Re: [PATCH net-next v3] net: stmmac: program the FSM timeout timer
From: zhu xin
Date: Tue Nov 17 2026 - 08:20:14 EST
Hi Maxime,
On Tue, Nov 17, 2026 at 6:23 PM Maxime Chevallier
<maxime.chevallier@xxxxxxxxxxx> wrote:
>
> Hi,
>
> On 11/16/26 12:49, Zxyan Zhu wrote:
> > stmmac enables the FSM timeout checker (TMOUTEN) but never programs the
> > MAC_FSM_ACT_Timer register. Per the databook the checker only fires once
> > NTMRMD is programmed with a non-zero duration; NTMRMD == 0 disables the
> > timer, so with TMOUTEN set but the timer left at reset a stuck FSM is
> > never reported and the advertised timeout protection is ineffective.
> >
> > Program MAC_FSM_ACT_Timer right after TMOUTEN is enabled, deriving the
> > TMR field (CSR clocks per 1us tic) from the CSR clock rate and defaulting
> > to a 1s timeout. A platform can override the default through the new
> > ntmrmd field in stmmac_safety_feature_cfg.
> >
> > The NTMRMD field holds a power-of-two microsecond timeout per the
> > databook: 0x0 disables it, and 0x1..0xa encode 1us, 1.024ms (~1ms),
> > 16.384ms (~16ms), 65.536ms (~64ms), 262.144ms (~256ms), 1.048s (~1s),
> > 4.194s (~4s), 16.777s (~16s), 33.554s (~32s), 67.108s (~64s).
> >
> > The issue was hit in testing on dwmac4 and dwxgmac, where a stuck FSM was
> > not reported through the safety interrupt despite TMOUTEN being set.
> >
> > Signed-off-by: Zxyan Zhu <zxyan0222@xxxxxxxxx>
> > ---
> > v3:
> > - Make the MAC_FSM_ACT_Timer write a read-modify-write so the other
> > fields (e.g. LTMRMD) are preserved.
> > - Warn when a valid CSR clock rate cannot be obtained.
> > - Document the NTMRMD encodings and fix 0x2 to 1ms (was 4ms).
> >
> > v2: https://lore.kernel.org/netdev/20261112053321.1103678-1-zxyan0222@xxxxxxxxx/
> > v1: https://lore.kernel.org/netdev/20261111143322.2234321-1-zxyan0222@xxxxxxxxx/
> >
> > drivers/net/ethernet/stmicro/stmmac/dwmac5.c | 22 +++++++++++++++++-
> > drivers/net/ethernet/stmicro/stmmac/dwmac5.h | 6 ++++-
> > .../net/ethernet/stmicro/stmmac/dwxgmac2.h | 3 +++
> > .../ethernet/stmicro/stmmac/dwxgmac2_core.c | 23 ++++++++++++++++++-
> > drivers/net/ethernet/stmicro/stmmac/hwif.h | 3 ++-
> > .../net/ethernet/stmicro/stmmac/stmmac_main.c | 11 ++++++++-
> > include/linux/stmmac.h | 17 ++++++++++++++
> > 7 files changed, 80 insertions(+), 5 deletions(-)
> >
> > diff --git a/drivers/net/ethernet/stmicro/stmmac/dwmac5.c b/drivers/net/ethernet/stmicro/stmmac/dwmac5.c
> > index f1bb981cab7c..c1e3d634f153 100644
> > --- a/drivers/net/ethernet/stmicro/stmmac/dwmac5.c
> > +++ b/drivers/net/ethernet/stmicro/stmmac/dwmac5.c
> > @@ -2,6 +2,7 @@
> > // Copyright (c) 2017 Synopsys, Inc. and/or its affiliates.
> > // stmmac Support for 5.xx Ethernet QoS cores
> >
> > +#include <linux/bitfield.h>
> > #include <linux/bitops.h>
> > #include <linux/iopoll.h>
> > #include "common.h"
> > @@ -184,7 +185,8 @@ static void dwmac5_handle_dma_err(struct net_device *ndev,
> > }
> >
> > int dwmac5_safety_feat_config(void __iomem *ioaddr, unsigned int asp,
> > - struct stmmac_safety_feature_cfg *safety_feat_cfg)
> > + struct stmmac_safety_feature_cfg *safety_feat_cfg,
> > + unsigned int csr_clk_mhz)
> > {
> > struct stmmac_safety_feature_cfg all_safety_feats = {
> > .tsoee = 1,
> > @@ -245,6 +247,24 @@ int dwmac5_safety_feat_config(void __iomem *ioaddr, unsigned int asp,
> > value |= TMOUTEN; /* FSM Timeout Feature */
> > writel(value, ioaddr + MAC_FSM_CONTROL);
> >
> > + /* Program the FSM active timer so the timeout checker actually fires.
> > + * TMR is the number of CSR clocks per 1us tic, i.e. csr_clk in MHz.
> > + * Default to a 1s timeout when the platform did not select one.
> > + */
> > + if (safety_feat_cfg->tmouten && csr_clk_mhz) {
> > + u32 ntmrmd = safety_feat_cfg->ntmrmd;
> > +
> > + ntmrmd = (ntmrmd >= STMMAC_NTMRMD_1US &&
> > + ntmrmd <= STMMAC_NTMRMD_64SEC) ?
> > + ntmrmd : STMMAC_NTMRMD_1SEC;
> > +
> > + value = readl(ioaddr + MAC_FSM_ACT_TIMER);
> > + value &= ~(MAC_FSM_ACT_TIMER_NTMRMD | MAC_FSM_ACT_TIMER_TMR);
> > + value |= FIELD_PREP(MAC_FSM_ACT_TIMER_NTMRMD, ntmrmd) |
> > + FIELD_PREP(MAC_FSM_ACT_TIMER_TMR, csr_clk_mhz);
> > + writel(value, ioaddr + MAC_FSM_ACT_TIMER);
> > + }
> > +
> > /* 4. Enable Data Parity Protection */
> > value = readl(ioaddr + MTL_DPP_CONTROL);
> > if (safety_feat_cfg->edpp)
> > diff --git a/drivers/net/ethernet/stmicro/stmmac/dwmac5.h b/drivers/net/ethernet/stmicro/stmmac/dwmac5.h
> > index 00b151b3b688..734e7630f28d 100644
> > --- a/drivers/net/ethernet/stmicro/stmmac/dwmac5.h
> > +++ b/drivers/net/ethernet/stmicro/stmmac/dwmac5.h
> > @@ -8,8 +8,11 @@
> > #define MAC_DPP_FSM_INT_STATUS 0x00000140
> > #define MAC_AXI_SLV_DPE_ADDR_STATUS 0x00000144
> > #define MAC_FSM_CONTROL 0x00000148
> > +#define MAC_FSM_ACT_TIMER 0x0000014C
> > #define PRTYEN BIT(1)
> > #define TMOUTEN BIT(0)
> > +#define MAC_FSM_ACT_TIMER_TMR GENMASK(9, 0)
> > +#define MAC_FSM_ACT_TIMER_NTMRMD GENMASK(19, 16)
> >
> > #define MAC_PPS_CONTROL 0x00000b70
> > #define PPS_MAXIDX(x) ((((x) + 1) * 8) - 1)
> > @@ -82,7 +85,8 @@
> > #define GMAC_INT_FPE_EN BIT(17)
> >
> > int dwmac5_safety_feat_config(void __iomem *ioaddr, unsigned int asp,
> > - struct stmmac_safety_feature_cfg *safety_cfg);
> > + struct stmmac_safety_feature_cfg *safety_cfg,
> > + unsigned int csr_clk_mhz);
> > int dwmac5_safety_feat_irq_status(struct net_device *ndev,
> > void __iomem *ioaddr, unsigned int asp,
> > struct stmmac_safety_stats *stats);
> > diff --git a/drivers/net/ethernet/stmicro/stmmac/dwxgmac2.h b/drivers/net/ethernet/stmicro/stmmac/dwxgmac2.h
> > index f8ab347f7b5b..1fd501728b18 100644
> > --- a/drivers/net/ethernet/stmicro/stmmac/dwxgmac2.h
> > +++ b/drivers/net/ethernet/stmicro/stmmac/dwxgmac2.h
> > @@ -169,8 +169,11 @@
> > #define XGMAC_HWFEAT_PCSEL GENMASK(1, 0)
> > #define XGMAC_MAC_DPP_FSM_INT_STATUS 0x00000150
> > #define XGMAC_MAC_FSM_CONTROL 0x00000158
> > +#define XGMAC_MAC_FSM_ACT_TIMER 0x0000015C
> > #define XGMAC_PRTYEN BIT(1)
> > #define XGMAC_TMOUTEN BIT(0)
> > +#define XGMAC_MAC_FSM_ACT_TIMER_TMR GENMASK(9, 0)
> > +#define XGMAC_MAC_FSM_ACT_TIMER_NTMRMD GENMASK(19, 16)
> > #define XGMAC_MDIO_ADDR 0x00000200
> > #define XGMAC_MDIO_DATA 0x00000204
> > #define XGMAC_MDIO_C22P 0x00000220
> > diff --git a/drivers/net/ethernet/stmicro/stmmac/dwxgmac2_core.c b/drivers/net/ethernet/stmicro/stmmac/dwxgmac2_core.c
> > index 7ad08651675e..b3f99a6dfcd7 100644
> > --- a/drivers/net/ethernet/stmicro/stmmac/dwxgmac2_core.c
> > +++ b/drivers/net/ethernet/stmicro/stmmac/dwxgmac2_core.c
> > @@ -4,6 +4,7 @@
> > * stmmac XGMAC support.
> > */
> >
> > +#include <linux/bitfield.h>
> > #include <linux/bitrev.h>
> > #include <linux/crc32.h>
> > #include <linux/iopoll.h>
> > @@ -837,7 +838,8 @@ static void dwxgmac3_handle_dma_err(struct net_device *ndev,
> >
> > static int
> > dwxgmac3_safety_feat_config(void __iomem *ioaddr, unsigned int asp,
> > - struct stmmac_safety_feature_cfg *safety_cfg)
> > + struct stmmac_safety_feature_cfg *safety_cfg,
> > + unsigned int csr_clk_mhz)
> > {
> > u32 value;
> >
> > @@ -873,6 +875,25 @@ dwxgmac3_safety_feat_config(void __iomem *ioaddr, unsigned int asp,
> > value |= XGMAC_TMOUTEN; /* FSM Timeout Feature */
> > writel(value, ioaddr + XGMAC_MAC_FSM_CONTROL);
> >
> > + /* Program the FSM active timer so the timeout checker actually fires.
> > + * TMR is the number of CSR clocks per 1us tic, i.e. csr_clk in MHz.
> > + * Default to a 1s timeout when the platform did not select one.
> > + */
> > + if (safety_cfg && safety_cfg->tmouten && csr_clk_mhz) {
> > + u32 ntmrmd = safety_cfg->ntmrmd;
> > +
> > + ntmrmd = (ntmrmd >= STMMAC_NTMRMD_1US &&
> > + ntmrmd <= STMMAC_NTMRMD_64SEC) ?
> > + ntmrmd : STMMAC_NTMRMD_1SEC;
> > +
> > + value = readl(ioaddr + XGMAC_MAC_FSM_ACT_TIMER);
> > + value &= ~(XGMAC_MAC_FSM_ACT_TIMER_NTMRMD |
> > + XGMAC_MAC_FSM_ACT_TIMER_TMR);
> > + value |= FIELD_PREP(XGMAC_MAC_FSM_ACT_TIMER_NTMRMD, ntmrmd) |
> > + FIELD_PREP(XGMAC_MAC_FSM_ACT_TIMER_TMR, csr_clk_mhz);
> > + writel(value, ioaddr + XGMAC_MAC_FSM_ACT_TIMER);
> > + }
> > +
> > /* 5. Enable Data Path Parity Protection */
> > value = readl(ioaddr + XGMAC_MTL_DPP_CONTROL);
> > /* already enabled by default, explicit enable it again */
> > diff --git a/drivers/net/ethernet/stmicro/stmmac/hwif.h b/drivers/net/ethernet/stmicro/stmmac/hwif.h
> > index 8ae95e91a36a..395655b01ea8 100644
> > --- a/drivers/net/ethernet/stmicro/stmmac/hwif.h
> > +++ b/drivers/net/ethernet/stmicro/stmmac/hwif.h
> > @@ -378,7 +378,8 @@ struct stmmac_ops {
> > bool srgmi_ral);
> > /* Safety Features */
> > int (*safety_feat_config)(void __iomem *ioaddr, unsigned int asp,
> > - struct stmmac_safety_feature_cfg *safety_cfg);
> > + struct stmmac_safety_feature_cfg *safety_cfg,
> > + unsigned int csr_clk_mhz);
> > int (*safety_feat_irq_status)(struct net_device *ndev,
> > void __iomem *ioaddr, unsigned int asp,
> > struct stmmac_safety_stats *stats);
> > diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
> > index 552840d29668..949ed3e5bb05 100644
> > --- a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
> > +++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c
> > @@ -3691,10 +3691,19 @@ static void stmmac_mtl_configuration(struct stmmac_priv *priv)
> >
> > static void stmmac_safety_feat_configuration(struct stmmac_priv *priv)
> > {
> > + unsigned int csr_clk_mhz = 0;
> > +
> > if (priv->dma_cap.asp) {
> > netdev_info(priv->dev, "Enabling Safety Features\n");
> > + if (priv->plat->stmmac_clk)
> > + csr_clk_mhz = clk_get_rate(priv->plat->stmmac_clk) /
> > + 1000000;
> > + if (!csr_clk_mhz)
> > + netdev_warn(priv->dev,
> > + "unable to get a valid CSR clock rate\n");
>
> I know it's the bot who asked for that, but I don't agree with this log here.
>
> FWIU it's legit to have ASP but no CSR clock (e.g. stmmac PCI)
Agreed, I'll drop the warning.
>
> > stmmac_safety_feat_config(priv, priv->ioaddr, priv->dma_cap.asp,
> > - priv->plat->safety_feat_cfg);
> > + priv->plat->safety_feat_cfg,
> > + csr_clk_mhz);
> > } else {
> > netdev_info(priv->dev, "No Safety Features support found\n");
> > }
> > diff --git a/include/linux/stmmac.h b/include/linux/stmmac.h
> > index 00be2df63d22..35c39cd6b630 100644
> > --- a/include/linux/stmmac.h
> > +++ b/include/linux/stmmac.h
> > @@ -168,8 +168,25 @@ struct stmmac_safety_feature_cfg {
> > u32 edpp;
> > u32 prtyen;
> > u32 tmouten;
> > + /* FSM normal-mode timeout value (MAC_FSM_ACT_Timer.NTMRMD). The TMR
> > + * field (CSR clocks per 1us) is derived from the CSR clock by stmmac.
> > + * 0 selects the default 1s timeout.
> > + */
>
> I'd rather this comment to be with the register value definitions :)
>
> > + u32 ntmrmd;
>
> I don't see any user for ntmrmd. If this is prep work for upcoming glue
> drivers, then this should come with the driver in question. If really
> needed, can't we just program a reasonable hardcoded default for the
> time being ?
Right, ntmrmd was indeed added for upcoming glue drivers and has no
in-tree user yet. I'll drop the field and just program a hardcoded default
for now.
What default would you prefer? The databook doesn't suggest a particular
value for NTMRMD. 1s (NTMRMD = 0x6) is my choice because FSM
transactions complete in microseconds, so 1s is comfortably above that
to avoid false timeouts, and it sits in the middle of the supported
range. I'm happy to use whatever you deem appropriate.
>
> Maxime
Thanks,
Zxyan