lkml.org 
[lkml]   [2010]   [Oct]   [7]   [last100]   RSS Feed
Views: [wrap][no wrap]   [headers]  [forward] 
 
Messages in this thread
    /
    From
    Date
    SubjectRE: [PATCH 1/2] drivers:bluetooth: TI_ST bluetooth driver
    Gustavo / Marcel,


    > -----Original Message-----
    > From: Gustavo F. Padovan [mailto:pao@profusion.mobi] On Behalf Of Gustavo F.
    > Padovan
    > Sent: Thursday, October 07, 2010 1:46 PM
    > To: Savoy, Pavan
    > Cc: linux-bluetooth@vger.kernel.org; marcel@holtmann.org; greg@kroah.com;
    > linux-kernel@vger.kernel.org
    > Subject: Re: [PATCH 1/2] drivers:bluetooth: TI_ST bluetooth driver
    >
    > Hi Pavan,
    >
    > Change the commit subject to "Bluetooth: TI_ST bluetooth driver"
    >
    > * pavan_savoy@ti.com <pavan_savoy@ti.com> [2010-10-07 14:47:16 -0400]:
    >
    > > From: Pavan Savoy <pavan_savoy@ti.com>
    > >
    > > This is the bluetooth protocol driver for the TI WiLink7 chipsets.
    > > Texas Instrument's WiLink chipsets combine wireless technologies
    > > like BT, FM, GPS and WLAN onto a single chip.
    > >
    > > This Bluetooth driver works on top of the TI_ST shared transport
    > > line discipline driver which also allows other drivers like
    > > FM V4L2 and GPS character driver to make use of the same UART interface.
    > >
    > > Signed-off-by: Pavan Savoy <pavan_savoy@ti.com>
    > > ---
    > > drivers/bluetooth/bt_ti.c | 489
    > +++++++++++++++++++++++++++++++++++++++++++++
    > > 1 files changed, 489 insertions(+), 0 deletions(-)
    > > create mode 100644 drivers/bluetooth/bt_ti.c
    >
    > We don't have filename with bt_.. in drivers/bluetooth/. Maybe ti_st.c
    > should be a better name, or something like that.
    >
    > >
    > > diff --git a/drivers/bluetooth/bt_ti.c b/drivers/bluetooth/bt_ti.c
    > > new file mode 100644
    > > index 0000000..dffbb56
    > > --- /dev/null
    > > +++ b/drivers/bluetooth/bt_ti.c
    > > @@ -0,0 +1,489 @@
    > > +/*
    > > + * Texas Instrument's Bluetooth Driver For Shared Transport.
    > > + *
    > > + * Bluetooth Driver acts as interface between HCI CORE and
    > > + * TI Shared Transport Layer.
    > > + *
    > > + * Copyright (C) 2009-2010 Texas Instruments
    > > + * Author: Raja Mani <raja_mani@ti.com>
    > > + * Pavan Savoy <pavan_savoy@ti.com>
    > > + *
    > > + * This program is free software; you can redistribute it and/or modify
    > > + * it under the terms of the GNU General Public License version 2 as
    > > + * published by the Free Software Foundation.
    > > + *
    > > + * 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 <linux/platform_device.h>
    > > +#include <net/bluetooth/bluetooth.h>
    > > +#include <net/bluetooth/hci_core.h>
    > > +
    > > +#include <linux/ti_wilink_st.h>
    > > +
    > > +/* Bluetooth Driver Version */
    > > +#define VERSION "1.0"
    > > +
    > > +/* Defines number of seconds to wait for reg completion
    > > + * callback getting called from ST (in case,registration
    > > + * with ST returns PENDING status)
    > > + */
    > > +#define BT_REGISTER_TIMEOUT 6000 /* 6 sec */
    > > +
    > > +/* BT driver's local status */
    > > +#define BT_DRV_RUNNING 0
    > > +#define BT_ST_REGISTERED 1
    > > +
    > > +/**
    > > + * struct hci_st - BT driver operation structure
    > > + * @hdev: hci device pointer which binds to bt driver
    > > + * @flags: used locally,to maintain various BT driver status
    > > + * @streg_cbdata: to hold ST registration callback status
    > > + * @st_write: write function pointer of ST driver
    > > + * @wait_for_btdrv_reg_completion - completion sync between hci_st_open
    > > + * and hci_st_registration_completion_cb.
    > > + */
    > > +struct hci_st {
    > > + struct hci_dev *hdev;
    > > + unsigned long flags;
    > > + char streg_cbdata;
    > > + long (*st_write) (struct sk_buff *);
    > > + struct completion wait_for_btdrv_reg_completion;
    > > +};
    > > +
    > > +static int reset;
    > > +
    > > +/* Increments HCI counters based on pocket ID (cmd,acl,sco) */
    > > +static inline void hci_st_tx_complete(struct hci_st *hst, int pkt_type)
    > > +{
    > > + struct hci_dev *hdev;
    > > + hdev = hst->hdev;
    > > +
    > > + /* Update HCI stat counters */
    > > + switch (pkt_type) {
    > > + case HCI_COMMAND_PKT:
    > > + hdev->stat.cmd_tx++;
    > > + break;
    > > +
    > > + case HCI_ACLDATA_PKT:
    > > + hdev->stat.acl_tx++;
    > > + break;
    > > +
    > > + case HCI_SCODATA_PKT:
    > > + hdev->stat.cmd_tx++;
    >
    > it should be sco_tx here.
    >
    > > + break;
    > > + }
    > > +}
    > > +
    > > +/* ------- Interfaces to Shared Transport ------ */
    > > +
    > > +/* Called by ST layer to indicate protocol registration completion
    > > + * status.hci_st_open() function will wait for signal from this
    > > + * API when st_register() function returns ST_PENDING.
    > > + */
    > > +static void hci_st_registration_completion_cb(void *priv_data, char data)
    >
    > That is not the hci layer, so rename this function (and the others) to
    > something that reflect where they are really doing.
    >
    > > +{
    > > + struct hci_st *lhst = (struct hci_st *)priv_data;
    > > + /* hci_st_open() function needs value of 'data' to know
    > > + * the registration status(success/fail),So have a back
    > > + * up of it.
    > > + */
    > > + lhst->streg_cbdata = data;
    > > +
    > > + /* Got a feedback from ST for BT driver registration
    > > + * request.Wackup hci_st_open() function to continue
    > > + * it's open operation.
    > > + */
    > > + complete(&lhst->wait_for_btdrv_reg_completion);
    > > +}
    > > +
    > > +/* Called by Shared Transport layer when receive data is
    > > + * available */
    > > +static long hci_st_receive(void *priv_data, struct sk_buff *skb)
    > > +{
    > > + int err;
    > > + int len;
    >
    > you can put err and len in the same line.
    >
    > > + struct hci_st *lhst = (struct hci_st *)priv_data;
    > > +
    > > + err = 0;
    > > + len = 0;
    >
    > and no need to set them to 0 here.
    >
    > > +
    > > + if (skb == NULL) {
    > > + BT_ERR("Invalid SKB received from ST");
    > > + return -EFAULT;
    > > + }
    >
    > We need a empty line here.
    >
    > > + if (!lhst) {
    > > + kfree_skb(skb);
    > > + BT_ERR("Invalid hci_st memory,freeing SKB");
    > > + return -EFAULT;
    > > + }
    >
    > And also here. Check the rest of the code for similar issues.
    >
    > > + if (!test_bit(BT_DRV_RUNNING, &lhst->flags)) {
    > > + kfree_skb(skb);
    > > + BT_ERR("Device is not running,freeing SKB");
    > > + return -EINVAL;
    > > + }
    >
    > If you are here, your device is running, right? Or am I missing
    > something?
    >
    > > +
    > > + len = skb->len;
    > > + skb->dev = (struct net_device *)lhst->hdev;
    > > +
    > > + /* Forward skb to HCI CORE layer */
    > > + err = hci_recv_frame(skb);
    > > + if (err) {
    > > + kfree_skb(skb);
    > > + BT_ERR("Unable to push skb to HCI CORE(%d),freeing SKB",
    > > + err);
    > > + return err;
    > > + }
    > > + lhst->hdev->stat.byte_rx += len;
    >
    > actually you even don't need len, just use skb->len
    >
    > > +
    > > + return 0;
    > > +}
    > > +
    > > +/* ------- Interfaces to HCI layer ------ */
    > > +
    > > +/* Called from HCI core to initialize the device */
    > > +static int hci_st_open(struct hci_dev *hdev)
    > > +{
    > > + static struct st_proto_s hci_st_proto;
    > > + unsigned long timeleft;
    > > + struct hci_st *hst;
    > > + int err;
    > > + err = 0;
    > > +
    > > + BT_DBG("%s %p", hdev->name, hdev);
    > > + hst = hdev->driver_data;
    > > +
    > > + /* Populate BT driver info required by ST */
    > > + memset(&hci_st_proto, 0, sizeof(hci_st_proto));
    > > +
    > > + /* BT driver ID */
    > > + hci_st_proto.type = ST_BT;
    > > +
    > > + /* Receive function which called from ST */
    > > + hci_st_proto.recv = hci_st_receive;
    > > +
    > > + /* Packet match function may used in future */
    > > + hci_st_proto.match_packet = NULL;
    >
    > It is already NULL, you dua a memset.
    >
    > > +
    > > + /* Callback to be called when registration is pending */
    > > + hci_st_proto.reg_complete_cb = hci_st_registration_completion_cb;
    > > +
    > > + /* This is write function pointer of ST. BT driver will make use of
    > this
    > > + * for sending any packets to chip. ST will assign and give to us, so
    > > + * make it as NULL */
    > > + hci_st_proto.write = NULL;
    >
    > Same here.
    >
    > > +
    > > + /* send in the hst to be received at registration complete callback
    > > + * and during st's receive
    > > + */
    > > + hci_st_proto.priv_data = hst;
    > > +
    > > + /* Register with ST layer */
    > > + err = st_register(&hci_st_proto);
    > > + if (err == -EINPROGRESS) {
    > > + /* Prepare wait-for-completion handler data structures.
    > > + * Needed to syncronize this and st_registration_completion_cb()
    > > + * functions.
    > > + */
    > > + init_completion(&hst->wait_for_btdrv_reg_completion);
    >
    > I'm not liking that, but I'll leave for Marcel and others comment.

    Thanks for the comments, I will take care of them all, but I want to
    tell why I do this here.
    "st_register" function takes time. It takes about 2 seconds for us
    to download the firmware. (or longer based on UART baud-rate).

    This is intended to happen in the process context. Say hciconfig hci0 up
    takes a while longer than usual or if someone doing an ioctl(HCIDEVUP)
    and this is also the reason I have a callback, as and when firmware download
    completes, I call the Bluetooth driver's callback.

    and I await marcel's comment on this.

    > > +
    > > + /* Reset ST registration callback status flag , this value
    > > + * will be updated in hci_st_registration_completion_cb()
    > > + * function whenever it called from ST driver.
    > > + */
    > > + hst->streg_cbdata = -EINPROGRESS;
    > > +
    > > + /* ST is busy with other protocol registration(may be busy with
    > > + * firmware download).So,Wait till the registration callback
    > > + * (passed as a argument to st_register() function) getting
    > > + * called from ST.
    > > + */
    > > + BT_DBG(" %s waiting for reg completion signal from ST",
    > > + __func__);
    > > +
    > > + timeleft =
    > > + wait_for_completion_timeout
    > > + (&hst->wait_for_btdrv_reg_completion,
    > > + msecs_to_jiffies(BT_REGISTER_TIMEOUT));
    > > + if (!timeleft) {
    > > + BT_ERR("Timeout(%d sec),didn't get reg"
    > > + "completion signal from ST",
    > > + BT_REGISTER_TIMEOUT / 1000);
    > > + return -ETIMEDOUT;
    > > + }
    > > +
    > > + /* Is ST registration callback called with ERROR value? */
    > > + if (hst->streg_cbdata != 0) {
    > > + BT_ERR("ST reg completion CB called with invalid"
    > > + "status %d", hst->streg_cbdata);
    > > + return -EAGAIN;
    > > + }
    > > + err = 0;
    > > + } else if (err == -1) {
    >
    > Use the proper error macro instead "-1"
    >
    > > + BT_ERR("st_register failed %d", err);
    > > + return -EAGAIN;
    > > + }
    > > +
    > > + /* Do we have proper ST write function? */
    > > + if (hci_st_proto.write != NULL) {
    > > + /* We need this pointer for sending any Bluetooth pkts */
    > > + hst->st_write = hci_st_proto.write;
    > > + } else {
    > > + BT_ERR("failed to get ST write func pointer");
    > > +
    > > + /* Undo registration with ST */
    > > + err = st_unregister(ST_BT);
    > > + if (err < 0)
    > > + BT_ERR("st_unregister failed %d", err);
    > > +
    > > + hst->st_write = NULL;
    > > + return -EAGAIN;
    > > + }
    > > +
    > > + /* Registration with ST layer is completed successfully,
    > > + * now chip is ready to accept commands from HCI CORE.
    > > + * Mark HCI Device flag as RUNNING
    > > + */
    > > + set_bit(HCI_RUNNING, &hdev->flags);
    > > +
    > > + /* Registration with ST successful */
    > > + set_bit(BT_ST_REGISTERED, &hst->flags);
    > > +
    > > + return err;
    > > +}
    > > +
    > > +/* Close device */
    > > +static int hci_st_close(struct hci_dev *hdev)
    > > +{
    > > + int err;
    > > + struct hci_st *hst;
    >
    > Skip a line after declarations.
    >
    > > + err = 0;
    >
    > you can set err to 0 in the declaration if you really need that.
    >
    > > +
    > > + hst = hdev->driver_data;
    > > + /* Unregister from ST layer */
    > > + if (test_and_clear_bit(BT_ST_REGISTERED, &hst->flags)) {
    > > + err = st_unregister(ST_BT);
    > > + if (err != 0) {
    > > + BT_ERR("st_unregister failed %d", err);
    > > + return -EBUSY;
    > > + }
    > > + }
    > > +
    > > + hst->st_write = NULL;
    > > +
    > > + /* ST layer would have moved chip to inactive state.
    > > + * So,clear HCI device RUNNING flag.
    > > + */
    > > + if (!test_and_clear_bit(HCI_RUNNING, &hdev->flags))
    > > + return 0;
    >
    > Looks you are screwing up the flags here, if it fails on st_unregister()
    > and returns HCI_RUNNING should keep set?
    >
    > > +
    > > + return err;
    >
    > Rethink how you are doing error handling here, it should no be
    > complicated like that.
    >
    > > +}
    > > +
    > > +/* Called from HCI CORE , Sends frames to Shared Transport */
    > > +static int hci_st_send_frame(struct sk_buff *skb)
    > > +{
    > > + struct hci_dev *hdev;
    > > + struct hci_st *hst;
    > > + long len;
    > > +
    > > + if (skb == NULL) {
    > > + BT_ERR("Invalid skb received from HCI CORE");
    > > + return -ENOMEM;
    > > + }
    > > + hdev = (struct hci_dev *)skb->dev;
    > > + if (!hdev) {
    > > + BT_ERR("SKB received for invalid HCI Device (hdev=NULL)");
    > > + return -ENODEV;
    > > + }
    > > + if (!test_bit(HCI_RUNNING, &hdev->flags)) {
    > > + BT_ERR("Device is not running");
    > > + return -EBUSY;
    > > + }
    > > +
    > > + hst = (struct hci_st *)hdev->driver_data;
    > > +
    > > + /* Prepend skb with frame type */
    > > + memcpy(skb_push(skb, 1), &bt_cb(skb)->pkt_type, 1);
    > > +
    > > + BT_DBG(" %s: type %d len %d", hdev->name, bt_cb(skb)->pkt_type,
    > > + skb->len);
    > > +
    > > + /* Insert skb to shared transport layer's transmit queue.
    > > + * Freeing skb memory is taken care in shared transport layer,
    > > + * so don't free skb memory here.
    > > + */
    > > + if (!hst->st_write) {
    > > + kfree_skb(skb);
    > > + BT_ERR(" Can't write to ST, st_write null?");
    > > + return -EAGAIN;
    > > + }
    > > + len = hst->st_write(skb);
    > > + if (len < 0) {
    > > + /* Something went wrong in st write , free skb memory */
    >
    > IMHO we don't need comments like that, clearly we now that something
    > went wrong.
    >
    > > + kfree_skb(skb);
    > > + BT_ERR(" ST write failed (%ld)", len);
    > > + return -EAGAIN;
    > > + }
    > > +
    > > + /* ST accepted our skb. So, Go ahead and do rest */
    > > + hdev->stat.byte_tx += len;
    > > + hci_st_tx_complete(hst, bt_cb(skb)->pkt_type);
    > > +
    > > + return 0;
    >
    > goto might be better to handle error here.
    >
    >
    > --
    > Gustavo F. Padovan
    > ProFUSION embedded systems - http://profusion.mobi


    \
     
     \ /
      Last update: 2010-10-07 23:57    [W:0.060 / U:60.660 seconds]
    ©2003-2016 Jasper Spaans. hosted at Digital OceanAdvertise on this site