[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <4F266D64.9030105@suse.cz>
Date: Mon, 30 Jan 2012 11:13:56 +0100
From: Jiri Slaby <jslaby@...e.cz>
To: Andrea Shepard <andrea@...sephoneslair.org>
CC: linux-kernel@...r.kernel.org, netdev@...r.kernel.org,
khc@...waw.pl, davem@...emloft.net, mmarek@...e.cz,
jkosina@...e.cz, joe@...ches.com, justinmattock@...il.com,
gregkh@...e.de, alan@...ux.intel.com, jdmason@...zu.us,
Jiri Slaby <jirislaby@...il.com>
Subject: Re: [22/22] Cyclades PC300 driver: omnibus patch from merge up to
final version (patches 01 through 20 inclusive)
On 01/30/2012 04:01 AM, Andrea Shepard wrote:
> --- a/drivers/net/wan/Kconfig
> +++ b/drivers/net/wan/Kconfig
> @@ -205,10 +205,8 @@ config WANXL_BUILD_FIRMWARE
>
> config PC300
> tristate "Cyclades-PC300 support (RS-232/V.35, X.21, T1/E1 boards)"
> - depends on HDLC && PCI && BROKEN
> + depends on HDLC && PCI
You should do this as the very last in the series. You break the build
by the very first patch.
> --- a/drivers/net/wan/pc300.h
> +++ b/drivers/net/wan/pc300.h
...
> @@ -143,13 +150,28 @@
> * Memory access functions/macros *
> * (required to support Alpha systems) *
> ***************************************/
> -#define cpc_writeb(port,val) {writeb((u8)(val),(port)); mb();}
> -#define cpc_writew(port,val) {writew((ushort)(val),(port)); mb();}
> -#define cpc_writel(port,val) {writel((u32)(val),(port)); mb();}
> +#ifdef __KERNEL__
Why is this in a header from drivers/ ? Perhaps you should move the
header if it is an interface. User does not need to see cpc_writeb et
al. anyway. And u8 and others are not defined there.
> +#define cpc_writeb(port, val) {writeb((u8)(val), \
> + (void __iomem *)(port)); mb(); }
> +#define cpc_writew(port, val) {writew((u16)(val), \
> + (void __iomem *)(port)); mb(); }
> +#define cpc_writel(port, val) {writel((u32)(val), \
> + (void __iomem *)(port)); mb(); }
> +
> +#define cpc_readb(port) readb((void __iomem *)(port))
> +#define cpc_readw(port) readw((void __iomem *)(port))
> +#define cpc_readl(port) readl((void __iomem *)(port))
>
> -#define cpc_readb(port) readb(port)
> -#define cpc_readw(port) readw(port)
> -#define cpc_readl(port) readl(port)
> +#else /* __KERNEL__ */
> +#define cpc_writeb(port, val) (*((u8 *)(port)) = (u8)(val))
> +#define cpc_writew(port, val) (*((u16 *)(port)) = (u16)(val))
> +#define cpc_writel(port, val) (*((u32 *)(port)) = (u32)(val))
> +
> +#define cpc_readb(port) (*((u8 *)(port)))
> +#define cpc_readw(port) (*((u16 *)(port)))
> +#define cpc_readl(port) (*((u32 *)(port)))
> +
> +#endif /* __KERNEL__ */
>
> /****** Data Structures *****************************************************/
>
> @@ -291,16 +349,31 @@ typedef struct pc300patterntst {
> u16 num_errors;
> } pc300patterntst_t;
>
> +#ifdef __KERNEL__
> +
> typedef struct pc300dev {
> struct pc300ch *chan;
> u8 trace_on;
> u32 line_on; /* DCD(X.21, RSV) / sync(TE) change counters */
> u32 line_off;
> +#ifdef __KERNEL__
Double ifdef?
> char name[16];
> - struct net_device *dev;
> + hdlc_device *hdlc;
> + struct net_device *netdev;
> +
> + void *private;
> + struct sk_buff *tx_skb;
> +
> + enum {
> + CPC_DMA_FULL,
> + CPC_DMA_ERROR,
> + TRANSMISSION_ACTIVE,
> + CHANNEL_CLOSED
> + } reason_stopped;
> #ifdef CONFIG_PC300_MLPPP
> void *cpc_tty; /* information to PC300 TTY driver */
> #endif
> +#endif /* __KERNEL__ */
> }pc300dev_t;
...
> @@ -346,6 +450,8 @@ typedef struct pc300chconf {
> u32 tslot_bitmap; /* bit[i]=1 => timeslot _i_ is active */
> } pc300chconf_t;
>
> +#ifdef __KERNEL__
> +
> typedef struct pc300ch {
> struct pc300 *card;
> int channel;
> @@ -365,8 +471,10 @@ typedef struct pc300 {
> spinlock_t card_lock;
> } pc300_t;
>
> +#endif /* __KERNEL__ */
> +
> typedef struct pc300conf {
> - pc300hw_t hw;
> + struct pc300hw_user hw;
> pc300chconf_t conf;
> } pc300conf_t;
>
> @@ -430,7 +538,19 @@ enum pc300_loopback_cmds {
> #define PC300_TX_QUEUE_LEN 100
> #define PC300_DEF_MTU 1600
>
> -/* Function Prototypes */
> -int cpc_open(struct net_device *dev);
> +#ifdef __KERNEL__
> +
> +int cpc_open(struct net_device *);
> +
> +#ifdef CONFIG_PC300_MLPPP
> +void cpc_tty_init(pc300dev_t *);
> +void cpc_tty_unregister_service(pc300dev_t *);
> +void cpc_tty_receive(pc300dev_t *);
> +void cpc_tty_trigger_poll(pc300dev_t *);
> +void cpc_tty_reset_var(void);
> +#endif /* CONFIG_PC300_MLPP */
> +
> +#endif /* __KERNEL__ */
>
> #endif /* _PC300_H */
> +
ETOOMANYIFDEFS. You should split the header. One part with the interface
-> include/, the other with internal structures -> leave here.
> --- a/drivers/net/wan/pc300_drv.c
> +++ b/drivers/net/wan/pc300_drv.c
> @@ -1,6 +1,6 @@
> #define USE_PCI_CLOCK
> static const char rcsid[] =
> -"Revision: 3.4.5 Date: 2002/03/07 ";
> +"Revision: 4.1.0 Date: 2004/02/20 ";
This is useless, isn't it? It doesn't mean anything after some time as
kernel evolves.
> @@ -16,6 +16,13 @@ static const char rcsid[] =
> * 2 of the License, or (at your option) any later version.
> *
> * Using tabstop = 4.
> + *
> + * Cyclades version 4.1.0 merged in, with new portability fixes,
> + * and ported to recent kernels by Andrea Shepard <andrea@...sephoneslair.org>
No, we have git log.
> + * Due to changes in certain ioctls necessary for portability, this
> + * version requires a new version of pc300utils, which may be found
> + * at: http://charon.persephoneslair.org/~andrea/software/pc300utils/
> *
> * $Log: pc300_drv.c,v $
> * Revision 3.23 2002/03/20 13:58:40 henrique
Another piece of cvs.
> @@ -238,21 +246,22 @@ static const char rcsid[] =
>
> #include "pc300.h"
>
> -#define CPC_LOCK(card,flags) \
> - do { \
> - spin_lock_irqsave(&card->card_lock, flags); \
> - } while (0)
> +#define CPC_LOCK(card, flags) \
> + { \
> + spin_lock_irqsave(&((card)->card_lock), (flags)); \
> + }
Nah. do-while has its meaning. Try this:
if (1)
CPC_LOCK(card,flags);
else
return;
Overall, you don't need these macros at all.
> -#define CPC_UNLOCK(card,flags) \
> - do { \
> - spin_unlock_irqrestore(&card->card_lock, flags); \
> - } while (0)
> +#define CPC_UNLOCK(card, flags) \
> + { \
> + spin_unlock_irqrestore(&((card)->card_lock), (flags)); \
> + }
...
> @@ -279,53 +288,128 @@ MODULE_DEVICE_TABLE(pci, cpc_pci_dev_id);
...
> -static void rx_dma_buf_check(pc300_t * card, int ch)
> +static void rx_dma_buf_check(pc300_t *card, int ch)
> {
> - volatile pcsca_bd_t __iomem *ptdescr;
> + pcsca_bd_t __iomem *ptdescr;
> int i;
> u16 first_bd = card->chan[ch].rx_first_bd;
> u16 last_bd = card->chan[ch].rx_last_bd;
> int ch_factor;
>
> ch_factor = ch * N_DMA_RX_BUF;
> - printk("#CH%d: f_bd = %d, l_bd = %d\n", ch, first_bd, last_bd);
> - for (i = 0, ptdescr = (card->hw.rambase +
> - DMA_RX_BD_BASE + ch_factor * sizeof(pcsca_bd_t));
> - i < N_DMA_RX_BUF; i++, ptdescr++) {
> + printk(KERN_DEBUG
> + "#CH%d: f_bd = %d, l_bd = %d\n", ch, first_bd, last_bd);
> + for (
> + i = 0,
> + ptdescr = (pcsca_bd_t *)
> + (card->hw.rambase + DMA_RX_BD_BASE +
> + ch_factor * sizeof(pcsca_bd_t));
> +
> + i < N_DMA_RX_BUF;
> +
> + i++,
> + ptdescr++
> + ) {
This is horrible.
> if (cpc_readb(&ptdescr->status) & DST_OSB)
> - printk ("\n CH%d RX%d: next=0x%x, ptbuf=0x%x, ST=0x%x, len=%d",
> - ch, i, cpc_readl(&ptdescr->next),
> - cpc_readl(&ptdescr->ptbuf),
> - cpc_readb(&ptdescr->status),
> - cpc_readw(&ptdescr->len));
> + printk(KERN_DEBUG
> + "\n CH%d RX%d: next=0x%08x, ptbuf=0x%08x, ST=0x%2x, len=%d",
> + ch, i, (u32) cpc_readl(&ptdescr->next),
> + (u32) cpc_readl(&ptdescr->ptbuf),
> + cpc_readb(&ptdescr->status),
> + cpc_readw(&ptdescr->len));
> }
> - printk("\n");
> + printk(KERN_DEBUG "\n");
> }
...
> -static void falc_intr_enable(pc300_t * card, int ch)
> +static void falc_intr_enable(pc300_t *card, int ch)
> {
> pc300ch_t *chan = (pc300ch_t *) & card->chan[ch];
> pc300chconf_t *conf = (pc300chconf_t *) & chan->conf;
> falc_t *pfalc = (falc_t *) & chan->falc;
> - void __iomem *falcbase = card->hw.falcbase;
> + uintptr_t falcbase = (uintptr_t)(card->hw.falcbase);
Why is that?
> @@ -1940,18 +2212,19 @@ static void cpc_net_rx(struct net_device *dev)
> continue;
> }
>
> - dev->stats.rx_bytes += rxb;
> + stats->rx_bytes += rxb;
>
> #ifdef PC300_DEBUG_RX
> - printk("%s R:", dev->name);
> + printk(KERN_DEBUG "%s R:", dev->name);
> for (i = 0; i < skb->len; i++)
> - printk(" %02x", *(skb->data + i));
> - printk("\n");
> + printk(KERN_DEBUG " %02x", *(skb->data + i));
> + printk(KERN_DEBUG "\n");
print_hex_dump_bytes();
> @@ -2486,73 +2847,157 @@ static void cpc_sca_status(pc300_t * card, int ch)
...
> static int cpc_ioctl(struct net_device *dev, struct ifreq *ifr, int cmd)
> {
> - pc300dev_t *d = (pc300dev_t *) dev_to_hdlc(dev)->priv;
> + pc300dev_t *d = (pc300dev_t *) (dev_to_hdlc(dev))->priv;
> pc300ch_t *chan = (pc300ch_t *) d->chan;
> pc300_t *card = (pc300_t *) chan->card;
> pc300conf_t conf_aux;
> pc300chconf_t *conf = (pc300chconf_t *) & chan->conf;
> int ch = chan->channel;
> - void __user *arg = ifr->ifr_data;
> + void *arg = (void *) ifr->ifr_data;
But it is a userspace ptr!
> @@ -3361,47 +3897,67 @@ static void cpc_init_card(pc300_t * card)
...
> - printk (" #%d, %dKB of RAM at 0x%08x, IRQ%d, channel %d.\n",
> - board_nbr, card->hw.ramsize / 1024,
> - card->hw.ramphys, card->hw.irq, i + 1);
> +#ifdef CONFIG_PHYS_ADDR_T_64BIT
> + printk(KERN_INFO " #%d, %dKB of RAM at 0x%016lx, IRQ %d, channel %d.\n",
> + board_nbr, card->hw.ramsize / 1024,
> + (unsigned long)(card->hw.ramphys),
> + card->hw.irq, i + 1);
> +#else /* !CONFIG_PHYS_ADDR_T_64BIT */
> + printk(KERN_INFO " #%d, %dKB of RAM at 0x%08x, IRQ %d, channel %d.\n",
> + board_nbr, card->hw.ramsize / 1024,
> + (unsigned int)(card->hw.ramphys),
> + card->hw.irq, i + 1);
> +#endif /* CONFIG_PHYS_ADDR_T_64BIT */
What about just a cast to ull?
> @@ -3495,29 +4081,63 @@ cpc_init_one(struct pci_dev *pdev, const struct pci_device_id *ent)
...
> - card->hw.plxbase = ioremap(card->hw.plxphys, card->hw.plxsize);
> - card->hw.rambase = ioremap(card->hw.ramphys, card->hw.alloc_ramsize);
> - card->hw.scabase = ioremap(card->hw.scaphys, card->hw.scasize);
> + err = pci_enable_device(pdev);
> + if (err != 0)
> + goto err_release_sca;
> +
> + card->hw.plxbase =
> + (void __iomem *) ioremap(card->hw.plxphys, card->hw.plxsize);
Why do you need the cast? You may use pci_ioremap_bar anyway.
> --- a/drivers/net/wan/pc300_tty.c
> +++ b/drivers/net/wan/pc300_tty.c
> @@ -481,11 +502,35 @@ static int cpc_tty_write(struct tty_struct *tty, const unsigned char *buf, int c
> return -EINVAL;
> }
>
> - if (cpc_tty_send_to_card(cpc_tty->pc300dev, (void*)buf, count)) {
> - /* failed to send */
> - CPC_TTY_DBG("%s: trasmition error\n", cpc_tty->name);
> - return 0;
> + if (from_user) {
Nowadays, it is always a kernel buffer. This check was removed years ago
already I suppose. Please do not re-add crap from out of tree drivers.
> + unsigned char *buf_tmp;
> +
> + buf_tmp = cpc_tty->buf_tx;
> + if (copy_from_user(buf_tmp, buf, count)) {
> + /* failed to copy from user */
> + CPC_TTY_DBG("%s: error in copy from user\n",
> + cpc_tty->name);
> + return -EINVAL;
> + }
> +
> + if (cpc_tty_send_to_card(cpc_tty->pc300dev,
> + (void *)buf_tmp, count)) {
> + /* failed to send */
> + CPC_TTY_DBG("%s: transmission error\n", cpc_tty->name);
> + return 0;
> + }
> + } else {
> + if (
> + cpc_tty_send_to_card(cpc_tty->pc300dev,
> + (void *)buf,
> + count)
> + ) {
> + /* failed to send */
> + CPC_TTY_DBG("%s: transmission error\n", cpc_tty->name);
> + return 0;
> + }
> }
> +
> return count;
> }
>
...
> @@ -618,9 +666,18 @@ static void cpc_tty_flush_buffer(struct tty_struct *tty)
>
> CPC_TTY_DBG("%s: call wake_up_interruptible\n",cpc_tty->name);
>
> - tty_wakeup(tty);
> - return;
> -}
> + wake_up_interruptible(&tty->write_wait);
> +
> + if (
> + (tty->flags & (1 << TTY_DO_WRITE_WAKEUP)) &&
> + tty->ldisc &&
> + tty->ldisc->ops && tty->ldisc->ops->write_wakeup
> + ) {
> + CPC_TTY_DBG("%s: call line disc. wake up\n",
> + cpc_tty->name);
> + tty->ldisc->ops->write_wakeup(tty);
> + }
> +}
Another instance of such crap.
> @@ -665,35 +722,35 @@ static void cpc_tty_hangup(struct tty_struct *tty)
> */
> static void cpc_tty_rx_work(struct work_struct *work)
> {
> - st_cpc_tty_area *cpc_tty;
> - unsigned long port;
> int i, j;
> - volatile st_cpc_rx_buf *buf;
> - char flags=0,flg_rx=1;
> - struct tty_ldisc *ld;
> + unsigned long port;
> + st_cpc_tty_area *cpc_tty;
> + st_cpc_rx_buf *buf;
> + char flags = 0, flg_rx = 1;
>
> if (cpc_tty_cnt == 0) return;
> -
> - for (i=0; (i < 4) && flg_rx ; i++) {
> - flg_rx = 0;
>
> + for (i = 0; (i < 4) && flg_rx; i++) {
> + flg_rx = 0;
> cpc_tty = container_of(work, st_cpc_tty_area, tty_rx_work);
> port = cpc_tty - cpc_tty_area;
>
> - for (j=0; j < CPC_TTY_NPORTS; j++) {
> + for (j = 0; j < CPC_TTY_NPORTS; j++) {
> cpc_tty = &cpc_tty_area[port];
> -
> - if ((buf=cpc_tty->buf_rx.first) != NULL) {
> - if (cpc_tty->tty) {
> - ld = tty_ldisc_ref(cpc_tty->tty);
> - if (ld) {
> - if (ld->ops->receive_buf) {
> - CPC_TTY_DBG("%s: call line disc. receive_buf\n",cpc_tty->name);
> - ld->ops->receive_buf(cpc_tty->tty, (char *)(buf->data), &flags, buf->size);
> - }
> - tty_ldisc_deref(ld);
> - }
> - }
> + buf = cpc_tty->buf_rx.first;
> + if (buf != NULL) {
> + if (cpc_tty->tty && cpc_tty->tty->ldisc &&
> + cpc_tty->tty->ldisc->ops &&
> + cpc_tty->tty->ldisc->ops->receive_buf) {
Third instance of crap. NACK!
> + CPC_TTY_DBG(
> + "%s: call line disc. receive_buf\n",
> + cpc_tty->name);
> + cpc_tty->tty->
> + ldisc->ops->
> + receive_buf(cpc_tty->tty,
> + (char *)(buf->data),
> + &flags, buf->size);
> + }
> cpc_tty->buf_rx.first = cpc_tty->buf_rx.first->next;
> kfree((void *)buf);
> buf = cpc_tty->buf_rx.first;
...
> @@ -789,13 +848,13 @@ void cpc_tty_receive(pc300dev_t *pc300dev)
> }
>
> new = kmalloc(rx_len + sizeof(st_cpc_rx_buf), GFP_ATOMIC);
> - if (!new) {
> + if (new == 0) {
Nah.
> @@ -821,8 +880,7 @@ void cpc_tty_receive(pc300dev_t *pc300dev)
> cpc_tty->name);
> cpc_tty_rx_disc_frame(pc300chan);
> rx_len = 0;
> - kfree(new);
> - new = NULL;
> + kfree((unsigned char *)new);
Nah. Isn't the assignment necessary (I didn't check)?
> @@ -831,15 +889,14 @@ void cpc_tty_receive(pc300dev_t *pc300dev)
> cpc_tty_rx_disc_frame(pc300chan);
> stats->rx_dropped++;
> rx_len = 0;
> - kfree(new);
> - new = NULL;
> + kfree((unsigned char *)new);
Nah.
> @@ -892,15 +950,29 @@ static void cpc_tty_tx_work(struct work_struct *work)
> {
> st_cpc_tty_area *cpc_tty =
> container_of(work, st_cpc_tty_area, tty_tx_work);
> - struct tty_struct *tty;
> + struct tty_struct *tty;
>
> CPC_TTY_DBG("%s: cpc_tty_tx_work init\n",cpc_tty->name);
>
> - if ((tty = cpc_tty->tty) == NULL) {
> - CPC_TTY_DBG("%s: the interface is not opened\n",cpc_tty->name);
> + tty = cpc_tty->tty;
> + if (tty == 0) {
> + CPC_TTY_DBG("%s: the interface is not opened\n",
> + cpc_tty->name);
> return;
> }
> - tty_wakeup(tty);
> +
> + if (
> + (tty->flags & (1 << TTY_DO_WRITE_WAKEUP)) &&
> + tty->ldisc &&
> + tty->ldisc->ops &&
> + tty->ldisc->ops->write_wakeup
> + ) {
> + CPC_TTY_DBG("%s:call line disc. wakeup\n",
> + cpc_tty->name);
> + tty->ldisc->ops->write_wakeup(tty);
> + }
> +
> + wake_up_interruptible(&tty->write_wait);
No!
> @@ -1032,38 +1108,40 @@ void cpc_tty_unregister_service(pc300dev_t *pc300dev)
> ulong flags;
> int res;
>
> - if ((cpc_tty= (st_cpc_tty_area *) pc300dev->cpc_tty) == NULL) {
> - CPC_TTY_DBG("%s: interface is not TTY\n", pc300dev->dev->name);
> + cpc_tty = (st_cpc_tty_area *) pc300dev->cpc_tty;
> + if (cpc_tty == 0) {
This is a pointer. 0 is *not* NULL in C.
> if (--cpc_tty_cnt == 0) {
> - if (serial_drv.refcount) {
> - CPC_TTY_DBG("%s: unregister is not possible, refcount=%d",
> - cpc_tty->name, serial_drv.refcount);
> - cpc_tty_cnt++;
> - cpc_tty_unreg_flag = 1;
> - return;
> - } else {
> - CPC_TTY_DBG("%s: unregister the tty driver\n", cpc_tty->name);
> - if ((res=tty_unregister_driver(&serial_drv))) {
> - CPC_TTY_DBG("%s: ERROR ->unregister the tty driver error=%d\n",
> - cpc_tty->name,res);
> - }
> + CPC_TTY_DBG("%s: unregister the tty driver\n",
> + cpc_tty->name);
> + res = tty_unregister_driver(&serial_drv);
> + if (res) {
> + CPC_TTY_DBG(
> + "%s: ERROR ->unregister the tty driver error=%d\n",
> + cpc_tty->name, res);
You may ignore the retval.
As you could see, I NACK the series. Please fix the issues.
thanks,
--
js
suse labs
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/
Powered by blists - more mailing lists