[Qemu-devel] [5344] Implement a HCI passthrough to host.

[Qemu-devel] [5344] Implement a HCI passthrough to host.

[Andrzej Zaborowski]

Revision: 5344
          http://svn.sv.gnu.org/viewvc/?view=rev&root=qemu&revision=5344
Author:   balrog
Date:     2008-09-28 23:49:55 +0000 (Sun, 28 Sep 2008)

Log Message:
-----------
Implement a HCI passthrough to host.

This allows using a host's physical HCI as one of the HCIs attached
to the virtual machine.  This brings various limitations because not
all commands/events are passed through by Linux kernel, some are
interpreted by the host's kernel for a speed gain.

Modified Paths:
--------------
    trunk/Makefile
    trunk/Makefile.target
    trunk/configure

Added Paths:
-----------
    trunk/bt-host.c

Modified: trunk/Makefile
===================================================================
--- trunk/Makefile	2008-09-28 23:34:13 UTC (rev 5343)
+++ trunk/Makefile	2008-09-28 23:49:55 UTC (rev 5344)
@@ -81,7 +81,7 @@
 OBJS+=usb.o usb-hub.o usb-linux.o usb-hid.o usb-msd.o usb-wacom.o
 OBJS+=usb-serial.o usb-net.o
 OBJS+=sd.o ssi-sd.o
-OBJS+=bt.o
+OBJS+=bt.o bt-host.o
 
 ifdef CONFIG_BRLAPI
 OBJS+= baum.o
@@ -166,6 +166,9 @@
 curses.o: curses.c keymaps.c curses_keys.h
 	$(CC) $(CFLAGS) $(CPPFLAGS) -c -o $@ $<
 
+bt-host.o: bt-host.c
+	$(CC) $(CFLAGS) $(CPPFLAGS) $(CONFIG_BLUEZ_CFLAGS) -c -o $@ $<
+
 audio/sdlaudio.o: audio/sdlaudio.c
 	$(CC) $(CFLAGS) $(CPPFLAGS) $(SDL_CFLAGS) -c -o $@ $<
 

Modified: trunk/Makefile.target
===================================================================
--- trunk/Makefile.target	2008-09-28 23:34:13 UTC (rev 5343)
+++ trunk/Makefile.target	2008-09-28 23:49:55 UTC (rev 5344)
@@ -524,6 +524,10 @@
 LIBS += $(CONFIG_VNC_TLS_LIBS)
 endif
 
+ifdef CONFIG_BLUEZ
+LIBS += $(CONFIG_BLUEZ_LIBS)
+endif
+
 # SCSI layer
 OBJS+= lsi53c895a.o esp.o
 

Added: trunk/bt-host.c
===================================================================
--- trunk/bt-host.c	                        (rev 0)
+++ trunk/bt-host.c	2008-09-28 23:49:55 UTC (rev 5344)
@@ -0,0 +1,198 @@
+/*
+ * Wrap a host Bluetooth HCI socket in a struct HCIInfo.
+ *
+ * Copyright (C) 2008 Andrzej Zaborowski  <balrog@zabor.org>
+ *
+ * This program is free software; you can redistribute it and/or
+ * modify it under the terms of the GNU General Public License as
+ * published by the Free Software Foundation; either version 2 or
+ * (at your option) version 3 of the License.
+ *
+ * This program is distributed in the hope that it will be useful,
+ * but WITHOUT ANY WARRANTY; without even the implied warranty of
+ * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
+ * GNU General Public License for more details.
+ *
+ * You should have received a copy of the GNU General Public License
+ * along with this program; if not, write to the Free Software
+ * Foundation, Inc., 59 Temple Place, Suite 330, Boston,
+ * MA 02111-1307 USA
+ */
+
+#include "qemu-common.h"
+#include "qemu-char.h"
+#include "sysemu.h"
+#include "net.h"
+
+#include <errno.h>
+#include <sys/ioctl.h>
+#include <sys/uio.h>
+#ifdef CONFIG_BLUEZ
+# include <bluetooth/bluetooth.h>
+# include <bluetooth/hci.h>
+# include <bluetooth/hci_lib.h>
+#else
+# include "hw/bt.h"
+# define HCI_MAX_FRAME_SIZE	1028
+#endif
+
+struct bt_host_hci_s {
+    struct HCIInfo hci;
+    int fd;
+
+    uint8_t hdr[HCI_MAX_FRAME_SIZE];
+    int len;
+};
+
+static void bt_host_send(struct HCIInfo *hci,
+                int type, const uint8_t *data, int len)
+{
+    struct bt_host_hci_s *s = (struct bt_host_hci_s *) hci;
+    uint8_t pkt = type;
+    struct iovec iv[2];
+    int ret;
+
+    iv[0].iov_base = &pkt;
+    iv[0].iov_len  = 1;
+    iv[1].iov_base = (void *) data;
+    iv[1].iov_len  = len;
+
+    while ((ret = writev(s->fd, iv, 2)) < 0)
+        if (errno != EAGAIN && errno != EINTR) {
+            fprintf(stderr, "qemu: error %i writing bluetooth packet.\n",
+                            errno);
+            return;
+        }
+}
+
+static void bt_host_cmd(struct HCIInfo *hci, const uint8_t *data, int len)
+{
+    bt_host_send(hci, HCI_COMMAND_PKT, data, len);
+}
+
+static void bt_host_acl(struct HCIInfo *hci, const uint8_t *data, int len)
+{
+    bt_host_send(hci, HCI_ACLDATA_PKT, data, len);
+}
+
+static void bt_host_sco(struct HCIInfo *hci, const uint8_t *data, int len)
+{
+    bt_host_send(hci, HCI_SCODATA_PKT, data, len);
+}
+
+static int bt_host_read_poll(void *opaque)
+{
+    struct bt_host_hci_s *s = (struct bt_host_hci_s *) opaque;
+
+    return !!s->hci.evt_recv;
+}
+
+static void bt_host_read(void *opaque)
+{
+    struct bt_host_hci_s *s = (struct bt_host_hci_s *) opaque;
+    uint8_t *pkt;
+    int pktlen;
+
+    /* Seems that we can't read only the header first and then the amount
+     * of data indicated in the header because Linux will discard everything
+     * that's not been read in one go.  */
+    s->len = read(s->fd, s->hdr, sizeof(s->hdr));
+
+    if (s->len < 0) {
+        fprintf(stderr, "qemu: error %i reading HCI frame\n", errno);
+        return;
+    }
+
+    pkt = s->hdr;
+    while (s->len --)
+        switch (*pkt ++) {
+        case HCI_EVENT_PKT:
+            if (s->len < 2)
+                goto bad_pkt;
+
+            pktlen = MIN(pkt[1] + 2, s->len);
+            s->hci.evt_recv(s->hci.opaque, pkt, pktlen);
+            s->len -= pktlen;
+            pkt += pktlen;
+
+            /* TODO: if this is an Inquiry Result event, it's also
+             * interpreted by Linux kernel before we received it, possibly
+             * we should clean the kernel Inquiry cache through
+             * ioctl(s->fd, HCI_INQUIRY, ...).  */
+            break;
+
+        case HCI_ACLDATA_PKT:
+            if (s->len < 4)
+                goto bad_pkt;
+
+            pktlen = MIN(((pkt[3] << 8) | pkt[2]) + 4, s->len);
+            s->hci.acl_recv(s->hci.opaque, pkt, pktlen);
+            s->len -= pktlen;
+            pkt += pktlen;
+            break;
+
+        case HCI_SCODATA_PKT:
+            if (s->len < 3)
+                goto bad_pkt;
+
+            pktlen = MIN(pkt[2] + 3, s->len);
+            s->len -= pktlen;
+            pkt += pktlen;
+
+        default:
+        bad_pkt:
+            fprintf(stderr, "qemu: bad HCI packet type %02x\n", pkt[-1]);
+        }
+}
+
+static int bt_host_bdaddr_set(struct HCIInfo *hci, const uint8_t *bd_addr)
+{
+    return -ENOTSUP;
+}
+
+struct HCIInfo *bt_host_hci(const char *id)
+{
+    struct bt_host_hci_s *s;
+    int fd = -1;
+#ifdef CONFIG_BLUEZ
+    int dev_id = hci_devid(id);
+    struct hci_filter flt;
+
+    if (dev_id < 0) {
+        fprintf(stderr, "qemu: `%s' not available\n", id);
+        return 0;
+    }
+
+    fd = hci_open_dev(dev_id);
+
+    /* XXX: can we ensure nobody else has the device opened?  */
+#endif
+
+    if (fd < 0) {
+        fprintf(stderr, "qemu: Can't open `%s': %s (%i)\n",
+                        id, strerror(errno), errno);
+        return 0;
+    }
+
+#ifdef CONFIG_BLUEZ
+    hci_filter_clear(&flt);
+    hci_filter_all_ptypes(&flt);
+    hci_filter_all_events(&flt);
+
+    if (setsockopt(fd, SOL_HCI, HCI_FILTER, &flt, sizeof(flt)) < 0) {
+        fprintf(stderr, "qemu: Can't set HCI filter on socket (%i)\n", errno);
+        return 0;
+    }
+#endif
+
+    s = qemu_mallocz(sizeof(struct bt_host_hci_s));
+    s->fd = fd;
+    s->hci.cmd_send = bt_host_cmd;
+    s->hci.sco_send = bt_host_sco;
+    s->hci.acl_send = bt_host_acl;
+    s->hci.bdaddr_set = bt_host_bdaddr_set;
+
+    qemu_set_fd_handler2(s->fd, bt_host_read_poll, bt_host_read, 0, s);
+
+    return &s->hci;
+}

Modified: trunk/configure
===================================================================
--- trunk/configure	2008-09-28 23:34:13 UTC (rev 5343)
+++ trunk/configure	2008-09-28 23:49:55 UTC (rev 5344)
@@ -110,6 +110,7 @@
 aio="yes"
 nptl="yes"
 mixemu="no"
+bluez="yes"
 signalfd="no"
 eventfd="no"
 
@@ -293,6 +294,8 @@
   ;;
   --disable-brlapi) brlapi="no"
   ;;
+  --disable-bluez) bluez="no"
+  ;;
   --enable-profiler) profiler="yes"
   ;;
   --enable-cocoa)
@@ -429,6 +432,7 @@
 echo "  --disable-brlapi         disable BrlAPI"
 echo "  --disable-vnc-tls        disable TLS encryption for VNC server"
 echo "  --disable-curses         disable curses output"
+echo "  --disable-bluez          disable bluez stack connectivity"
 echo "  --disable-nptl           disable usermode NPTL support"
 echo "  --enable-system          enable all system emulation targets"
 echo "  --disable-system         disable all system emulation targets"
@@ -891,6 +895,16 @@
 fi # test "$curses"
 
 ##########################################
+# bluez support probe
+if test "$bluez" = "yes" ; then
+  `pkg-config bluez` || bluez="no"
+fi
+if test "$bluez" = "yes" ; then
+  bluez_cflags=`pkg-config --cflags bluez`
+  bluez_libs=`pkg-config --libs bluez`
+fi
+
+##########################################
 # AIO probe
 if test "$aio" = "yes" ; then
   aio=no
@@ -1254,6 +1268,12 @@
   echo "#define CONFIG_BRLAPI 1" >> $config_h
   echo "BRLAPI_LIBS=-lbrlapi" >> $config_mak
 fi
+if test "$bluez" = "yes" ; then
+  echo "CONFIG_BLUEZ=yes" >> $config_mak
+  echo "CONFIG_BLUEZ_CFLAGS=$bluez_cflags" >> $config_mak
+  echo "CONFIG_BLUEZ_LIBS=$bluez_libs" >> $config_mak
+  echo "#define CONFIG_BLUEZ 1" >> $config_h
+fi
 if test "$aio" = "yes" ; then
   echo "#define CONFIG_AIO 1" >> $config_h
   echo "CONFIG_AIO=yes" >> $config_mak

Re: [Qemu-devel] [5344] Implement a HCI passthrough to host.

[Blue Swirl]

On 9/29/08, Andrzej Zaborowski <balrogg@gmail.com> wrote:
> Revision: 5344
>           http://svn.sv.gnu.org/viewvc/?view=rev&root=qemu&revision=5344
>  Author:   balrog
>  Date:     2008-09-28 23:49:55 +0000 (Sun, 28 Sep 2008)
>

>  +# bluez support probe
>  +if test "$bluez" = "yes" ; then
>  +  `pkg-config bluez` || bluez="no"
>  +fi

I think probing should be done more like curses or AIO probes.

Re: [Qemu-devel] [5344] Implement a HCI passthrough to host.

[Anthony Liguori]

Blue Swirl wrote:
> On 9/29/08, Andrzej Zaborowski <balrogg@gmail.com> wrote:
>   
>> Revision: 5344
>>           http://svn.sv.gnu.org/viewvc/?view=rev&root=qemu&revision=5344
>>  Author:   balrog
>>  Date:     2008-09-28 23:49:55 +0000 (Sun, 28 Sep 2008)
>>
>>     
>
>   
>>  +# bluez support probe
>>  +if test "$bluez" = "yes" ; then
>>  +  `pkg-config bluez` || bluez="no"
>>  +fi
>>     
>
> I think probing should be done more like curses or AIO probes.
>   

Mainly because this breaks cross-compilation.  See VNC TLS detection for 
how to detect something with pkg-config without breaking cross compilation.

Regards,

Anthony Liguori

Re: [Qemu-devel] [5344] Implement a HCI passthrough to host.

[Anthony Liguori]

Andrzej Zaborowski wrote:
> Revision: 5344
>           http://svn.sv.gnu.org/viewvc/?view=rev&root=qemu&revision=5344
> Author:   balrog
> Date:     2008-09-28 23:49:55 +0000 (Sun, 28 Sep 2008)
>
> Log Message:
> -----------
> Implement a HCI passthrough to host.
>
> This allows using a host's physical HCI as one of the HCIs attached
> to the virtual machine.  This brings various limitations because not
> all commands/events are passed through by Linux kernel, some are
> interpreted by the host's kernel for a speed gain.
>   

This broke the Windows build.  The whole file should probably be build 
conditionally on CONFIG_BLUEZ.

Regards,

Anthony Liguori

Re: [Qemu-devel] [5344] Implement a HCI passthrough to host.

[andrzej zaborowski]

2008/9/29 Anthony Liguori <anthony@codemonkey.ws>:
> Andrzej Zaborowski wrote:
>>
>> Revision: 5344
>>          http://svn.sv.gnu.org/viewvc/?view=rev&root=qemu&revision=5344
>> Author:   balrog
>> Date:     2008-09-28 23:49:55 +0000 (Sun, 28 Sep 2008)
>>
>> Log Message:
>> -----------
>> Implement a HCI passthrough to host.
>>
>> This allows using a host's physical HCI as one of the HCIs attached
>> to the virtual machine.  This brings various limitations because not
>> all commands/events are passed through by Linux kernel, some are
>> interpreted by the host's kernel for a speed gain.
>>
>
> This broke the Windows build.  The whole file should probably be build
> conditionally on CONFIG_BLUEZ.

Is that because of uio.h?  writev() seems to be posix.  Does win32
have it under a different name?

Added a stronger check for bluez, hopefully will cross-compile better now.

Cheers

Re: [Qemu-devel] [5344] Implement a HCI passthrough to host.

[Anthony Liguori]

andrzej zaborowski wrote:
> 2008/9/29 Anthony Liguori <anthony@codemonkey.ws>:
>   
>> Andrzej Zaborowski wrote:
>>     
>>> Revision: 5344
>>>          http://svn.sv.gnu.org/viewvc/?view=rev&root=qemu&revision=5344
>>> Author:   balrog
>>> Date:     2008-09-28 23:49:55 +0000 (Sun, 28 Sep 2008)
>>>
>>> Log Message:
>>> -----------
>>> Implement a HCI passthrough to host.
>>>
>>> This allows using a host's physical HCI as one of the HCIs attached
>>> to the virtual machine.  This brings various limitations because not
>>> all commands/events are passed through by Linux kernel, some are
>>> interpreted by the host's kernel for a speed gain.
>>>
>>>       
>> This broke the Windows build.  The whole file should probably be build
>> conditionally on CONFIG_BLUEZ.
>>     
>
> Is that because of uio.h?  writev() seems to be posix.  Does win32
> have it under a different name?
>   

Yeah, there is no uio in win32.  win32 isn't posix fwiw so a lot is missing.

> Added a stronger check for bluez, hopefully will cross-compile better now.
>   

I think the best solution is to not compile in the entire file if 
CONFIG_BLUEZ isn't set.  Then we don't have to worry about dealing with 
things like uio.

Regards,

Anthony Liguori

> Cheers
>
>
>   

Re: [Qemu-devel] [5344] Implement a HCI passthrough to host.

[Blue Swirl]

On 9/30/08, andrzej zaborowski <balrogg@gmail.com> wrote:
> 2008/9/29 Anthony Liguori <anthony@codemonkey.ws>:
>
> > Andrzej Zaborowski wrote:
>  >>
>  >> Revision: 5344
>  >>          http://svn.sv.gnu.org/viewvc/?view=rev&root=qemu&revision=5344
>  >> Author:   balrog
>  >> Date:     2008-09-28 23:49:55 +0000 (Sun, 28 Sep 2008)
>  >>
>  >> Log Message:
>  >> -----------
>  >> Implement a HCI passthrough to host.
>  >>
>  >> This allows using a host's physical HCI as one of the HCIs attached
>  >> to the virtual machine.  This brings various limitations because not
>  >> all commands/events are passed through by Linux kernel, some are
>  >> interpreted by the host's kernel for a speed gain.
>  >>
>  >
>  > This broke the Windows build.  The whole file should probably be build
>  > conditionally on CONFIG_BLUEZ.
>
>
> Is that because of uio.h?  writev() seems to be posix.  Does win32
>  have it under a different name?
>
>  Added a stronger check for bluez, hopefully will cross-compile better now.

You are still relying that the host pkg-config exists and gives
information that is valid for cross-compiling.

Re: [Qemu-devel] [5344] Implement a HCI passthrough to host.

[andrzej zaborowski]

2008/9/30 Blue Swirl <blauwirbel@gmail.com>:
> On 9/30/08, andrzej zaborowski <balrogg@gmail.com> wrote:
>> 2008/9/29 Anthony Liguori <anthony@codemonkey.ws>:
>>
>> > Andrzej Zaborowski wrote:
>>  >>
>>  >> Revision: 5344
>>  >>          http://svn.sv.gnu.org/viewvc/?view=rev&root=qemu&revision=5344
>>  >> Author:   balrog
>>  >> Date:     2008-09-28 23:49:55 +0000 (Sun, 28 Sep 2008)
>>  >>
>>  >> Log Message:
>>  >> -----------
>>  >> Implement a HCI passthrough to host.
>>  >>
>>  >> This allows using a host's physical HCI as one of the HCIs attached
>>  >> to the virtual machine.  This brings various limitations because not
>>  >> all commands/events are passed through by Linux kernel, some are
>>  >> interpreted by the host's kernel for a speed gain.
>>  >>
>>  >
>>  > This broke the Windows build.  The whole file should probably be build
>>  > conditionally on CONFIG_BLUEZ.
>>
>>
>> Is that because of uio.h?  writev() seems to be posix.  Does win32
>>  have it under a different name?
>>
>>  Added a stronger check for bluez, hopefully will cross-compile better now.
>
> You are still relying that the host pkg-config exists and gives
> information that is valid for cross-compiling.

Not really -- the information will be (likely) wrong for
cross-compiling and configure will disable bluez, which is what we
want -- unless the user sets PKG_CONFIG_PATH correctly.

Cheers

Re: [Qemu-devel] [5344] Implement a HCI passthrough to host.

[Anthony Liguori]

Blue Swirl wrote:
> On 9/30/08, andrzej zaborowski <balrogg@gmail.com> wrote:
>   
>> 2008/9/29 Anthony Liguori <anthony@codemonkey.ws>:
>>
>>     
>>> Andrzej Zaborowski wrote:
>>>       
>>  >>
>>  >> Revision: 5344
>>  >>          http://svn.sv.gnu.org/viewvc/?view=rev&root=qemu&revision=5344
>>  >> Author:   balrog
>>  >> Date:     2008-09-28 23:49:55 +0000 (Sun, 28 Sep 2008)
>>  >>
>>  >> Log Message:
>>  >> -----------
>>  >> Implement a HCI passthrough to host.
>>  >>
>>  >> This allows using a host's physical HCI as one of the HCIs attached
>>  >> to the virtual machine.  This brings various limitations because not
>>  >> all commands/events are passed through by Linux kernel, some are
>>  >> interpreted by the host's kernel for a speed gain.
>>  >>
>>  >
>>  > This broke the Windows build.  The whole file should probably be build
>>  > conditionally on CONFIG_BLUEZ.
>>
>>
>> Is that because of uio.h?  writev() seems to be posix.  Does win32
>>  have it under a different name?
>>
>>  Added a stronger check for bluez, hopefully will cross-compile better now.
>>     
>
> You are still relying that the host pkg-config exists and gives
> information that is valid for cross-compiling.
>   

When cross compiling, you should set you PKG_CONFIG_PATH variable so 
that the appropriate set of pkg-config configs are found.

Regards,

Anthony Liguori