Re: [PATCH v2] mfd: loongson-se: Add multi-node support
From: Huacai Chen
Date: Mon Apr 27 2026 - 05:37:30 EST
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.
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
> >>
> >>
>
>