Re: [PATCH v16 0/9] Linux RISC-V AIA Support

From: Dawei Li
Date: Thu Mar 14 2024 - 01:48:28 EST


Hi Anup,

Thanks for the great work on AIA!

Firstly I must apologize for the top posting cuz I miss the series mail for
my inbox. And for the record, I am replying to [1].

After I checkout riscv_aia_v16 branch at https://github.com/avpatel/linux.git,
some call traces about spurious interrupt are screaming:

[ 1.244292] irq 11: nobody cared (try booting with the "irqpoll" option)
[ 1.244707] CPU: 0 PID: 0 Comm: swapper/0 Not tainted 6.8.0-rc6-00055-gd7d4d086b79a #14
[ 1.244932] Hardware name: riscv-virtio,qemu (DT)
[ 1.245130] Call Trace:
[ 1.245337] [<ffffffff80005f32>] dump_backtrace+0x1c/0x24
[ 1.245559] [<ffffffff809023f4>] show_stack+0x2c/0x38
[ 1.245761] [<ffffffff8090ec8c>] dump_stack_lvl+0x3c/0x54
[ 1.245979] [<ffffffff8090ecb8>] dump_stack+0x14/0x1c
[ 1.246116] [<ffffffff8090368c>] __report_bad_irq+0x3e/0xae
[ 1.246276] [<ffffffff8006a938>] note_interrupt+0x1f4/0x23e
[ 1.246382] [<ffffffff800679e2>] handle_irq_event_percpu+0x42/0x50
[ 1.246508] [<ffffffff80067a2c>] handle_irq_event+0x3c/0x68
[ 1.246628] [<ffffffff8006b7a4>] handle_fasteoi_irq+0xac/0x176
[ 1.246788] [<ffffffff80066c3c>] generic_handle_domain_irq+0x1c/0x2a
[ 1.246928] [<ffffffff80473f0e>] imsic_handle_irq+0x72/0x12a
[ 1.247041] [<ffffffff80066c3c>] generic_handle_domain_irq+0x1c/0x2a
[ 1.247165] [<ffffffff80472470>] riscv_intc_aia_irq+0x26/0x38
[ 1.247291] [<ffffffff8090ed0a>] handle_riscv_irq+0x4a/0x74
[ 1.247430] [<ffffffff80918af6>] call_on_irq_stack+0x32/0x40
[ 1.247626] handlers:
[ 1.247721] [<000000004ebb6fe3>] vm_interrupt
[ 1.248314] Disabling IRQ #11
[ 1.303510] irq 12: nobody cared (try booting with the "irqpoll" option)
[ 1.303700] CPU: 1 PID: 0 Comm: swapper/1 Not tainted 6.8.0-rc6-00055-gd7d4d086b79a #14
[ 1.304768] Hardware name: riscv-virtio,qemu (DT)
[ 1.305460] Call Trace:
[ 1.305868] [<ffffffff80005f32>] dump_backtrace+0x1c/0x24
[ 1.306621] [<ffffffff809023f4>] show_stack+0x2c/0x38
[ 1.307295] [<ffffffff8090ec8c>] dump_stack_lvl+0x3c/0x54
[ 1.308025] [<ffffffff8090ecb8>] dump_stack+0x14/0x1c
[ 1.308698] [<ffffffff8090368c>] __report_bad_irq+0x3e/0xae
[ 1.309698] [<ffffffff8006a938>] note_interrupt+0x1f4/0x23e
[ 1.310568] [<ffffffff800679e2>] handle_irq_event_percpu+0x42/0x50
[ 1.311542] [<ffffffff80067a2c>] handle_irq_event+0x3c/0x68
[ 1.312298] [<ffffffff8006b7a4>] handle_fasteoi_irq+0xac/0x176
[ 1.313127] [<ffffffff80066c3c>] generic_handle_domain_irq+0x1c/0x2a
[ 1.315375] [<ffffffff80473f0e>] imsic_handle_irq+0x72/0x12a
[ 1.316163] [<ffffffff80066c3c>] generic_handle_domain_irq+0x1c/0x2a
[ 1.317078] [<ffffffff80472470>] riscv_intc_aia_irq+0x26/0x38
[ 1.318036] [<ffffffff8090ed0a>] handle_riscv_irq+0x4a/0x74
[ 1.318934] [<ffffffff8090f3e6>] do_irq+0x4a/0x4c
[ 1.319831] [<ffffffff809189ac>] ret_from_exception+0x0/0x64
[ 1.320860] [<ffffffff8091845c>] _raw_spin_unlock_irqrestore+0x1a/0x2e
[ 1.321887] [<ffffffff80473d06>] imsic_local_timer_callback+0x50/0x5c
[ 1.322957] [<ffffffff8008ade4>] call_timer_fn.isra.0+0x14/0x5e
[ 1.323781] [<ffffffff8008b44a>] run_timer_softirq+0x388/0x3a8
[ 1.324579] [<ffffffff80918c7e>] __do_softirq+0x100/0x252
[ 1.325304] [<ffffffff80018464>] irq_exit_rcu+0x7c/0xa6
[ 1.326126] [<ffffffff8090ed24>] handle_riscv_irq+0x64/0x74
[ 1.327535] [<ffffffff80918af6>] call_on_irq_stack+0x32/0x40
[ 1.328703] handlers:
[ 1.330724] [<000000008b6c4f15>] serial8250_interrupt
[ 1.331573] Disabling IRQ #12
[ 1.350231] Run /sbin/init as init process

And I take a look on /proc/interrupts, spurious interrupts are happening:
~ # cat /proc/interrupts
CPU0 CPU1 CPU2 CPU3 CPU4 CPU5 CPU6 CPU7 CPU8 CPU9 CPU10 CPU11 CPU12 CPU13 CPU14 CPU15
10: 83 222 350 163 71 101 193 68 54 60 174 50 51 48 48 46 RISC-V INTC 5 Edge riscv-timer
11: 100001 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 APLIC-MSI-d000000.aplic 8 Level -fasteoi virtio0
12: 0 100000 0 0 0 0 0 0 0 0 0 0 0 0 0 0 APLIC-MSI-d000000.aplic 10 Level -fasteoi ttyS0
13: 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 APLIC-MSI-d000000.aplic 11 Level -fasteoi 101000.rtc
IPI0: 3 8 4 7 7 6 5 4 7 7 7 5 6 10 4 5 Rescheduling interrupts
IPI1: 727 517 431 448 497 456 514 476 488 466 455 456 443 443 439 433 Function call interrupts
IPI2: 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 CPU stop interrupts
IPI3: 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 CPU stop (for crash dump) interrupts
IPI4: 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 IRQ work interrupts
IPI5: 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 Timer broadcast interrupts

But when I switched back to V15, everything is fine.

Here are some of info _maybe_ helpful:
kernel: d7d4d086b79a11ad8bbcaf732c7830faa3f3192f
qemu: e1007b6bab5cf97705bf4f2aaec1f607787355b8
qemu cmdline: qemu-system-riscv64 -M virt,aia=aplic-imsic -m 256M -nographic -kernel ~/src/linux/arch/riscv/boot/Image -drive file=rootfs.img,format=raw,id=hd0 -device virtio-blk-device,drive=hd0 -append "root=/dev/vda rw console=ttyS0" -smp 16
config: defconfig(make ARCH=riscv defconfig)

And I am happy to provide any extra info if needed.

After I take a git diff and check your changelog:
[Quoting from [1]]
- Simplified aplic_msi_irq_eoi() in PATCH7 based on the section "4.9.2 Special
consideration for level-sensitive interrupt sources" of the RISC-V AIA
specification and also provided handler name.

I apologize if I miss something.

[1] https://lore.kernel.org/lkml/20240307140307.646078-1-apatel@xxxxxxxxxxxxxxxx/

Thanks,

Dawei