qemu-devel.nongnu.org archive mirror
 help / color / mirror / Atom feed
From: "Andreas Färber" <afaerber@suse.de>
To: Igor Mammedov <imammedo@redhat.com>
Cc: Alexander Graf <agraf@suse.de>,
	Marcelo Tosatti <mtosatti@redhat.com>,
	qemu-devel@nongnu.org, kvm <kvm@vger.kernel.org>,
	Avi Kivity <avi@redhat.com>
Subject: Re: [Qemu-devel] [PATCH qom-cpu v2 4/7] cpus: Pass CPUState to qemu_cpu_is_self()
Date: Tue, 30 Oct 2012 21:08:34 +0100	[thread overview]
Message-ID: <509033C2.7050401@suse.de> (raw)
In-Reply-To: <20121016024923.6b29da4a@thinkpad.mammed.net>

Am 16.10.2012 02:49, schrieb Igor Mammedov:
> On Fri, 12 Oct 2012 03:26:40 +0200
> Andreas Färber <afaerber@suse.de> wrote:
> 
>> Change return type to bool, move to include/qemu/cpu.h and
>> add documentation.
>>
>> Signed-off-by: Andreas Färber <afaerber@suse.de>
>> ---
>>  cpus.c             |   10 ++++------
>>  exec.c             |    3 ++-
>>  hw/apic.c          |    6 ++++--
>>  include/qemu/cpu.h |   10 ++++++++++
>>  kvm-all.c          |    4 +++-
>>  qemu-common.h      |    1 -
>>  target-i386/kvm.c  |    6 ++++--
>>  7 Dateien geändert, 27 Zeilen hinzugefügt(+), 13 Zeilen entfernt(-)
>>
>> diff --git a/cpus.c b/cpus.c
>> index 750a76f..849ea8a 100644
>> --- a/cpus.c
>> +++ b/cpus.c
>> @@ -638,9 +638,10 @@ void qemu_init_cpu_loop(void)
>>  
>>  void run_on_cpu(CPUArchState *env, void (*func)(void *data), void *data)
>>  {
>> +    CPUState *cpu = ENV_GET_CPU(env);
>>      struct qemu_work_item wi;
>>  
>> -    if (qemu_cpu_is_self(env)) {
>> +    if (qemu_cpu_is_self(cpu)) {
>>          func(data);
>>          return;
>>      }
>> @@ -855,7 +856,7 @@ static void qemu_cpu_kick_thread(CPUArchState *env)
>>          exit(1);
>>      }
>>  #else /* _WIN32 */
>> -    if (!qemu_cpu_is_self(env)) {
>> +    if (!qemu_cpu_is_self(cpu)) {
>>          SuspendThread(cpu->hThread);
>>          cpu_signal(0);
>>          ResumeThread(cpu->hThread);
>> @@ -890,11 +891,8 @@ void qemu_cpu_kick_self(void)
>>  #endif
>>  }
>>  
>> -int qemu_cpu_is_self(void *_env)
>> +bool qemu_cpu_is_self(CPUState *cpu)
>>  {
>> -    CPUArchState *env = _env;
>> -    CPUState *cpu = ENV_GET_CPU(env);
>> -
>>      return qemu_thread_is_self(cpu->thread);
>>  }
>>  
>> diff --git a/exec.c b/exec.c
>> index 7899042..e21be32 100644
>> --- a/exec.c
>> +++ b/exec.c
>> @@ -1685,6 +1685,7 @@ static void cpu_unlink_tb(CPUArchState *env)
>>  /* mask must never be zero, except for A20 change call */
>>  static void tcg_handle_interrupt(CPUArchState *env, int mask)
>>  {
>> +    CPUState *cpu = ENV_GET_CPU(env);
> Is there any chance to get rid of expensive cast on this call path?

I've spent some time investigating this and it would be rather complicated:

* We can't change the argument to CPUState yet (pushing the
responsibility to the caller) since interrupt_request is being accessed
til the not-yet-committable end of original part 4 series, the TB is
being flushed, icount being handled, and it is being used for the first
if, so not movable to a deeper block.

* A pointer to this function (or the KVM version below) is stored as
cpu_interrupt_handler and accessed through cpu_interrupt(), which for
softmmu calls the handler and for *-user does something similar to the
TCG version. This function is being used in a lot of places, so adding a
CPUState argument in addition to the CPUArchState would touch a lot of
places for an interim solution.

* QOM makes no guarantees about subclass layout (.env placement), so
going from CPUState to CPUArchState involves quite some ugly #ifdef'fery
that I am trying to avoid, going only the I-know-my-type direction.

I'll therefore accept your Reviewed-by to apply this and will keep
working during the Soft Freeze on the follow-ups that allow changing
this and others completely over to CPUState. Alex has just given green
light for the ppc prerequisites so I can apply them together with xtensa
to qom-cpu and post the next batch if I haven't already.

Regards,
Andreas

> 
>>      int old_mask;
>>  
>>      old_mask = env->interrupt_request;
>> @@ -1694,7 +1695,7 @@ static void tcg_handle_interrupt(CPUArchState *env, int mask)
>>       * If called from iothread context, wake the target cpu in
>>       * case its halted.
>>       */
>> -    if (!qemu_cpu_is_self(env)) {
>> +    if (!qemu_cpu_is_self(cpu)) {
>>          qemu_cpu_kick(env);
>>          return;
>>      }
>> diff --git a/hw/apic.c b/hw/apic.c
>> index ccf2819..1b4cd2f 100644
>> --- a/hw/apic.c
>> +++ b/hw/apic.c
>> @@ -107,7 +107,7 @@ static void apic_sync_vapic(APICCommonState *s, int sync_type)
>>          length = offsetof(VAPICState, enabled) - offsetof(VAPICState, isr);
>>  
>>          if (sync_type & SYNC_TO_VAPIC) {
>> -            assert(qemu_cpu_is_self(&s->cpu->env));
>> +            assert(qemu_cpu_is_self(CPU(s->cpu)));
>>  
>>              vapic_state.tpr = s->tpr;
>>              vapic_state.enabled = 1;
>> @@ -363,10 +363,12 @@ static int apic_irq_pending(APICCommonState *s)
>>  /* signal the CPU if an irq is pending */
>>  static void apic_update_irq(APICCommonState *s)
>>  {
>> +    CPUState *cpu = CPU(s->cpu);
>> +
>>      if (!(s->spurious_vec & APIC_SV_ENABLE)) {
>>          return;
>>      }
>> -    if (!qemu_cpu_is_self(&s->cpu->env)) {
>> +    if (!qemu_cpu_is_self(cpu)) {
>>          cpu_interrupt(&s->cpu->env, CPU_INTERRUPT_POLL);
>>      } else if (apic_irq_pending(s) > 0) {
>>          cpu_interrupt(&s->cpu->env, CPU_INTERRUPT_HARD);
>> diff --git a/include/qemu/cpu.h b/include/qemu/cpu.h
>> index ad706a6..7be983d 100644
>> --- a/include/qemu/cpu.h
>> +++ b/include/qemu/cpu.h
>> @@ -78,5 +78,15 @@ struct CPUState {
>>   */
>>  void cpu_reset(CPUState *cpu);
>>  
>> +/**
>> + * qemu_cpu_is_self:
>> + * @cpu: The vCPU to check against.
>> + *
>> + * Checks whether the caller is executing on the vCPU thread.
>> + *
>> + * Returns: %true if called from @cpu's thread, %false otherwise.
>> + */
>> +bool qemu_cpu_is_self(CPUState *cpu);
>> +
>>  
>>  #endif
>> diff --git a/kvm-all.c b/kvm-all.c
>> index 92a7137..db01aeb 100644
>> --- a/kvm-all.c
>> +++ b/kvm-all.c
>> @@ -854,9 +854,11 @@ static MemoryListener kvm_memory_listener = {
>>  
>>  static void kvm_handle_interrupt(CPUArchState *env, int mask)
>>  {
>> +    CPUState *cpu = ENV_GET_CPU(env);
>> +
>>      env->interrupt_request |= mask;
>>  
>> -    if (!qemu_cpu_is_self(env)) {
>> +    if (!qemu_cpu_is_self(cpu)) {
>>          qemu_cpu_kick(env);
>>      }
>>  }
>> diff --git a/qemu-common.h b/qemu-common.h
>> index b54612b..2094742 100644
>> --- a/qemu-common.h
>> +++ b/qemu-common.h
>> @@ -326,7 +326,6 @@ int cpu_load(QEMUFile *f, void *opaque, int version_id);
>>  /* Unblock cpu */
>>  void qemu_cpu_kick(void *env);
>>  void qemu_cpu_kick_self(void);
>> -int qemu_cpu_is_self(void *env);
>>  
>>  /* work queue */
>>  struct qemu_work_item {
>> diff --git a/target-i386/kvm.c b/target-i386/kvm.c
>> index 5b18383..cf3d2f1 100644
>> --- a/target-i386/kvm.c
>> +++ b/target-i386/kvm.c
>> @@ -1552,9 +1552,10 @@ static int kvm_get_debugregs(CPUX86State *env)
>>  
>>  int kvm_arch_put_registers(CPUX86State *env, int level)
>>  {
>> +    CPUState *cpu = ENV_GET_CPU(env);
>>      int ret;
>>  
>> -    assert(cpu_is_stopped(env) || qemu_cpu_is_self(env));
>> +    assert(cpu_is_stopped(env) || qemu_cpu_is_self(cpu));
>>  
>>      ret = kvm_getput_regs(env, 1);
>>      if (ret < 0) {
>> @@ -1609,9 +1610,10 @@ int kvm_arch_put_registers(CPUX86State *env, int level)
>>  
>>  int kvm_arch_get_registers(CPUX86State *env)
>>  {
>> +    CPUState *cpu = ENV_GET_CPU(env);
>>      int ret;
>>  
>> -    assert(cpu_is_stopped(env) || qemu_cpu_is_self(env));
>> +    assert(cpu_is_stopped(env) || qemu_cpu_is_self(cpu));
>>  
>>      ret = kvm_getput_regs(env, 0);
>>      if (ret < 0) {
>> -- 
>> 1.7.10.4
>>
>>
> Reviewed-by: Igor Mammedov <imammedo@redhat.com>
> 


-- 
SUSE LINUX Products GmbH, Maxfeldstr. 5, 90409 Nürnberg, Germany
GF: Jeff Hawn, Jennifer Guild, Felix Imendörffer; HRB 16746 AG Nürnberg

  reply	other threads:[~2012-10-30 20:09 UTC|newest]

Thread overview: 18+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2012-10-12  1:26 [Qemu-devel] [PATCH qom-cpu v2 0/7] QOM CPUState, part 4a: CPU_COMMON, continued Andreas Färber
2012-10-12  1:26 ` [Qemu-devel] [PATCH qom-cpu v2 1/7] target-i386: Inline APIC cpu_env property setting Andreas Färber
2012-10-15 23:22   ` Igor Mammedov
2012-10-16 16:02   ` Andreas Färber
2012-10-16 16:04     ` Paolo Bonzini
2012-10-17 12:50     ` Jan Kiszka
2012-10-12  1:26 ` [Qemu-devel] [PATCH qom-cpu v2 2/7] apic: Store X86CPU in APICCommonState Andreas Färber
2012-10-15 23:28   ` Igor Mammedov
2012-10-12  1:26 ` [Qemu-devel] [PATCH qom-cpu v2 3/7] target-i386: Pass X86CPU to cpu_x86_load_seg_cache_sipi() Andreas Färber
2012-10-15 23:29   ` Igor Mammedov
2012-10-12  1:26 ` [Qemu-devel] [PATCH qom-cpu v2 4/7] cpus: Pass CPUState to qemu_cpu_is_self() Andreas Färber
2012-10-16  0:49   ` Igor Mammedov
2012-10-30 20:08     ` Andreas Färber [this message]
2012-10-30 20:33   ` Andreas Färber
2012-10-12  1:26 ` [Qemu-devel] [PATCH qom-cpu v2 5/7] cpus: Pass CPUState to qemu_cpu_kick_thread() Andreas Färber
2012-10-16  0:53   ` Igor Mammedov
2012-10-12  1:26 ` [Qemu-devel] [PATCH qom-cpu v2 6/7] cpu: Move created field to CPUState Andreas Färber
2012-10-12  1:26 ` [Qemu-devel] [PATCH qom-cpu v2 7/7] cpu: Move stop " Andreas Färber

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=509033C2.7050401@suse.de \
    --to=afaerber@suse.de \
    --cc=agraf@suse.de \
    --cc=avi@redhat.com \
    --cc=imammedo@redhat.com \
    --cc=kvm@vger.kernel.org \
    --cc=mtosatti@redhat.com \
    --cc=qemu-devel@nongnu.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).