kernel-fxtec-pro1x/drivers/net/phy/phy.c

1005 lines
24 KiB
C
Raw Normal View History

/*
* drivers/net/phy/phy.c
*
* Framework for configuring and reading PHY devices
* Based on code in sungem_phy.c and gianfar_phy.c
*
* Author: Andy Fleming
*
* Copyright (c) 2004 Freescale Semiconductor, Inc.
PHYLIB: IRQ event workqueue handling fixes Keep track of disable_irq_nosync() invocations and call enable_irq() the right number of times if work has been cancelled that would include them. Now that the call to flush_work_keventd() (problematic because of rtnl_mutex being held) has been replaced by cancel_work_sync() another issue has arisen and been left unresolved. As the MDIO bus cannot be accessed from the interrupt context the PHY interrupt handler uses disable_irq_nosync() to prevent from looping and schedules some work to be done as a softirq, which, apart from handling the state change of the originating PHY, is responsible for reenabling the interrupt. Now if the interrupt line is shared by another device and a call to the softirq handler has been cancelled, that call to enable_irq() never happens and the other device cannot use its interrupt anymore as its stuck disabled. I decided to use a counter rather than a flag because there may be more than one call to phy_change() cancelled in the queue -- a real one and a fake one triggered by free_irq() if DEBUG_SHIRQ is used, if nothing else. Therefore because of its nesting property enable_irq() has to be called the right number of times to match the number disable_irq_nosync() was called and restore the original state. This DEBUG_SHIRQ feature is also the reason why free_irq() has to be called before cancel_work_sync(). While at it I updated the comment about phy_stop_interrupts() being called from `keventd' -- this is no longer relevant as the use of cancel_work_sync() makes such an approach unnecessary. OTOH a similar comment referring to flush_scheduled_work() in phy_stop() still applies as using cancel_work_sync() there would be dangerous. Checked with checkpatch.pl and at the run time (with and without DEBUG_SHIRQ). Signed-off-by: Maciej W. Rozycki <macro@linux-mips.org> Cc: Andy Fleming <afleming@freescale.com> Cc: Jeff Garzik <jgarzik@pobox.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Jeff Garzik <jeff@garzik.org>
2007-09-28 23:42:14 -06:00
* Copyright (c) 2006, 2007 Maciej W. Rozycki
*
* 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 of the License, or (at your
* option) any later version.
*
*/
#include <linux/kernel.h>
#include <linux/string.h>
#include <linux/errno.h>
#include <linux/unistd.h>
#include <linux/slab.h>
#include <linux/interrupt.h>
#include <linux/init.h>
#include <linux/delay.h>
#include <linux/netdevice.h>
#include <linux/etherdevice.h>
#include <linux/skbuff.h>
#include <linux/mm.h>
#include <linux/module.h>
#include <linux/mii.h>
#include <linux/ethtool.h>
#include <linux/phy.h>
#include <linux/timer.h>
#include <linux/workqueue.h>
PHYLIB: IRQ event workqueue handling fixes Keep track of disable_irq_nosync() invocations and call enable_irq() the right number of times if work has been cancelled that would include them. Now that the call to flush_work_keventd() (problematic because of rtnl_mutex being held) has been replaced by cancel_work_sync() another issue has arisen and been left unresolved. As the MDIO bus cannot be accessed from the interrupt context the PHY interrupt handler uses disable_irq_nosync() to prevent from looping and schedules some work to be done as a softirq, which, apart from handling the state change of the originating PHY, is responsible for reenabling the interrupt. Now if the interrupt line is shared by another device and a call to the softirq handler has been cancelled, that call to enable_irq() never happens and the other device cannot use its interrupt anymore as its stuck disabled. I decided to use a counter rather than a flag because there may be more than one call to phy_change() cancelled in the queue -- a real one and a fake one triggered by free_irq() if DEBUG_SHIRQ is used, if nothing else. Therefore because of its nesting property enable_irq() has to be called the right number of times to match the number disable_irq_nosync() was called and restore the original state. This DEBUG_SHIRQ feature is also the reason why free_irq() has to be called before cancel_work_sync(). While at it I updated the comment about phy_stop_interrupts() being called from `keventd' -- this is no longer relevant as the use of cancel_work_sync() makes such an approach unnecessary. OTOH a similar comment referring to flush_scheduled_work() in phy_stop() still applies as using cancel_work_sync() there would be dangerous. Checked with checkpatch.pl and at the run time (with and without DEBUG_SHIRQ). Signed-off-by: Maciej W. Rozycki <macro@linux-mips.org> Cc: Andy Fleming <afleming@freescale.com> Cc: Jeff Garzik <jgarzik@pobox.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Jeff Garzik <jeff@garzik.org>
2007-09-28 23:42:14 -06:00
#include <asm/atomic.h>
#include <asm/io.h>
#include <asm/irq.h>
#include <asm/uaccess.h>
/**
* phy_print_status - Convenience function to print out the current phy status
* @phydev: the phy_device struct
*/
void phy_print_status(struct phy_device *phydev)
{
pr_info("PHY: %s - Link is %s", phydev->dev.bus_id,
phydev->link ? "Up" : "Down");
if (phydev->link)
printk(" - %d/%s", phydev->speed,
DUPLEX_FULL == phydev->duplex ?
"Full" : "Half");
printk("\n");
}
EXPORT_SYMBOL(phy_print_status);
/**
* phy_read - Convenience function for reading a given PHY register
* @phydev: the phy_device struct
* @regnum: register number to read
*
* NOTE: MUST NOT be called from interrupt context,
* because the bus read/write functions may wait for an interrupt
* to conclude the operation.
*/
int phy_read(struct phy_device *phydev, u16 regnum)
{
int retval;
struct mii_bus *bus = phydev->bus;
BUG_ON(in_interrupt());
mutex_lock(&bus->mdio_lock);
retval = bus->read(bus, phydev->addr, regnum);
mutex_unlock(&bus->mdio_lock);
return retval;
}
EXPORT_SYMBOL(phy_read);
/**
* phy_write - Convenience function for writing a given PHY register
* @phydev: the phy_device struct
* @regnum: register number to write
* @val: value to write to @regnum
*
* NOTE: MUST NOT be called from interrupt context,
* because the bus read/write functions may wait for an interrupt
* to conclude the operation.
*/
int phy_write(struct phy_device *phydev, u16 regnum, u16 val)
{
int err;
struct mii_bus *bus = phydev->bus;
BUG_ON(in_interrupt());
mutex_lock(&bus->mdio_lock);
err = bus->write(bus, phydev->addr, regnum, val);
mutex_unlock(&bus->mdio_lock);
return err;
}
EXPORT_SYMBOL(phy_write);
/**
* phy_clear_interrupt - Ack the phy device's interrupt
* @phydev: the phy_device struct
*
* If the @phydev driver has an ack_interrupt function, call it to
* ack and clear the phy device's interrupt.
*
* Returns 0 on success on < 0 on error.
*/
int phy_clear_interrupt(struct phy_device *phydev)
{
int err = 0;
if (phydev->drv->ack_interrupt)
err = phydev->drv->ack_interrupt(phydev);
return err;
}
/**
* phy_config_interrupt - configure the PHY device for the requested interrupts
* @phydev: the phy_device struct
* @interrupts: interrupt flags to configure for this @phydev
*
* Returns 0 on success on < 0 on error.
*/
int phy_config_interrupt(struct phy_device *phydev, u32 interrupts)
{
int err = 0;
phydev->interrupts = interrupts;
if (phydev->drv->config_intr)
err = phydev->drv->config_intr(phydev);
return err;
}
/**
* phy_aneg_done - return auto-negotiation status
* @phydev: target phy_device struct
*
* Description: Reads the status register and returns 0 either if
* auto-negotiation is incomplete, or if there was an error.
* Returns BMSR_ANEGCOMPLETE if auto-negotiation is done.
*/
static inline int phy_aneg_done(struct phy_device *phydev)
{
int retval;
retval = phy_read(phydev, MII_BMSR);
return (retval < 0) ? retval : (retval & BMSR_ANEGCOMPLETE);
}
/* A structure for mapping a particular speed and duplex
* combination to a particular SUPPORTED and ADVERTISED value */
struct phy_setting {
int speed;
int duplex;
u32 setting;
};
/* A mapping of all SUPPORTED settings to speed/duplex */
2006-03-03 19:33:57 -07:00
static const struct phy_setting settings[] = {
{
.speed = 10000,
.duplex = DUPLEX_FULL,
.setting = SUPPORTED_10000baseT_Full,
},
{
.speed = SPEED_1000,
.duplex = DUPLEX_FULL,
.setting = SUPPORTED_1000baseT_Full,
},
{
.speed = SPEED_1000,
.duplex = DUPLEX_HALF,
.setting = SUPPORTED_1000baseT_Half,
},
{
.speed = SPEED_100,
.duplex = DUPLEX_FULL,
.setting = SUPPORTED_100baseT_Full,
},
{
.speed = SPEED_100,
.duplex = DUPLEX_HALF,
.setting = SUPPORTED_100baseT_Half,
},
{
.speed = SPEED_10,
.duplex = DUPLEX_FULL,
.setting = SUPPORTED_10baseT_Full,
},
{
.speed = SPEED_10,
.duplex = DUPLEX_HALF,
.setting = SUPPORTED_10baseT_Half,
},
};
#define MAX_NUM_SETTINGS ARRAY_SIZE(settings)
/**
* phy_find_setting - find a PHY settings array entry that matches speed & duplex
* @speed: speed to match
* @duplex: duplex to match
*
* Description: Searches the settings array for the setting which
* matches the desired speed and duplex, and returns the index
* of that setting. Returns the index of the last setting if
* none of the others match.
*/
static inline int phy_find_setting(int speed, int duplex)
{
int idx = 0;
while (idx < ARRAY_SIZE(settings) &&
(settings[idx].speed != speed ||
settings[idx].duplex != duplex))
idx++;
return idx < MAX_NUM_SETTINGS ? idx : MAX_NUM_SETTINGS - 1;
}
/**
* phy_find_valid - find a PHY setting that matches the requested features mask
* @idx: The first index in settings[] to search
* @features: A mask of the valid settings
*
* Description: Returns the index of the first valid setting less
* than or equal to the one pointed to by idx, as determined by
* the mask in features. Returns the index of the last setting
* if nothing else matches.
*/
static inline int phy_find_valid(int idx, u32 features)
{
while (idx < MAX_NUM_SETTINGS && !(settings[idx].setting & features))
idx++;
return idx < MAX_NUM_SETTINGS ? idx : MAX_NUM_SETTINGS - 1;
}
/**
* phy_sanitize_settings - make sure the PHY is set to supported speed and duplex
* @phydev: the target phy_device struct
*
* Description: Make sure the PHY is set to supported speeds and
* duplexes. Drop down by one in this order: 1000/FULL,
* 1000/HALF, 100/FULL, 100/HALF, 10/FULL, 10/HALF.
*/
void phy_sanitize_settings(struct phy_device *phydev)
{
u32 features = phydev->supported;
int idx;
/* Sanitize settings based on PHY capabilities */
if ((features & SUPPORTED_Autoneg) == 0)
phydev->autoneg = AUTONEG_DISABLE;
idx = phy_find_valid(phy_find_setting(phydev->speed, phydev->duplex),
features);
phydev->speed = settings[idx].speed;
phydev->duplex = settings[idx].duplex;
}
EXPORT_SYMBOL(phy_sanitize_settings);
/**
* phy_ethtool_sset - generic ethtool sset function, handles all the details
* @phydev: target phy_device struct
* @cmd: ethtool_cmd
*
* A few notes about parameter checking:
* - We don't set port or transceiver, so we don't care what they
* were set to.
* - phy_start_aneg() will make sure forced settings are sane, and
* choose the next best ones from the ones selected, so we don't
* care if ethtool tries to give us bad values.
*/
int phy_ethtool_sset(struct phy_device *phydev, struct ethtool_cmd *cmd)
{
if (cmd->phy_address != phydev->addr)
return -EINVAL;
/* We make sure that we don't pass unsupported
* values in to the PHY */
cmd->advertising &= phydev->supported;
/* Verify the settings we care about. */
if (cmd->autoneg != AUTONEG_ENABLE && cmd->autoneg != AUTONEG_DISABLE)
return -EINVAL;
if (cmd->autoneg == AUTONEG_ENABLE && cmd->advertising == 0)
return -EINVAL;
if (cmd->autoneg == AUTONEG_DISABLE
&& ((cmd->speed != SPEED_1000
&& cmd->speed != SPEED_100
&& cmd->speed != SPEED_10)
|| (cmd->duplex != DUPLEX_HALF
&& cmd->duplex != DUPLEX_FULL)))
return -EINVAL;
phydev->autoneg = cmd->autoneg;
phydev->speed = cmd->speed;
phydev->advertising = cmd->advertising;
if (AUTONEG_ENABLE == cmd->autoneg)
phydev->advertising |= ADVERTISED_Autoneg;
else
phydev->advertising &= ~ADVERTISED_Autoneg;
phydev->duplex = cmd->duplex;
/* Restart the PHY */
phy_start_aneg(phydev);
return 0;
}
EXPORT_SYMBOL(phy_ethtool_sset);
int phy_ethtool_gset(struct phy_device *phydev, struct ethtool_cmd *cmd)
{
cmd->supported = phydev->supported;
cmd->advertising = phydev->advertising;
cmd->speed = phydev->speed;
cmd->duplex = phydev->duplex;
cmd->port = PORT_MII;
cmd->phy_address = phydev->addr;
cmd->transceiver = XCVR_EXTERNAL;
cmd->autoneg = phydev->autoneg;
return 0;
}
EXPORT_SYMBOL(phy_ethtool_gset);
/**
* phy_mii_ioctl - generic PHY MII ioctl interface
* @phydev: the phy_device struct
* @mii_data: MII ioctl data
* @cmd: ioctl cmd to execute
*
* Note that this function is currently incompatible with the
* PHYCONTROL layer. It changes registers without regard to
* current state. Use at own risk.
*/
int phy_mii_ioctl(struct phy_device *phydev,
struct mii_ioctl_data *mii_data, int cmd)
{
u16 val = mii_data->val_in;
switch (cmd) {
case SIOCGMIIPHY:
mii_data->phy_id = phydev->addr;
break;
case SIOCGMIIREG:
mii_data->val_out = phy_read(phydev, mii_data->reg_num);
break;
case SIOCSMIIREG:
if (!capable(CAP_NET_ADMIN))
return -EPERM;
if (mii_data->phy_id == phydev->addr) {
switch(mii_data->reg_num) {
case MII_BMCR:
if ((val & (BMCR_RESET|BMCR_ANENABLE)) == 0)
phydev->autoneg = AUTONEG_DISABLE;
else
phydev->autoneg = AUTONEG_ENABLE;
if ((!phydev->autoneg) && (val & BMCR_FULLDPLX))
phydev->duplex = DUPLEX_FULL;
else
phydev->duplex = DUPLEX_HALF;
if ((!phydev->autoneg) &&
(val & BMCR_SPEED1000))
phydev->speed = SPEED_1000;
else if ((!phydev->autoneg) &&
(val & BMCR_SPEED100))
phydev->speed = SPEED_100;
break;
case MII_ADVERTISE:
phydev->advertising = val;
break;
default:
/* do nothing */
break;
}
}
phy_write(phydev, mii_data->reg_num, val);
if (mii_data->reg_num == MII_BMCR
&& val & BMCR_RESET
&& phydev->drv->config_init) {
phy_scan_fixups(phydev);
phydev->drv->config_init(phydev);
}
break;
default:
return -ENOTTY;
}
return 0;
}
EXPORT_SYMBOL(phy_mii_ioctl);
/**
* phy_start_aneg - start auto-negotiation for this PHY device
* @phydev: the phy_device struct
*
* Description: Sanitizes the settings (if we're not autonegotiating
* them), and then calls the driver's config_aneg function.
* If the PHYCONTROL Layer is operating, we change the state to
* reflect the beginning of Auto-negotiation or forcing.
*/
int phy_start_aneg(struct phy_device *phydev)
{
int err;
mutex_lock(&phydev->lock);
if (AUTONEG_DISABLE == phydev->autoneg)
phy_sanitize_settings(phydev);
err = phydev->drv->config_aneg(phydev);
if (err < 0)
goto out_unlock;
if (phydev->state != PHY_HALTED) {
if (AUTONEG_ENABLE == phydev->autoneg) {
phydev->state = PHY_AN;
phydev->link_timeout = PHY_AN_TIMEOUT;
} else {
phydev->state = PHY_FORCING;
phydev->link_timeout = PHY_FORCE_TIMEOUT;
}
}
out_unlock:
mutex_unlock(&phydev->lock);
return err;
}
EXPORT_SYMBOL(phy_start_aneg);
static void phy_change(struct work_struct *work);
static void phy_state_machine(struct work_struct *work);
static void phy_timer(unsigned long data);
/**
* phy_start_machine - start PHY state machine tracking
* @phydev: the phy_device struct
* @handler: callback function for state change notifications
*
* Description: The PHY infrastructure can run a state machine
* which tracks whether the PHY is starting up, negotiating,
* etc. This function starts the timer which tracks the state
* of the PHY. If you want to be notified when the state changes,
* pass in the callback @handler, otherwise, pass NULL. If you
* want to maintain your own state machine, do not call this
* function.
*/
void phy_start_machine(struct phy_device *phydev,
void (*handler)(struct net_device *))
{
phydev->adjust_state = handler;
INIT_WORK(&phydev->state_queue, phy_state_machine);
init_timer(&phydev->phy_timer);
phydev->phy_timer.function = &phy_timer;
phydev->phy_timer.data = (unsigned long) phydev;
mod_timer(&phydev->phy_timer, jiffies + HZ);
}
/**
* phy_stop_machine - stop the PHY state machine tracking
* @phydev: target phy_device struct
*
* Description: Stops the state machine timer, sets the state to UP
* (unless it wasn't up yet). This function must be called BEFORE
* phy_detach.
*/
void phy_stop_machine(struct phy_device *phydev)
{
del_timer_sync(&phydev->phy_timer);
cancel_work_sync(&phydev->state_queue);
mutex_lock(&phydev->lock);
if (phydev->state > PHY_UP)
phydev->state = PHY_UP;
mutex_unlock(&phydev->lock);
phydev->adjust_state = NULL;
}
/**
* phy_force_reduction - reduce PHY speed/duplex settings by one step
* @phydev: target phy_device struct
*
* Description: Reduces the speed/duplex settings by one notch,
* in this order--
* 1000/FULL, 1000/HALF, 100/FULL, 100/HALF, 10/FULL, 10/HALF.
* The function bottoms out at 10/HALF.
*/
static void phy_force_reduction(struct phy_device *phydev)
{
int idx;
idx = phy_find_setting(phydev->speed, phydev->duplex);
idx++;
idx = phy_find_valid(idx, phydev->supported);
phydev->speed = settings[idx].speed;
phydev->duplex = settings[idx].duplex;
pr_info("Trying %d/%s\n", phydev->speed,
DUPLEX_FULL == phydev->duplex ?
"FULL" : "HALF");
}
/**
* phy_error - enter HALTED state for this PHY device
* @phydev: target phy_device struct
*
* Moves the PHY to the HALTED state in response to a read
* or write error, and tells the controller the link is down.
* Must not be called from interrupt context, or while the
* phydev->lock is held.
*/
static void phy_error(struct phy_device *phydev)
{
mutex_lock(&phydev->lock);
phydev->state = PHY_HALTED;
mutex_unlock(&phydev->lock);
}
/**
* phy_interrupt - PHY interrupt handler
* @irq: interrupt line
* @phy_dat: phy_device pointer
*
* Description: When a PHY interrupt occurs, the handler disables
* interrupts, and schedules a work task to clear the interrupt.
*/
IRQ: Maintain regs pointer globally rather than passing to IRQ handlers Maintain a per-CPU global "struct pt_regs *" variable which can be used instead of passing regs around manually through all ~1800 interrupt handlers in the Linux kernel. The regs pointer is used in few places, but it potentially costs both stack space and code to pass it around. On the FRV arch, removing the regs parameter from all the genirq function results in a 20% speed up of the IRQ exit path (ie: from leaving timer_interrupt() to leaving do_IRQ()). Where appropriate, an arch may override the generic storage facility and do something different with the variable. On FRV, for instance, the address is maintained in GR28 at all times inside the kernel as part of general exception handling. Having looked over the code, it appears that the parameter may be handed down through up to twenty or so layers of functions. Consider a USB character device attached to a USB hub, attached to a USB controller that posts its interrupts through a cascaded auxiliary interrupt controller. A character device driver may want to pass regs to the sysrq handler through the input layer which adds another few layers of parameter passing. I've build this code with allyesconfig for x86_64 and i386. I've runtested the main part of the code on FRV and i386, though I can't test most of the drivers. I've also done partial conversion for powerpc and MIPS - these at least compile with minimal configurations. This will affect all archs. Mostly the changes should be relatively easy. Take do_IRQ(), store the regs pointer at the beginning, saving the old one: struct pt_regs *old_regs = set_irq_regs(regs); And put the old one back at the end: set_irq_regs(old_regs); Don't pass regs through to generic_handle_irq() or __do_IRQ(). In timer_interrupt(), this sort of change will be necessary: - update_process_times(user_mode(regs)); - profile_tick(CPU_PROFILING, regs); + update_process_times(user_mode(get_irq_regs())); + profile_tick(CPU_PROFILING); I'd like to move update_process_times()'s use of get_irq_regs() into itself, except that i386, alone of the archs, uses something other than user_mode(). Some notes on the interrupt handling in the drivers: (*) input_dev() is now gone entirely. The regs pointer is no longer stored in the input_dev struct. (*) finish_unlinks() in drivers/usb/host/ohci-q.c needs checking. It does something different depending on whether it's been supplied with a regs pointer or not. (*) Various IRQ handler function pointers have been moved to type irq_handler_t. Signed-Off-By: David Howells <dhowells@redhat.com> (cherry picked from 1b16e7ac850969f38b375e511e3fa2f474a33867 commit)
2006-10-05 07:55:46 -06:00
static irqreturn_t phy_interrupt(int irq, void *phy_dat)
{
struct phy_device *phydev = phy_dat;
if (PHY_HALTED == phydev->state)
return IRQ_NONE; /* It can't be ours. */
/* The MDIO bus is not allowed to be written in interrupt
* context, so we need to disable the irq here. A work
* queue will write the PHY to disable and clear the
* interrupt, and then reenable the irq line. */
disable_irq_nosync(irq);
PHYLIB: IRQ event workqueue handling fixes Keep track of disable_irq_nosync() invocations and call enable_irq() the right number of times if work has been cancelled that would include them. Now that the call to flush_work_keventd() (problematic because of rtnl_mutex being held) has been replaced by cancel_work_sync() another issue has arisen and been left unresolved. As the MDIO bus cannot be accessed from the interrupt context the PHY interrupt handler uses disable_irq_nosync() to prevent from looping and schedules some work to be done as a softirq, which, apart from handling the state change of the originating PHY, is responsible for reenabling the interrupt. Now if the interrupt line is shared by another device and a call to the softirq handler has been cancelled, that call to enable_irq() never happens and the other device cannot use its interrupt anymore as its stuck disabled. I decided to use a counter rather than a flag because there may be more than one call to phy_change() cancelled in the queue -- a real one and a fake one triggered by free_irq() if DEBUG_SHIRQ is used, if nothing else. Therefore because of its nesting property enable_irq() has to be called the right number of times to match the number disable_irq_nosync() was called and restore the original state. This DEBUG_SHIRQ feature is also the reason why free_irq() has to be called before cancel_work_sync(). While at it I updated the comment about phy_stop_interrupts() being called from `keventd' -- this is no longer relevant as the use of cancel_work_sync() makes such an approach unnecessary. OTOH a similar comment referring to flush_scheduled_work() in phy_stop() still applies as using cancel_work_sync() there would be dangerous. Checked with checkpatch.pl and at the run time (with and without DEBUG_SHIRQ). Signed-off-by: Maciej W. Rozycki <macro@linux-mips.org> Cc: Andy Fleming <afleming@freescale.com> Cc: Jeff Garzik <jgarzik@pobox.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Jeff Garzik <jeff@garzik.org>
2007-09-28 23:42:14 -06:00
atomic_inc(&phydev->irq_disable);
schedule_work(&phydev->phy_queue);
return IRQ_HANDLED;
}
/**
* phy_enable_interrupts - Enable the interrupts from the PHY side
* @phydev: target phy_device struct
*/
int phy_enable_interrupts(struct phy_device *phydev)
{
int err;
err = phy_clear_interrupt(phydev);
if (err < 0)
return err;
err = phy_config_interrupt(phydev, PHY_INTERRUPT_ENABLED);
return err;
}
EXPORT_SYMBOL(phy_enable_interrupts);
/**
* phy_disable_interrupts - Disable the PHY interrupts from the PHY side
* @phydev: target phy_device struct
*/
int phy_disable_interrupts(struct phy_device *phydev)
{
int err;
/* Disable PHY interrupts */
err = phy_config_interrupt(phydev, PHY_INTERRUPT_DISABLED);
if (err)
goto phy_err;
/* Clear the interrupt */
err = phy_clear_interrupt(phydev);
if (err)
goto phy_err;
return 0;
phy_err:
phy_error(phydev);
return err;
}
EXPORT_SYMBOL(phy_disable_interrupts);
/**
* phy_start_interrupts - request and enable interrupts for a PHY device
* @phydev: target phy_device struct
*
* Description: Request the interrupt for the given PHY.
* If this fails, then we set irq to PHY_POLL.
* Otherwise, we enable the interrupts in the PHY.
* This should only be called with a valid IRQ number.
* Returns 0 on success or < 0 on error.
*/
int phy_start_interrupts(struct phy_device *phydev)
{
int err = 0;
INIT_WORK(&phydev->phy_queue, phy_change);
PHYLIB: IRQ event workqueue handling fixes Keep track of disable_irq_nosync() invocations and call enable_irq() the right number of times if work has been cancelled that would include them. Now that the call to flush_work_keventd() (problematic because of rtnl_mutex being held) has been replaced by cancel_work_sync() another issue has arisen and been left unresolved. As the MDIO bus cannot be accessed from the interrupt context the PHY interrupt handler uses disable_irq_nosync() to prevent from looping and schedules some work to be done as a softirq, which, apart from handling the state change of the originating PHY, is responsible for reenabling the interrupt. Now if the interrupt line is shared by another device and a call to the softirq handler has been cancelled, that call to enable_irq() never happens and the other device cannot use its interrupt anymore as its stuck disabled. I decided to use a counter rather than a flag because there may be more than one call to phy_change() cancelled in the queue -- a real one and a fake one triggered by free_irq() if DEBUG_SHIRQ is used, if nothing else. Therefore because of its nesting property enable_irq() has to be called the right number of times to match the number disable_irq_nosync() was called and restore the original state. This DEBUG_SHIRQ feature is also the reason why free_irq() has to be called before cancel_work_sync(). While at it I updated the comment about phy_stop_interrupts() being called from `keventd' -- this is no longer relevant as the use of cancel_work_sync() makes such an approach unnecessary. OTOH a similar comment referring to flush_scheduled_work() in phy_stop() still applies as using cancel_work_sync() there would be dangerous. Checked with checkpatch.pl and at the run time (with and without DEBUG_SHIRQ). Signed-off-by: Maciej W. Rozycki <macro@linux-mips.org> Cc: Andy Fleming <afleming@freescale.com> Cc: Jeff Garzik <jgarzik@pobox.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Jeff Garzik <jeff@garzik.org>
2007-09-28 23:42:14 -06:00
atomic_set(&phydev->irq_disable, 0);
if (request_irq(phydev->irq, phy_interrupt,
IRQF_SHARED,
"phy_interrupt",
phydev) < 0) {
printk(KERN_WARNING "%s: Can't get IRQ %d (PHY)\n",
phydev->bus->name,
phydev->irq);
phydev->irq = PHY_POLL;
return 0;
}
err = phy_enable_interrupts(phydev);
return err;
}
EXPORT_SYMBOL(phy_start_interrupts);
/**
* phy_stop_interrupts - disable interrupts from a PHY device
* @phydev: target phy_device struct
*/
int phy_stop_interrupts(struct phy_device *phydev)
{
int err;
err = phy_disable_interrupts(phydev);
if (err)
phy_error(phydev);
PHYLIB: IRQ event workqueue handling fixes Keep track of disable_irq_nosync() invocations and call enable_irq() the right number of times if work has been cancelled that would include them. Now that the call to flush_work_keventd() (problematic because of rtnl_mutex being held) has been replaced by cancel_work_sync() another issue has arisen and been left unresolved. As the MDIO bus cannot be accessed from the interrupt context the PHY interrupt handler uses disable_irq_nosync() to prevent from looping and schedules some work to be done as a softirq, which, apart from handling the state change of the originating PHY, is responsible for reenabling the interrupt. Now if the interrupt line is shared by another device and a call to the softirq handler has been cancelled, that call to enable_irq() never happens and the other device cannot use its interrupt anymore as its stuck disabled. I decided to use a counter rather than a flag because there may be more than one call to phy_change() cancelled in the queue -- a real one and a fake one triggered by free_irq() if DEBUG_SHIRQ is used, if nothing else. Therefore because of its nesting property enable_irq() has to be called the right number of times to match the number disable_irq_nosync() was called and restore the original state. This DEBUG_SHIRQ feature is also the reason why free_irq() has to be called before cancel_work_sync(). While at it I updated the comment about phy_stop_interrupts() being called from `keventd' -- this is no longer relevant as the use of cancel_work_sync() makes such an approach unnecessary. OTOH a similar comment referring to flush_scheduled_work() in phy_stop() still applies as using cancel_work_sync() there would be dangerous. Checked with checkpatch.pl and at the run time (with and without DEBUG_SHIRQ). Signed-off-by: Maciej W. Rozycki <macro@linux-mips.org> Cc: Andy Fleming <afleming@freescale.com> Cc: Jeff Garzik <jgarzik@pobox.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Jeff Garzik <jeff@garzik.org>
2007-09-28 23:42:14 -06:00
free_irq(phydev->irq, phydev);
/*
PHYLIB: IRQ event workqueue handling fixes Keep track of disable_irq_nosync() invocations and call enable_irq() the right number of times if work has been cancelled that would include them. Now that the call to flush_work_keventd() (problematic because of rtnl_mutex being held) has been replaced by cancel_work_sync() another issue has arisen and been left unresolved. As the MDIO bus cannot be accessed from the interrupt context the PHY interrupt handler uses disable_irq_nosync() to prevent from looping and schedules some work to be done as a softirq, which, apart from handling the state change of the originating PHY, is responsible for reenabling the interrupt. Now if the interrupt line is shared by another device and a call to the softirq handler has been cancelled, that call to enable_irq() never happens and the other device cannot use its interrupt anymore as its stuck disabled. I decided to use a counter rather than a flag because there may be more than one call to phy_change() cancelled in the queue -- a real one and a fake one triggered by free_irq() if DEBUG_SHIRQ is used, if nothing else. Therefore because of its nesting property enable_irq() has to be called the right number of times to match the number disable_irq_nosync() was called and restore the original state. This DEBUG_SHIRQ feature is also the reason why free_irq() has to be called before cancel_work_sync(). While at it I updated the comment about phy_stop_interrupts() being called from `keventd' -- this is no longer relevant as the use of cancel_work_sync() makes such an approach unnecessary. OTOH a similar comment referring to flush_scheduled_work() in phy_stop() still applies as using cancel_work_sync() there would be dangerous. Checked with checkpatch.pl and at the run time (with and without DEBUG_SHIRQ). Signed-off-by: Maciej W. Rozycki <macro@linux-mips.org> Cc: Andy Fleming <afleming@freescale.com> Cc: Jeff Garzik <jgarzik@pobox.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Jeff Garzik <jeff@garzik.org>
2007-09-28 23:42:14 -06:00
* Cannot call flush_scheduled_work() here as desired because
* of rtnl_lock(), but we do not really care about what would
* be done, except from enable_irq(), so cancel any work
* possibly pending and take care of the matter below.
*/
cancel_work_sync(&phydev->phy_queue);
PHYLIB: IRQ event workqueue handling fixes Keep track of disable_irq_nosync() invocations and call enable_irq() the right number of times if work has been cancelled that would include them. Now that the call to flush_work_keventd() (problematic because of rtnl_mutex being held) has been replaced by cancel_work_sync() another issue has arisen and been left unresolved. As the MDIO bus cannot be accessed from the interrupt context the PHY interrupt handler uses disable_irq_nosync() to prevent from looping and schedules some work to be done as a softirq, which, apart from handling the state change of the originating PHY, is responsible for reenabling the interrupt. Now if the interrupt line is shared by another device and a call to the softirq handler has been cancelled, that call to enable_irq() never happens and the other device cannot use its interrupt anymore as its stuck disabled. I decided to use a counter rather than a flag because there may be more than one call to phy_change() cancelled in the queue -- a real one and a fake one triggered by free_irq() if DEBUG_SHIRQ is used, if nothing else. Therefore because of its nesting property enable_irq() has to be called the right number of times to match the number disable_irq_nosync() was called and restore the original state. This DEBUG_SHIRQ feature is also the reason why free_irq() has to be called before cancel_work_sync(). While at it I updated the comment about phy_stop_interrupts() being called from `keventd' -- this is no longer relevant as the use of cancel_work_sync() makes such an approach unnecessary. OTOH a similar comment referring to flush_scheduled_work() in phy_stop() still applies as using cancel_work_sync() there would be dangerous. Checked with checkpatch.pl and at the run time (with and without DEBUG_SHIRQ). Signed-off-by: Maciej W. Rozycki <macro@linux-mips.org> Cc: Andy Fleming <afleming@freescale.com> Cc: Jeff Garzik <jgarzik@pobox.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Jeff Garzik <jeff@garzik.org>
2007-09-28 23:42:14 -06:00
/*
* If work indeed has been cancelled, disable_irq() will have
* been left unbalanced from phy_interrupt() and enable_irq()
* has to be called so that other devices on the line work.
*/
while (atomic_dec_return(&phydev->irq_disable) >= 0)
enable_irq(phydev->irq);
return err;
}
EXPORT_SYMBOL(phy_stop_interrupts);
/**
* phy_change - Scheduled by the phy_interrupt/timer to handle PHY changes
* @work: work_struct that describes the work to be done
*/
static void phy_change(struct work_struct *work)
{
int err;
struct phy_device *phydev =
container_of(work, struct phy_device, phy_queue);
err = phy_disable_interrupts(phydev);
if (err)
goto phy_err;
mutex_lock(&phydev->lock);
if ((PHY_RUNNING == phydev->state) || (PHY_NOLINK == phydev->state))
phydev->state = PHY_CHANGELINK;
mutex_unlock(&phydev->lock);
PHYLIB: IRQ event workqueue handling fixes Keep track of disable_irq_nosync() invocations and call enable_irq() the right number of times if work has been cancelled that would include them. Now that the call to flush_work_keventd() (problematic because of rtnl_mutex being held) has been replaced by cancel_work_sync() another issue has arisen and been left unresolved. As the MDIO bus cannot be accessed from the interrupt context the PHY interrupt handler uses disable_irq_nosync() to prevent from looping and schedules some work to be done as a softirq, which, apart from handling the state change of the originating PHY, is responsible for reenabling the interrupt. Now if the interrupt line is shared by another device and a call to the softirq handler has been cancelled, that call to enable_irq() never happens and the other device cannot use its interrupt anymore as its stuck disabled. I decided to use a counter rather than a flag because there may be more than one call to phy_change() cancelled in the queue -- a real one and a fake one triggered by free_irq() if DEBUG_SHIRQ is used, if nothing else. Therefore because of its nesting property enable_irq() has to be called the right number of times to match the number disable_irq_nosync() was called and restore the original state. This DEBUG_SHIRQ feature is also the reason why free_irq() has to be called before cancel_work_sync(). While at it I updated the comment about phy_stop_interrupts() being called from `keventd' -- this is no longer relevant as the use of cancel_work_sync() makes such an approach unnecessary. OTOH a similar comment referring to flush_scheduled_work() in phy_stop() still applies as using cancel_work_sync() there would be dangerous. Checked with checkpatch.pl and at the run time (with and without DEBUG_SHIRQ). Signed-off-by: Maciej W. Rozycki <macro@linux-mips.org> Cc: Andy Fleming <afleming@freescale.com> Cc: Jeff Garzik <jgarzik@pobox.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Jeff Garzik <jeff@garzik.org>
2007-09-28 23:42:14 -06:00
atomic_dec(&phydev->irq_disable);
enable_irq(phydev->irq);
/* Reenable interrupts */
if (PHY_HALTED != phydev->state)
err = phy_config_interrupt(phydev, PHY_INTERRUPT_ENABLED);
if (err)
goto irq_enable_err;
return;
irq_enable_err:
disable_irq(phydev->irq);
PHYLIB: IRQ event workqueue handling fixes Keep track of disable_irq_nosync() invocations and call enable_irq() the right number of times if work has been cancelled that would include them. Now that the call to flush_work_keventd() (problematic because of rtnl_mutex being held) has been replaced by cancel_work_sync() another issue has arisen and been left unresolved. As the MDIO bus cannot be accessed from the interrupt context the PHY interrupt handler uses disable_irq_nosync() to prevent from looping and schedules some work to be done as a softirq, which, apart from handling the state change of the originating PHY, is responsible for reenabling the interrupt. Now if the interrupt line is shared by another device and a call to the softirq handler has been cancelled, that call to enable_irq() never happens and the other device cannot use its interrupt anymore as its stuck disabled. I decided to use a counter rather than a flag because there may be more than one call to phy_change() cancelled in the queue -- a real one and a fake one triggered by free_irq() if DEBUG_SHIRQ is used, if nothing else. Therefore because of its nesting property enable_irq() has to be called the right number of times to match the number disable_irq_nosync() was called and restore the original state. This DEBUG_SHIRQ feature is also the reason why free_irq() has to be called before cancel_work_sync(). While at it I updated the comment about phy_stop_interrupts() being called from `keventd' -- this is no longer relevant as the use of cancel_work_sync() makes such an approach unnecessary. OTOH a similar comment referring to flush_scheduled_work() in phy_stop() still applies as using cancel_work_sync() there would be dangerous. Checked with checkpatch.pl and at the run time (with and without DEBUG_SHIRQ). Signed-off-by: Maciej W. Rozycki <macro@linux-mips.org> Cc: Andy Fleming <afleming@freescale.com> Cc: Jeff Garzik <jgarzik@pobox.com> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Jeff Garzik <jeff@garzik.org>
2007-09-28 23:42:14 -06:00
atomic_inc(&phydev->irq_disable);
phy_err:
phy_error(phydev);
}
/**
* phy_stop - Bring down the PHY link, and stop checking the status
* @phydev: target phy_device struct
*/
void phy_stop(struct phy_device *phydev)
{
mutex_lock(&phydev->lock);
if (PHY_HALTED == phydev->state)
goto out_unlock;
if (phydev->irq != PHY_POLL) {
/* Disable PHY Interrupts */
phy_config_interrupt(phydev, PHY_INTERRUPT_DISABLED);
/* Clear any pending interrupts */
phy_clear_interrupt(phydev);
}
phydev->state = PHY_HALTED;
out_unlock:
mutex_unlock(&phydev->lock);
/*
* Cannot call flush_scheduled_work() here as desired because
* of rtnl_lock(), but PHY_HALTED shall guarantee phy_change()
* will not reenable interrupts.
*/
}
/**
* phy_start - start or restart a PHY device
* @phydev: target phy_device struct
*
* Description: Indicates the attached device's readiness to
* handle PHY-related work. Used during startup to start the
* PHY, and after a call to phy_stop() to resume operation.
* Also used to indicate the MDIO bus has cleared an error
* condition.
*/
void phy_start(struct phy_device *phydev)
{
mutex_lock(&phydev->lock);
switch (phydev->state) {
case PHY_STARTING:
phydev->state = PHY_PENDING;
break;
case PHY_READY:
phydev->state = PHY_UP;
break;
case PHY_HALTED:
phydev->state = PHY_RESUMING;
default:
break;
}
mutex_unlock(&phydev->lock);
}
EXPORT_SYMBOL(phy_stop);
EXPORT_SYMBOL(phy_start);
/**
* phy_state_machine - Handle the state machine
* @work: work_struct that describes the work to be done
*
* Description: Scheduled by the state_queue workqueue each time
* phy_timer is triggered.
*/
static void phy_state_machine(struct work_struct *work)
{
struct phy_device *phydev =
container_of(work, struct phy_device, state_queue);
int needs_aneg = 0;
int err = 0;
mutex_lock(&phydev->lock);
if (phydev->adjust_state)
phydev->adjust_state(phydev->attached_dev);
switch(phydev->state) {
case PHY_DOWN:
case PHY_STARTING:
case PHY_READY:
case PHY_PENDING:
break;
case PHY_UP:
needs_aneg = 1;
phydev->link_timeout = PHY_AN_TIMEOUT;
break;
case PHY_AN:
err = phy_read_status(phydev);
if (err < 0)
break;
/* If the link is down, give up on
* negotiation for now */
if (!phydev->link) {
phydev->state = PHY_NOLINK;
netif_carrier_off(phydev->attached_dev);
phydev->adjust_link(phydev->attached_dev);
break;
}
/* Check if negotiation is done. Break
* if there's an error */
err = phy_aneg_done(phydev);
if (err < 0)
break;
/* If AN is done, we're running */
if (err > 0) {
phydev->state = PHY_RUNNING;
netif_carrier_on(phydev->attached_dev);
phydev->adjust_link(phydev->attached_dev);
} else if (0 == phydev->link_timeout--) {
int idx;
needs_aneg = 1;
/* If we have the magic_aneg bit,
* we try again */
if (phydev->drv->flags & PHY_HAS_MAGICANEG)
break;
/* The timer expired, and we still
* don't have a setting, so we try
* forcing it until we find one that
* works, starting from the fastest speed,
* and working our way down */
idx = phy_find_valid(0, phydev->supported);
phydev->speed = settings[idx].speed;
phydev->duplex = settings[idx].duplex;
phydev->autoneg = AUTONEG_DISABLE;
pr_info("Trying %d/%s\n", phydev->speed,
DUPLEX_FULL ==
phydev->duplex ?
"FULL" : "HALF");
}
break;
case PHY_NOLINK:
err = phy_read_status(phydev);
if (err)
break;
if (phydev->link) {
phydev->state = PHY_RUNNING;
netif_carrier_on(phydev->attached_dev);
phydev->adjust_link(phydev->attached_dev);
}
break;
case PHY_FORCING:
err = genphy_update_link(phydev);
if (err)
break;
if (phydev->link) {
phydev->state = PHY_RUNNING;
netif_carrier_on(phydev->attached_dev);
} else {
if (0 == phydev->link_timeout--) {
phy_force_reduction(phydev);
needs_aneg = 1;
}
}
phydev->adjust_link(phydev->attached_dev);
break;
case PHY_RUNNING:
/* Only register a CHANGE if we are
* polling */
if (PHY_POLL == phydev->irq)
phydev->state = PHY_CHANGELINK;
break;
case PHY_CHANGELINK:
err = phy_read_status(phydev);
if (err)
break;
if (phydev->link) {
phydev->state = PHY_RUNNING;
netif_carrier_on(phydev->attached_dev);
} else {
phydev->state = PHY_NOLINK;
netif_carrier_off(phydev->attached_dev);
}
phydev->adjust_link(phydev->attached_dev);
if (PHY_POLL != phydev->irq)
err = phy_config_interrupt(phydev,
PHY_INTERRUPT_ENABLED);
break;
case PHY_HALTED:
if (phydev->link) {
phydev->link = 0;
netif_carrier_off(phydev->attached_dev);
phydev->adjust_link(phydev->attached_dev);
}
break;
case PHY_RESUMING:
err = phy_clear_interrupt(phydev);
if (err)
break;
err = phy_config_interrupt(phydev,
PHY_INTERRUPT_ENABLED);
if (err)
break;
if (AUTONEG_ENABLE == phydev->autoneg) {
err = phy_aneg_done(phydev);
if (err < 0)
break;
/* err > 0 if AN is done.
* Otherwise, it's 0, and we're
* still waiting for AN */
if (err > 0) {
phydev->state = PHY_RUNNING;
} else {
phydev->state = PHY_AN;
phydev->link_timeout = PHY_AN_TIMEOUT;
}
} else
phydev->state = PHY_RUNNING;
break;
}
mutex_unlock(&phydev->lock);
if (needs_aneg)
err = phy_start_aneg(phydev);
if (err < 0)
phy_error(phydev);
mod_timer(&phydev->phy_timer, jiffies + PHY_STATE_TIME * HZ);
}
/* PHY timer which schedules the state machine work */
static void phy_timer(unsigned long data)
{
struct phy_device *phydev = (struct phy_device *)data;
/*
* PHY I/O operations can potentially sleep so we ensure that
* it's done from a process context
*/
schedule_work(&phydev->state_queue);
}