Re: [Xen-devel] [PATCH v3 04/13] xen/arm: support for guest SGI

[Julien Grall <julien.grall@xxxxxxxxxx>]

On 04/26/2013 01:12 PM, Sander Bogaert wrote:

> On 24 April 2013 21:07, Stefano Stabellini
> <stefano.stabellini@xxxxxxxxxxxxx> wrote:
>> Trap writes to GICD_SGIR, parse the requests, inject SGIs into the right
>> guest vcpu.
>>
>> Signed-off-by: Stefano Stabellini <stefano.stabellini@xxxxxxxxxxxxx>
>>
>> Changes in v3:
>> - make use of cpumask_from_bitmap.
>> ---
>>  xen/arch/arm/vgic.c       |   55 
>> ++++++++++++++++++++++++++++++++++++++++----
>>  xen/include/asm-arm/gic.h |    3 ++
>>  2 files changed, 53 insertions(+), 5 deletions(-)
>>
>> diff --git a/xen/arch/arm/vgic.c b/xen/arch/arm/vgic.c
>> index b30da78..8912aab 100644
>> --- a/xen/arch/arm/vgic.c
>> +++ b/xen/arch/arm/vgic.c
>> @@ -370,6 +370,7 @@ static void vgic_enable_irqs(struct vcpu *v, uint32_t r, 
>> int n)
>>
>>  static int vgic_distr_mmio_write(struct vcpu *v, mmio_info_t *info)
>>  {
>> +    struct domain *d = v->domain;
>>      struct hsr_dabt dabt = info->dabt;
>>      struct cpu_user_regs *regs = guest_cpu_user_regs();
>>      register_t *r = select_user_reg(regs, dabt.reg);
>> @@ -498,11 +499,55 @@ static int vgic_distr_mmio_write(struct vcpu *v, 
>> mmio_info_t *info)
>>          goto write_ignore;
>>
>>      case GICD_SGIR:
>> -        if ( dabt.size != 2 ) goto bad_width;
>> -        printk("vGICD: unhandled write %#"PRIregister" to ICFGR%d\n",
>> -               *r, gicd_reg - GICD_ICFGR);
>> -        return 0;
>> -
>> +        {
>> +            cpumask_t vcpu_mask;
>> +            int virtual_irq;
>> +            int filter;
>> +            int vcpuid;
>> +            struct vcpu *vt;
>> +            int i;
>> +            unsigned long bits;
>> +
>> +            if ( dabt.size != 2 ) goto bad_width;
>> +
>> +            filter = (*r & GICD_SGI_TARGET_LIST_MASK);
>> +            virtual_irq = (*r & GICD_SGI_INTID_MASK);
>> +
>> +            cpumask_clear(&vcpu_mask);
>> +            switch ( filter )
>> +            {
>> +                case GICD_SGI_TARGET_LIST:
>> +                    bits = (*r & GICD_SGI_TARGET_MASK) >> 
>> GICD_SGI_TARGET_SHIFT;
>> +                    vcpu_mask = cpumask_from_bitmap(&bits, 8);
>> +                    break;
>> +                case GICD_SGI_TARGET_OTHERS:
>> +                    for ( i = 0; i < d->max_vcpus; i++ )
>> +                    {
>> +                        if ( i != current->vcpu_id && d->vcpu[i] != NULL )
>> +                            cpumask_set_cpu(i, &vcpu_mask);
>> +                    }
>> +                case GICD_SGI_TARGET_SELF:
>> +                    vcpu_mask = *cpumask_of(current->vcpu_id);
>> +                    break;
>> +                default:
>> +                    printk("vGICD: unhandled GICD_SGIR write %x with wrong 
>> TargetListFilter field\n", *r);
>> +                    return 0;
>> +            }
>> +
>> +            for_each_cpu( vcpuid, &vcpu_mask )
>> +            {
>> +                if ( vcpuid >= d->max_vcpus || (vt = d->vcpu[vcpuid]) == 
>> NULL ||
>> +                        virtual_irq >= 16 )
>> +                {
>> +                    printk("vGICD: GICD_SGIR write r=%x vcpu_mask=%lx, 
>> wrong CPUTargetList\n",
>> +                           *r, *cpumask_bits(&vcpu_mask));
>> +                    return 0;
>> +                }
>> +                vgic_vcpu_inject_irq(vt, virtual_irq, 1);
>> +            }
>> +            return 1;
>> +        }
>> +
>>      case GICD_CPENDSGIR ... GICD_CPENDSGIRN:
>>          if ( dabt.size != 0 && dabt.size != 2 ) goto bad_width;
>>          printk("vGICD: unhandled %s write %#"PRIregister" to 
>> ICPENDSGIR%d\n",
>> diff --git a/xen/include/asm-arm/gic.h b/xen/include/asm-arm/gic.h
>> index 92711d5..4f2c8b8 100644
>> --- a/xen/include/asm-arm/gic.h
>> +++ b/xen/include/asm-arm/gic.h
>> @@ -51,12 +51,15 @@
>>  #define GICD_SPENDSGIRN (0xF2C/4)
>>  #define GICD_ICPIDR2    (0xFE8/4)
>>
>> +#define GICD_SGI_TARGET_LIST_SHIFT   (24)
>> +#define GICD_SGI_TARGET_LIST_MASK    (0x3UL << GICD_SGI_TARGET_LIST_SHIFT)
>>  #define GICD_SGI_TARGET_LIST   (0UL<<24)
>>  #define GICD_SGI_TARGET_OTHERS (1UL<<24)
>>  #define GICD_SGI_TARGET_SELF   (2UL<<24)
>>  #define GICD_SGI_TARGET_SHIFT  (16)
>>  #define GICD_SGI_TARGET_MASK   (0xFFUL<<GICD_SGI_TARGET_SHIFT)
>>  #define GICD_SGI_GROUP1        (1UL<<15)
>> +#define GICD_SGI_INTID_MASK    (0xFUL)
>>
>>  #define GICC_CTLR       (0x0000/4)
>>  #define GICC_PMR        (0x0004/4)
>> --
>> 1.7.2.5
>>
>>
> 
> Hi,
> 
> I've been running crashme ( executing random instructions ) on Xen on
> Arm installed on an Arndale board. I'm using Julien's dom0 and Xen
> branch. The Xen branch wasn't updated in a few days.
> 
> I managed to crash Xen from dom0 userspace:
> (XEN)
> (XEN) ****************************************
> (XEN) Panic on CPU 1:
> (XEN) Unhandled SGI 2 on CPU1
> (XEN) ****************************************
> (XEN)
> (XEN) Reboot in five seconds...
> 
> Tracking this back to the source:
> xen/arch/arm/gic.c:648 static void do_sgi
> xen/arch/arm/gic.c:671 void gic_interrupt
>

> From what I understood from the source all sgi's except 0 & 1
> would/will cause this panic since they are just not handled. I was
> wondering if this patch would fix that because going over it didn't
> convince me it would. The reason I'm asking and not just trying it
> because it's not easy to reproduce a crashme crash.


Could you give me the latest commit ID/tag you use? On the latest
arndale branch, SGI 2 should be handled. Xen use this SGI to call
function on another CPU.

-- 
Julien

_______________________________________________
Xen-devel mailing list
Xen-devel@xxxxxxxxxxxxx
http://lists.xen.org/xen-devel