jcs.org
qemu with hax to log dma reads & writes
jcs.org/2018/11/12/vfio
usb-host: workaround libusb bug
libusb seems to no allways call the completion callback for requests
canceled (which it is supposed to do according to the docs). So add
a limit to avoid qemu waiting forever.
Tested-by: BALATON Zoltan <balaton@eik.bme.hu>
Signed-off-by: Gerd Hoffmann <kraxel@redhat.com>
Message-Id: <20200529072225.3195-1-kraxel@redhat.com>
+14
+14
hw/usb/host-libusb.c
+14
hw/usb/host-libusb.c
···
972
972
static void usb_host_abort_xfers(USBHostDevice *s)
973
973
{
974
974
USBHostRequest *r, *rtmp;
975
+
int limit = 100;
975
976
976
977
QTAILQ_FOREACH_SAFE(r, &s->requests, next, rtmp) {
977
978
usb_host_req_abort(r);
···
982
983
memset(&tv, 0, sizeof(tv));
983
984
tv.tv_usec = 2500;
984
985
libusb_handle_events_timeout(ctx, &tv);
986
+
if (--limit == 0) {
987
+
/*
988
+
* Don't wait forever for libusb calling the complete
989
+
* callback (which will unlink and free the request).
990
+
*
991
+
* Leaking memory here, to make sure libusb will not
992
+
* access memory which we have released already.
993
+
*/
994
+
QTAILQ_FOREACH_SAFE(r, &s->requests, next, rtmp) {
995
+
QTAILQ_REMOVE(&s->requests, r, next);
996
+
}
997
+
return;
998
+
}
985
999
}
986
1000
}
987
1001