linux/drivers/usb/host/ehci-hcd.c

1247 lines
34 KiB
C
Raw Normal View History

/*
* Copyright (c) 2000-2004 by David Brownell
*
* 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.
*
* 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., 675 Mass Ave, Cambridge, MA 02139, USA.
*/
#include <linux/module.h>
#include <linux/pci.h>
#include <linux/dmapool.h>
#include <linux/kernel.h>
#include <linux/delay.h>
#include <linux/ioport.h>
#include <linux/sched.h>
#include <linux/slab.h>
#include <linux/vmalloc.h>
#include <linux/errno.h>
#include <linux/init.h>
#include <linux/timer.h>
#include <linux/list.h>
#include <linux/interrupt.h>
#include <linux/usb.h>
#include <linux/moduleparam.h>
#include <linux/dma-mapping.h>
#include <linux/debugfs.h>
#include "../core/hcd.h"
#include <asm/byteorder.h>
#include <asm/io.h>
#include <asm/irq.h>
#include <asm/system.h>
#include <asm/unaligned.h>
/*-------------------------------------------------------------------------*/
/*
* EHCI hc_driver implementation ... experimental, incomplete.
* Based on the final 1.0 register interface specification.
*
* USB 2.0 shows up in upcoming www.pcmcia.org technology.
* First was PCMCIA, like ISA; then CardBus, which is PCI.
* Next comes "CardBay", using USB 2.0 signals.
*
* Contains additional contributions by Brad Hards, Rory Bolt, and others.
* Special thanks to Intel and VIA for providing host controllers to
* test this driver on, and Cypress (including In-System Design) for
* providing early devices for those host controllers to talk to!
*/
#define DRIVER_AUTHOR "David Brownell"
#define DRIVER_DESC "USB 2.0 'Enhanced' Host Controller (EHCI) Driver"
static const char hcd_name [] = "ehci_hcd";
#undef VERBOSE_DEBUG
#undef EHCI_URB_TRACE
#ifdef DEBUG
#define EHCI_STATS
#endif
/* magic numbers that can affect system performance */
#define EHCI_TUNE_CERR 3 /* 0-3 qtd retries; 0 == don't stop */
#define EHCI_TUNE_RL_HS 4 /* nak throttle; see 4.9 */
#define EHCI_TUNE_RL_TT 0
#define EHCI_TUNE_MULT_HS 1 /* 1-3 transactions/uframe; 4.10.3 */
#define EHCI_TUNE_MULT_TT 1
#define EHCI_TUNE_FLS 2 /* (small) 256 frame schedule */
#define EHCI_IAA_MSECS 10 /* arbitrary */
#define EHCI_IO_JIFFIES (HZ/10) /* io watchdog > irq_thresh */
#define EHCI_ASYNC_JIFFIES (HZ/20) /* async idle timeout */
#define EHCI_SHRINK_FRAMES 5 /* async qh unlink delay */
/* Initial IRQ latency: faster than hw default */
static int log2_irq_thresh = 0; // 0 to 6
module_param (log2_irq_thresh, int, S_IRUGO);
MODULE_PARM_DESC (log2_irq_thresh, "log2 IRQ latency, 1-64 microframes");
/* initial park setting: slower than hw default */
static unsigned park = 0;
module_param (park, uint, S_IRUGO);
MODULE_PARM_DESC (park, "park setting; 1-3 back-to-back async packets");
/* for flakey hardware, ignore overcurrent indicators */
static int ignore_oc = 0;
module_param (ignore_oc, bool, S_IRUGO);
MODULE_PARM_DESC (ignore_oc, "ignore bogus hardware overcurrent indications");
#define INTR_MASK (STS_IAA | STS_FATAL | STS_PCD | STS_ERR | STS_INT)
/*-------------------------------------------------------------------------*/
#include "ehci.h"
#include "ehci-dbg.c"
/*-------------------------------------------------------------------------*/
static void
timer_action(struct ehci_hcd *ehci, enum ehci_timer_action action)
{
/* Don't override timeouts which shrink or (later) disable
* the async ring; just the I/O watchdog. Note that if a
* SHRINK were pending, OFF would never be requested.
*/
if (timer_pending(&ehci->watchdog)
&& ((BIT(TIMER_ASYNC_SHRINK) | BIT(TIMER_ASYNC_OFF))
& ehci->actions))
return;
if (!test_and_set_bit(action, &ehci->actions)) {
unsigned long t;
switch (action) {
case TIMER_IO_WATCHDOG:
if (!ehci->need_io_watchdog)
return;
t = EHCI_IO_JIFFIES;
break;
case TIMER_ASYNC_OFF:
t = EHCI_ASYNC_JIFFIES;
break;
/* case TIMER_ASYNC_SHRINK: */
default:
/* add a jiffie since we synch against the
* 8 KHz uframe counter.
*/
t = DIV_ROUND_UP(EHCI_SHRINK_FRAMES * HZ, 1000) + 1;
break;
}
mod_timer(&ehci->watchdog, t + jiffies);
}
}
/*-------------------------------------------------------------------------*/
/*
* handshake - spin reading hc until handshake completes or fails
* @ptr: address of hc register to be read
* @mask: bits to look at in result of read
* @done: value of those bits when handshake succeeds
* @usec: timeout in microseconds
*
* Returns negative errno, or zero on success
*
* Success happens when the "mask" bits have the specified value (hardware
* handshake done). There are two failure modes: "usec" have passed (major
* hardware flakeout), or the register reads as all-ones (hardware removed).
*
* That last failure should_only happen in cases like physical cardbus eject
* before driver shutdown. But it also seems to be caused by bugs in cardbus
* bridge shutdown: shutting down the bridge before the devices using it.
*/
static int handshake (struct ehci_hcd *ehci, void __iomem *ptr,
u32 mask, u32 done, int usec)
{
u32 result;
do {
result = ehci_readl(ehci, ptr);
if (result == ~(u32)0) /* card removed */
return -ENODEV;
result &= mask;
if (result == done)
return 0;
udelay (1);
usec--;
} while (usec > 0);
return -ETIMEDOUT;
}
/* force HC to halt state from unknown (EHCI spec section 2.3) */
static int ehci_halt (struct ehci_hcd *ehci)
{
u32 temp = ehci_readl(ehci, &ehci->regs->status);
/* disable any irqs left enabled by previous code */
ehci_writel(ehci, 0, &ehci->regs->intr_enable);
if ((temp & STS_HALT) != 0)
return 0;
temp = ehci_readl(ehci, &ehci->regs->command);
temp &= ~CMD_RUN;
ehci_writel(ehci, temp, &ehci->regs->command);
return handshake (ehci, &ehci->regs->status,
STS_HALT, STS_HALT, 16 * 125);
}
static int handshake_on_error_set_halt(struct ehci_hcd *ehci, void __iomem *ptr,
u32 mask, u32 done, int usec)
{
int error;
error = handshake(ehci, ptr, mask, done, usec);
if (error) {
ehci_halt(ehci);
ehci_to_hcd(ehci)->state = HC_STATE_HALT;
ehci_err(ehci, "force halt; handhake %p %08x %08x -> %d\n",
ptr, mask, done, error);
}
return error;
}
/* put TDI/ARC silicon into EHCI mode */
static void tdi_reset (struct ehci_hcd *ehci)
{
u32 __iomem *reg_ptr;
u32 tmp;
reg_ptr = (u32 __iomem *)(((u8 __iomem *)ehci->regs) + USBMODE);
tmp = ehci_readl(ehci, reg_ptr);
tmp |= USBMODE_CM_HC;
/* The default byte access to MMR space is LE after
* controller reset. Set the required endian mode
* for transfer buffers to match the host microprocessor
*/
if (ehci_big_endian_mmio(ehci))
tmp |= USBMODE_BE;
ehci_writel(ehci, tmp, reg_ptr);
}
/* reset a non-running (STS_HALT == 1) controller */
static int ehci_reset (struct ehci_hcd *ehci)
{
int retval;
u32 command = ehci_readl(ehci, &ehci->regs->command);
command |= CMD_RESET;
dbg_cmd (ehci, "reset", command);
ehci_writel(ehci, command, &ehci->regs->command);
ehci_to_hcd(ehci)->state = HC_STATE_HALT;
ehci->next_statechange = jiffies;
retval = handshake (ehci, &ehci->regs->command,
CMD_RESET, 0, 250 * 1000);
if (ehci->has_hostpc) {
ehci_writel(ehci, USBMODE_EX_HC | USBMODE_EX_VBPS,
(u32 __iomem *)(((u8 *)ehci->regs) + USBMODE_EX));
ehci_writel(ehci, TXFIFO_DEFAULT,
(u32 __iomem *)(((u8 *)ehci->regs) + TXFILLTUNING));
}
if (retval)
return retval;
if (ehci_is_TDI(ehci))
tdi_reset (ehci);
return retval;
}
/* idle the controller (from running) */
static void ehci_quiesce (struct ehci_hcd *ehci)
{
u32 temp;
#ifdef DEBUG
if (!HC_IS_RUNNING (ehci_to_hcd(ehci)->state))
BUG ();
#endif
/* wait for any schedule enables/disables to take effect */
temp = ehci_readl(ehci, &ehci->regs->command) << 10;
temp &= STS_ASS | STS_PSS;
if (handshake_on_error_set_halt(ehci, &ehci->regs->status,
STS_ASS | STS_PSS, temp, 16 * 125))
return;
/* then disable anything that's still active */
temp = ehci_readl(ehci, &ehci->regs->command);
temp &= ~(CMD_ASE | CMD_IAAD | CMD_PSE);
ehci_writel(ehci, temp, &ehci->regs->command);
/* hardware can take 16 microframes to turn off ... */
handshake_on_error_set_halt(ehci, &ehci->regs->status,
STS_ASS | STS_PSS, 0, 16 * 125);
}
/*-------------------------------------------------------------------------*/
static void end_unlink_async(struct ehci_hcd *ehci);
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 21:55:46 +08:00
static void ehci_work(struct ehci_hcd *ehci);
#include "ehci-hub.c"
#include "ehci-mem.c"
#include "ehci-q.c"
#include "ehci-sched.c"
/*-------------------------------------------------------------------------*/
static void ehci_iaa_watchdog(unsigned long param)
{
struct ehci_hcd *ehci = (struct ehci_hcd *) param;
unsigned long flags;
spin_lock_irqsave (&ehci->lock, flags);
/* Lost IAA irqs wedge things badly; seen first with a vt8235.
* So we need this watchdog, but must protect it against both
* (a) SMP races against real IAA firing and retriggering, and
* (b) clean HC shutdown, when IAA watchdog was pending.
*/
if (ehci->reclaim
&& !timer_pending(&ehci->iaa_watchdog)
&& HC_IS_RUNNING(ehci_to_hcd(ehci)->state)) {
u32 cmd, status;
/* If we get here, IAA is *REALLY* late. It's barely
* conceivable that the system is so busy that CMD_IAAD
* is still legitimately set, so let's be sure it's
* clear before we read STS_IAA. (The HC should clear
* CMD_IAAD when it sets STS_IAA.)
*/
cmd = ehci_readl(ehci, &ehci->regs->command);
if (cmd & CMD_IAAD)
ehci_writel(ehci, cmd & ~CMD_IAAD,
&ehci->regs->command);
/* If IAA is set here it either legitimately triggered
* before we cleared IAAD above (but _way_ late, so we'll
* still count it as lost) ... or a silicon erratum:
* - VIA seems to set IAA without triggering the IRQ;
* - IAAD potentially cleared without setting IAA.
*/
status = ehci_readl(ehci, &ehci->regs->status);
if ((status & STS_IAA) || !(cmd & CMD_IAAD)) {
COUNT (ehci->stats.lost_iaa);
ehci_writel(ehci, STS_IAA, &ehci->regs->status);
}
ehci_vdbg(ehci, "IAA watchdog: status %x cmd %x\n",
status, cmd);
end_unlink_async(ehci);
}
spin_unlock_irqrestore(&ehci->lock, flags);
}
static void ehci_watchdog(unsigned long param)
{
struct ehci_hcd *ehci = (struct ehci_hcd *) param;
unsigned long flags;
spin_lock_irqsave(&ehci->lock, flags);
/* stop async processing after it's idled a bit */
if (test_bit (TIMER_ASYNC_OFF, &ehci->actions))
start_unlink_async (ehci, ehci->async);
/* ehci could run by timer, without IRQs ... */
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 21:55:46 +08:00
ehci_work (ehci);
spin_unlock_irqrestore (&ehci->lock, flags);
}
/* On some systems, leaving remote wakeup enabled prevents system shutdown.
* The firmware seems to think that powering off is a wakeup event!
* This routine turns off remote wakeup and everything else, on all ports.
*/
static void ehci_turn_off_all_ports(struct ehci_hcd *ehci)
{
int port = HCS_N_PORTS(ehci->hcs_params);
while (port--)
ehci_writel(ehci, PORT_RWC_BITS,
&ehci->regs->port_status[port]);
}
/*
* Halt HC, turn off all ports, and let the BIOS use the companion controllers.
* Should be called with ehci->lock held.
*/
static void ehci_silence_controller(struct ehci_hcd *ehci)
{
ehci_halt(ehci);
ehci_turn_off_all_ports(ehci);
/* make BIOS/etc use companion controller during reboot */
ehci_writel(ehci, 0, &ehci->regs->configured_flag);
/* unblock posted writes */
ehci_readl(ehci, &ehci->regs->configured_flag);
}
/* ehci_shutdown kick in for silicon on any bus (not just pci, etc).
* This forcibly disables dma and IRQs, helping kexec and other cases
* where the next system software may expect clean state.
*/
static void ehci_shutdown(struct usb_hcd *hcd)
{
struct ehci_hcd *ehci = hcd_to_ehci(hcd);
del_timer_sync(&ehci->watchdog);
del_timer_sync(&ehci->iaa_watchdog);
spin_lock_irq(&ehci->lock);
ehci_silence_controller(ehci);
spin_unlock_irq(&ehci->lock);
}
static void ehci_port_power (struct ehci_hcd *ehci, int is_on)
{
unsigned port;
if (!HCS_PPC (ehci->hcs_params))
return;
ehci_dbg (ehci, "...power%s ports...\n", is_on ? "up" : "down");
for (port = HCS_N_PORTS (ehci->hcs_params); port > 0; )
(void) ehci_hub_control(ehci_to_hcd(ehci),
is_on ? SetPortFeature : ClearPortFeature,
USB_PORT_FEAT_POWER,
port--, NULL, 0);
/* Flush those writes */
ehci_readl(ehci, &ehci->regs->command);
msleep(20);
}
/*-------------------------------------------------------------------------*/
/*
* ehci_work is called from some interrupts, timers, and so on.
* it calls driver completion functions, after dropping ehci->lock.
*/
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 21:55:46 +08:00
static void ehci_work (struct ehci_hcd *ehci)
{
timer_action_done (ehci, TIMER_IO_WATCHDOG);
/* another CPU may drop ehci->lock during a schedule scan while
* it reports urb completions. this flag guards against bogus
* attempts at re-entrant schedule scanning.
*/
if (ehci->scanning)
return;
ehci->scanning = 1;
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 21:55:46 +08:00
scan_async (ehci);
if (ehci->next_uframe != -1)
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 21:55:46 +08:00
scan_periodic (ehci);
ehci->scanning = 0;
/* the IO watchdog guards against hardware or driver bugs that
* misplace IRQs, and should let us run completely without IRQs.
* such lossage has been observed on both VT6202 and VT8235.
*/
if (HC_IS_RUNNING (ehci_to_hcd(ehci)->state) &&
(ehci->async->qh_next.ptr != NULL ||
ehci->periodic_sched != 0))
timer_action (ehci, TIMER_IO_WATCHDOG);
}
/*
* Called when the ehci_hcd module is removed.
*/
static void ehci_stop (struct usb_hcd *hcd)
{
struct ehci_hcd *ehci = hcd_to_ehci (hcd);
ehci_dbg (ehci, "stop\n");
/* no more interrupts ... */
del_timer_sync (&ehci->watchdog);
del_timer_sync(&ehci->iaa_watchdog);
spin_lock_irq(&ehci->lock);
if (HC_IS_RUNNING (hcd->state))
ehci_quiesce (ehci);
ehci_silence_controller(ehci);
ehci_reset (ehci);
spin_unlock_irq(&ehci->lock);
remove_companion_file(ehci);
remove_debug_files (ehci);
/* root hub is shut down separately (first, when possible) */
spin_lock_irq (&ehci->lock);
if (ehci->async)
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 21:55:46 +08:00
ehci_work (ehci);
spin_unlock_irq (&ehci->lock);
ehci_mem_cleanup (ehci);
#ifdef EHCI_STATS
ehci_dbg (ehci, "irq normal %ld err %ld reclaim %ld (lost %ld)\n",
ehci->stats.normal, ehci->stats.error, ehci->stats.reclaim,
ehci->stats.lost_iaa);
ehci_dbg (ehci, "complete %ld unlink %ld\n",
ehci->stats.complete, ehci->stats.unlink);
#endif
dbg_status (ehci, "ehci_stop completed",
ehci_readl(ehci, &ehci->regs->status));
}
/* one-time init, only for memory state */
static int ehci_init(struct usb_hcd *hcd)
{
struct ehci_hcd *ehci = hcd_to_ehci(hcd);
u32 temp;
int retval;
u32 hcc_params;
struct ehci_qh_hw *hw;
spin_lock_init(&ehci->lock);
/*
* keep io watchdog by default, those good HCDs could turn off it later
*/
ehci->need_io_watchdog = 1;
init_timer(&ehci->watchdog);
ehci->watchdog.function = ehci_watchdog;
ehci->watchdog.data = (unsigned long) ehci;
init_timer(&ehci->iaa_watchdog);
ehci->iaa_watchdog.function = ehci_iaa_watchdog;
ehci->iaa_watchdog.data = (unsigned long) ehci;
/*
* hw default: 1K periodic list heads, one per frame.
* periodic_size can shrink by USBCMD update if hcc_params allows.
*/
ehci->periodic_size = DEFAULT_I_TDPS;
INIT_LIST_HEAD(&ehci->cached_itd_list);
if ((retval = ehci_mem_init(ehci, GFP_KERNEL)) < 0)
return retval;
/* controllers may cache some of the periodic schedule ... */
hcc_params = ehci_readl(ehci, &ehci->caps->hcc_params);
if (HCC_ISOC_CACHE(hcc_params)) // full frame cache
ehci->i_thresh = 8;
else // N microframes cached
ehci->i_thresh = 2 + HCC_ISOC_THRES(hcc_params);
ehci->reclaim = NULL;
ehci->next_uframe = -1;
ehci->clock_frame = -1;
/*
* dedicate a qh for the async ring head, since we couldn't unlink
* a 'real' qh without stopping the async schedule [4.8]. use it
* as the 'reclamation list head' too.
* its dummy is used in hw_alt_next of many tds, to prevent the qh
* from automatically advancing to the next td after short reads.
*/
ehci->async->qh_next.qh = NULL;
hw = ehci->async->hw;
hw->hw_next = QH_NEXT(ehci, ehci->async->qh_dma);
hw->hw_info1 = cpu_to_hc32(ehci, QH_HEAD);
hw->hw_token = cpu_to_hc32(ehci, QTD_STS_HALT);
hw->hw_qtd_next = EHCI_LIST_END(ehci);
ehci->async->qh_state = QH_STATE_LINKED;
hw->hw_alt_next = QTD_NEXT(ehci, ehci->async->dummy->qtd_dma);
/* clear interrupt enables, set irq latency */
if (log2_irq_thresh < 0 || log2_irq_thresh > 6)
log2_irq_thresh = 0;
temp = 1 << (16 + log2_irq_thresh);
if (HCC_CANPARK(hcc_params)) {
/* HW default park == 3, on hardware that supports it (like
* NVidia and ALI silicon), maximizes throughput on the async
* schedule by avoiding QH fetches between transfers.
*
* With fast usb storage devices and NForce2, "park" seems to
* make problems: throughput reduction (!), data errors...
*/
if (park) {
park = min(park, (unsigned) 3);
temp |= CMD_PARK;
temp |= park << 8;
}
ehci_dbg(ehci, "park %d\n", park);
}
if (HCC_PGM_FRAMELISTLEN(hcc_params)) {
/* periodic schedule size can be smaller than default */
temp &= ~(3 << 2);
temp |= (EHCI_TUNE_FLS << 2);
switch (EHCI_TUNE_FLS) {
case 0: ehci->periodic_size = 1024; break;
case 1: ehci->periodic_size = 512; break;
case 2: ehci->periodic_size = 256; break;
default: BUG();
}
}
ehci->command = temp;
return 0;
}
/* start HC running; it's halted, ehci_init() has been run (once) */
static int ehci_run (struct usb_hcd *hcd)
{
struct ehci_hcd *ehci = hcd_to_ehci (hcd);
int retval;
u32 temp;
u32 hcc_params;
hcd->uses_new_polling = 1;
hcd->poll_rh = 0;
/* EHCI spec section 4.1 */
if ((retval = ehci_reset(ehci)) != 0) {
ehci_mem_cleanup(ehci);
return retval;
}
ehci_writel(ehci, ehci->periodic_dma, &ehci->regs->frame_list);
ehci_writel(ehci, (u32)ehci->async->qh_dma, &ehci->regs->async_next);
/*
* hcc_params controls whether ehci->regs->segment must (!!!)
* be used; it constrains QH/ITD/SITD and QTD locations.
* pci_pool consistent memory always uses segment zero.
* streaming mappings for I/O buffers, like pci_map_single(),
* can return segments above 4GB, if the device allows.
*
* NOTE: the dma mask is visible through dma_supported(), so
* drivers can pass this info along ... like NETIF_F_HIGHDMA,
* Scsi_Host.highmem_io, and so forth. It's readonly to all
* host side drivers though.
*/
hcc_params = ehci_readl(ehci, &ehci->caps->hcc_params);
if (HCC_64BIT_ADDR(hcc_params)) {
ehci_writel(ehci, 0, &ehci->regs->segment);
#if 0
// this is deeply broken on almost all architectures
if (!dma_set_mask(hcd->self.controller, DMA_BIT_MASK(64)))
ehci_info(ehci, "enabled 64bit DMA\n");
#endif
}
// Philips, Intel, and maybe others need CMD_RUN before the
// root hub will detect new devices (why?); NEC doesn't
ehci->command &= ~(CMD_LRESET|CMD_IAAD|CMD_PSE|CMD_ASE|CMD_RESET);
ehci->command |= CMD_RUN;
ehci_writel(ehci, ehci->command, &ehci->regs->command);
dbg_cmd (ehci, "init", ehci->command);
/*
* Start, enabling full USB 2.0 functionality ... usb 1.1 devices
* are explicitly handed to companion controller(s), so no TT is
* involved with the root hub. (Except where one is integrated,
* and there's no companion controller unless maybe for USB OTG.)
*
* Turning on the CF flag will transfer ownership of all ports
* from the companions to the EHCI controller. If any of the
* companions are in the middle of a port reset at the time, it
* could cause trouble. Write-locking ehci_cf_port_reset_rwsem
* guarantees that no resets are in progress. After we set CF,
* a short delay lets the hardware catch up; new resets shouldn't
* be started before the port switching actions could complete.
*/
down_write(&ehci_cf_port_reset_rwsem);
hcd->state = HC_STATE_RUNNING;
ehci_writel(ehci, FLAG_CF, &ehci->regs->configured_flag);
ehci_readl(ehci, &ehci->regs->command); /* unblock posted writes */
msleep(5);
up_write(&ehci_cf_port_reset_rwsem);
temp = HC_VERSION(ehci_readl(ehci, &ehci->caps->hc_capbase));
ehci_info (ehci,
"USB %x.%x started, EHCI %x.%02x%s\n",
((ehci->sbrn & 0xf0)>>4), (ehci->sbrn & 0x0f),
temp >> 8, temp & 0xff,
ignore_oc ? ", overcurrent ignored" : "");
ehci_writel(ehci, INTR_MASK,
&ehci->regs->intr_enable); /* Turn On Interrupts */
/* GRR this is run-once init(), being done every time the HC starts.
* So long as they're part of class devices, we can't do it init()
* since the class device isn't created that early.
*/
create_debug_files(ehci);
create_companion_file(ehci);
return 0;
}
/*-------------------------------------------------------------------------*/
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 21:55:46 +08:00
static irqreturn_t ehci_irq (struct usb_hcd *hcd)
{
struct ehci_hcd *ehci = hcd_to_ehci (hcd);
u32 status, masked_status, pcd_status = 0, cmd;
int bh;
spin_lock (&ehci->lock);
status = ehci_readl(ehci, &ehci->regs->status);
/* e.g. cardbus physical eject */
if (status == ~(u32) 0) {
ehci_dbg (ehci, "device removed\n");
goto dead;
}
masked_status = status & INTR_MASK;
if (!masked_status) { /* irq sharing? */
spin_unlock(&ehci->lock);
return IRQ_NONE;
}
/* clear (just) interrupts */
ehci_writel(ehci, masked_status, &ehci->regs->status);
cmd = ehci_readl(ehci, &ehci->regs->command);
bh = 0;
#ifdef VERBOSE_DEBUG
/* unrequested/ignored: Frame List Rollover */
dbg_status (ehci, "irq", status);
#endif
/* INT, ERR, and IAA interrupt rates can be throttled */
/* normal [4.15.1.2] or error [4.15.1.1] completion */
if (likely ((status & (STS_INT|STS_ERR)) != 0)) {
if (likely ((status & STS_ERR) == 0))
COUNT (ehci->stats.normal);
else
COUNT (ehci->stats.error);
bh = 1;
}
/* complete the unlinking of some qh [4.15.2.3] */
if (status & STS_IAA) {
/* guard against (alleged) silicon errata */
if (cmd & CMD_IAAD) {
ehci_writel(ehci, cmd & ~CMD_IAAD,
&ehci->regs->command);
ehci_dbg(ehci, "IAA with IAAD still set?\n");
}
if (ehci->reclaim) {
COUNT(ehci->stats.reclaim);
end_unlink_async(ehci);
} else
ehci_dbg(ehci, "IAA with nothing to reclaim?\n");
}
/* remote wakeup [4.3.1] */
if (status & STS_PCD) {
unsigned i = HCS_N_PORTS (ehci->hcs_params);
/* kick root hub later */
pcd_status = status;
/* resume root hub? */
if (!(cmd & CMD_RUN))
EHCI: Fix root-hub and port suspend/resume problems This patch (as738b) fixes numerous problems in the controller/root-hub suspend/resume/remote-wakeup support in ehci-hcd: The bus_resume() routine should wake up only the ports that were suspended by bus_suspend(). Ports that were already suspended should remain that way. The interrupt mask is used to detect loss of power in the bus_resume() routine (if the mask is 0 then power was lost). However bus_suspend() always sets the mask to 0. Instead the mask should retain its normal value, with port-change-detect interrupts disabled if remote wakeup is turned off. The interrupt mask should be reset to its correct value at the end of bus_resume() regardless of whether power was lost. bus_resume() reinitializes the operational registers if power was lost. However those registers are not in the aux power well, hence they can lose their values whenever the controller is put into D3. They should always be reinitialized. When a port-change interrupt occurs and the root hub is suspended, the interrupt handler should request a root-hub resume instead of starting up the controller all by itself. There's no need for the interrupt handler to request a root-hub resume every time a suspended port sends a remote-wakeup request. The pci_resume() method doesn't need to check for connected ports when deciding whether or not to reset the controller. It can make that decision based on whether Vaux power was maintained. Even when the controller does not need to be reset, pci_resume() must undo the effect of pci_suspend() by re-enabling the interrupt mask. If power was lost, pci_resume() must not call ehci_run(). At this point the root hub is still supposed to be suspended, not running. It's enough to rewrite the command register and set the configured_flag. Signed-off-by: Alan Stern <stern@rowland.harvard.edu> Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
2006-11-10 03:42:16 +08:00
usb_hcd_resume_root_hub(hcd);
while (i--) {
int pstatus = ehci_readl(ehci,
&ehci->regs->port_status [i]);
if (pstatus & PORT_OWNER)
continue;
if (!(test_bit(i, &ehci->suspended_ports) &&
((pstatus & PORT_RESUME) ||
!(pstatus & PORT_SUSPEND)) &&
(pstatus & PORT_PE) &&
ehci->reset_done[i] == 0))
continue;
/* start 20 msec resume signaling from this port,
* and make khubd collect PORT_STAT_C_SUSPEND to
* stop that signaling.
*/
ehci->reset_done [i] = jiffies + msecs_to_jiffies (20);
ehci_dbg (ehci, "port %d remote wakeup\n", i + 1);
mod_timer(&hcd->rh_timer, ehci->reset_done[i]);
}
}
/* PCI errors [4.15.2.4] */
if (unlikely ((status & STS_FATAL) != 0)) {
ehci_err(ehci, "fatal error\n");
dbg_cmd(ehci, "fatal", cmd);
dbg_status(ehci, "fatal", status);
ehci_halt(ehci);
dead:
ehci_reset(ehci);
ehci_writel(ehci, 0, &ehci->regs->configured_flag);
/* generic layer kills/unlinks all urbs, then
* uses ehci_stop to clean up the rest
*/
bh = 1;
}
if (bh)
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 21:55:46 +08:00
ehci_work (ehci);
spin_unlock (&ehci->lock);
if (pcd_status)
usb_hcd_poll_rh_status(hcd);
return IRQ_HANDLED;
}
/*-------------------------------------------------------------------------*/
/*
* non-error returns are a promise to giveback() the urb later
* we drop ownership so next owner (or urb unlink) can get it
*
* urb + dev is in hcd.self.controller.urb_list
* we're queueing TDs onto software and hardware lists
*
* hcd-specific init for hcpriv hasn't been done yet
*
* NOTE: control, bulk, and interrupt share the same code to append TDs
* to a (possibly active) QH, and the same QH scanning code.
*/
static int ehci_urb_enqueue (
struct usb_hcd *hcd,
struct urb *urb,
gfp_t mem_flags
) {
struct ehci_hcd *ehci = hcd_to_ehci (hcd);
struct list_head qtd_list;
INIT_LIST_HEAD (&qtd_list);
switch (usb_pipetype (urb->pipe)) {
case PIPE_CONTROL:
/* qh_completions() code doesn't handle all the fault cases
* in multi-TD control transfers. Even 1KB is rare anyway.
*/
if (urb->transfer_buffer_length > (16 * 1024))
return -EMSGSIZE;
/* FALLTHROUGH */
/* case PIPE_BULK: */
default:
if (!qh_urb_transaction (ehci, urb, &qtd_list, mem_flags))
return -ENOMEM;
return submit_async(ehci, urb, &qtd_list, mem_flags);
case PIPE_INTERRUPT:
if (!qh_urb_transaction (ehci, urb, &qtd_list, mem_flags))
return -ENOMEM;
return intr_submit(ehci, urb, &qtd_list, mem_flags);
case PIPE_ISOCHRONOUS:
if (urb->dev->speed == USB_SPEED_HIGH)
return itd_submit (ehci, urb, mem_flags);
else
return sitd_submit (ehci, urb, mem_flags);
}
}
static void unlink_async (struct ehci_hcd *ehci, struct ehci_qh *qh)
{
/* failfast */
if (!HC_IS_RUNNING(ehci_to_hcd(ehci)->state) && ehci->reclaim)
end_unlink_async(ehci);
/* if it's not linked then there's nothing to do */
if (qh->qh_state != QH_STATE_LINKED)
;
/* defer till later if busy */
else if (ehci->reclaim) {
struct ehci_qh *last;
for (last = ehci->reclaim;
last->reclaim;
last = last->reclaim)
continue;
qh->qh_state = QH_STATE_UNLINK_WAIT;
last->reclaim = qh;
/* start IAA cycle */
} else
start_unlink_async (ehci, qh);
}
/* remove from hardware lists
* completions normally happen asynchronously
*/
static int ehci_urb_dequeue(struct usb_hcd *hcd, struct urb *urb, int status)
{
struct ehci_hcd *ehci = hcd_to_ehci (hcd);
struct ehci_qh *qh;
unsigned long flags;
int rc;
spin_lock_irqsave (&ehci->lock, flags);
rc = usb_hcd_check_unlink_urb(hcd, urb, status);
if (rc)
goto done;
switch (usb_pipetype (urb->pipe)) {
// case PIPE_CONTROL:
// case PIPE_BULK:
default:
qh = (struct ehci_qh *) urb->hcpriv;
if (!qh)
break;
switch (qh->qh_state) {
case QH_STATE_LINKED:
case QH_STATE_COMPLETING:
unlink_async(ehci, qh);
break;
case QH_STATE_UNLINK:
case QH_STATE_UNLINK_WAIT:
/* already started */
break;
case QH_STATE_IDLE:
/* QH might be waiting for a Clear-TT-Buffer */
qh_completions(ehci, qh);
break;
}
break;
case PIPE_INTERRUPT:
qh = (struct ehci_qh *) urb->hcpriv;
if (!qh)
break;
switch (qh->qh_state) {
case QH_STATE_LINKED:
intr_deschedule (ehci, qh);
/* FALL THROUGH */
case QH_STATE_IDLE:
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 21:55:46 +08:00
qh_completions (ehci, qh);
break;
default:
ehci_dbg (ehci, "bogus qh %p state %d\n",
qh, qh->qh_state);
goto done;
}
/* reschedule QH iff another request is queued */
if (!list_empty (&qh->qtd_list)
&& HC_IS_RUNNING (hcd->state)) {
rc = qh_schedule(ehci, qh);
/* An error here likely indicates handshake failure
* or no space left in the schedule. Neither fault
* should happen often ...
*
* FIXME kill the now-dysfunctional queued urbs
*/
if (rc != 0)
ehci_err(ehci,
"can't reschedule qh %p, err %d",
qh, rc);
}
break;
case PIPE_ISOCHRONOUS:
// itd or sitd ...
// wait till next completion, do it then.
// completion irqs can wait up to 1024 msec,
break;
}
done:
spin_unlock_irqrestore (&ehci->lock, flags);
return rc;
}
/*-------------------------------------------------------------------------*/
// bulk qh holds the data toggle
static void
ehci_endpoint_disable (struct usb_hcd *hcd, struct usb_host_endpoint *ep)
{
struct ehci_hcd *ehci = hcd_to_ehci (hcd);
unsigned long flags;
struct ehci_qh *qh, *tmp;
/* ASSERT: any requests/urbs are being unlinked */
/* ASSERT: nobody can be submitting urbs for this any more */
rescan:
spin_lock_irqsave (&ehci->lock, flags);
qh = ep->hcpriv;
if (!qh)
goto done;
/* endpoints can be iso streams. for now, we don't
* accelerate iso completions ... so spin a while.
*/
if (qh->hw->hw_info1 == 0) {
ehci_vdbg (ehci, "iso delay\n");
goto idle_timeout;
}
if (!HC_IS_RUNNING (hcd->state))
qh->qh_state = QH_STATE_IDLE;
switch (qh->qh_state) {
case QH_STATE_LINKED:
for (tmp = ehci->async->qh_next.qh;
tmp && tmp != qh;
tmp = tmp->qh_next.qh)
continue;
/* periodic qh self-unlinks on empty */
if (!tmp)
goto nogood;
unlink_async (ehci, qh);
/* FALL THROUGH */
case QH_STATE_UNLINK: /* wait for hw to finish? */
case QH_STATE_UNLINK_WAIT:
idle_timeout:
spin_unlock_irqrestore (&ehci->lock, flags);
schedule_timeout_uninterruptible(1);
goto rescan;
case QH_STATE_IDLE: /* fully unlinked */
if (qh->clearing_tt)
goto idle_timeout;
if (list_empty (&qh->qtd_list)) {
qh_put (qh);
break;
}
/* else FALL THROUGH */
default:
nogood:
/* caller was supposed to have unlinked any requests;
* that's not our job. just leak this memory.
*/
ehci_err (ehci, "qh %p (#%02x) state %d%s\n",
qh, ep->desc.bEndpointAddress, qh->qh_state,
list_empty (&qh->qtd_list) ? "" : "(has tds)");
break;
}
ep->hcpriv = NULL;
done:
spin_unlock_irqrestore (&ehci->lock, flags);
return;
}
static void
ehci_endpoint_reset(struct usb_hcd *hcd, struct usb_host_endpoint *ep)
{
struct ehci_hcd *ehci = hcd_to_ehci(hcd);
struct ehci_qh *qh;
int eptype = usb_endpoint_type(&ep->desc);
int epnum = usb_endpoint_num(&ep->desc);
int is_out = usb_endpoint_dir_out(&ep->desc);
unsigned long flags;
if (eptype != USB_ENDPOINT_XFER_BULK && eptype != USB_ENDPOINT_XFER_INT)
return;
spin_lock_irqsave(&ehci->lock, flags);
qh = ep->hcpriv;
/* For Bulk and Interrupt endpoints we maintain the toggle state
* in the hardware; the toggle bits in udev aren't used at all.
* When an endpoint is reset by usb_clear_halt() we must reset
* the toggle bit in the QH.
*/
if (qh) {
usb_settoggle(qh->dev, epnum, is_out, 0);
if (!list_empty(&qh->qtd_list)) {
WARN_ONCE(1, "clear_halt for a busy endpoint\n");
} else if (qh->qh_state == QH_STATE_LINKED) {
/* The toggle value in the QH can't be updated
* while the QH is active. Unlink it now;
* re-linking will call qh_refresh().
*/
if (eptype == USB_ENDPOINT_XFER_BULK) {
unlink_async(ehci, qh);
} else {
intr_deschedule(ehci, qh);
(void) qh_schedule(ehci, qh);
}
}
}
spin_unlock_irqrestore(&ehci->lock, flags);
}
static int ehci_get_frame (struct usb_hcd *hcd)
{
struct ehci_hcd *ehci = hcd_to_ehci (hcd);
return (ehci_readl(ehci, &ehci->regs->frame_index) >> 3) %
ehci->periodic_size;
}
/*-------------------------------------------------------------------------*/
MODULE_DESCRIPTION(DRIVER_DESC);
MODULE_AUTHOR (DRIVER_AUTHOR);
MODULE_LICENSE ("GPL");
#ifdef CONFIG_PCI
#include "ehci-pci.c"
#define PCI_DRIVER ehci_pci_driver
#endif
#ifdef CONFIG_USB_EHCI_FSL
#include "ehci-fsl.c"
#define PLATFORM_DRIVER ehci_fsl_driver
#endif
#ifdef CONFIG_SOC_AU1200
#include "ehci-au1xxx.c"
#define PLATFORM_DRIVER ehci_hcd_au1xxx_driver
#endif
#ifdef CONFIG_PPC_PS3
#include "ehci-ps3.c"
#define PS3_SYSTEM_BUS_DRIVER ps3_ehci_driver
#endif
#ifdef CONFIG_USB_EHCI_HCD_PPC_OF
#include "ehci-ppc-of.c"
#define OF_PLATFORM_DRIVER ehci_hcd_ppc_of_driver
#endif
#ifdef CONFIG_PLAT_ORION
#include "ehci-orion.c"
#define PLATFORM_DRIVER ehci_orion_driver
#endif
#ifdef CONFIG_ARCH_IXP4XX
#include "ehci-ixp4xx.c"
#define PLATFORM_DRIVER ixp4xx_ehci_driver
#endif
#ifdef CONFIG_USB_W90X900_EHCI
#include "ehci-w90x900.c"
#define PLATFORM_DRIVER ehci_hcd_w90x900_driver
#endif
#ifdef CONFIG_ARCH_AT91
#include "ehci-atmel.c"
#define PLATFORM_DRIVER ehci_atmel_driver
#endif
#if !defined(PCI_DRIVER) && !defined(PLATFORM_DRIVER) && \
!defined(PS3_SYSTEM_BUS_DRIVER) && !defined(OF_PLATFORM_DRIVER)
#error "missing bus glue for ehci-hcd"
#endif
static int __init ehci_hcd_init(void)
{
int retval = 0;
if (usb_disabled())
return -ENODEV;
printk(KERN_INFO "%s: " DRIVER_DESC "\n", hcd_name);
set_bit(USB_EHCI_LOADED, &usb_hcds_loaded);
if (test_bit(USB_UHCI_LOADED, &usb_hcds_loaded) ||
test_bit(USB_OHCI_LOADED, &usb_hcds_loaded))
printk(KERN_WARNING "Warning! ehci_hcd should always be loaded"
" before uhci_hcd and ohci_hcd, not after\n");
pr_debug("%s: block sizes: qh %Zd qtd %Zd itd %Zd sitd %Zd\n",
hcd_name,
sizeof(struct ehci_qh), sizeof(struct ehci_qtd),
sizeof(struct ehci_itd), sizeof(struct ehci_sitd));
#ifdef DEBUG
ehci_debug_root = debugfs_create_dir("ehci", usb_debug_root);
if (!ehci_debug_root) {
retval = -ENOENT;
goto err_debug;
}
#endif
#ifdef PLATFORM_DRIVER
retval = platform_driver_register(&PLATFORM_DRIVER);
if (retval < 0)
goto clean0;
#endif
#ifdef PCI_DRIVER
retval = pci_register_driver(&PCI_DRIVER);
if (retval < 0)
goto clean1;
#endif
#ifdef PS3_SYSTEM_BUS_DRIVER
retval = ps3_ehci_driver_register(&PS3_SYSTEM_BUS_DRIVER);
if (retval < 0)
goto clean2;
#endif
#ifdef OF_PLATFORM_DRIVER
retval = of_register_platform_driver(&OF_PLATFORM_DRIVER);
if (retval < 0)
goto clean3;
#endif
return retval;
#ifdef OF_PLATFORM_DRIVER
/* of_unregister_platform_driver(&OF_PLATFORM_DRIVER); */
clean3:
#endif
#ifdef PS3_SYSTEM_BUS_DRIVER
ps3_ehci_driver_unregister(&PS3_SYSTEM_BUS_DRIVER);
clean2:
#endif
#ifdef PCI_DRIVER
pci_unregister_driver(&PCI_DRIVER);
clean1:
#endif
#ifdef PLATFORM_DRIVER
platform_driver_unregister(&PLATFORM_DRIVER);
clean0:
#endif
#ifdef DEBUG
debugfs_remove(ehci_debug_root);
ehci_debug_root = NULL;
err_debug:
#endif
clear_bit(USB_EHCI_LOADED, &usb_hcds_loaded);
return retval;
}
module_init(ehci_hcd_init);
static void __exit ehci_hcd_cleanup(void)
{
#ifdef OF_PLATFORM_DRIVER
of_unregister_platform_driver(&OF_PLATFORM_DRIVER);
#endif
#ifdef PLATFORM_DRIVER
platform_driver_unregister(&PLATFORM_DRIVER);
#endif
#ifdef PCI_DRIVER
pci_unregister_driver(&PCI_DRIVER);
#endif
#ifdef PS3_SYSTEM_BUS_DRIVER
ps3_ehci_driver_unregister(&PS3_SYSTEM_BUS_DRIVER);
#endif
#ifdef DEBUG
debugfs_remove(ehci_debug_root);
#endif
clear_bit(USB_EHCI_LOADED, &usb_hcds_loaded);
}
module_exit(ehci_hcd_cleanup);