Re: [PATCH v3] net: stmmac: fix FPE events losing

From: Paolo Abeni
Date: Thu Nov 30 2023 - 09:13:56 EST


On Thu, 2023-11-30 at 15:09 +0100, Paolo Abeni wrote:
> On Thu, 2023-11-30 at 16:09 +0300, Serge Semin wrote:
> > Hi Paolo
> >
> > On Thu, Nov 30, 2023 at 10:55:34AM +0100, Paolo Abeni wrote:
> > > On Tue, 2023-11-28 at 05:56 +0000, Jianheng Zhang wrote:
> > > > The status bits of register MAC_FPE_CTRL_STS are clear on read. Using
> > > > 32-bit read for MAC_FPE_CTRL_STS in dwmac5_fpe_configure() and
> > > > dwmac5_fpe_send_mpacket() clear the status bits. Then the stmmac interrupt
> > > > handler missing FPE event status and leads to FPE handshaking failure and
> > > > retries.
> > > > To avoid clear status bits of MAC_FPE_CTRL_STS in dwmac5_fpe_configure()
> > > > and dwmac5_fpe_send_mpacket(), add fpe_csr to stmmac_fpe_cfg structure to
> > > > cache the control bits of MAC_FPE_CTRL_STS and to avoid reading
> > > > MAC_FPE_CTRL_STS in those methods.
> > > >
> > > > Fixes: 5a5586112b92 ("net: stmmac: support FPE link partner hand-shaking procedure")
> > > > Reviewed-by: Serge Semin <fancer.lancer@xxxxxxxxx>
> > > > Signed-off-by: Jianheng Zhang <jianheng@xxxxxxxxxxxx>
> > > > ---
> > > > drivers/net/ethernet/stmicro/stmmac/dwmac5.c | 45 +++++++++-------------
> > > > drivers/net/ethernet/stmicro/stmmac/dwmac5.h | 4 +-
> > > > .../net/ethernet/stmicro/stmmac/dwxgmac2_core.c | 3 +-
> > > > drivers/net/ethernet/stmicro/stmmac/hwif.h | 4 +-
> > > > drivers/net/ethernet/stmicro/stmmac/stmmac_main.c | 8 +++-
> > > > drivers/net/ethernet/stmicro/stmmac/stmmac_tc.c | 1 +
> > > > include/linux/stmmac.h | 1 +
> > > > 7 files changed, 36 insertions(+), 30 deletions(-)
> > > >
> > > > diff --git a/drivers/net/ethernet/stmicro/stmmac/dwmac5.c b/drivers/net/ethernet/stmicro/stmmac/dwmac5.c
> > > > index e95d35f..8fd1675 100644
> > > > --- a/drivers/net/ethernet/stmicro/stmmac/dwmac5.c
> > > > +++ b/drivers/net/ethernet/stmicro/stmmac/dwmac5.c
> > > > @@ -710,28 +710,22 @@ void dwmac5_est_irq_status(void __iomem *ioaddr, struct net_device *dev,
> > > > }
> > > > }
> > > >
> > > > -void dwmac5_fpe_configure(void __iomem *ioaddr, u32 num_txq, u32 num_rxq,
> > > > +void dwmac5_fpe_configure(void __iomem *ioaddr, struct stmmac_fpe_cfg *cfg,
> > > > + u32 num_txq, u32 num_rxq,
> > > > bool enable)
> > > > {
> > > > u32 value;
> > > >
> > > > - if (!enable) {
> > > > - value = readl(ioaddr + MAC_FPE_CTRL_STS);
> > > > -
> > > > - value &= ~EFPE;
> > > > -
> > > > - writel(value, ioaddr + MAC_FPE_CTRL_STS);
> > > > - return;
> > > > + if (enable) {
> > > > + cfg->fpe_csr = EFPE;
> > > > + value = readl(ioaddr + GMAC_RXQ_CTRL1);
> > > > + value &= ~GMAC_RXQCTRL_FPRQ;
> > > > + value |= (num_rxq - 1) << GMAC_RXQCTRL_FPRQ_SHIFT;
> > > > + writel(value, ioaddr + GMAC_RXQ_CTRL1);
> > > > + } else {
> > > > + cfg->fpe_csr = 0;
> > > > }
> > > > -
> > > > - value = readl(ioaddr + GMAC_RXQ_CTRL1);
> > > > - value &= ~GMAC_RXQCTRL_FPRQ;
> > > > - value |= (num_rxq - 1) << GMAC_RXQCTRL_FPRQ_SHIFT;
> > > > - writel(value, ioaddr + GMAC_RXQ_CTRL1);
> > > > -
> > > > - value = readl(ioaddr + MAC_FPE_CTRL_STS);
> > > > - value |= EFPE;
> > > > - writel(value, ioaddr + MAC_FPE_CTRL_STS);
> > > > + writel(cfg->fpe_csr, ioaddr + MAC_FPE_CTRL_STS);
> > > > }
> > > >
> > > > int dwmac5_fpe_irq_status(void __iomem *ioaddr, struct net_device *dev)
> > > > @@ -741,6 +735,9 @@ int dwmac5_fpe_irq_status(void __iomem *ioaddr, struct net_device *dev)
> > > >
> > > > status = FPE_EVENT_UNKNOWN;
> > > >
> > > > + /* Reads from the MAC_FPE_CTRL_STS register should only be performed
> > > > + * here, since the status flags of MAC_FPE_CTRL_STS are "clear on read"
> > > > + */
> > > > value = readl(ioaddr + MAC_FPE_CTRL_STS);
> > > >
> > > > if (value & TRSP) {
> > > > @@ -766,19 +763,15 @@ int dwmac5_fpe_irq_status(void __iomem *ioaddr, struct net_device *dev)
> > > > return status;
> > > > }
> > > >
> > > > -void dwmac5_fpe_send_mpacket(void __iomem *ioaddr, enum stmmac_mpacket_type type)
> > > > +void dwmac5_fpe_send_mpacket(void __iomem *ioaddr, struct stmmac_fpe_cfg *cfg,
> > > > + enum stmmac_mpacket_type type)
> > > > {
> > > > - u32 value;
> > > > + u32 value = cfg->fpe_csr;
> > > >
> > > > - value = readl(ioaddr + MAC_FPE_CTRL_STS);
> > > > -
> > > > - if (type == MPACKET_VERIFY) {
> > > > - value &= ~SRSP;
> > > > + if (type == MPACKET_VERIFY)
> > > > value |= SVER;
> > > > - } else {
> > > > - value &= ~SVER;
> > > > + else if (type == MPACKET_RESPONSE)
> > > > value |= SRSP;
> > > > - }
> > > >
> > > > writel(value, ioaddr + MAC_FPE_CTRL_STS);
> > > > }
> > >
> >
> > > It's unclear to me why it's not necessary to preserve the SVER/SRSP
> > > bits across MAC_FPE_CTRL_STS writes. I guess they are not part of the
> > > status bits? perhaps an explicit comment somewhere will help?
> >
> > The SRSP and SVER are self-cleared flags with no effect on zero
> > writing. Their responsibility is to emit the Respond and Verify
> > mPackets respectively. As soon as the packets are sent, the flags will
> > be reset by hardware automatically. So no, they aren't a part of the
> > status bits.
> >
> > Note since 'value' now isn't read from the MAC_FPE_CTRL_STS register,
> > there is no point in clearing up these flags in the local variable
> > because 'value' has now them cleared by default.
> >
> > Not sure whether a comment about that is required, since the described
> > behavior is well documented in the Synopsys HW-manual.
>
> Thanks for the explanation, it clarifies the things to me. I agree
> there is no need for a patch change.

I'm sorry, I have to take back the last sentence: the submitter and SoB
email address still don't match. @Jianheng: please fix it for good at
the next iteration.

Cheers,

Paolo