, HOST_CTRL_INT_STATUS_REG);
> > 
> >  	/* Disable All ISR0/1 Sources */
> > -	reg = PCIE_ISR0_ALL_MASK;
> > -	reg &= ~PCIE_ISR0_MSI_INT_PENDING;
> > -	advk_writel(pcie, reg, PCIE_ISR0_MASK_REG);
> > -
> > +	advk_writel(pcie, PCIE_ISR0_ALL_MASK, PCIE_ISR0_MASK_REG);
> >  	advk_writel(pcie, PCIE_ISR1_ALL_MASK, PCIE_ISR1_MASK_REG);
> > 
> >  	/* Unmask all MSIs */
> > -	advk_writel(pcie, 0, PCIE_MSI_MASK_REG);
> > +	advk_writel(pcie, ~(u32)PCIE_MSI_ALL_MASK, PCIE_MSI_MASK_REG);
> > +
> > +	/* Unmask summary MSI interrupt */
> > +	reg = advk_readl(pcie, PCIE_ISR0_MASK_REG);
> > +	reg &= ~PCIE_ISR0_MSI_INT_PENDING;
> > +	advk_writel(pcie, reg, PCIE_ISR0_MASK_REG);
> > 
> >  	/* Enable summary interrupt for GIC SPI source */
> >  	reg = PCIE_IRQ_ALL_MASK & (~PCIE_IRQ_ENABLE_INTS_MASK);
> > @@ -1177,7 +1181,7 @@ static void advk_pcie_handle_msi(struct advk_pcie
> > *pcie)
> > 
> >  	msi_mask = advk_readl(pcie, PCIE_MSI_MASK_REG);
> >  	msi_val = advk_readl(pcie, PCIE_MSI_STATUS_REG);
> > -	msi_status = msi_val & ~msi_mask;
> > +	msi_status = msi_val & ((~msi_mask) & PCIE_MSI_ALL_MASK);
> > 
> >  	for (msi_idx = 0; msi_idx < MSI_IRQ_NUM; msi_idx++) {
> >  		if (!(BIT(msi_idx) & msi_status))
> 
> This still begs the question: why would you ever enable all
> MSIs the first place? They should be enabled on request only.

This is going to be fixed in followup patches. Just Lorenzo asked to
send smaller patch series, so I have to split that followup fix into
the next round.

>         M.
> -- 
> Jazz is not dead. It just smells funny...