Re: [PATCH v2 0/5] PCI: dwc: improve msi handling

From: Jon Hunter
Date: Fri Sep 25 2020 - 11:13:24 EST

Hi Jisheng,

On 25/09/2020 10:27, Jisheng Zhang wrote:

...

>> Could you please try below patch?
>>
>>
>> diff --git a/drivers/pci/controller/dwc/pcie-designware-host.c b/drivers/pci/controller/dwc/pcie-designware-host.c
>> index bf25d783b5c5..7e5dc54d060e 100644
>> --- a/drivers/pci/controller/dwc/pcie-designware-host.c
>> +++ b/drivers/pci/controller/dwc/pcie-designware-host.c
>> @@ -197,7 +197,6 @@ static struct irq_chip dw_pci_msi_bottom_irq_chip = {
>> .name = "DWPCI-MSI",
>> .irq_ack = dw_pci_bottom_ack,
>> .irq_compose_msi_msg = dw_pci_setup_msi_msg,
>> - .irq_set_affinity = dw_pci_msi_set_affinity,
>> .irq_mask = dw_pci_bottom_mask,
>> .irq_unmask = dw_pci_bottom_unmask,
>> };
>
> A complete patch w/o compiler warning:
>
> diff --git a/drivers/pci/controller/dwc/pcie-designware-host.c b/drivers/pci/controller/dwc/pcie-designware-host.c
> index bf25d783b5c5..18f719cfed0b 100644
> --- a/drivers/pci/controller/dwc/pcie-designware-host.c
> +++ b/drivers/pci/controller/dwc/pcie-designware-host.c
> @@ -137,12 +137,6 @@ static void dw_pci_setup_msi_msg(struct irq_data *d, struct msi_msg *msg)
> (int)d->hwirq, msg->address_hi, msg->address_lo);
> }
>
> -static int dw_pci_msi_set_affinity(struct irq_data *d,
> - const struct cpumask *mask, bool force)
> -{
> - return -EINVAL;
> -}
> -
> static void dw_pci_bottom_mask(struct irq_data *d)
> {
> struct pcie_port *pp = irq_data_get_irq_chip_data(d);
> @@ -197,7 +191,6 @@ static struct irq_chip dw_pci_msi_bottom_irq_chip = {
> .name = "DWPCI-MSI",
> .irq_ack = dw_pci_bottom_ack,
> .irq_compose_msi_msg = dw_pci_setup_msi_msg,
> - .irq_set_affinity = dw_pci_msi_set_affinity,
> .irq_mask = dw_pci_bottom_mask,
> .irq_unmask = dw_pci_bottom_unmask,
> };
>


Thanks I was not expecting this to work because ...

int irq_do_set_affinity(struct irq_data *data, const struct cpumask *mask,
bool force)
{
struct irq_desc *desc = irq_data_to_desc(data);
struct irq_chip *chip = irq_data_get_irq_chip(data);
int ret;

if (!chip || !chip->irq_set_affinity)
return -EINVAL;

However, with your patch Tegra crashes on boot ...

[ 11.613853] Unable to handle kernel NULL pointer dereference at virtual address 0000000000000000
[ 11.622500] Mem abort info:
[ 11.622515] ESR = 0x86000004
[ 11.622524] EC = 0x21: IABT (current EL), IL = 32 bits
[ 11.622540] SET = 0, FnV = 0
[ 11.636544] EA = 0, S1PTW = 0
[ 11.636554] user pgtable: 4k pages, 48-bit VAs, pgdp=000000046a28e000
[ 11.636559] [0000000000000000] pgd=0000000000000000, p4d=0000000000000000
[ 11.652652] Internal error: Oops: 86000004 [#1] PREEMPT SMP
[ 11.652658] Modules linked in: pwm_tegra phy_tegra194_p2u crct10dif_ce lm90 pwm_fan tegra_bpmp_thermal pcie_tegra194 ip_tables x_tables ipv6
[ 11.670525] CPU: 3 PID: 138 Comm: kworker/3:3 Not tainted 5.9.0-rc4-dirty #12
[ 11.670534] Hardware name: NVIDIA Jetson AGX Xavier Developer Kit (DT)
[ 11.683967] Workqueue: events deferred_probe_work_func
[ 11.683974] pstate: 60c00089 (nZCv daIf +PAN +UAO BTYPE=--)
[ 11.683985] pc : 0x0
[ 11.696669] lr : msi_domain_set_affinity+0x44/0xc0
[ 11.696672] sp : ffff800012bcb390
[ 11.696680] x29: ffff800012bcb390 x28: ffff0003e3033c20
[ 11.709891] x27: ffff0003e76cfe58 x26: 0000000000000000
[ 11.709900] x25: ffff800011d7e850 x24: ffff800011d7e878
[ 11.709908] x23: 0000000000000000 x22: ffff0003e76cfe00
[ 11.709914] x21: ffff0003e76cfe58 x20: ffff0003e76cfe58
[ 11.709921] x19: ffff800011b19000 x18: ffffffffffffffff
[ 11.709927] x17: 0000000000000000 x16: 0000000000000000
[ 11.741262] x15: ffff800011b19948 x14: 0000000000000040
[ 11.741267] x13: 0000000000000228 x12: 0000000000000030
[ 11.741272] x11: 0101010101010101 x10: 0000000000000040
[ 11.741277] x9 : 0000000000000000 x8 : 0000000000000004
[ 11.741281] x7 : ffffffffffffffff x6 : 00000000000000ff
[ 11.767374] x5 : 0000000000000000 x4 : 0000000000000000
[ 11.767379] x3 : 0000000000000000 x2 : 0000000000000000
[ 11.767384] x1 : ffff800011d7e898 x0 : ffff0003e262bf00
[ 11.767406] Call trace:
[ 11.767410] 0x0
[ 11.767424] irq_do_set_affinity+0x4c/0x178
[ 11.791400] irq_setup_affinity+0x124/0x1b0
[ 11.791423] irq_startup+0x6c/0x118
[ 11.791434] __setup_irq+0x810/0x8a0
[ 11.802510] request_threaded_irq+0xdc/0x188
[ 11.802517] pcie_pme_probe+0x98/0x110
[ 11.802536] pcie_port_probe_service+0x34/0x60
[ 11.814799] really_probe+0x110/0x400
[ 11.814809] driver_probe_device+0x54/0xb8
[ 11.822438] __device_attach_driver+0x90/0xc0
[ 11.822463] bus_for_each_drv+0x70/0xc8
[ 11.822471] __device_attach+0xec/0x150
[ 11.834307] device_initial_probe+0x10/0x18
[ 11.834311] bus_probe_device+0x94/0xa0
[ 11.834315] device_add+0x464/0x730
[ 11.834338] device_register+0x1c/0x28
[ 11.834349] pcie_port_device_register+0x2d0/0x3e8
[ 11.854056] pcie_portdrv_probe+0x34/0xd8
[ 11.854063] local_pci_probe+0x3c/0xa0
[ 11.854088] pci_device_probe+0x128/0x1c8
[ 11.854103] really_probe+0x110/0x400
[ 11.869283] driver_probe_device+0x54/0xb8
[ 11.869311] __device_attach_driver+0x90/0xc0
[ 11.877638] bus_for_each_drv+0x70/0xc8
[ 11.877645] __device_attach+0xec/0x150
[ 11.877669] device_attach+0x10/0x18
[ 11.877680] pci_bus_add_device+0x4c/0xb0
[ 11.892642] pci_bus_add_devices+0x44/0x90
[ 11.892646] dw_pcie_host_init+0x370/0x4f8
[ 11.892653] tegra_pcie_dw_probe+0x5e8/0xb50 [pcie_tegra194]
[ 11.892661] platform_drv_probe+0x50/0xa8
[ 11.910179] really_probe+0x110/0x400
[ 11.910183] driver_probe_device+0x54/0xb8
[ 11.910186] __device_attach_driver+0x90/0xc0
[ 11.910213] bus_for_each_drv+0x70/0xc8
[ 11.910240] __device_attach+0xec/0x150
[ 11.929689] device_initial_probe+0x10/0x18
[ 11.929694] bus_probe_device+0x94/0xa0
[ 11.929719] deferred_probe_work_func+0x6c/0xa0
[ 11.929730] process_one_work+0x1cc/0x360
[ 11.946008] worker_thread+0x48/0x450
[ 11.949602] kthread+0x120/0x150
[ 11.952803] ret_from_fork+0x10/0x1c
[ 11.956332] Code: bad PC value
[ 11.959360] ---[ end trace 03c30e252fe4e40b ]---

To be honest, I am not sure I completely understand why it crashes here.

Cheers
Jon

--
nvpublic