| Message ID | 20100521175520.GA25276@redhat.com |
|---|---|
| State | New |
| Headers | show |
On 05/21/10 19:55, Shahar Havivi wrote: > Remove usb_host_device_release and using usb_host_close to handle usb_del command. > Gerd, What do you think about the usb_cleanup()? We need a mechanism to handle this for sure. I don't like that usb-specific approach very much though. I think we should either do that at qdev level, then at exit walk the whole device tree and call cleanup functions (if present). So every device has the chance to do cleanups when needed. Or we could have a exit notifier, which can be used for device (and also other) cleanup work. I tend to think that a exit notifier will be better. We probably have only a few devices which actually have to do some cleanup work (usb passthrough, maybe pci passthrough too), so building qdev infrastructure for that feels a bit like overkill. And exit notifiers are more generic, i.e. it will also work for non-device stuff. cheers, Gerd
On Tue, May 25, 2010 at 10:58:50AM +0200, Gerd Hoffmann wrote: > Date: Tue, 25 May 2010 10:58:50 +0200 > From: Gerd Hoffmann <kraxel@redhat.com> > To: Shahar Havivi <shaharh@redhat.com> > CC: "David S. Ahern" <daahern@cisco.com>, > Markus Armbruster <armbru@redhat.com>, qemu-devel@nongnu.org > Subject: Re: [PATCH] Release usb devices on shutdown and usb_del command > > On 05/21/10 19:55, Shahar Havivi wrote: > >Remove usb_host_device_release and using usb_host_close to handle usb_del command. > >Gerd, What do you think about the usb_cleanup()? > > We need a mechanism to handle this for sure. I don't like that > usb-specific approach very much though. > > I think we should either do that at qdev level, then at exit walk > the whole device tree and call cleanup functions (if present). So > every device has the chance to do cleanups when needed. > > Or we could have a exit notifier, which can be used for device (and > also other) cleanup work. > > I tend to think that a exit notifier will be better. We probably > have only a few devices which actually have to do some cleanup work > (usb passthrough, maybe pci passthrough too), so building qdev > infrastructure for that feels a bit like overkill. And exit > notifiers are more generic, i.e. it will also work for non-device > stuff. > > cheers, > Gerd > We can add at exit to the first device that added in usb-linux.c via usb_host_device_open() or usb_host_initfn() and the exit will iterate all usb devices in linux only, How is that solution?
diff --git a/hw/usb.h b/hw/usb.h index 00d2802..7ddf63c 100644 --- a/hw/usb.h +++ b/hw/usb.h @@ -258,6 +258,7 @@ void usb_send_msg(USBDevice *dev, int msg); USBDevice *usb_host_device_open(const char *devname); int usb_host_device_close(const char *devname); void usb_host_info(Monitor *mon); +void usb_cleanup(void); /* usb-hid.c */ void usb_hid_datain_cb(USBDevice *dev, void *opaque, void (*datain)(void *)); diff --git a/usb-bsd.c b/usb-bsd.c index 48567a3..c3eb891 100644 --- a/usb-bsd.c +++ b/usb-bsd.c @@ -634,3 +634,8 @@ int usb_host_device_close(const char *devname) { return 0; } + +void usb_cleanup(void) +{ + return 0; +} diff --git a/usb-linux.c b/usb-linux.c index 88273ff..98909b7 100644 --- a/usb-linux.c +++ b/usb-linux.c @@ -286,6 +286,17 @@ static void async_cancel(USBPacket *unused, void *opaque) } } +void usb_cleanup(void) +{ + struct USBHostDevice *s; + + QTAILQ_FOREACH(s, &hostdevs, next) { + if (s->fd != -1) { + ioctl(s->fd, USBDEVFS_RESET); + } + } +} + static int usb_host_claim_interfaces(USBHostDevice *dev, int configuration) { int dev_descr_len, config_descr_len; @@ -991,6 +1002,7 @@ static int usb_host_close(USBHostDevice *dev) async_complete(dev); dev->closing = 0; usb_device_detach(&dev->dev); + ioctl(dev->fd, USBDEVFS_RESET); close(dev->fd); dev->fd = -1; return 0; diff --git a/usb-stub.c b/usb-stub.c index 9c3fcea..bb513de 100644 --- a/usb-stub.c +++ b/usb-stub.c @@ -50,3 +50,8 @@ int usb_host_device_close(const char *devname) { return 0; } + +void usb_cleanup(void) +{ + return 0; +} diff --git a/vl.c b/vl.c index d77b47c..e3f4dc9 100644 --- a/vl.c +++ b/vl.c @@ -3914,6 +3914,7 @@ int main(int argc, char **argv, char **envp) main_loop(); quit_timers(); net_cleanup(); + usb_cleanup(); return 0; }
Remove usb_host_device_release and using usb_host_close to handle usb_del command. Gerd, What do you think about the usb_cleanup()? If it will be in usb-linux.c we will have to ad atexit to each device, if we usb-bus.c we will have to implement in bsd and stub... Signed-off-by: Shahar Havivi <shaharh@redhat.com> --- hw/usb.h | 1 + usb-bsd.c | 5 +++++ usb-linux.c | 12 ++++++++++++ usb-stub.c | 5 +++++ vl.c | 1 + 5 files changed, 24 insertions(+), 0 deletions(-)