From mboxrd@z Thu Jan 1 00:00:00 1970 Received: from eggs.gnu.org ([2001:4830:134:3::10]:43655) by lists.gnu.org with esmtp (Exim 4.71) (envelope-from ) id 1ZkaXp-0004x4-5d for qemu-devel@nongnu.org; Fri, 09 Oct 2015 12:30:10 -0400 Received: from Debian-exim by eggs.gnu.org with spam-scanned (Exim 4.71) (envelope-from ) id 1ZkaWj-0004pc-Vz for qemu-devel@nongnu.org; Fri, 09 Oct 2015 12:29:20 -0400 Date: Sat, 10 Oct 2015 00:27:00 +0800 From: Fam Zheng Message-ID: <20151009162700.GA11073@ad.nay.redhat.com> References: <1444369526-2227-1-git-send-email-famz@redhat.com> <1444369526-2227-6-git-send-email-famz@redhat.com> <20151009143126.GK3956@noname.redhat.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <20151009143126.GK3956@noname.redhat.com> Subject: Re: [Qemu-devel] [PATCH 05/12] aio: introduce aio_{disable, enable}_clients List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , To: Kevin Wolf Cc: Paolo Bonzini , qemu-block@nongnu.org, qemu-devel@nongnu.org, Stefan Hajnoczi On Fri, 10/09 16:31, Kevin Wolf wrote: > Am 09.10.2015 um 07:45 hat Fam Zheng geschrieben: > > Signed-off-by: Fam Zheng > > --- > > aio-posix.c | 3 ++- > > aio-win32.c | 3 ++- > > async.c | 42 ++++++++++++++++++++++++++++++++++++++++++ > > include/block/aio.h | 30 ++++++++++++++++++++++++++++++ > > 4 files changed, 76 insertions(+), 2 deletions(-) > > > > diff --git a/aio-posix.c b/aio-posix.c > > index d25fcfc..a261892 100644 > > --- a/aio-posix.c > > +++ b/aio-posix.c > > @@ -261,7 +261,8 @@ bool aio_poll(AioContext *ctx, bool blocking) > > > > /* fill pollfds */ > > QLIST_FOREACH(node, &ctx->aio_handlers, node) { > > - if (!node->deleted && node->pfd.events) { > > + if (!node->deleted && node->pfd.events > > + && !aio_type_disabled(ctx, node->type)) { > > add_pollfd(node); > > } > > } > > diff --git a/aio-win32.c b/aio-win32.c > > index f5ecf57..66cff60 100644 > > --- a/aio-win32.c > > +++ b/aio-win32.c > > @@ -309,7 +309,8 @@ bool aio_poll(AioContext *ctx, bool blocking) > > /* fill fd sets */ > > count = 0; > > QLIST_FOREACH(node, &ctx->aio_handlers, node) { > > - if (!node->deleted && node->io_notify) { > > + if (!node->deleted && node->io_notify > > + && !aio_type_disabled(ctx, node->type)) { > > events[count++] = event_notifier_get_handle(node->e); > > } > > } > > diff --git a/async.c b/async.c > > index 244bf79..855b9d5 100644 > > --- a/async.c > > +++ b/async.c > > @@ -361,3 +361,45 @@ void aio_context_release(AioContext *ctx) > > { > > rfifolock_unlock(&ctx->lock); > > } > > + > > +bool aio_type_disabled(AioContext *ctx, int type) > > +{ > > + int i = 1; > > + int n = 0; > > + > > + while (type) { > > + bool b = type & 0x1; > > + type >>= 1; > > + n++; > > Any specific reason for leaving client_disable_counters[0] unused? No, I should have started from 0. > > > + i <<= 1; > > i is never read. > > > + if (!b) { > > + continue; > > + } > > + if (ctx->client_disable_counters[n]) { > > + return true; > > + } > > + } > > + return false; > > +} > > In general I wonder whether this function really needs to take a mask > with possibly multiple set bits instead of just a single type. Previous versions used to have more types than "internal" and "external", so it has been a mask. So yes, I think a single type will be better now. > > > +void aio_disable_enable_clients(AioContext *ctx, int clients_mask, > > + bool is_disable) > > +{ > > + int i = 1; > > + int n = 0; > > + aio_context_acquire(ctx); > > + > > + while (clients_mask) { > > + bool b = clients_mask & 0x1; > > + clients_mask >>= 1; > > + n++; > > + i <<= 1; > > This i isn't used either. > > > + if (!b) { > > + continue; > > + } > > + if (ctx->client_disable_counters[n]) { > > + return true; > > + } > > Wait, why are you checking the state instead of setting it? Oops, apparent I screwed my workspaces as I do remember coding this assignment. And I must have used a wrong command when building the tree so that I don't even catch the compiling error. :( > > How did you test this series? So far only smoke testing and qemu-iotests, because I don't have a good idea of testifying the transaction's atomicity. Any suggestions? > > > + } > > + aio_context_release(ctx); > > +} > > diff --git a/include/block/aio.h b/include/block/aio.h > > index 60e796b..b687152 100644 > > --- a/include/block/aio.h > > +++ b/include/block/aio.h > > @@ -122,6 +122,8 @@ struct AioContext { > > > > /* TimerLists for calling timers - one per clock type */ > > QEMUTimerListGroup tlg; > > + > > + int client_disable_counters[sizeof(int)]; > > }; > > sizeof return the size in bytes. I think you mean bits here? You're right, will fix it. Fam