[<prev] [next>] [<thread-prev] [day] [month] [year] [list]
Message-ID: <aXOdDgLT97N9MrJj@lizhi-Precision-Tower-5810>
Date: Fri, 23 Jan 2026 11:08:46 -0500
From: Frank Li <Frank.li@....com>
To: adrianhoyin.ng@...era.com
Cc: alexandre.belloni@...tlin.com, linux-i3c@...ts.infradead.org,
linux-kernel@...r.kernel.org
Subject: Re: [PATCH v2 2/2] i3c: dw-i3c-master: fix SIR reject bit mapping
for dynamic addresses
On Fri, Jan 23, 2026 at 03:35:12PM +0800, adrianhoyin.ng@...era.com wrote:
> From: Adrian Ng Ho Yin <adrianhoyin.ng@...era.com>
>
> The IBI_SIR_REQ_REJECT register is a 32-bit bitmap indexed by the
> dynamic address of each I3C slave. The DesignWare controller derives
> the bit index by folding the 7-bit dynamic address into a 5-bit value,
> using the sum of the lower 5 bits and the upper 2 bits, modulo 32.
>
> The current implementation incorrectly uses the device table index
> when updating the SIR reject mask, which can result in rejecting or
> accepting IBIs for the wrong device.
>
> Compute the SIR reject bit index directly from the dynamic address,
> as defined by the controller specification, and use it consistently
> when updating the reject mask.
>
> Signed-off-by: Adrian Ng Ho Yin <adrianhoyin.ng@...era.com>
> ---
> drivers/i3c/master/dw-i3c-master.c | 26 ++++++++++++++++++++++++--
> 1 file changed, 24 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/i3c/master/dw-i3c-master.c b/drivers/i3c/master/dw-i3c-master.c
> index 15101f41b9e6..ffb1edd32097 100644
> --- a/drivers/i3c/master/dw-i3c-master.c
> +++ b/drivers/i3c/master/dw-i3c-master.c
> @@ -205,6 +205,11 @@
> #define EXTENDED_CAPABILITY 0xe8
> #define SLAVE_CONFIG 0xec
>
> +#define DYN_ADDR_LO_BITS 5
> +#define DYN_ADDR_LO_MASK GENMASK(4, 0)
> +#define DYN_ADDR_HI_MASK GENMASK(6, 5)
> +#define IBI_SIR_BIT_MOD 32 /* 32-bit vector */
> +
> #define DW_I3C_DEV_NACK_RETRY_CNT_MAX 0x3
> #define DEV_ADDR_TABLE_DEV_NACK_RETRY_MASK GENMASK(30, 29)
> #define DEV_ADDR_TABLE_DYNAMIC_MASK GENMASK(23, 16)
> @@ -217,6 +222,7 @@
> #define DEV_ADDR_TABLE_DYNAMIC_ADDR(x) FIELD_PREP(DEV_ADDR_TABLE_DYNAMIC_MASK, x)
> #define DEV_ADDR_TABLE_STATIC_ADDR(x) FIELD_PREP(DEV_ADDR_TABLE_STATIC_MASK, x)
> #define DEV_ADDR_TABLE_LOC(start, idx) ((start) + ((idx) << 2))
> +#define DEV_ADDR_TABLE_GET_DYNAMIC_ADDR(x) FIELD_GET(DEV_ADDR_TABLE_DYNAMIC_MASK, x)
>
> #define I3C_BUS_SDR1_SCL_RATE 8000000
> #define I3C_BUS_SDR2_SCL_RATE 6000000
> @@ -264,6 +270,14 @@ struct dw_i3c_drvdata {
> u32 flags;
> };
>
> +static inline u32 get_ibi_sir_bit_index(u8 addr)
> +{
> + u32 lo = addr & DYN_ADDR_LO_MASK;
> + u32 hi = (addr & DYN_ADDR_HI_MASK) >> DYN_ADDR_LO_BITS;
Use GET_FIELD(), so needn't define DYN_ADDR_LO_BITS.
Frank
> +
> + return (lo + hi) % IBI_SIR_BIT_MOD;
> +}
> +
> static bool dw_i3c_master_supports_ccc_cmd(struct i3c_master_controller *m,
> const struct i3c_ccc_cmd *cmd)
> {
> @@ -1235,11 +1249,19 @@ static void dw_i3c_master_set_sir_enabled(struct dw_i3c_master *master,
> {
> u32 dat_entry, reg;
> bool global;
> + u8 dynamic_addr;
>
> dat_entry = DEV_ADDR_TABLE_LOC(master->datstartaddr, idx);
>
> guard(spinlock_irqsave)(&master->devs_lock);
> reg = readl(master->regs + dat_entry);
> + dynamic_addr = DEV_ADDR_TABLE_GET_DYNAMIC_ADDR(reg);
> +
> + if (!dynamic_addr)
> + dev_warn(master->dev,
> + "<%s> unassigned slave device, dynamic addr:%x\n",
> + __func__, dynamic_addr);
> +
> if (enable) {
> reg &= ~DEV_ADDR_TABLE_SIR_REJECT;
> if (dev->info.bcr & I3C_BCR_IBI_PAYLOAD)
> @@ -1252,11 +1274,11 @@ static void dw_i3c_master_set_sir_enabled(struct dw_i3c_master *master,
>
> if (enable) {
> global = (master->sir_rej_mask == IBI_REQ_REJECT_ALL);
> - master->sir_rej_mask &= ~BIT(idx);
> + master->sir_rej_mask &= ~BIT(get_ibi_sir_bit_index(dynamic_addr));
> } else {
> bool hj_rejected = !!(readl(master->regs + DEVICE_CTRL) & DEV_CTRL_HOT_JOIN_NACK);
>
> - master->sir_rej_mask |= BIT(idx);
> + master->sir_rej_mask |= BIT(get_ibi_sir_bit_index(dynamic_addr));
> global = (master->sir_rej_mask == IBI_REQ_REJECT_ALL) && hj_rejected;
> }
> writel(master->sir_rej_mask, master->regs + IBI_SIR_REQ_REJECT);
> --
> 2.49.GIT
>
Powered by blists - more mailing lists