qemu-devel.nongnu.org archive mirror
 help / color / mirror / Atom feed
From: Paolo Bonzini <pbonzini@redhat.com>
To: fred.konrad@greensocs.com, qemu-devel@nongnu.org, mttcg@greensocs.com
Cc: peter.maydell@linaro.org, a.spyridakis@virtualopensystems.com,
	mark.burton@greensocs.com, agraf@suse.de,
	alistair.francis@xilinx.com, guillaume.delbergue@greensocs.com,
	alex.bennee@linaro.org
Subject: Re: [Qemu-devel] [RFC PATCH V6 06/18] tcg: remove tcg_halt_cond global variable.
Date: Fri, 26 Jun 2015 17:02:50 +0200	[thread overview]
Message-ID: <558D699A.2010804@redhat.com> (raw)
In-Reply-To: <1435330053-18733-7-git-send-email-fred.konrad@greensocs.com>



On 26/06/2015 16:47, fred.konrad@greensocs.com wrote:
> From: KONRAD Frederic <fred.konrad@greensocs.com>
> 
> This removes tcg_halt_cond global variable.
> We need one QemuCond per virtual cpu for multithread TCG.
> 
> Signed-off-by: KONRAD Frederic <fred.konrad@greensocs.com>
> ---
>  cpus.c | 18 +++++++-----------
>  1 file changed, 7 insertions(+), 11 deletions(-)
> 
> diff --git a/cpus.c b/cpus.c
> index 2d62a35..79383df 100644
> --- a/cpus.c
> +++ b/cpus.c
> @@ -813,7 +813,6 @@ static unsigned iothread_requesting_mutex;
>  static QemuThread io_thread;
>  
>  static QemuThread *tcg_cpu_thread;
> -static QemuCond *tcg_halt_cond;
>  
>  /* cpu creation */
>  static QemuCond qemu_cpu_cond;
> @@ -919,15 +918,13 @@ static void qemu_wait_io_event_common(CPUState *cpu)
>      cpu->thread_kicked = false;
>  }
>  
> -static void qemu_tcg_wait_io_event(void)
> +static void qemu_tcg_wait_io_event(CPUState *cpu)
>  {
> -    CPUState *cpu;
> -
>      while (all_cpu_threads_idle()) {
>         /* Start accounting real time to the virtual clock if the CPUs
>            are idle.  */
>          qemu_clock_warp(QEMU_CLOCK_VIRTUAL);
> -        qemu_cond_wait(tcg_halt_cond, &qemu_global_mutex);
> +        qemu_cond_wait(cpu->halt_cond, &qemu_global_mutex);
>      }
>  
>      while (iothread_requesting_mutex) {
> @@ -1047,7 +1044,7 @@ static void *qemu_tcg_cpu_thread_fn(void *arg)
>  
>      /* wait for initial kick-off after machine start */
>      while (first_cpu->stopped) {
> -        qemu_cond_wait(tcg_halt_cond, &qemu_global_mutex);
> +        qemu_cond_wait(first_cpu->halt_cond, &qemu_global_mutex);
>  
>          /* process any pending work */
>          CPU_FOREACH(cpu) {
> @@ -1068,7 +1065,7 @@ static void *qemu_tcg_cpu_thread_fn(void *arg)
>                  qemu_clock_notify(QEMU_CLOCK_VIRTUAL);
>              }
>          }
> -        qemu_tcg_wait_io_event();
> +        qemu_tcg_wait_io_event(QTAILQ_FIRST(&cpus));

Does this work (for non-multithreaded TCG) if tcg_thread_fn is waiting
on the "wrong" condition variable?  For example if all CPUs are idle and
the second CPU wakes up, qemu_tcg_wait_io_event won't be kicked out of
the wait.

I think you need to have a CPUThread struct like this:

   struct CPUThread {
       QemuThread thread;
       QemuCond halt_cond;
   };

and in CPUState have a CPUThread * field instead of the thread and
halt_cond fields.

Then single-threaded TCG can point all CPUStates to the same instance of
the struct, while multi-threaded TCG can point each CPUState to a
different struct.

Paolo

>      }
>  
>      return NULL;
> @@ -1235,12 +1232,12 @@ static void qemu_tcg_init_vcpu(CPUState *cpu)
>  
>      tcg_cpu_address_space_init(cpu, cpu->as);
>  
> +    cpu->halt_cond = g_malloc0(sizeof(QemuCond));
> +    qemu_cond_init(cpu->halt_cond);
> +
>      /* share a single thread for all cpus with TCG */
>      if (!tcg_cpu_thread) {
>          cpu->thread = g_malloc0(sizeof(QemuThread));
> -        cpu->halt_cond = g_malloc0(sizeof(QemuCond));
> -        qemu_cond_init(cpu->halt_cond);
> -        tcg_halt_cond = cpu->halt_cond;
>          snprintf(thread_name, VCPU_THREAD_NAME_SIZE, "CPU %d/TCG",
>                   cpu->cpu_index);
>          qemu_thread_create(cpu->thread, thread_name, qemu_tcg_cpu_thread_fn,
> @@ -1254,7 +1251,6 @@ static void qemu_tcg_init_vcpu(CPUState *cpu)
>          tcg_cpu_thread = cpu->thread;
>      } else {
>          cpu->thread = tcg_cpu_thread;
> -        cpu->halt_cond = tcg_halt_cond;
>      }

  reply	other threads:[~2015-06-26 15:03 UTC|newest]

Thread overview: 82+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2015-06-26 14:47 [Qemu-devel] [RFC PATCH V6 00/18] Multithread TCG fred.konrad
2015-06-26 14:47 ` [Qemu-devel] [RFC PATCH V6 01/18] cpu: make cpu_thread_is_idle public fred.konrad
2015-07-07  9:47   ` Alex Bennée
2015-07-07 11:43     ` Frederic Konrad
2015-06-26 14:47 ` [Qemu-devel] [RFC PATCH V6 02/18] replace spinlock by QemuMutex fred.konrad
2015-07-07 10:15   ` Alex Bennée
2015-07-07 10:22     ` Paolo Bonzini
2015-07-07 11:48       ` Frederic Konrad
2015-07-07 12:34         ` Paolo Bonzini
2015-07-07 13:06           ` Frederic Konrad
2015-07-07 11:46     ` Frederic Konrad
2015-06-26 14:47 ` [Qemu-devel] [RFC PATCH V6 03/18] remove unused spinlock fred.konrad
2015-06-26 14:53   ` Paolo Bonzini
2015-06-26 15:29     ` Frederic Konrad
2015-06-26 15:46       ` Paolo Bonzini
2015-06-26 14:47 ` [Qemu-devel] [RFC PATCH V6 04/18] add support for spin lock on POSIX systems exclusively fred.konrad
2015-06-26 14:55   ` Paolo Bonzini
2015-06-26 15:31     ` Frederic Konrad
2015-06-26 14:47 ` [Qemu-devel] [RFC PATCH V6 05/18] protect TBContext with tb_lock fred.konrad
2015-06-26 14:56   ` Paolo Bonzini
2015-06-26 15:39     ` Frederic Konrad
2015-06-26 15:45       ` Paolo Bonzini
2015-06-26 16:20   ` Paolo Bonzini
2015-07-07 12:22   ` Alex Bennée
2015-07-07 13:16     ` Frederic Konrad
2015-06-26 14:47 ` [Qemu-devel] [RFC PATCH V6 06/18] tcg: remove tcg_halt_cond global variable fred.konrad
2015-06-26 15:02   ` Paolo Bonzini [this message]
2015-06-26 15:41     ` Frederic Konrad
2015-07-07 12:27       ` Alex Bennée
2015-07-07 13:17         ` Frederic Konrad
2015-06-26 14:47 ` [Qemu-devel] [RFC PATCH V6 07/18] Drop global lock during TCG code execution fred.konrad
2015-06-26 14:56   ` Jan Kiszka
2015-06-26 15:08     ` Paolo Bonzini
2015-06-26 15:36     ` Frederic Konrad
2015-06-26 15:42       ` Jan Kiszka
2015-06-26 16:11         ` Frederic Konrad
2015-07-07 12:33       ` Alex Bennée
2015-07-07 13:18         ` Frederic Konrad
2015-06-26 14:47 ` [Qemu-devel] [RFC PATCH V6 08/18] cpu: remove exit_request global fred.konrad
2015-06-26 15:03   ` Paolo Bonzini
2015-07-07 13:04   ` Alex Bennée
2015-07-07 13:25     ` Frederic Konrad
2015-06-26 14:47 ` [Qemu-devel] [RFC PATCH V6 09/18] cpu: add a tcg_executing flag fred.konrad
2015-07-07 13:23   ` Alex Bennée
2015-07-07 13:30     ` Frederic Konrad
2015-06-26 14:47 ` [Qemu-devel] [RFC PATCH V6 10/18] tcg: switch on multithread fred.konrad
2015-07-07 13:40   ` Alex Bennée
2015-06-26 14:47 ` [Qemu-devel] [RFC PATCH V6 11/18] cpus: make qemu_cpu_kick_thread public fred.konrad
2015-07-07 15:11   ` Alex Bennée
2015-06-26 14:47 ` [Qemu-devel] [RFC PATCH V6 12/18] Use atomic cmpxchg to atomically check the exclusive value in a STREX fred.konrad
2015-06-26 14:47 ` [Qemu-devel] [RFC PATCH V6 13/18] cpu: introduce async_run_safe_work_on_cpu fred.konrad
2015-06-26 15:35   ` Paolo Bonzini
2015-06-26 16:09     ` Frederic Konrad
2015-06-26 16:23       ` Paolo Bonzini
2015-06-26 16:36         ` Frederic Konrad
2015-06-26 14:47 ` [Qemu-devel] [RFC PATCH V6 14/18] add a callback when tb_invalidate is called fred.konrad
2015-06-26 16:20   ` Paolo Bonzini
2015-06-26 16:40     ` Frederic Konrad
2015-07-07 15:32   ` Alex Bennée
2015-06-26 14:47 ` [Qemu-devel] [RFC PATCH V6 15/18] cpu: introduce tlb_flush*_all fred.konrad
2015-06-26 15:15   ` Paolo Bonzini
2015-06-26 15:54     ` Frederic Konrad
2015-06-26 16:01       ` Paolo Bonzini
2015-06-26 16:08         ` Peter Maydell
2015-06-26 16:30           ` Frederic Konrad
2015-06-26 16:31             ` Paolo Bonzini
2015-06-26 16:35               ` Frederic Konrad
2015-06-26 16:39                 ` Paolo Bonzini
2015-07-06 14:29             ` Mark Burton
2015-07-07 16:12               ` Alex Bennée
2015-06-26 16:54           ` Paolo Bonzini
2015-07-08 15:35           ` Frederic Konrad
2015-07-07 15:52   ` Alex Bennée
2015-06-26 14:47 ` [Qemu-devel] [RFC PATCH V6 16/18] arm: use tlb_flush*_all fred.konrad
2015-07-07 16:14   ` Alex Bennée
2015-06-26 14:47 ` [Qemu-devel] [RFC PATCH V6 17/18] translate-all: introduces tb_flush_safe fred.konrad
2015-07-07 16:16   ` Alex Bennée
2015-06-26 14:47 ` [Qemu-devel] [RFC PATCH V6 18/18] translate-all: (wip) use tb_flush_safe when we can't alloc more tb fred.konrad
2015-06-26 16:21   ` Paolo Bonzini
2015-06-26 16:38     ` Frederic Konrad
2015-07-07 16:17   ` Alex Bennée
2015-07-07 16:23     ` Frederic Konrad

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=558D699A.2010804@redhat.com \
    --to=pbonzini@redhat.com \
    --cc=a.spyridakis@virtualopensystems.com \
    --cc=agraf@suse.de \
    --cc=alex.bennee@linaro.org \
    --cc=alistair.francis@xilinx.com \
    --cc=fred.konrad@greensocs.com \
    --cc=guillaume.delbergue@greensocs.com \
    --cc=mark.burton@greensocs.com \
    --cc=mttcg@greensocs.com \
    --cc=peter.maydell@linaro.org \
    --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).