[Qemu-devel] Disabling IRQ error

[Qemu-devel] Disabling IRQ error

[Xie Xianshan]

hi everyone,

I`m getting the nobody cared disabling IRQ error, when i raised external 
interrupts IRQ3 to the Openpic in QEMU.
(Actually, any external interrupts irq i raised can reproduce this 
error, but internal interrupts work fine)

And this IRQ3 is sharing irq with usb card.

I have tried to resolve this issue as follows, but nothing changed.
  1)tried to boot with "irqpoll" option.
  2)tried to stop raising the irq for usb card.
  3)tried to boot with "kernel_irqchip" option

Related log from dmesg:
[ 2079.800787] irq 19: nobody cared (try booting with the "irqpoll" option)
[ 2079.800891] Call Trace:
[ 2079.801303] [d7ff3f40] [c0007780] show_stack+0x7c/0x1a0 (unreliable)
[ 2079.801398] [d7ff3f80] [c007ad48] __report_bad_irq+0x5c/0xe0
[ 2079.801439] [d7ff3fa0] [c007af58] note_interrupt+0x18c/0x240
[ 2079.801466] [d7ff3fd0] [c007be04] handle_fasteoi_irq+0xf8/0x158
[ 2079.801492] [d7ff3ff0] [c000d9d4] call_handle_irq+0x18/0x28
[ 2079.801582] [d7ff5ec0] [c0004df0] do_IRQ+0xf0/0x16c
[ 2079.801609] [d7ff5ee0] [c000e9bc] ret_from_except+0x0/0x18
[ 2079.801652] --- Exception: 501 at __do_softirq+0x94/0x18c
[ 2079.801664]     LR = __do_softirq+0x54/0x18c
[ 2079.801698] [d7ff5ff0] [c000d9ac] call_do_softirq+0x14/0x24
[ 2079.801725] [d788fc50] [c0004bc0] do_softirq+0x74/0xa0
[ 2079.801751] [d788fc70] [c004350c] irq_exit+0x3c/0x8c
[ 2079.801775] [d788fc80] [c0004e3c] do_IRQ+0x13c/0x16c
[ 2079.801800] [d788fca0] [c000e9bc] ret_from_except+0x0/0x18
[ 2079.802566] --- Exception: 501 at bbc_dma_exec+0xa08/0xef4 [bbc_driver]
[ 2079.802580]     LR = bbc_dma_exec+0x970/0xef4 [bbc_driver]
[ 2079.802620] [d788fdf0] [d94d4f28] 
bbc_ioctl_dma_read_write+0x2bc/0x464 [bbc_driver]
[ 2079.802661] [d788fe70] [d94cae84] bbc_ioctl+0x2a0/0x36c [bbc_driver]
[ 2079.802754] [d788feb0] [c00d1f04] do_vfs_ioctl+0x6b8/0x760
[ 2079.802782] [d788ff10] [c00d2014] sys_ioctl+0x68/0xa8
[ 2079.802807] [d788ff40] [c000e368] ret_from_syscall+0x0/0x3c
[ 2079.802891] --- Exception: c01 at 0xfaffda8
[ 2079.802901]     LR = 0xfb8ec20
[ 2079.802936] handlers:
[ 2079.803034] [<c0268d70>] (usb_hcd_irq+0x0/0xac)
[ 2079.803120] [<d94cc1bc>] (bbc_interrupt_handler+0x0/0x65c [bbc_driver])
[ 2079.803181] Disabling IRQ #19

And interrupt information from /proc/interrupts and stat:
# cat /proc/interrupts
            CPU0
  18:       6923    OpenPIC         Level        ohci_hcd:usb2, 
ohci_hcd:usb3, drvbbc_ldc
  19:     100001    OpenPIC         Level        ehci_hcd:usb1, drvbbc
  20:          0    OpenPIC         Edge         internal_error
  21:          0    OpenPIC         Level        NMI
  22:          0    OpenPIC         Level        SP_I2C_Handler
  42:       1148    OpenPIC         Level        serial
  43:        174    OpenPIC         Level        i2c-mpc, i2c-mpc
  59:          0    OpenPIC         Level        fsl_espi
  72:      51888    OpenPIC         Level        mmc0
LOC:     143330   Local timer interrupts
SPU:          0   Spurious interrupts
CNT:          0   Performance monitoring interrupts
MCE:          0   Machine check exceptions

#cat /proc/stat
cpu  4929 0 5784 353692 736 0 460 0 0 0
cpu0 4929 0 5784 353692 736 0 460 0 0 0
intr 394342 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 6929 100001 0 0 0 0 0 0 
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 1574 174 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 
0 0 0 0 0 0 0 0 0 0 0 0 53202 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 
0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0
ctxt 357415
btime 1378108122
processes 1648
procs_running 1
procs_blocked 0
softirq 329977 0 154040 0 0 1145 0 17497 0 337 156958


I`ve no idea what the problem is and how to fix it.
Do you have experience to this?

Thanks in advance for any advice.


Simen

Re: [Qemu-devel] Disabling IRQ error

[Max Filippov]

On Tue, Sep 10, 2013 at 11:25 AM, Xie Xianshan <xiexs@cn.fujitsu.com> wrote:
> hi everyone,
>
> I`m getting the nobody cared disabling IRQ error, when i raised external
> interrupts IRQ3 to the Openpic in QEMU.
> (Actually, any external interrupts irq i raised can reproduce this error,
> but internal interrupts work fine)
>
> And this IRQ3 is sharing irq with usb card.

Could you please explain what you mean by "I raised external interrupt",
what you generally try to achieve and what behaviour you expected?

Your description reminds me of this thread:
https://lists.gnu.org/archive/html/qemu-devel/2013-08/msg04063.html

-- 
Thanks.
-- Max

Re: [Qemu-devel] Disabling IRQ error

[Xie Xianshan]

Hi Max,
   Thanks for your reply.
   And I am sorry for my unclear description.
   I want to add a new device "fpga" for e500, and trigger an interrupt 
IRQ3 while the register BB_INTR_REG which belongs to device "fpga" is 
wrote by the device driver of "fpga".
   For e500, IRQ3 is an external interrupt irq.

   According the debug log, the disabling error is encoutered during 
writing BB_INTR_REG register.
   - write BB_INTR_REG register
       - qemu_irq_raise() is called.
       - after serval minutes,
         the error message about disabling irq is displayed.
   - continue the next execution without error(with poll?)


My sample code is as follows:
----------------------------------------------
hw/ppce500_fpga.c
----------------------------------------------
typedef struct FPGAState {
     SysBusDevice busdev;
     MemoryRegion iomem;
     qemu_irq irq;

}FPGAState;

static void fpga_write(FPGAState *s, unsigned int offset, uint32_t 
value, unsigned size) {
     switch(offset) {
     case BB_INTR_REG:
         qemu_irq_raise(s->irq);
         break;
     }
}

static int ppce500_fpga_initfn(SysBusDevice *dev) {
     FPGAState *s;
     s = FROM_SYSBUS(FPGAState, SYS_BUS_DEVICE(dev));
     sysbus_init_irq(dev, &s->irq);
     ...
}

----------------------------------------------
hw/ppc/e500.c
----------------------------------------------
void ppce500_init(PPCE500Params *params) {
     ...
     qemu_irq *mpic;

     for (i = 0; i < 256; i++) {
         mpic[i] = qdev_get_gpio_in(dev, i);
     }
     ...
     /* add for FPGA */
     dev_fpga = qdev_create(NULL, "fpga");
     dev_fpga->id = "fpga";
     qdev_init_nofail(dev_fpga);
     s = SYS_BUS_DEVICE(dev_fpga);
     memory_region_add_subregion(fpga_space, FPGA_REGS_OFFSET,
             sysbus_mmio_get_region(s, 0));
     sysbus_connect_irq(s, 0, mpic[3]);
     ....
}


Thanks

  Simen


于 2013/09/10 16:23, Max Filippov 写道:
> On Tue, Sep 10, 2013 at 11:25 AM, Xie Xianshan<xiexs@cn.fujitsu.com>  wrote:
>> hi everyone,
>>
>> I`m getting the nobody cared disabling IRQ error, when i raised external
>> interrupts IRQ3 to the Openpic in QEMU.
>> (Actually, any external interrupts irq i raised can reproduce this error,
>> but internal interrupts work fine)
>>
>> And this IRQ3 is sharing irq with usb card.
>
> Could you please explain what you mean by "I raised external interrupt",
> what you generally try to achieve and what behaviour you expected?
>
> Your description reminds me of this thread:
> https://lists.gnu.org/archive/html/qemu-devel/2013-08/msg04063.html
>


-- 
Best Regards
Xie Xianshan
--------------------------------------------------
Xie Xianshan
Development Dept.I
Nanjing Fujitsu Nanda Software Tech. Co., Ltd.(FNST)
No. 6 Wenzhu Road, Nanjing, 210012, China
PHONE: +86+25-86630566-9555
FUJITSU INTERNAL: 7998-9555
MAIL: xiexs@cn.fujitsu.com
--------------------------------------------------
This communication is for use by the intended recipient(s) only and may
contain information that is privileged, confidential and exempt from
disclosure under applicable law. If you are not an intended recipient of
this communication, you are hereby notified that any dissemination,
distribution or copying hereof is strictly prohibited.  If you have
received this communication in error, please notify me by reply e-mail,
permanently delete this communication from your system, and destroy any
hard copies you may have printed

Re: [Qemu-devel] Disabling IRQ error

[Max Filippov]

On Wed, Sep 11, 2013 at 12:12 PM, Xie Xianshan <xiexs@cn.fujitsu.com> wrote:
>   I want to add a new device "fpga" for e500, and trigger an interrupt IRQ3
> while the register BB_INTR_REG which belongs to device "fpga" is wrote by
> the device driver of "fpga".
>   For e500, IRQ3 is an external interrupt irq.
>   According the debug log, the disabling error is encoutered during writing
> BB_INTR_REG register.
>   - write BB_INTR_REG register
>       - qemu_irq_raise() is called.
>       - after serval minutes,
>         the error message about disabling irq is displayed.
>   - continue the next execution without error(with poll?)

So your device raises IRQ, but it doesn't lower it. Real devices
usually don't do that, they either generate a short pulse on the
IRQ line (in case of edge-triggered IRQ) or raise IRQ line on
some event and then lower it on a command from its driver
(level-triggered IRQ).

You can do the following to make your device behave that way:
- make your fpga device capable of lowering its IRQ, e.g. by adding
  another register:

> static void fpga_write(FPGAState *s, unsigned int offset, uint32_t value,
> unsigned size) {
>     switch(offset) {
>     case BB_INTR_REG:
>         qemu_irq_raise(s->irq);
>         break;
       case BB_INTC_REG:
           qemu_irq_lower(s->irq);
           break;
>     }
> }

- provide an interrupt service routine in the linux driver for your fpga
  device that would check whether the interrupt was caused by its
  device, and if so lower the device's IRQ.

Thanks.
-- Max

Re: [Qemu-devel] Disabling IRQ error

[Xie Xianshan]

Hi Max,
     Thanks for your patience and help.
     I`ve tried to do what you said, but the problem doesn`t go away.
And actually i cannot add a new register to the fpga device, because the 
fpga device i`m emulating already exists in the real world.
So i cannot change anything about hardware properties and linux driver 
for this device.

     By the way, how did you finally fix your problem?

Thanks,
  Simen

于 2013/09/11 17:29, Max Filippov 写道:
> On Wed, Sep 11, 2013 at 12:12 PM, Xie Xianshan<xiexs@cn.fujitsu.com>  wrote:
>>    I want to add a new device "fpga" for e500, and trigger an interrupt IRQ3
>> while the register BB_INTR_REG which belongs to device "fpga" is wrote by
>> the device driver of "fpga".
>>    For e500, IRQ3 is an external interrupt irq.
>>    According the debug log, the disabling error is encoutered during writing
>> BB_INTR_REG register.
>>    - write BB_INTR_REG register
>>        - qemu_irq_raise() is called.
>>        - after serval minutes,
>>          the error message about disabling irq is displayed.
>>    - continue the next execution without error(with poll?)
>
> So your device raises IRQ, but it doesn't lower it. Real devices
> usually don't do that, they either generate a short pulse on the
> IRQ line (in case of edge-triggered IRQ) or raise IRQ line on
> some event and then lower it on a command from its driver
> (level-triggered IRQ).
>
> You can do the following to make your device behave that way:
> - make your fpga device capable of lowering its IRQ, e.g. by adding
>    another register:
>
>> static void fpga_write(FPGAState *s, unsigned int offset, uint32_t value,
>> unsigned size) {
>>      switch(offset) {
>>      case BB_INTR_REG:
>>          qemu_irq_raise(s->irq);
>>          break;
>         case BB_INTC_REG:
>             qemu_irq_lower(s->irq);
>             break;
>>      }
>> }
>
> - provide an interrupt service routine in the linux driver for your fpga
>    device that would check whether the interrupt was caused by its
>    device, and if so lower the device's IRQ.
>
> Thanks.
> -- Max
>
>


-- 
Best Regards
Xie Xianshan
--------------------------------------------------
Xie Xianshan
Development Dept.I
Nanjing Fujitsu Nanda Software Tech. Co., Ltd.(FNST)
No. 6 Wenzhu Road, Nanjing, 210012, China
PHONE: +86+25-86630566-9555
FUJITSU INTERNAL: 7998-9555
MAIL: xiexs@cn.fujitsu.com
--------------------------------------------------
This communication is for use by the intended recipient(s) only and may
contain information that is privileged, confidential and exempt from
disclosure under applicable law. If you are not an intended recipient of
this communication, you are hereby notified that any dissemination,
distribution or copying hereof is strictly prohibited.  If you have
received this communication in error, please notify me by reply e-mail,
permanently delete this communication from your system, and destroy any
hard copies you may have printed

Re: [Qemu-devel] Disabling IRQ error

[Max Filippov]

On Thu, Sep 12, 2013 at 11:49 AM, Xie Xianshan <xiexs@cn.fujitsu.com> wrote:
> Hi Max,
>     Thanks for your patience and help.
>     I`ve tried to do what you said, but the problem doesn`t go away.
> And actually i cannot add a new register to the fpga device, because the
> fpga device i`m emulating already exists in the real world.
> So i cannot change anything about hardware properties and linux driver for
> this device.

Then I don't get its IRQ logic. Does it mean an IRQ to be edge-triggered?
Its model should use qemu_irq_pulse then instead of qemu_irq_raise and
its IRQ line should be connected to edge-sensing input of interrupt controller.
Input that you have used is also used to sense PCI IRQ and is level-sensing.

>     By the way, how did you finally fix your problem?

I didn't have any. (:

-- 
Thanks.
-- Max

Re: [Qemu-devel] Disabling IRQ error

[Xie Xianshan]

Dear Max,
    > Does it mean an IRQ to be edge-triggered?
    No, it is a level-sensitive and active-high interrupt.
This is why i tried to use qemu_irq_raise() to trigger IRQ.


Thanks,
Simen


>> Hi Max,
>>      Thanks for your patience and help.
>>      I`ve tried to do what you said, but the problem doesn`t go away.
>> And actually i cannot add a new register to the fpga device, because the
>> fpga device i`m emulating already exists in the real world.
>> So i cannot change anything about hardware properties and linux driver for
>> this device.
>
> Then I don't get its IRQ logic. Does it mean an IRQ to be edge-triggered?
> Its model should use qemu_irq_pulse then instead of qemu_irq_raise and
> its IRQ line should be connected to edge-sensing input of interrupt controller.
> Input that you have used is also used to sense PCI IRQ and is level-sensing.
>
>>      By the way, how did you finally fix your problem?
>
> I didn't have any. (:
>

Re: [Qemu-devel] Disabling IRQ error

[Max Filippov]

On Thu, Sep 12, 2013 at 2:51 PM, Xie Xianshan <xiexs@cn.fujitsu.com> wrote:
> Dear Max,
>
>    > Does it mean an IRQ to be edge-triggered?
>    No, it is a level-sensitive and active-high interrupt.
> This is why i tried to use qemu_irq_raise() to trigger IRQ.

Ok, back to your original question:

> I`m getting the nobody cared disabling IRQ error, when i raised external interrupts IRQ3 to the Openpic in QEMU.

Your linux driver should return anything except IRQ_NONE from its ISR
when it detects IRQ from your device. As I understand once you raise IRQ
you don't lower it, so the driver must always return anything except
IRQ_NONE. If it doesn't do so you will see the error message above.

-- 
Thanks.
-- Max

Re: [Qemu-devel] Disabling IRQ error

[Xie Xianshan]

Dear Max,
     Sorry for the late reply.
     Thanks for your advice about lowering the irqs after raising them, 
and that fixed my problem.
     Thanks again for your kindness.

Thanks,
Simen

于 2013/09/11 17:29, Max Filippov 写道:
> On Wed, Sep 11, 2013 at 12:12 PM, Xie Xianshan<xiexs@cn.fujitsu.com>  wrote:
>>    I want to add a new device "fpga" for e500, and trigger an interrupt IRQ3
>> while the register BB_INTR_REG which belongs to device "fpga" is wrote by
>> the device driver of "fpga".
>>    For e500, IRQ3 is an external interrupt irq.
>>    According the debug log, the disabling error is encoutered during writing
>> BB_INTR_REG register.
>>    - write BB_INTR_REG register
>>        - qemu_irq_raise() is called.
>>        - after serval minutes,
>>          the error message about disabling irq is displayed.
>>    - continue the next execution without error(with poll?)
>
> So your device raises IRQ, but it doesn't lower it. Real devices
> usually don't do that, they either generate a short pulse on the
> IRQ line (in case of edge-triggered IRQ) or raise IRQ line on
> some event and then lower it on a command from its driver
> (level-triggered IRQ).
>
> You can do the following to make your device behave that way:
> - make your fpga device capable of lowering its IRQ, e.g. by adding
>    another register:
>
>> static void fpga_write(FPGAState *s, unsigned int offset, uint32_t value,
>> unsigned size) {
>>      switch(offset) {
>>      case BB_INTR_REG:
>>          qemu_irq_raise(s->irq);
>>          break;
>         case BB_INTC_REG:
>             qemu_irq_lower(s->irq);
>             break;
>>      }
>> }
>
> - provide an interrupt service routine in the linux driver for your fpga
>    device that would check whether the interrupt was caused by its
>    device, and if so lower the device's IRQ.
>
> Thanks.
> -- Max
>
>
>