qemu-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Qemu-devel] [PATCH qom-next 07/12] target-i386: move cpu halted dec


From: Igor Mammedov
Subject: Re: [Qemu-devel] [PATCH qom-next 07/12] target-i386: move cpu halted decision into x86_cpu_reset
Date: Wed, 30 May 2012 17:21:49 +0200
User-agent: Mozilla/5.0 (X11; Linux x86_64; rv:12.0) Gecko/20120430 Thunderbird/12.0.1

On 05/30/2012 04:54 PM, Andreas Färber wrote:
Am 30.05.2012 16:13, schrieb Igor Mammedov:
On 05/30/2012 02:11 PM, Andreas Färber wrote:
Am 30.05.2012 00:10, schrieb Igor Mammedov:
From: Igor Mammedov<address@hidden>

MP initialization protocol differs between cpu families, and for P6 and
onward models it is up to CPU to decide if it will be BSP using this
protocol, so try to model this. However there is no point in
implementing
MP initialization protocol in qemu. Thus first CPU is always marked
as BSP.

This patch:
   - moves decision to designate BSP from board into cpu, making cpu
self-sufficient in this regard. Later it will allow to cleanup hw/pc.c
and remove cpu_reset and wrappers from there.
   - stores flag that CPU is BSP in IA32_APIC_BASE to model behavior
described in Inted SDM vol 3a part 1 chapter 8.4.1
   - uses MSR_IA32_APICBASE_BSP flag in apic_base for checking if cpu
is BSP

patch is based on Jan Kiszka's proposal:
      http://thread.gmane.org/gmane.comp.emulators.qemu/100806

v2:
    - fix build for i386-linux-user
        spotted-by: Peter Maydell<address@hidden>

Signed-off-by: Igor Mammedov<address@hidden>
---
   hw/apic.h            |    2 +-
   hw/apic_common.c     |   18 ++++++++++++------
   hw/pc.c              |    9 ---------
   target-i386/cpu.c    |    9 +++++++++
   target-i386/helper.c |    1 -
   target-i386/kvm.c    |    5 +++--
   6 files changed, 25 insertions(+), 19 deletions(-)

diff --git a/hw/apic.h b/hw/apic.h
index 62179ce..d961ed4 100644
--- a/hw/apic.h
+++ b/hw/apic.h
@@ -20,9 +20,9 @@ void apic_init_reset(DeviceState *s);
   void apic_sipi(DeviceState *s);
   void apic_handle_tpr_access_report(DeviceState *d, target_ulong ip,
                                      TPRAccess access);
+void apic_designate_bsp(DeviceState *d);

   /* pc.c */
-int cpu_is_bsp(CPUX86State *env);
   DeviceState *cpu_get_current_apic(void);

   #endif
diff --git a/hw/apic_common.c b/hw/apic_common.c
index 46a9ff7..98ad6f0 100644
--- a/hw/apic_common.c
+++ b/hw/apic_common.c
@@ -43,8 +43,8 @@ uint64_t cpu_get_apic_base(DeviceState *d)
           trace_cpu_get_apic_base((uint64_t)s->apicbase);
           return s->apicbase;
       } else {
-        trace_cpu_get_apic_base(0);
-        return 0;
+        trace_cpu_get_apic_base(MSR_IA32_APICBASE_BSP);
+        return MSR_IA32_APICBASE_BSP;
       }
   }

@@ -201,22 +201,28 @@ void apic_init_reset(DeviceState *d)
       s->timer_expiry = -1;
   }

+void apic_designate_bsp(DeviceState *d)
+{
+    if (d) {

This check looks odd. The function call is already guarded with
cpu_index == 0. What other case would lead to d == NULL and can't we
check that at the call site?

cpu_index == 0 doesn't imply that cpu has apic hence the check. And we
for sure
can check it call site, would you like to do it there?
PS:
  there are many checks for this condition in APIC code, so may be it should
  be there style-wise at least?

What I referred to as odd was the indentation of the actual implementation.

So I'd be fine with either checking at the call site or

     if (d == NULL) {
         return;
     }

     ...

+        APICCommonState *s = APIC_COMMON(d);

If this is a QOM cast macro, it will implicitly assert d != NULL, no?

it actually will lead to null pointer dereference, like this:
   OBJECT_CHECK(type, obj, name) \
     ((type *)object_dynamic_cast_assert(OBJECT(obj), (name)))
       =>   object_dynamic_cast(obj, typename)
         =>   object_is_type(obj, target_type)
           =>   type_is_ancestor(obj->class->type, target_type);
                                   ^^^

Mind to send a patch fixing that? :-)
I'd suppose the _is_... functions should probably return false and the
cast should assert. Anthony?

Maybe asserting right at the beginning of  object_dynamic_cast_assert would be 
better
than hiding it somewhere deep in call chain, because later someone could add 
access to obj
before *_is_* functions and with explicit assert in the beginnig of 
object_dynamic_cast_assert()
it would be hard to mess things up.


Andreas


--
-----
 Igor



reply via email to

[Prev in Thread] Current Thread [Next in Thread]