Re: [PATCH v2] mfd: loongson-se: Add multi-node support

[Huacai Chen]

On Mon, Apr 27, 2026 at 5:52 PM Qunqin Zhao <zhaoqunqin@xxxxxxxxxxx> wrote:
>
>
> 在 2026/4/27 下午5:37, Huacai Chen 写道:
> > On Mon, Apr 27, 2026 at 5:24 PM Qunqin Zhao <zhaoqunqin@xxxxxxxxxxx> wrote:
> >>
> >> 在 2026/4/27 下午5:02, Huacai Chen 写道:
> >>> Hi, Qunqin,
> >>>
> >>> On Mon, Apr 27, 2026 at 4:55 PM Qunqin Zhao <zhaoqunqin@xxxxxxxxxxx> wrote:
> >>>> On the Loongson platform, each node is equipped with a security engine
> >>>> device. However, due to a hardware flaw, only the device on node 0 can
> >>>> trigger interrupts. Therefore, interrupts from other nodes are forwarded
> >>>> by node 0. We need to check in the interrupt handler of node 0 whether
> >>>> this interrupt is intended for other nodes.
> >>> Multi-node or multi-package? In my opinion SE has no relationship with
> >>> NUMA node, so maybe package?
> >> Here is the output of lscpu from my machine:
> >>
> >> [loongson@localhost ~]$ lscpu
> >> Architecture:          loongarch64
> >>     CPU op-mode(s):      32-bit, 64-bit
> >>     Address sizes:       48 bits physical, 48 bits virtual
> >>     Byte Order:          Little Endian
> >> CPU(s):                128
> >>     On-line CPU(s) list: 0-127
> >> Model name:            Loongson-3C6000/D
> >>     CPU family:          Loongson-64bit
> >>     Model:               0x11
> >>     Thread(s) per core:  2
> >>     Core(s) per socket:  32
> >>     Socket(s):           2
> >>     BogoMIPS:            4200.00
> >>     Flags:               cpucfg lam ual fpu lsx lasx crc32 complex crypto
> >> lvz lbt_x86 lbt_arm lbt_mips
> >> Caches (sum of all):
> >>     L1d:                 4 MiB (64 instances)
> >>     L1i:                 4 MiB (64 instances)
> >>     L2:                  16 MiB (64 instances)
> >>     L3:                  128 MiB (4 instances)
> >> NUMA:
> >>     NUMA node(s):        4
> >>     NUMA node0 CPU(s):   0-31
> >>     NUMA node1 CPU(s):   32-63
> >>     NUMA node2 CPU(s):   64-95
> >>     NUMA node3 CPU(s):   96-127
> >>
> >> There are four SE devices in my system, one for each NUMA node.
> > For Loongson-3C6000 node is the same as package. You should consider
> > Loongson-3C5000L, one package contains four nodes.
>
> I am not familiar with the SE-related components on the 3C5000L, and
> this driver is not compatible with the 5000 series.
Whether it is compatible to Loongson-3C5000L is not important. The
importance is package is not always equal to node, and we should
consider whether SE is per-node or per-package.

Huacai

>
> Qunqin,
>
> Thanks.
>
> >
> > Huacai
> >
> >> Qunqin,
> >>
> >> Thanks
> >>
> >>> Huacai
> >>>
> >>>> Signed-off-by: Qunqin Zhao <zhaoqunqin@xxxxxxxxxxx>
> >>>> ---
> >>>> Changes in v2:
> >>>>           -Resending due to no feedback for one month.
> >>>>           -Rebased on top of latest mainline (7.1-rc1) to ensure the patch
> >>>>            applies cleanly.
> >>>>           -No functional changes since the previous submission.
> >>>>
> >>>> Link to v1:
> >>>> https://lore.kernel.org/all/20260226102225.19516-1-zhaoqunqin@xxxxxxxxxxx/#t
> >>>>
> >>>>    drivers/mfd/loongson-se.c       | 38 +++++++++++++++++++++++++++------
> >>>>    include/linux/mfd/loongson-se.h |  3 +++
> >>>>    2 files changed, 35 insertions(+), 6 deletions(-)
> >>>>
> >>>> diff --git a/drivers/mfd/loongson-se.c b/drivers/mfd/loongson-se.c
> >>>> index 3902ba377d6..40e18c21268 100644
> >>>> --- a/drivers/mfd/loongson-se.c
> >>>> +++ b/drivers/mfd/loongson-se.c
> >>>> @@ -37,6 +37,9 @@ struct loongson_se_controller_cmd {
> >>>>           u32 info[7];
> >>>>    };
> >>>>
> >>>> +static DECLARE_COMPLETION(node0);
> >>>> +static struct loongson_se *se_node[SE_MAX_NODES];
> >>>> +
> >>>>    static int loongson_se_poll(struct loongson_se *se, u32 int_bit)
> >>>>    {
> >>>>           u32 status;
> >>>> @@ -133,8 +136,8 @@ EXPORT_SYMBOL_GPL(loongson_se_init_engine);
> >>>>    static irqreturn_t se_irq_handler(int irq, void *dev_id)
> >>>>    {
> >>>>           struct loongson_se *se = dev_id;
> >>>> -       u32 int_status;
> >>>> -       int id;
> >>>> +       u32 int_status, node_irq = 0;
> >>>> +       int id, node;
> >>>>
> >>>>           spin_lock(&se->dev_lock);
> >>>>
> >>>> @@ -147,6 +150,11 @@ static irqreturn_t se_irq_handler(int irq, void *dev_id)
> >>>>                   writel(SE_INT_CONTROLLER, se->base + SE_S2LINT_CL);
> >>>>           }
> >>>>
> >>>> +       if (int_status & SE_INT_OTHER_NODE) {
> >>>> +               int_status &= ~SE_INT_OTHER_NODE;
> >>>> +               node_irq = 1;
> >>>> +       }
> >>>> +
> >>>>           /* For engines */
> >>>>           while (int_status) {
> >>>>                   id = __ffs(int_status);
> >>>> @@ -157,6 +165,14 @@ static irqreturn_t se_irq_handler(int irq, void *dev_id)
> >>>>
> >>>>           spin_unlock(&se->dev_lock);
> >>>>
> >>>> +       if (node_irq) {
> >>>> +               writel(SE_INT_OTHER_NODE, se->base + SE_S2LINT_CL);
> >>>> +               for (node = 1; node < SE_MAX_NODES; node++) {
> >>>> +                       if (se_node[node])
> >>>> +                               se_irq_handler(irq, se_node[node]);
> >>>> +               }
> >>>> +       }
> >>>> +
> >>>>           return IRQ_HANDLED;
> >>>>    }
> >>>>
> >>>> @@ -189,6 +205,7 @@ static int loongson_se_probe(struct platform_device *pdev)
> >>>>           struct loongson_se *se;
> >>>>           int nr_irq, irq, err, i;
> >>>>           dma_addr_t paddr;
> >>>> +       int node = dev_to_node(dev);
> >>>>
> >>>>           se = devm_kmalloc(dev, sizeof(*se), GFP_KERNEL);
> >>>>           if (!se)
> >>>> @@ -213,9 +230,16 @@ static int loongson_se_probe(struct platform_device *pdev)
> >>>>
> >>>>           writel(SE_INT_ALL, se->base + SE_S2LINT_EN);
> >>>>
> >>>> -       nr_irq = platform_irq_count(pdev);
> >>>> -       if (nr_irq <= 0)
> >>>> -               return -ENODEV;
> >>>> +       if (node == 0 || node == NUMA_NO_NODE) {
> >>>> +               nr_irq = platform_irq_count(pdev);
> >>>> +               if (nr_irq <= 0)
> >>>> +                       return -ENODEV;
> >>>> +       } else {
> >>>> +               /* Only the device on node 0 can trigger interrupts */
> >>>> +               nr_irq = 0;
> >>>> +               wait_for_completion_interruptible(&node0);
> >>>> +               se_node[node] = se;
> >>>> +       }
> >>>>
> >>>>           for (i = 0; i < nr_irq; i++) {
> >>>>                   irq = platform_get_irq(pdev, i);
> >>>> @@ -228,7 +252,9 @@ static int loongson_se_probe(struct platform_device *pdev)
> >>>>           if (err)
> >>>>                   return err;
> >>>>
> >>>> -       return devm_mfd_add_devices(dev, PLATFORM_DEVID_NONE, engines,
> >>>> +       complete_all(&node0);
> >>>> +
> >>>> +       return devm_mfd_add_devices(dev, PLATFORM_DEVID_AUTO, engines,
> >>>>                                       ARRAY_SIZE(engines), NULL, 0, NULL);
> >>>>    }
> >>>>
> >>>> diff --git a/include/linux/mfd/loongson-se.h b/include/linux/mfd/loongson-se.h
> >>>> index 07afa0c2524..a80e06eb017 100644
> >>>> --- a/include/linux/mfd/loongson-se.h
> >>>> +++ b/include/linux/mfd/loongson-se.h
> >>>> @@ -20,6 +20,9 @@
> >>>>
> >>>>    #define SE_INT_ALL                     0xffffffff
> >>>>    #define SE_INT_CONTROLLER              BIT(0)
> >>>> +#define SE_INT_OTHER_NODE              BIT(31)
> >>>> +
> >>>> +#define SE_MAX_NODES                   8
> >>>>
> >>>>    #define SE_ENGINE_MAX                  16
> >>>>    #define SE_ENGINE_RNG                  1
> >>>>
> >>>> base-commit: 254f49634ee16a731174d2ae34bc50bd5f45e731
> >>>> --
> >>>> 2.47.2
> >>>>
> >>>>
> >>
>