| Message ID | 1270140161-17216-5-git-send-email-agraf@suse.de |
|---|---|
| State | New |
| Headers | show |
On (Thu) Apr 01 2010 [18:42:39], Alexander Graf wrote: > When using virtio-console on s390, the input doesn't work. > > The root of the problem is rather simple. What happens is the following: > > 1) create character device for stdio > 2) char device is done creating, sends OPENED event > 3) virtio-console adds handlers > 4) no event comes because the char device is open already > 5) virtio-console doesn't accept input because it didn't > receive an OPENED event > > To make that sure virtio-console gets notified that the character device > is open even when it's been open from the beginning, this patch introduces > a variable that keeps track of the opened state. If the device is open when > the event handlers get installed, we just notify the handler. > > This fixes input with virtio-console on s390. > > Signed-off-by: Alexander Graf <agraf@suse.de> Acked-by: Amit Shah <amit.shah@redhat.com> Amit
On Thu, Apr 01, 2010 at 06:42:39PM +0200, Alexander Graf wrote: > When using virtio-console on s390, the input doesn't work. > > The root of the problem is rather simple. What happens is the following: > > 1) create character device for stdio > 2) char device is done creating, sends OPENED event > 3) virtio-console adds handlers > 4) no event comes because the char device is open already > 5) virtio-console doesn't accept input because it didn't > receive an OPENED event > > To make that sure virtio-console gets notified that the character device > is open even when it's been open from the beginning, this patch introduces > a variable that keeps track of the opened state. If the device is open when > the event handlers get installed, we just notify the handler. > > This fixes input with virtio-console on s390. Thanks, applied. > Signed-off-by: Alexander Graf <agraf@suse.de> > --- > qemu-char.c | 20 ++++++++++++++++++++ > qemu-char.h | 1 + > 2 files changed, 21 insertions(+), 0 deletions(-) > > diff --git a/qemu-char.c b/qemu-char.c > index 6ad6609..a9d9442 100644 > --- a/qemu-char.c > +++ b/qemu-char.c > @@ -109,6 +109,16 @@ static QTAILQ_HEAD(CharDriverStateHead, CharDriverState) chardevs = > > static void qemu_chr_event(CharDriverState *s, int event) > { > + /* Keep track if the char device is open */ > + switch (event) { > + case CHR_EVENT_OPENED: > + s->opened = 1; > + break; > + case CHR_EVENT_CLOSED: > + s->opened = 0; > + break; > + } > + > if (!s->chr_event) > return; > s->chr_event(s->handler_opaque, event); > @@ -193,6 +203,12 @@ void qemu_chr_add_handlers(CharDriverState *s, > s->handler_opaque = opaque; > if (s->chr_update_read_handler) > s->chr_update_read_handler(s); > + > + /* We're connecting to an already opened device, so let's make sure we > + also get the open event */ > + if (s->opened) { > + qemu_chr_generic_open(s); > + } > } > > static int null_chr_write(CharDriverState *chr, const uint8_t *buf, int len) > @@ -475,6 +491,10 @@ static CharDriverState *qemu_chr_open_mux(CharDriverState *drv) > chr->chr_write = mux_chr_write; > chr->chr_update_read_handler = mux_chr_update_read_handler; > chr->chr_accept_input = mux_chr_accept_input; > + > + /* Muxes are always open on creation */ > + qemu_chr_generic_open(chr); > + > return chr; > } > > diff --git a/qemu-char.h b/qemu-char.h > index 3a9427b..e3a0783 100644 > --- a/qemu-char.h > +++ b/qemu-char.h > @@ -67,6 +67,7 @@ struct CharDriverState { > QEMUBH *bh; > char *label; > char *filename; > + int opened; > QTAILQ_ENTRY(CharDriverState) next; > }; > > -- > 1.6.0.2 > > > >
diff --git a/qemu-char.c b/qemu-char.c index 6ad6609..a9d9442 100644 --- a/qemu-char.c +++ b/qemu-char.c @@ -109,6 +109,16 @@ static QTAILQ_HEAD(CharDriverStateHead, CharDriverState) chardevs = static void qemu_chr_event(CharDriverState *s, int event) { + /* Keep track if the char device is open */ + switch (event) { + case CHR_EVENT_OPENED: + s->opened = 1; + break; + case CHR_EVENT_CLOSED: + s->opened = 0; + break; + } + if (!s->chr_event) return; s->chr_event(s->handler_opaque, event); @@ -193,6 +203,12 @@ void qemu_chr_add_handlers(CharDriverState *s, s->handler_opaque = opaque; if (s->chr_update_read_handler) s->chr_update_read_handler(s); + + /* We're connecting to an already opened device, so let's make sure we + also get the open event */ + if (s->opened) { + qemu_chr_generic_open(s); + } } static int null_chr_write(CharDriverState *chr, const uint8_t *buf, int len) @@ -475,6 +491,10 @@ static CharDriverState *qemu_chr_open_mux(CharDriverState *drv) chr->chr_write = mux_chr_write; chr->chr_update_read_handler = mux_chr_update_read_handler; chr->chr_accept_input = mux_chr_accept_input; + + /* Muxes are always open on creation */ + qemu_chr_generic_open(chr); + return chr; } diff --git a/qemu-char.h b/qemu-char.h index 3a9427b..e3a0783 100644 --- a/qemu-char.h +++ b/qemu-char.h @@ -67,6 +67,7 @@ struct CharDriverState { QEMUBH *bh; char *label; char *filename; + int opened; QTAILQ_ENTRY(CharDriverState) next; };
When using virtio-console on s390, the input doesn't work. The root of the problem is rather simple. What happens is the following: 1) create character device for stdio 2) char device is done creating, sends OPENED event 3) virtio-console adds handlers 4) no event comes because the char device is open already 5) virtio-console doesn't accept input because it didn't receive an OPENED event To make that sure virtio-console gets notified that the character device is open even when it's been open from the beginning, this patch introduces a variable that keeps track of the opened state. If the device is open when the event handlers get installed, we just notify the handler. This fixes input with virtio-console on s390. Signed-off-by: Alexander Graf <agraf@suse.de> --- qemu-char.c | 20 ++++++++++++++++++++ qemu-char.h | 1 + 2 files changed, 21 insertions(+), 0 deletions(-)