All of lore.kernel.org
 help / color / mirror / Atom feed
From: Anthony Liguori <aliguori@us.ibm.com>
To: Paolo Bonzini <pbonzini@redhat.com>, qemu-devel@nongnu.org
Cc: amit.shah@redhat.com, peter.crosthwaite@xilinx.com, gson@gson.org
Subject: Re: [Qemu-devel] [PATCH] qemu-char: eliminate busy waiting on can_read returning zero
Date: Fri, 05 Apr 2013 07:54:39 -0500	[thread overview]
Message-ID: <87ip41f88g.fsf@codemonkey.ws> (raw)
In-Reply-To: <1365147607-18298-1-git-send-email-pbonzini@redhat.com>

Paolo Bonzini <pbonzini@redhat.com> writes:

> The character backend refactoring introduced an undesirable busy wait.
> The busy wait happens if can_read returns zero and there is data available
> on the character device's file descriptor.  Then, the I/O watch will
> fire continuously and, with TCG, the CPU thread will never run.
>
>     1) Char backend asks front end if it can write
>     2) Front end says no
>     3) poll() finds the char backend's descriptor is available
>     4) Goto (1)
>
> What we really want is this (note that step 3 avoids the busy wait):
>
>     1) Char backend asks front end if it can write
>     2) Front end says no
>     3) poll() goes on without char backend's descriptor
>     4) Goto (1) until qemu_chr_accept_input() called
>
>     5) Char backend asks front end if it can write
>     6) Front end says yes
>     7) poll() finds the char backend's descriptor is available
>     8) Backend handler called
>
> After this patch, the IOWatchPoll source and the watch source are
> separated.  The IOWatchPoll is simply a hook that runs during the prepare
> phase on each main loop iteration.  The hook adds/removes the actual
> source depending on the return value from can_read.
>
> A simple reproducer is
>
>     qemu-system-i386 -serial mon:stdio
>
> ... followed by banging on the terminal as much as you can. :)  Without
> this patch, emulation will hang.
>
> Signed-off-by: Paolo Bonzini <pbonzini@redhat.com>

I guess this works with migration because we assume that after migration
the main loop will do a complete run?  Is this a safe assumption or does
there need to be a qemu_notify_event() somewhere after migration to make
sure this doesn't cause a hang?

Regards,

Anthony Liguori

> ---
>         This supersedes Peter's patch.
>
>  qemu-char.c  | 64 +++++++++++++++++++++-------------------------------------------
>  1 file changed, 21 insertions(+), 43 deletions(-)
>
> diff --git a/qemu-char.c b/qemu-char.c
> index e5eb8dd..d4239b5 100644
> --- a/qemu-char.c
> +++ b/qemu-char.c
> @@ -594,65 +594,52 @@ int recv_all(int fd, void *_buf, int len1, bool single_read)
>  
>  typedef struct IOWatchPoll
>  {
> +    GSource parent;
> +
>      GSource *src;
> -    int max_size;
> +    bool active;
>  
>      IOCanReadHandler *fd_can_read;
>      void *opaque;
> -
> -    QTAILQ_ENTRY(IOWatchPoll) node;
>  } IOWatchPoll;
>  
> -static QTAILQ_HEAD(, IOWatchPoll) io_watch_poll_list =
> -    QTAILQ_HEAD_INITIALIZER(io_watch_poll_list);
> -
>  static IOWatchPoll *io_watch_poll_from_source(GSource *source)
>  {
> -    IOWatchPoll *i;
> -
> -    QTAILQ_FOREACH(i, &io_watch_poll_list, node) {
> -        if (i->src == source) {
> -            return i;
> -        }
> -    }
> -
> -    return NULL;
> +    return container_of(source, IOWatchPoll, parent);
>  }
>  
>  static gboolean io_watch_poll_prepare(GSource *source, gint *timeout_)
>  {
>      IOWatchPoll *iwp = io_watch_poll_from_source(source);
> -
> -    iwp->max_size = iwp->fd_can_read(iwp->opaque);
> -    if (iwp->max_size == 0) {
> +    bool active = iwp->fd_can_read(iwp->opaque) > 0;
> +    if (iwp->active == active) {
>          return FALSE;
>      }
>  
> -    return g_io_watch_funcs.prepare(source, timeout_);
> +    iwp->active = active;
> +    if (active) {
> +        g_source_attach(iwp->src, NULL);
> +    } else {
> +        g_source_remove(g_source_get_id(iwp->src));
> +    }
> +    return FALSE;
>  }
>  
>  static gboolean io_watch_poll_check(GSource *source)
>  {
> -    IOWatchPoll *iwp = io_watch_poll_from_source(source);
> -
> -    if (iwp->max_size == 0) {
> -        return FALSE;
> -    }
> -
> -    return g_io_watch_funcs.check(source);
> +    return FALSE;
>  }
>  
>  static gboolean io_watch_poll_dispatch(GSource *source, GSourceFunc callback,
>                                         gpointer user_data)
>  {
> -    return g_io_watch_funcs.dispatch(source, callback, user_data);
> +    abort();
>  }
>  
>  static void io_watch_poll_finalize(GSource *source)
>  {
>      IOWatchPoll *iwp = io_watch_poll_from_source(source);
> -    QTAILQ_REMOVE(&io_watch_poll_list, iwp, node);
> -    g_io_watch_funcs.finalize(source);
> +    g_source_unref(iwp->src);
>  }
>  
>  static GSourceFuncs io_watch_poll_funcs = {
> @@ -669,24 +657,15 @@ static guint io_add_watch_poll(GIOChannel *channel,
>                                 gpointer user_data)
>  {
>      IOWatchPoll *iwp;
> -    GSource *src;
> -    guint tag;
> -
> -    src = g_io_create_watch(channel, G_IO_IN | G_IO_ERR | G_IO_HUP);
> -    g_source_set_funcs(src, &io_watch_poll_funcs);
> -    g_source_set_callback(src, (GSourceFunc)fd_read, user_data, NULL);
> -    tag = g_source_attach(src, NULL);
> -    g_source_unref(src);
>  
> -    iwp = g_malloc0(sizeof(*iwp));
> -    iwp->src = src;
> -    iwp->max_size = 0;
> +    iwp = (IOWatchPoll *) g_source_new(&io_watch_poll_funcs, sizeof(IOWatchPoll));
> +    iwp->active = FALSE;
>      iwp->fd_can_read = fd_can_read;
>      iwp->opaque = user_data;
> +    iwp->src = g_io_create_watch(channel, G_IO_IN | G_IO_ERR | G_IO_HUP);
> +    g_source_set_callback(iwp->src, (GSourceFunc)fd_read, user_data, NULL);
>  
> -    QTAILQ_INSERT_HEAD(&io_watch_poll_list, iwp, node);
> -
> -    return tag;
> +    return g_source_attach(&iwp->parent, NULL);
>  }
>  
>  #ifndef _WIN32

  parent reply	other threads:[~2013-04-05 12:54 UTC|newest]

Thread overview: 5+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2013-04-05  7:40 [Qemu-devel] [PATCH] qemu-char: eliminate busy waiting on can_read returning zero Paolo Bonzini
2013-04-05 12:34 ` Anthony Liguori
2013-04-05 12:54 ` Anthony Liguori [this message]
2013-04-05 13:01   ` Paolo Bonzini
2013-04-05 13:46 ` Anthony Liguori

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=87ip41f88g.fsf@codemonkey.ws \
    --to=aliguori@us.ibm.com \
    --cc=amit.shah@redhat.com \
    --cc=gson@gson.org \
    --cc=pbonzini@redhat.com \
    --cc=peter.crosthwaite@xilinx.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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.