Merge /spare/repo/linux-2.6/

Documentation/networking/phy.txt

@@ -0,0 +1,288 @@
+
+-------
+PHY Abstraction Layer
+(Updated 2005-07-21)
+
+Purpose
+
+ Most network devices consist of set of registers which provide an interface
+ to a MAC layer, which communicates with the physical connection through a
+ PHY.  The PHY concerns itself with negotiating link parameters with the link
+ partner on the other side of the network connection (typically, an ethernet
+ cable), and provides a register interface to allow drivers to determine what
+ settings were chosen, and to configure what settings are allowed.
+
+ While these devices are distinct from the network devices, and conform to a
+ standard layout for the registers, it has been common practice to integrate
+ the PHY management code with the network driver.  This has resulted in large
+ amounts of redundant code.  Also, on embedded systems with multiple (and
+ sometimes quite different) ethernet controllers connected to the same 
+ management bus, it is difficult to ensure safe use of the bus.
+
+ Since the PHYs are devices, and the management busses through which they are
+ accessed are, in fact, busses, the PHY Abstraction Layer treats them as such.
+ In doing so, it has these goals:
+
+   1) Increase code-reuse
+   2) Increase overall code-maintainability
+   3) Speed development time for new network drivers, and for new systems
+ 
+ Basically, this layer is meant to provide an interface to PHY devices which
+ allows network driver writers to write as little code as possible, while
+ still providing a full feature set.
+
+The MDIO bus
+
+ Most network devices are connected to a PHY by means of a management bus.
+ Different devices use different busses (though some share common interfaces).
+ In order to take advantage of the PAL, each bus interface needs to be
+ registered as a distinct device.
+
+ 1) read and write functions must be implemented.  Their prototypes are:
+
+     int write(struct mii_bus *bus, int mii_id, int regnum, u16 value);
+     int read(struct mii_bus *bus, int mii_id, int regnum);
+
+   mii_id is the address on the bus for the PHY, and regnum is the register
+   number.  These functions are guaranteed not to be called from interrupt
+   time, so it is safe for them to block, waiting for an interrupt to signal
+   the operation is complete
+ 
+ 2) A reset function is necessary.  This is used to return the bus to an
+   initialized state.
+
+ 3) A probe function is needed.  This function should set up anything the bus
+   driver needs, setup the mii_bus structure, and register with the PAL using
+   mdiobus_register.  Similarly, there's a remove function to undo all of
+   that (use mdiobus_unregister).
+ 
+ 4) Like any driver, the device_driver structure must be configured, and init
+   exit functions are used to register the driver.
+
+ 5) The bus must also be declared somewhere as a device, and registered.
+
+ As an example for how one driver implemented an mdio bus driver, see
+ drivers/net/gianfar_mii.c and arch/ppc/syslib/mpc85xx_devices.c
+
+Connecting to a PHY
+
+ Sometime during startup, the network driver needs to establish a connection
+ between the PHY device, and the network device.  At this time, the PHY's bus
+ and drivers need to all have been loaded, so it is ready for the connection.
+ At this point, there are several ways to connect to the PHY:
+
+ 1) The PAL handles everything, and only calls the network driver when
+   the link state changes, so it can react.
+
+ 2) The PAL handles everything except interrupts (usually because the
+   controller has the interrupt registers).
+
+ 3) The PAL handles everything, but checks in with the driver every second,
+   allowing the network driver to react first to any changes before the PAL
+   does.
+ 
+ 4) The PAL serves only as a library of functions, with the network device
+   manually calling functions to update status, and configure the PHY
+
+
+Letting the PHY Abstraction Layer do Everything
+
+ If you choose option 1 (The hope is that every driver can, but to still be
+ useful to drivers that can't), connecting to the PHY is simple:
+
+ First, you need a function to react to changes in the link state.  This
+ function follows this protocol:
+
+   static void adjust_link(struct net_device *dev);
+ 
+ Next, you need to know the device name of the PHY connected to this device. 
+ The name will look something like, "phy0:0", where the first number is the
+ bus id, and the second is the PHY's address on that bus.
+ 
+ Now, to connect, just call this function:
+ 
+   phydev = phy_connect(dev, phy_name, &adjust_link, flags);
+
+ phydev is a pointer to the phy_device structure which represents the PHY.  If
+ phy_connect is successful, it will return the pointer.  dev, here, is the
+ pointer to your net_device.  Once done, this function will have started the
+ PHY's software state machine, and registered for the PHY's interrupt, if it
+ has one.  The phydev structure will be populated with information about the
+ current state, though the PHY will not yet be truly operational at this
+ point.
+
+ flags is a u32 which can optionally contain phy-specific flags.
+ This is useful if the system has put hardware restrictions on
+ the PHY/controller, of which the PHY needs to be aware.
+
+ Now just make sure that phydev->supported and phydev->advertising have any
+ values pruned from them which don't make sense for your controller (a 10/100
+ controller may be connected to a gigabit capable PHY, so you would need to
+ mask off SUPPORTED_1000baseT*).  See include/linux/ethtool.h for definitions
+ for these bitfields. Note that you should not SET any bits, or the PHY may
+ get put into an unsupported state.
+
+ Lastly, once the controller is ready to handle network traffic, you call
+ phy_start(phydev).  This tells the PAL that you are ready, and configures the
+ PHY to connect to the network.  If you want to handle your own interrupts,
+ just set phydev->irq to PHY_IGNORE_INTERRUPT before you call phy_start.
+ Similarly, if you don't want to use interrupts, set phydev->irq to PHY_POLL.
+
+ When you want to disconnect from the network (even if just briefly), you call
+ phy_stop(phydev).
+
+Keeping Close Tabs on the PAL
+
+ It is possible that the PAL's built-in state machine needs a little help to
+ keep your network device and the PHY properly in sync.  If so, you can
+ register a helper function when connecting to the PHY, which will be called
+ every second before the state machine reacts to any changes.  To do this, you
+ need to manually call phy_attach() and phy_prepare_link(), and then call
+ phy_start_machine() with the second argument set to point to your special
+ handler.
+
+ Currently there are no examples of how to use this functionality, and testing
+ on it has been limited because the author does not have any drivers which use
+ it (they all use option 1).  So Caveat Emptor.
+
+Doing it all yourself
+
+ There's a remote chance that the PAL's built-in state machine cannot track
+ the complex interactions between the PHY and your network device.  If this is
+ so, you can simply call phy_attach(), and not call phy_start_machine or
+ phy_prepare_link().  This will mean that phydev->state is entirely yours to
+ handle (phy_start and phy_stop toggle between some of the states, so you
+ might need to avoid them).
+
+ An effort has been made to make sure that useful functionality can be
+ accessed without the state-machine running, and most of these functions are
+ descended from functions which did not interact with a complex state-machine.
+ However, again, no effort has been made so far to test running without the
+ state machine, so tryer beware.
+
+ Here is a brief rundown of the functions:
+
+ int phy_read(struct phy_device *phydev, u16 regnum);
+ int phy_write(struct phy_device *phydev, u16 regnum, u16 val);
+
+   Simple read/write primitives.  They invoke the bus's read/write function
+   pointers.
+
+ void phy_print_status(struct phy_device *phydev);
+ 
+   A convenience function to print out the PHY status neatly.
+
+ int phy_clear_interrupt(struct phy_device *phydev);
+ int phy_config_interrupt(struct phy_device *phydev, u32 interrupts);
+   
+   Clear the PHY's interrupt, and configure which ones are allowed,
+   respectively.  Currently only supports all on, or all off.
+ 
+ int phy_enable_interrupts(struct phy_device *phydev);
+ int phy_disable_interrupts(struct phy_device *phydev);
+
+   Functions which enable/disable PHY interrupts, clearing them
+   before and after, respectively.
+
+ int phy_start_interrupts(struct phy_device *phydev);
+ int phy_stop_interrupts(struct phy_device *phydev);
+
+   Requests the IRQ for the PHY interrupts, then enables them for
+   start, or disables then frees them for stop.
+
+ struct phy_device * phy_attach(struct net_device *dev, const char *phy_id,
+		 u32 flags);
+
+   Attaches a network device to a particular PHY, binding the PHY to a generic
+   driver if none was found during bus initialization.  Passes in
+   any phy-specific flags as needed.
+
+ int phy_start_aneg(struct phy_device *phydev);
+   
+   Using variables inside the phydev structure, either configures advertising
+   and resets autonegotiation, or disables autonegotiation, and configures
+   forced settings.
+
+ static inline int phy_read_status(struct phy_device *phydev);
+
+   Fills the phydev structure with up-to-date information about the current
+   settings in the PHY.
+
+ void phy_sanitize_settings(struct phy_device *phydev)
+   
+   Resolves differences between currently desired settings, and
+   supported settings for the given PHY device.  Does not make
+   the changes in the hardware, though.
+
+ int phy_ethtool_sset(struct phy_device *phydev, struct ethtool_cmd *cmd);
+ int phy_ethtool_gset(struct phy_device *phydev, struct ethtool_cmd *cmd);
+
+   Ethtool convenience functions.
+
+ int phy_mii_ioctl(struct phy_device *phydev,
+                 struct mii_ioctl_data *mii_data, int cmd);
+
+   The MII ioctl.  Note that this function will completely screw up the state
+   machine if you write registers like BMCR, BMSR, ADVERTISE, etc.  Best to
+   use this only to write registers which are not standard, and don't set off
+   a renegotiation.
+
+
+PHY Device Drivers
+
+ With the PHY Abstraction Layer, adding support for new PHYs is
+ quite easy.  In some cases, no work is required at all!  However,
+ many PHYs require a little hand-holding to get up-and-running.
+
+Generic PHY driver
+
+ If the desired PHY doesn't have any errata, quirks, or special
+ features you want to support, then it may be best to not add
+ support, and let the PHY Abstraction Layer's Generic PHY Driver
+ do all of the work.  
+
+Writing a PHY driver
+
+ If you do need to write a PHY driver, the first thing to do is
+ make sure it can be matched with an appropriate PHY device.
+ This is done during bus initialization by reading the device's
+ UID (stored in registers 2 and 3), then comparing it to each
+ driver's phy_id field by ANDing it with each driver's
+ phy_id_mask field.  Also, it needs a name.  Here's an example:
+
+   static struct phy_driver dm9161_driver = {
+         .phy_id         = 0x0181b880,
+	 .name           = "Davicom DM9161E",
+	 .phy_id_mask    = 0x0ffffff0,
+	 ...
+   }
+
+ Next, you need to specify what features (speed, duplex, autoneg,
+ etc) your PHY device and driver support.  Most PHYs support
+ PHY_BASIC_FEATURES, but you can look in include/mii.h for other
+ features.
+
+ Each driver consists of a number of function pointers:
+
+   config_init: configures PHY into a sane state after a reset.
+     For instance, a Davicom PHY requires descrambling disabled.
+   probe: Does any setup needed by the driver
+   suspend/resume: power management
+   config_aneg: Changes the speed/duplex/negotiation settings
+   read_status: Reads the current speed/duplex/negotiation settings
+   ack_interrupt: Clear a pending interrupt
+   config_intr: Enable or disable interrupts
+   remove: Does any driver take-down
+
+ Of these, only config_aneg and read_status are required to be
+ assigned by the driver code.  The rest are optional.  Also, it is
+ preferred to use the generic phy driver's versions of these two
+ functions if at all possible: genphy_read_status and
+ genphy_config_aneg.  If this is not possible, it is likely that
+ you only need to perform some actions before and after invoking
+ these functions, and so your functions will wrap the generic
+ ones.
+
+ Feel free to look at the Marvell, Cicada, and Davicom drivers in
+ drivers/net/phy/ for examples (the lxt and qsemi drivers have
+ not been tested as of this writing)

Makefile

@@ -2,7 +2,7 @@ VERSION = 2
 PATCHLEVEL = 6
 SUBLEVEL = 13
 EXTRAVERSION =
-NAME=Woozy Numbat
+NAME=Affluent Albatross
 
 # *DOCUMENTATION*
 # To see a list of typical targets execute "make help"

arch/alpha/kernel/signal.c

@@ -566,13 +566,12 @@ handle_signal(int sig, struct k_sigaction *ka, siginfo_t *info,
 	if (ka->sa.sa_flags & SA_RESETHAND)
 		ka->sa.sa_handler = SIG_DFL;
 
-	if (!(ka->sa.sa_flags & SA_NODEFER)) {
-		spin_lock_irq(&current->sighand->siglock);
-		sigorsets(&current->blocked,&current->blocked,&ka->sa.sa_mask);
+	spin_lock_irq(&current->sighand->siglock);
+	sigorsets(&current->blocked,&current->blocked,&ka->sa.sa_mask);
+	if (!(ka->sa.sa_flags & SA_NODEFER)) 
 		sigaddset(&current->blocked,sig);
-		recalc_sigpending();
-		spin_unlock_irq(&current->sighand->siglock);
-	}
+	recalc_sigpending();
+	spin_unlock_irq(&current->sighand->siglock);
 }
 
 static inline void

arch/arm/Kconfig

@@ -635,10 +635,6 @@ config PM
 	  and the Battery Powered Linux mini-HOWTO, available from
 	  <http://www.tldp.org/docs.html#howto>.
 
-	  Note that, even if you say N here, Linux on the x86 architecture
-	  will issue the hlt instruction if nothing is to be done, thereby
-	  sending the processor to sleep and saving power.
-
 config APM
 	tristate "Advanced Power Management Emulation"
 	depends on PM
@@ -650,12 +646,6 @@ config APM
 	  battery status information, and user-space programs will receive
 	  notification of APM "events" (e.g. battery status change).
 
-	  If you select "Y" here, you can disable actual use of the APM
-	  BIOS by passing the "apm=off" option to the kernel at boot time.
-
-	  Note that the APM support is almost completely disabled for
-	  machines with more than one CPU.
-
 	  In order to use APM, you will need supporting software. For location
 	  and more information, read <file:Documentation/pm.txt> and the
 	  Battery Powered Linux mini-HOWTO, available from
@@ -665,39 +655,12 @@ config APM
 	  manpage ("man 8 hdparm") for that), and it doesn't turn off
 	  VESA-compliant "green" monitors.
 
-	  This driver does not support the TI 4000M TravelMate and the ACER
-	  486/DX4/75 because they don't have compliant BIOSes. Many "green"
-	  desktop machines also don't have compliant BIOSes, and this driver
-	  may cause those machines to panic during the boot phase.
-
 	  Generally, if you don't have a battery in your machine, there isn't
 	  much point in using this driver and you should say N. If you get
 	  random kernel OOPSes or reboots that don't seem to be related to
 	  anything, try disabling/enabling this option (or disabling/enabling
 	  APM in your BIOS).
 
-	  Some other things you should try when experiencing seemingly random,
-	  "weird" problems:
-
-	  1) make sure that you have enough swap space and that it is
-	  enabled.
-	  2) pass the "no-hlt" option to the kernel
-	  3) switch on floating point emulation in the kernel and pass
-	  the "no387" option to the kernel
-	  4) pass the "floppy=nodma" option to the kernel
-	  5) pass the "mem=4M" option to the kernel (thereby disabling
-	  all but the first 4 MB of RAM)
-	  6) make sure that the CPU is not over clocked.
-	  7) read the sig11 FAQ at <http://www.bitwizard.nl/sig11/>
-	  8) disable the cache from your BIOS settings
-	  9) install a fan for the video card or exchange video RAM
-	  10) install a better fan for the CPU
-	  11) exchange RAM chips
-	  12) exchange the motherboard.
-
-	  To compile this driver as a module, choose M here: the
-	  module will be called apm.
-
 endmenu
 
 source "net/Kconfig"
@@ -752,6 +715,8 @@ source "drivers/hwmon/Kconfig"
 
 source "drivers/misc/Kconfig"
 
+source "drivers/mfd/Kconfig"
+
 source "drivers/media/Kconfig"
 
 source "drivers/video/Kconfig"

arch/arm/common/Kconfig

@@ -1,6 +1,9 @@
 config ICST525
 	bool
 
+config ARM_GIC
+	bool
+
 config ICST307
 	bool
 

arch/arm/common/Makefile

@@ -4,6 +4,7 @@
 
 obj-y				+= rtctime.o
 obj-$(CONFIG_ARM_AMBA)		+= amba.o
+obj-$(CONFIG_ARM_GIC)		+= gic.o
 obj-$(CONFIG_ICST525)		+= icst525.o
 obj-$(CONFIG_ICST307)		+= icst307.o
 obj-$(CONFIG_SA1111)		+= sa1111.o

arch/arm/common/gic.c

@@ -0,0 +1,166 @@
+/*
+ *  linux/arch/arm/common/gic.c
+ *
+ *  Copyright (C) 2002 ARM Limited, All Rights Reserved.
+ *
+ * This program is free software; you can redistribute it and/or modify
+ * it under the terms of the GNU General Public License version 2 as
+ * published by the Free Software Foundation.
+ *
+ * Interrupt architecture for the GIC:
+ *
+ * o There is one Interrupt Distributor, which receives interrupts
+ *   from system devices and sends them to the Interrupt Controllers.
+ *
+ * o There is one CPU Interface per CPU, which sends interrupts sent
+ *   by the Distributor, and interrupts generated locally, to the
+ *   associated CPU.
+ *
+ * Note that IRQs 0-31 are special - they are local to each CPU.
+ * As such, the enable set/clear, pending set/clear and active bit
+ * registers are banked per-cpu for these sources.
+ */
+#include <linux/init.h>
+#include <linux/kernel.h>
+#include <linux/list.h>
+#include <linux/smp.h>
+
+#include <asm/irq.h>
+#include <asm/io.h>
+#include <asm/mach/irq.h>
+#include <asm/hardware/gic.h>
+
+static void __iomem *gic_dist_base;
+static void __iomem *gic_cpu_base;
+
+/*
+ * Routines to acknowledge, disable and enable interrupts
+ *
+ * Linux assumes that when we're done with an interrupt we need to
+ * unmask it, in the same way we need to unmask an interrupt when
+ * we first enable it.
+ *
+ * The GIC has a seperate notion of "end of interrupt" to re-enable
+ * an interrupt after handling, in order to support hardware
+ * prioritisation.
+ *
+ * We can make the GIC behave in the way that Linux expects by making
+ * our "acknowledge" routine disable the interrupt, then mark it as
+ * complete.
+ */
+static void gic_ack_irq(unsigned int irq)
+{
+	u32 mask = 1 << (irq % 32);
+	writel(mask, gic_dist_base + GIC_DIST_ENABLE_CLEAR + (irq / 32) * 4);
+	writel(irq, gic_cpu_base + GIC_CPU_EOI);
+}
+
+static void gic_mask_irq(unsigned int irq)
+{
+	u32 mask = 1 << (irq % 32);
+	writel(mask, gic_dist_base + GIC_DIST_ENABLE_CLEAR + (irq / 32) * 4);
+}
+
+static void gic_unmask_irq(unsigned int irq)
+{
+	u32 mask = 1 << (irq % 32);
+	writel(mask, gic_dist_base + GIC_DIST_ENABLE_SET + (irq / 32) * 4);
+}
+
+static void gic_set_cpu(struct irqdesc *desc, unsigned int irq, unsigned int cpu)
+{
+	void __iomem *reg = gic_dist_base + GIC_DIST_TARGET + (irq & ~3);
+	unsigned int shift = (irq % 4) * 8;
+	u32 val;
+
+	val = readl(reg) & ~(0xff << shift);
+	val |= 1 << (cpu + shift);
+	writel(val, reg);
+}
+
+static struct irqchip gic_chip = {
+	.ack		= gic_ack_irq,
+	.mask		= gic_mask_irq,
+	.unmask		= gic_unmask_irq,
+#ifdef CONFIG_SMP
+	.set_cpu	= gic_set_cpu,
+#endif
+};
+
+void __init gic_dist_init(void __iomem *base)
+{
+	unsigned int max_irq, i;
+	u32 cpumask = 1 << smp_processor_id();
+
+	cpumask |= cpumask << 8;
+	cpumask |= cpumask << 16;
+
+	gic_dist_base = base;
+
+	writel(0, base + GIC_DIST_CTRL);
+
+	/*
+	 * Find out how many interrupts are supported.
+	 */
+	max_irq = readl(base + GIC_DIST_CTR) & 0x1f;
+	max_irq = (max_irq + 1) * 32;
+
+	/*
+	 * The GIC only supports up to 1020 interrupt sources.
+	 * Limit this to either the architected maximum, or the
+	 * platform maximum.
+	 */
+	if (max_irq > max(1020, NR_IRQS))
+		max_irq = max(1020, NR_IRQS);
+
+	/*
+	 * Set all global interrupts to be level triggered, active low.
+	 */
+	for (i = 32; i < max_irq; i += 16)
+		writel(0, base + GIC_DIST_CONFIG + i * 4 / 16);
+
+	/*
+	 * Set all global interrupts to this CPU only.
+	 */
+	for (i = 32; i < max_irq; i += 4)
+		writel(cpumask, base + GIC_DIST_TARGET + i * 4 / 4);
+
+	/*
+	 * Set priority on all interrupts.
+	 */
+	for (i = 0; i < max_irq; i += 4)
+		writel(0xa0a0a0a0, base + GIC_DIST_PRI + i * 4 / 4);
+
+	/*
+	 * Disable all interrupts.
+	 */
+	for (i = 0; i < max_irq; i += 32)
+		writel(0xffffffff, base + GIC_DIST_ENABLE_CLEAR + i * 4 / 32);
+
+	/*
+	 * Setup the Linux IRQ subsystem.
+	 */
+	for (i = 29; i < max_irq; i++) {
+		set_irq_chip(i, &gic_chip);
+		set_irq_handler(i, do_level_IRQ);
+		set_irq_flags(i, IRQF_VALID | IRQF_PROBE);
+	}
+
+	writel(1, base + GIC_DIST_CTRL);
+}
+
+void __cpuinit gic_cpu_init(void __iomem *base)
+{
+	gic_cpu_base = base;
+	writel(0xf0, base + GIC_CPU_PRIMASK);
+	writel(1, base + GIC_CPU_CTRL);
+}
+
+#ifdef CONFIG_SMP
+void gic_raise_softirq(cpumask_t cpumask, unsigned int irq)
+{
+	unsigned long map = *cpus_addr(cpumask);
+
+	writel(map << 16 | irq, gic_dist_base + GIC_DIST_SOFTINT);
+}
+#endif

arch/arm/kernel/signal.c

@@ -658,11 +658,12 @@ handle_signal(unsigned long sig, struct k_sigaction *ka,
 	/*
 	 * Block the signal if we were unsuccessful.
 	 */
-	if (ret != 0 || !(ka->sa.sa_flags & SA_NODEFER)) {
+	if (ret != 0) {
 		spin_lock_irq(&tsk->sighand->siglock);
 		sigorsets(&tsk->blocked, &tsk->blocked,
 			  &ka->sa.sa_mask);
-		sigaddset(&tsk->blocked, sig);
+		if (!(ka->sa.sa_flags & SA_NODEFER))
+			sigaddset(&tsk->blocked, sig);
 		recalc_sigpending();
 		spin_unlock_irq(&tsk->sighand->siglock);
 	}

arch/arm/mach-sa1100/assabet.c

@@ -35,6 +35,7 @@
 #include <asm/mach/map.h>
 #include <asm/mach/serial_sa1100.h>
 #include <asm/arch/assabet.h>
+#include <asm/arch/mcp.h>
 
 #include "generic.h"
 
@@ -198,6 +199,11 @@ static struct irda_platform_data assabet_irda_data = {
 	.set_speed	= assabet_irda_set_speed,
 };
 
+static struct mcp_plat_data assabet_mcp_data = {
+	.mccr0		= MCCR0_ADM,
+	.sclk_rate	= 11981000,
+};
+
 static void __init assabet_init(void)
 {
 	/*
@@ -246,6 +252,7 @@ static void __init assabet_init(void)
 	sa11x0_set_flash_data(&assabet_flash_data, assabet_flash_resources,
 			      ARRAY_SIZE(assabet_flash_resources));
 	sa11x0_set_irda_data(&assabet_irda_data);
+	sa11x0_set_mcp_data(&assabet_mcp_data);
 }
 
 /*

arch/arm/mach-sa1100/cerf.c

@@ -29,6 +29,7 @@
 #include <asm/mach/serial_sa1100.h>
 
 #include <asm/arch/cerf.h>
+#include <asm/arch/mcp.h>
 #include "generic.h"
 
 static struct resource cerfuart2_resources[] = {
@@ -116,10 +117,16 @@ static void __init cerf_map_io(void)
 	GPDR |= CERF_GPIO_CF_RESET;
 }
 
+static struct mcp_plat_data cerf_mcp_data = {
+	.mccr0		= MCCR0_ADM,
+	.sclk_rate	= 11981000,
+};
+
 static void __init cerf_init(void)
 {
 	platform_add_devices(cerf_devices, ARRAY_SIZE(cerf_devices));
 	sa11x0_set_flash_data(&cerf_flash_data, &cerf_flash_resource, 1);
+	sa11x0_set_mcp_data(&cerf_mcp_data);
 }
 
 MACHINE_START(CERF, "Intrinsyc CerfBoard/CerfCube")

arch/arm/mach-sa1100/generic.c

@@ -221,6 +221,11 @@ static struct platform_device sa11x0mcp_device = {
 	.resource	= sa11x0mcp_resources,
 };
 
+void sa11x0_set_mcp_data(struct mcp_plat_data *data)
+{
+	sa11x0mcp_device.dev.platform_data = data;
+}
+
 static struct resource sa11x0ssp_resources[] = {
 	[0] = {
 		.start	= 0x80070000,

arch/arm/mach-sa1100/generic.h

@@ -34,5 +34,8 @@ struct resource;
 extern void sa11x0_set_flash_data(struct flash_platform_data *flash,
 				  struct resource *res, int nr);
 
+struct sa11x0_ssp_plat_ops;
+extern void sa11x0_set_ssp_data(struct sa11x0_ssp_plat_ops *ops);
+
 struct irda_platform_data;
 void sa11x0_set_irda_data(struct irda_platform_data *irda);

arch/arm/mach-sa1100/lart.c

@@ -13,12 +13,23 @@
 #include <asm/mach/arch.h>
 #include <asm/mach/map.h>
 #include <asm/mach/serial_sa1100.h>
+#include <asm/arch/mcp.h>
 
 #include "generic.h"
 
 
 #warning "include/asm/arch-sa1100/ide.h needs fixing for lart"
 
+static struct mcp_plat_data lart_mcp_data = {
+	.mccr0		= MCCR0_ADM,
+	.sclk_rate	= 11981000,
+};
+
+static void __init lart_init(void)
+{
+	sa11x0_set_mcp_data(&lart_mcp_data);
+}
+
 static struct map_desc lart_io_desc[] __initdata = {
  /* virtual     physical    length      type */
   { 0xe8000000, 0x00000000, 0x00400000, MT_DEVICE }, /* main flash memory */
@@ -47,5 +58,6 @@ MACHINE_START(LART, "LART")
 	.boot_params	= 0xc0000100,
 	.map_io		= lart_map_io,
 	.init_irq	= sa1100_init_irq,
+	.init_machine	= lart_init,
 	.timer		= &sa1100_timer,
 MACHINE_END

arch/arm/mach-sa1100/shannon.c

@@ -18,6 +18,7 @@
 #include <asm/mach/flash.h>
 #include <asm/mach/map.h>
 #include <asm/mach/serial_sa1100.h>
+#include <asm/arch/mcp.h>
 #include <asm/arch/shannon.h>
 
 #include "generic.h"
@@ -52,9 +53,15 @@ static struct resource shannon_flash_resource = {
 	.flags		= IORESOURCE_MEM,
 };
 
+static struct mcp_plat_data shannon_mcp_data = {
+	.mccr0		= MCCR0_ADM,
+	.sclk_rate	= 11981000,
+};
+
 static void __init shannon_init(void)
 {
 	sa11x0_set_flash_data(&shannon_flash_data, &shannon_flash_resource, 1);
+	sa11x0_set_mcp_data(&shannon_mcp_data);
 }
 
 static void __init shannon_map_io(void)

arch/arm/mach-sa1100/simpad.c

@@ -23,6 +23,7 @@
 #include <asm/mach/flash.h>
 #include <asm/mach/map.h>
 #include <asm/mach/serial_sa1100.h>
+#include <asm/arch/mcp.h>
 #include <asm/arch/simpad.h>
 
 #include <linux/serial_core.h>
@@ -123,6 +124,11 @@ static struct resource simpad_flash_resources [] = {
 	}
 };
 
+static struct mcp_plat_data simpad_mcp_data = {
+	.mccr0		= MCCR0_ADM,
+	.sclk_rate	= 11981000,
+};
+
 
 
 static void __init simpad_map_io(void)
@@ -157,6 +163,7 @@ static void __init simpad_map_io(void)
 
 	sa11x0_set_flash_data(&simpad_flash_data, simpad_flash_resources,
 			      ARRAY_SIZE(simpad_flash_resources));
+	sa11x0_set_mcp_data(&simpad_mcp_data);
 }
 
 static void simpad_power_off(void)

arch/arm26/kernel/signal.c

@@ -454,14 +454,13 @@ handle_signal(unsigned long sig, siginfo_t *info, sigset_t *oldset,
 		if (ka->sa.sa_flags & SA_ONESHOT)
 			ka->sa.sa_handler = SIG_DFL;
 
-		if (!(ka->sa.sa_flags & SA_NODEFER)) {
-			spin_lock_irq(&tsk->sighand->siglock);
-			sigorsets(&tsk->blocked, &tsk->blocked,
-				  &ka->sa.sa_mask);
+		spin_lock_irq(&tsk->sighand->siglock);
+		sigorsets(&tsk->blocked, &tsk->blocked,
+			  &ka->sa.sa_mask);
+		if (!(ka->sa.sa_flags & SA_NODEFER))
 			sigaddset(&tsk->blocked, sig);
-			recalc_sigpending();
-			spin_unlock_irq(&tsk->sighand->siglock);
-		}
+		recalc_sigpending();
+		spin_unlock_irq(&tsk->sighand->siglock);
 		return;
 	}
 

arch/cris/arch-v10/kernel/signal.c

@@ -517,13 +517,12 @@ handle_signal(int canrestart, unsigned long sig,
 	if (ka->sa.sa_flags & SA_ONESHOT)
 		ka->sa.sa_handler = SIG_DFL;
 
-	if (!(ka->sa.sa_flags & SA_NODEFER)) {
-		spin_lock_irq(&current->sighand->siglock);
-		sigorsets(&current->blocked,&current->blocked,&ka->sa.sa_mask);
+	spin_lock_irq(&current->sighand->siglock);
+	sigorsets(&current->blocked,&current->blocked,&ka->sa.sa_mask);
+	if (!(ka->sa.sa_flags & SA_NODEFER))
 		sigaddset(&current->blocked,sig);
-		recalc_sigpending();
-		spin_unlock_irq(&current->sighand->siglock);
-	}
+	recalc_sigpending();
+	spin_unlock_irq(&current->sighand->siglock);
 }
 
 /*

arch/cris/arch-v32/kernel/signal.c

@@ -568,13 +568,12 @@ handle_signal(int canrestart, unsigned long sig,
 	if (ka->sa.sa_flags & SA_ONESHOT)
 		ka->sa.sa_handler = SIG_DFL;
 
-	if (!(ka->sa.sa_flags & SA_NODEFER)) {
-		spin_lock_irq(&current->sighand->siglock);
-		sigorsets(&current->blocked,&current->blocked,&ka->sa.sa_mask);
+	spin_lock_irq(&current->sighand->siglock);
+	sigorsets(&current->blocked,&current->blocked,&ka->sa.sa_mask);
+	if (!(ka->sa.sa_flags & SA_NODEFER))
 		sigaddset(&current->blocked,sig);
-		recalc_sigpending();
-		spin_unlock_irq(&current->sighand->siglock);
-	}
+	recalc_sigpending();
+	spin_unlock_irq(&current->sighand->siglock);
 }
 
 /*

arch/frv/kernel/signal.c

@@ -506,13 +506,12 @@ static void handle_signal(unsigned long sig, siginfo_t *info,
 	else
 		setup_frame(sig, ka, oldset, regs);
 
-	if (!(ka->sa.sa_flags & SA_NODEFER)) {
-		spin_lock_irq(&current->sighand->siglock);
-		sigorsets(&current->blocked, &current->blocked, &ka->sa.sa_mask);
+	spin_lock_irq(&current->sighand->siglock);
+	sigorsets(&current->blocked, &current->blocked, &ka->sa.sa_mask);
+	if (!(ka->sa.sa_flags & SA_NODEFER))
 		sigaddset(&current->blocked, sig);
-		recalc_sigpending();
-		spin_unlock_irq(&current->sighand->siglock);
-	}
+	recalc_sigpending();
+	spin_unlock_irq(&current->sighand->siglock);
 } /* end handle_signal() */
 
 /*****************************************************************************/

arch/h8300/kernel/signal.c

@@ -488,13 +488,12 @@ handle_signal(unsigned long sig, siginfo_t *info, struct k_sigaction *ka,
 	else
 		setup_frame(sig, ka, oldset, regs);
 
-	if (!(ka->sa.sa_flags & SA_NODEFER)) {
-		spin_lock_irq(&current->sighand->siglock);
-		sigorsets(&current->blocked,&current->blocked,&ka->sa.sa_mask);
+	spin_lock_irq(&current->sighand->siglock);
+	sigorsets(&current->blocked,&current->blocked,&ka->sa.sa_mask);
+	if (!(ka->sa.sa_flags & SA_NODEFER))
 		sigaddset(&current->blocked,sig);
-		recalc_sigpending();
-		spin_unlock_irq(&current->sighand->siglock);
-	}
+	recalc_sigpending();
+	spin_unlock_irq(&current->sighand->siglock);
 }
 
 /*

arch/i386/kernel/signal.c

@@ -577,10 +577,11 @@ handle_signal(unsigned long sig, siginfo_t *info, struct k_sigaction *ka,
 	else
 		ret = setup_frame(sig, ka, oldset, regs);
 
-	if (ret && !(ka->sa.sa_flags & SA_NODEFER)) {
+	if (ret) {
 		spin_lock_irq(&current->sighand->siglock);
 		sigorsets(&current->blocked,&current->blocked,&ka->sa.sa_mask);
-		sigaddset(&current->blocked,sig);
+		if (!(ka->sa.sa_flags & SA_NODEFER))
+			sigaddset(&current->blocked,sig);
 		recalc_sigpending();
 		spin_unlock_irq(&current->sighand->siglock);
 	}

arch/ia64/kernel/signal.c

@@ -467,15 +467,12 @@ handle_signal (unsigned long sig, struct k_sigaction *ka, siginfo_t *info, sigse
 		if (!setup_frame(sig, ka, info, oldset, scr))
 			return 0;
 
-	if (!(ka->sa.sa_flags & SA_NODEFER)) {
-		spin_lock_irq(&current->sighand->siglock);
-		{
-			sigorsets(&current->blocked, &current->blocked, &ka->sa.sa_mask);
-			sigaddset(&current->blocked, sig);
-			recalc_sigpending();
-		}
-		spin_unlock_irq(&current->sighand->siglock);
-	}
+	spin_lock_irq(&current->sighand->siglock);
+	sigorsets(&current->blocked, &current->blocked, &ka->sa.sa_mask);
+	if (!(ka->sa.sa_flags & SA_NODEFER))
+		sigaddset(&current->blocked, sig);
+	recalc_sigpending();
+	spin_unlock_irq(&current->sighand->siglock);
 	return 1;
 }
 

arch/m32r/kernel/signal.c

@@ -341,13 +341,12 @@ handle_signal(unsigned long sig, struct k_sigaction *ka, siginfo_t *info,
 	/* Set up the stack frame */
 	setup_rt_frame(sig, ka, info, oldset, regs);
 
-	if (!(ka->sa.sa_flags & SA_NODEFER)) {
-		spin_lock_irq(&current->sighand->siglock);
-		sigorsets(&current->blocked,&current->blocked,&ka->sa.sa_mask);
+	spin_lock_irq(&current->sighand->siglock);
+	sigorsets(&current->blocked,&current->blocked,&ka->sa.sa_mask);
+	if (!(ka->sa.sa_flags & SA_NODEFER))
 		sigaddset(&current->blocked,sig);
-		recalc_sigpending();
-		spin_unlock_irq(&current->sighand->siglock);
-	}
+	recalc_sigpending();
+	spin_unlock_irq(&current->sighand->siglock);
 }
 
 /*

arch/m68knommu/kernel/signal.c

@@ -732,13 +732,12 @@ handle_signal(int sig, struct k_sigaction *ka, siginfo_t *info,
 	if (ka->sa.sa_flags & SA_ONESHOT)
 		ka->sa.sa_handler = SIG_DFL;
 
-	if (!(ka->sa.sa_flags & SA_NODEFER)) {
-		spin_lock_irq(&current->sighand->siglock);
-		sigorsets(&current->blocked,&current->blocked,&ka->sa.sa_mask);
+	spin_lock_irq(&current->sighand->siglock);
+	sigorsets(&current->blocked,&current->blocked,&ka->sa.sa_mask);
+	if (!(ka->sa.sa_flags & SA_NODEFER))
 		sigaddset(&current->blocked,sig);
-		recalc_sigpending();
-		spin_unlock_irq(&current->sighand->siglock);
-	}
+	recalc_sigpending();
+	spin_unlock_irq(&current->sighand->siglock);
 }
 
 /*

arch/mips/kernel/irixsig.c

@@ -155,13 +155,12 @@ static inline void handle_signal(unsigned long sig, siginfo_t *info,
 	else
 		setup_irix_frame(ka, regs, sig, oldset);
 
-	if (!(ka->sa.sa_flags & SA_NODEFER)) {
-		spin_lock_irq(&current->sighand->siglock);
-		sigorsets(&current->blocked,&current->blocked,&ka->sa.sa_mask);
+	spin_lock_irq(&current->sighand->siglock);
+	sigorsets(&current->blocked,&current->blocked,&ka->sa.sa_mask);
+	if (!(ka->sa.sa_flags & SA_NODEFER))
 		sigaddset(&current->blocked,sig);
-		recalc_sigpending();
-		spin_unlock_irq(&current->sighand->siglock);
-	}
+	recalc_sigpending();
+	spin_unlock_irq(&current->sighand->siglock);
 }
 
 asmlinkage int do_irix_signal(sigset_t *oldset, struct pt_regs *regs)

arch/mips/kernel/signal.c

@@ -425,13 +425,12 @@ static inline void handle_signal(unsigned long sig, siginfo_t *info,
 		setup_frame(ka, regs, sig, oldset);
 #endif
 
-	if (!(ka->sa.sa_flags & SA_NODEFER)) {
-		spin_lock_irq(&current->sighand->siglock);
-		sigorsets(&current->blocked,&current->blocked,&ka->sa.sa_mask);
+	spin_lock_irq(&current->sighand->siglock);
+	sigorsets(&current->blocked,&current->blocked,&ka->sa.sa_mask);
+	if (!(ka->sa.sa_flags & SA_NODEFER))
 		sigaddset(&current->blocked,sig);
-		recalc_sigpending();
-		spin_unlock_irq(&current->sighand->siglock);
-	}
+	recalc_sigpending();
+	spin_unlock_irq(&current->sighand->siglock);
 }
 
 extern int do_signal32(sigset_t *oldset, struct pt_regs *regs);

arch/mips/kernel/signal32.c

@@ -751,13 +751,12 @@ static inline void handle_signal(unsigned long sig, siginfo_t *info,
 	else
 		setup_frame(ka, regs, sig, oldset);
 
-	if (!(ka->sa.sa_flags & SA_NODEFER)) {
-		spin_lock_irq(&current->sighand->siglock);
-		sigorsets(&current->blocked,&current->blocked,&ka->sa.sa_mask);
+	spin_lock_irq(&current->sighand->siglock);
+	sigorsets(&current->blocked,&current->blocked,&ka->sa.sa_mask);
+	if (!(ka->sa.sa_flags & SA_NODEFER))
 		sigaddset(&current->blocked,sig);
-		recalc_sigpending();
-		spin_unlock_irq(&current->sighand->siglock);
-	}
+	recalc_sigpending();
+	spin_unlock_irq(&current->sighand->siglock);
 }
 
 int do_signal32(sigset_t *oldset, struct pt_regs *regs)

arch/parisc/kernel/signal.c

@@ -517,13 +517,12 @@ handle_signal(unsigned long sig, siginfo_t *info, struct k_sigaction *ka,
 	if (!setup_rt_frame(sig, ka, info, oldset, regs, in_syscall))
 		return 0;
 
-	if (!(ka->sa.sa_flags & SA_NODEFER)) {
-		spin_lock_irq(&current->sighand->siglock);
-		sigorsets(&current->blocked,&current->blocked,&ka->sa.sa_mask);
+	spin_lock_irq(&current->sighand->siglock);
+	sigorsets(&current->blocked,&current->blocked,&ka->sa.sa_mask);
+	if (!(ka->sa.sa_flags & SA_NODEFER))
 		sigaddset(&current->blocked,sig);
-		recalc_sigpending();
-		spin_unlock_irq(&current->sighand->siglock);
-	}
+	recalc_sigpending();
+	spin_unlock_irq(&current->sighand->siglock);
 	return 1;
 }
 

arch/ppc/kernel/signal.c

@@ -759,13 +759,12 @@ int do_signal(sigset_t *oldset, struct pt_regs *regs)
 	else
 		handle_signal(signr, &ka, &info, oldset, regs, newsp);
 
-	if (!(ka.sa.sa_flags & SA_NODEFER)) {
-		spin_lock_irq(&current->sighand->siglock);
-		sigorsets(&current->blocked,&current->blocked,&ka.sa.sa_mask);
+	spin_lock_irq(&current->sighand->siglock);
+	sigorsets(&current->blocked,&current->blocked,&ka.sa.sa_mask);
+	if (!(ka.sa.sa_flags & SA_NODEFER))
 		sigaddset(&current->blocked, signr);
-		recalc_sigpending();
-		spin_unlock_irq(&current->sighand->siglock);
-	}
+	recalc_sigpending();
+	spin_unlock_irq(&current->sighand->siglock);
 
 	return 1;
 }

arch/ppc64/kernel/signal.c

@@ -481,10 +481,11 @@ static int handle_signal(unsigned long sig, struct k_sigaction *ka,
 	/* Set up Signal Frame */
 	ret = setup_rt_frame(sig, ka, info, oldset, regs);
 
-	if (ret && !(ka->sa.sa_flags & SA_NODEFER)) {
+	if (ret) {
 		spin_lock_irq(&current->sighand->siglock);
 		sigorsets(&current->blocked, &current->blocked, &ka->sa.sa_mask);
-		sigaddset(&current->blocked,sig);
+		if (!(ka->sa.sa_flags & SA_NODEFER))
+			sigaddset(&current->blocked,sig);
 		recalc_sigpending();
 		spin_unlock_irq(&current->sighand->siglock);
 	}

arch/ppc64/kernel/signal32.c

@@ -976,11 +976,12 @@ int do_signal32(sigset_t *oldset, struct pt_regs *regs)
 	else
 		ret = handle_signal32(signr, &ka, &info, oldset, regs, newsp);
 
-	if (ret && !(ka.sa.sa_flags & SA_NODEFER)) {
+	if (ret) {
 		spin_lock_irq(&current->sighand->siglock);
 		sigorsets(&current->blocked, &current->blocked,
 			  &ka.sa.sa_mask);
-		sigaddset(&current->blocked, signr);
+		if (!(ka.sa.sa_flags & SA_NODEFER))
+			sigaddset(&current->blocked, signr);
 		recalc_sigpending();
 		spin_unlock_irq(&current->sighand->siglock);
 	}

arch/s390/kernel/compat_signal.c

@@ -637,12 +637,11 @@ handle_signal32(unsigned long sig, struct k_sigaction *ka,
 	else
 		setup_frame32(sig, ka, oldset, regs);
 
-	if (!(ka->sa.sa_flags & SA_NODEFER)) {
-		spin_lock_irq(&current->sighand->siglock);
-		sigorsets(&current->blocked,&current->blocked,&ka->sa.sa_mask);
+	spin_lock_irq(&current->sighand->siglock);
+	sigorsets(&current->blocked,&current->blocked,&ka->sa.sa_mask);
+	if (!(ka->sa.sa_flags & SA_NODEFER))
 		sigaddset(&current->blocked,sig);
-		recalc_sigpending();
-		spin_unlock_irq(&current->sighand->siglock);
-	}
+	recalc_sigpending();
+	spin_unlock_irq(&current->sighand->siglock);
 }
 

arch/s390/kernel/signal.c

@@ -429,13 +429,12 @@ handle_signal(unsigned long sig, struct k_sigaction *ka,
 	else
 		setup_frame(sig, ka, oldset, regs);
 
-	if (!(ka->sa.sa_flags & SA_NODEFER)) {
-		spin_lock_irq(&current->sighand->siglock);
-		sigorsets(&current->blocked,&current->blocked,&ka->sa.sa_mask);
+	spin_lock_irq(&current->sighand->siglock);
+	sigorsets(&current->blocked,&current->blocked,&ka->sa.sa_mask);
+	if (!(ka->sa.sa_flags & SA_NODEFER))
 		sigaddset(&current->blocked,sig);
-		recalc_sigpending();
-		spin_unlock_irq(&current->sighand->siglock);
-	}
+	recalc_sigpending();
+	spin_unlock_irq(&current->sighand->siglock);
 }
 
 /*

arch/sh/kernel/signal.c

@@ -546,13 +546,12 @@ handle_signal(unsigned long sig, struct k_sigaction *ka, siginfo_t *info,
 	if (ka->sa.sa_flags & SA_ONESHOT)
 		ka->sa.sa_handler = SIG_DFL;
 
-	if (!(ka->sa.sa_flags & SA_NODEFER)) {
-		spin_lock_irq(&current->sighand->siglock);
-		sigorsets(&current->blocked,&current->blocked,&ka->sa.sa_mask);
+	spin_lock_irq(&current->sighand->siglock);
+	sigorsets(&current->blocked,&current->blocked,&ka->sa.sa_mask);
+	if (!(ka->sa.sa_flags & SA_NODEFER))
 		sigaddset(&current->blocked,sig);
-		recalc_sigpending();
-		spin_unlock_irq(&current->sighand->siglock);
-	}
+	recalc_sigpending();
+	spin_unlock_irq(&current->sighand->siglock);
 }
 
 /*

arch/sh64/kernel/signal.c

@@ -664,13 +664,12 @@ handle_signal(unsigned long sig, siginfo_t *info, struct k_sigaction *ka,
 	else
 		setup_frame(sig, ka, oldset, regs);
 
-	if (!(ka->sa.sa_flags & SA_NODEFER)) {
-		spin_lock_irq(&current->sighand->siglock);
-		sigorsets(&current->blocked,&current->blocked,&ka->sa.sa_mask);
+	spin_lock_irq(&current->sighand->siglock);
+	sigorsets(&current->blocked,&current->blocked,&ka->sa.sa_mask);
+	if (!(ka->sa.sa_flags & SA_NODEFER))
 		sigaddset(&current->blocked,sig);
-		recalc_sigpending();
-		spin_unlock_irq(&current->sighand->siglock);
-	}
+	recalc_sigpending();
+	spin_unlock_irq(&current->sighand->siglock);
 }
 
 /*

arch/sparc/kernel/setup.c

@@ -32,7 +32,6 @@
 #include <linux/spinlock.h>
 #include <linux/root_dev.h>
 
-#include <asm/segment.h>
 #include <asm/system.h>
 #include <asm/io.h>
 #include <asm/processor.h>

arch/sparc/kernel/signal.c

@@ -1034,13 +1034,12 @@ handle_signal(unsigned long signr, struct k_sigaction *ka,
 		else
 			setup_frame(&ka->sa, regs, signr, oldset, info);
 	}
-	if (!(ka->sa.sa_flags & SA_NOMASK)) {
-		spin_lock_irq(&current->sighand->siglock);
-		sigorsets(&current->blocked,&current->blocked,&ka->sa.sa_mask);
+	spin_lock_irq(&current->sighand->siglock);
+	sigorsets(&current->blocked,&current->blocked,&ka->sa.sa_mask);
+	if (!(ka->sa.sa_flags & SA_NOMASK))
 		sigaddset(&current->blocked, signr);
-		recalc_sigpending();
-		spin_unlock_irq(&current->sighand->siglock);
-	}
+	recalc_sigpending();
+	spin_unlock_irq(&current->sighand->siglock);
 }
 
 static inline void syscall_restart(unsigned long orig_i0, struct pt_regs *regs,

arch/sparc/kernel/tick14.c

@@ -19,7 +19,6 @@
 #include <linux/interrupt.h>
 
 #include <asm/oplib.h>
-#include <asm/segment.h>
 #include <asm/timer.h>
 #include <asm/mostek.h>
 #include <asm/system.h>

arch/sparc/kernel/time.c

@@ -32,7 +32,6 @@
 #include <linux/profile.h>
 
 #include <asm/oplib.h>
-#include <asm/segment.h>
 #include <asm/timer.h>
 #include <asm/mostek.h>
 #include <asm/system.h>

arch/sparc/mm/fault.c

@@ -23,7 +23,6 @@
 #include <linux/module.h>
 
 #include <asm/system.h>
-#include <asm/segment.h>
 #include <asm/page.h>
 #include <asm/pgtable.h>
 #include <asm/memreg.h>

arch/sparc/mm/init.c

@@ -25,7 +25,6 @@
 #include <linux/bootmem.h>
 
 #include <asm/system.h>
-#include <asm/segment.h>
 #include <asm/vac-ops.h>
 #include <asm/page.h>
 #include <asm/pgtable.h>

arch/sparc64/kernel/entry.S

@@ -21,6 +21,7 @@
 #include <asm/visasm.h>
 #include <asm/estate.h>
 #include <asm/auxio.h>
+#include <asm/sfafsr.h>
 
 #define curptr      g6
 
@@ -690,79 +691,102 @@ netbsd_syscall:
 	retl
 	 nop
 
-	/* These next few routines must be sure to clear the
-	 * SFSR FaultValid bit so that the fast tlb data protection
-	 * handler does not flush the wrong context and lock up the
-	 * box.
+	/* We need to carefully read the error status, ACK
+	 * the errors, prevent recursive traps, and pass the
+	 * information on to C code for logging.
+	 *
+	 * We pass the AFAR in as-is, and we encode the status
+	 * information as described in asm-sparc64/sfafsr.h
 	 */
-	.globl		__do_data_access_exception
-	.globl		__do_data_access_exception_tl1
-__do_data_access_exception_tl1:
-	rdpr		%pstate, %g4
-	wrpr		%g4, PSTATE_MG|PSTATE_AG, %pstate
-	mov		TLB_SFSR, %g3
-	mov		DMMU_SFAR, %g5
-	ldxa		[%g3] ASI_DMMU, %g4	! Get SFSR
-	ldxa		[%g5] ASI_DMMU, %g5	! Get SFAR
-	stxa		%g0, [%g3] ASI_DMMU	! Clear SFSR.FaultValid bit
+	.globl		__spitfire_access_error
+__spitfire_access_error:
+	/* Disable ESTATE error reporting so that we do not
+	 * take recursive traps and RED state the processor.
+	 */
+	stxa		%g0, [%g0] ASI_ESTATE_ERROR_EN
 	membar		#Sync
-	ba,pt		%xcc, winfix_dax
-	 rdpr		%tpc, %g3
-__do_data_access_exception:
-	rdpr		%pstate, %g4
-	wrpr		%g4, PSTATE_MG|PSTATE_AG, %pstate
-	mov		TLB_SFSR, %g3
-	mov		DMMU_SFAR, %g5
-	ldxa		[%g3] ASI_DMMU, %g4	! Get SFSR
-	ldxa		[%g5] ASI_DMMU, %g5	! Get SFAR
-	stxa		%g0, [%g3] ASI_DMMU	! Clear SFSR.FaultValid bit
-	membar		#Sync
-	sethi		%hi(109f), %g7
-	ba,pt		%xcc, etrap
-109:	 or		%g7, %lo(109b), %g7
-	mov		%l4, %o1
-	mov		%l5, %o2
-	call		data_access_exception
-	 add		%sp, PTREGS_OFF, %o0
-	ba,pt		%xcc, rtrap
-	 clr		%l6
 
-	.globl		__do_instruction_access_exception
-	.globl		__do_instruction_access_exception_tl1
-__do_instruction_access_exception_tl1:
-	rdpr		%pstate, %g4
-	wrpr		%g4, PSTATE_MG|PSTATE_AG, %pstate
-	mov		TLB_SFSR, %g3
-	mov		DMMU_SFAR, %g5
-	ldxa		[%g3] ASI_DMMU, %g4	! Get SFSR
-	ldxa		[%g5] ASI_DMMU, %g5	! Get SFAR
-	stxa		%g0, [%g3] ASI_IMMU	! Clear FaultValid bit
+	mov		UDBE_UE, %g1
+	ldxa		[%g0] ASI_AFSR, %g4	! Get AFSR
+
+	/* __spitfire_cee_trap branches here with AFSR in %g4 and
+	 * UDBE_CE in %g1.  It only clears ESTATE_ERR_CE in the
+	 * ESTATE Error Enable register.
+	 */
+__spitfire_cee_trap_continue:
+	ldxa		[%g0] ASI_AFAR, %g5	! Get AFAR
+
+	rdpr		%tt, %g3
+	and		%g3, 0x1ff, %g3		! Paranoia
+	sllx		%g3, SFSTAT_TRAP_TYPE_SHIFT, %g3
+	or		%g4, %g3, %g4
+	rdpr		%tl, %g3
+	cmp		%g3, 1
+	mov		1, %g3
+	bleu		%xcc, 1f
+	 sllx		%g3, SFSTAT_TL_GT_ONE_SHIFT, %g3
+
+	or		%g4, %g3, %g4
+
+	/* Read in the UDB error register state, clearing the
+	 * sticky error bits as-needed.  We only clear them if
+	 * the UE bit is set.  Likewise, __spitfire_cee_trap
+	 * below will only do so if the CE bit is set.
+	 *
+	 * NOTE: UltraSparc-I/II have high and low UDB error
+	 *       registers, corresponding to the two UDB units
+	 *       present on those chips.  UltraSparc-IIi only
+	 *       has a single UDB, called "SDB" in the manual.
+	 *       For IIi the upper UDB register always reads
+	 *       as zero so for our purposes things will just
+	 *       work with the checks below.
+	 */
+1:	ldxa		[%g0] ASI_UDBH_ERROR_R, %g3
+	and		%g3, 0x3ff, %g7		! Paranoia
+	sllx		%g7, SFSTAT_UDBH_SHIFT, %g7
+	or		%g4, %g7, %g4
+	andcc		%g3, %g1, %g3		! UDBE_UE or UDBE_CE
+	be,pn		%xcc, 1f
+	 nop
+	stxa		%g3, [%g0] ASI_UDB_ERROR_W
 	membar		#Sync
-	sethi		%hi(109f), %g7
+
+1:	mov		0x18, %g3
+	ldxa		[%g3] ASI_UDBL_ERROR_R, %g3
+	and		%g3, 0x3ff, %g7		! Paranoia
+	sllx		%g7, SFSTAT_UDBL_SHIFT, %g7
+	or		%g4, %g7, %g4
+	andcc		%g3, %g1, %g3		! UDBE_UE or UDBE_CE
+	be,pn		%xcc, 1f
+	 nop
+	mov		0x18, %g7
+	stxa		%g3, [%g7] ASI_UDB_ERROR_W
+	membar		#Sync
+
+1:	/* Ok, now that we've latched the error state,
+	 * clear the sticky bits in the AFSR.
+	 */
+	stxa		%g4, [%g0] ASI_AFSR
+	membar		#Sync
+
+	rdpr		%tl, %g2
+	cmp		%g2, 1
+	rdpr		%pil, %g2
+	bleu,pt		%xcc, 1f
+	 wrpr		%g0, 15, %pil
+
 	ba,pt		%xcc, etraptl1
-109:	 or		%g7, %lo(109b), %g7
-	mov		%l4, %o1
-	mov		%l5, %o2
-	call		instruction_access_exception_tl1
-	 add		%sp, PTREGS_OFF, %o0
-	ba,pt		%xcc, rtrap
-	 clr		%l6
+	 rd		%pc, %g7
 
-__do_instruction_access_exception:
-	rdpr		%pstate, %g4
-	wrpr		%g4, PSTATE_MG|PSTATE_AG, %pstate
-	mov		TLB_SFSR, %g3
-	mov		DMMU_SFAR, %g5
-	ldxa		[%g3] ASI_DMMU, %g4	! Get SFSR
-	ldxa		[%g5] ASI_DMMU, %g5	! Get SFAR
-	stxa		%g0, [%g3] ASI_IMMU	! Clear FaultValid bit
-	membar		#Sync
-	sethi		%hi(109f), %g7
-	ba,pt		%xcc, etrap
-109:	 or		%g7, %lo(109b), %g7
-	mov		%l4, %o1
+	ba,pt		%xcc, 2f
+	 nop
+
+1:	ba,pt		%xcc, etrap_irq
+	 rd		%pc, %g7
+
+2:	mov		%l4, %o1
 	mov		%l5, %o2
-	call		instruction_access_exception
+	call		spitfire_access_error
 	 add		%sp, PTREGS_OFF, %o0
 	ba,pt		%xcc, rtrap
 	 clr		%l6
@@ -784,79 +808,124 @@ __do_instruction_access_exception:
 	 * as it is the only situation where we can safely record
 	 * and log.  For trap level >1 we just clear the CE bit
 	 * in the AFSR and return.
-	 */
-
-	/* Our trap handling infrastructure allows us to preserve
-	 * two 64-bit values during etrap for arguments to
-	 * subsequent C code.  Therefore we encode the information
-	 * as follows:
 	 *
-	 * value 1) Full 64-bits of AFAR
-	 * value 2) Low 33-bits of AFSR, then bits 33-->42
-	 *          are UDBL error status and bits 43-->52
-	 *          are UDBH error status
+	 * This is just like __spiftire_access_error above, but it
+	 * specifically handles correctable errors.  If an
+	 * uncorrectable error is indicated in the AFSR we
+	 * will branch directly above to __spitfire_access_error
+	 * to handle it instead.  Uncorrectable therefore takes
+	 * priority over correctable, and the error logging
+	 * C code will notice this case by inspecting the
+	 * trap type.
 	 */
-	.align	64
-	.globl	cee_trap
-cee_trap:
-	ldxa	[%g0] ASI_AFSR, %g1		! Read AFSR
-	ldxa	[%g0] ASI_AFAR, %g2		! Read AFAR
-	sllx	%g1, 31, %g1			! Clear reserved bits
-	srlx	%g1, 31, %g1			! in AFSR
+	.globl		__spitfire_cee_trap
+__spitfire_cee_trap:
+	ldxa		[%g0] ASI_AFSR, %g4	! Get AFSR
+	mov		1, %g3
+	sllx		%g3, SFAFSR_UE_SHIFT, %g3
+	andcc		%g4, %g3, %g0		! Check for UE
+	bne,pn		%xcc, __spitfire_access_error
+	 nop
 
-	/* NOTE: UltraSparc-I/II have high and low UDB error
-	 *       registers, corresponding to the two UDB units
-	 *       present on those chips.  UltraSparc-IIi only
-	 *       has a single UDB, called "SDB" in the manual.
-	 *       For IIi the upper UDB register always reads
-	 *       as zero so for our purposes things will just
-	 *       work with the checks below.
+	/* Ok, in this case we only have a correctable error.
+	 * Indicate we only wish to capture that state in register
+	 * %g1, and we only disable CE error reporting unlike UE
+	 * handling which disables all errors.
 	 */
-	ldxa	[%g0] ASI_UDBL_ERROR_R, %g3	! Read UDB-Low error status
-	andcc	%g3, (1 << 8), %g4		! Check CE bit
-	sllx	%g3, (64 - 10), %g3		! Clear reserved bits
-	srlx	%g3, (64 - 10), %g3		! in UDB-Low error status
+	ldxa		[%g0] ASI_ESTATE_ERROR_EN, %g3
+	andn		%g3, ESTATE_ERR_CE, %g3
+	stxa		%g3, [%g0] ASI_ESTATE_ERROR_EN
+	membar		#Sync
 
-	sllx	%g3, (33 + 0), %g3		! Shift up to encoding area
-	or	%g1, %g3, %g1			! Or it in
-	be,pn	%xcc, 1f			! Branch if CE bit was clear
+	/* Preserve AFSR in %g4, indicate UDB state to capture in %g1 */
+	ba,pt		%xcc, __spitfire_cee_trap_continue
+	 mov		UDBE_CE, %g1
+
+	.globl		__spitfire_data_access_exception
+	.globl		__spitfire_data_access_exception_tl1
+__spitfire_data_access_exception_tl1:
+	rdpr		%pstate, %g4
+	wrpr		%g4, PSTATE_MG|PSTATE_AG, %pstate
+	mov		TLB_SFSR, %g3
+	mov		DMMU_SFAR, %g5
+	ldxa		[%g3] ASI_DMMU, %g4	! Get SFSR
+	ldxa		[%g5] ASI_DMMU, %g5	! Get SFAR
+	stxa		%g0, [%g3] ASI_DMMU	! Clear SFSR.FaultValid bit
+	membar		#Sync
+	rdpr		%tt, %g3
+	cmp		%g3, 0x80		! first win spill/fill trap
+	blu,pn		%xcc, 1f
+	 cmp		%g3, 0xff		! last win spill/fill trap
+	bgu,pn		%xcc, 1f
 	 nop
-	stxa	%g4, [%g0] ASI_UDB_ERROR_W	! Clear CE sticky bit in UDBL
-	membar	#Sync				! Synchronize ASI stores
-1:	mov	0x18, %g5			! Addr of UDB-High error status
-	ldxa	[%g5] ASI_UDBH_ERROR_R, %g3	! Read it
+	ba,pt		%xcc, winfix_dax
+	 rdpr		%tpc, %g3
+1:	sethi		%hi(109f), %g7
+	ba,pt		%xcc, etraptl1
+109:	 or		%g7, %lo(109b), %g7
+	mov		%l4, %o1
+	mov		%l5, %o2
+	call		spitfire_data_access_exception_tl1
+	 add		%sp, PTREGS_OFF, %o0
+	ba,pt		%xcc, rtrap
+	 clr		%l6
 
-	andcc	%g3, (1 << 8), %g4		! Check CE bit
-	sllx	%g3, (64 - 10), %g3		! Clear reserved bits
-	srlx	%g3, (64 - 10), %g3		! in UDB-High error status
-	sllx	%g3, (33 + 10), %g3		! Shift up to encoding area
-	or	%g1, %g3, %g1			! Or it in
-	be,pn	%xcc, 1f			! Branch if CE bit was clear
-	 nop
-	nop
+__spitfire_data_access_exception:
+	rdpr		%pstate, %g4
+	wrpr		%g4, PSTATE_MG|PSTATE_AG, %pstate
+	mov		TLB_SFSR, %g3
+	mov		DMMU_SFAR, %g5
+	ldxa		[%g3] ASI_DMMU, %g4	! Get SFSR
+	ldxa		[%g5] ASI_DMMU, %g5	! Get SFAR
+	stxa		%g0, [%g3] ASI_DMMU	! Clear SFSR.FaultValid bit
+	membar		#Sync
+	sethi		%hi(109f), %g7
+	ba,pt		%xcc, etrap
+109:	 or		%g7, %lo(109b), %g7
+	mov		%l4, %o1
+	mov		%l5, %o2
+	call		spitfire_data_access_exception
+	 add		%sp, PTREGS_OFF, %o0
+	ba,pt		%xcc, rtrap
+	 clr		%l6
 
-	stxa	%g4, [%g5] ASI_UDB_ERROR_W	! Clear CE sticky bit in UDBH
-	membar	#Sync				! Synchronize ASI stores
-1:	mov	1, %g5				! AFSR CE bit is
-	sllx	%g5, 20, %g5			! bit 20
-	stxa	%g5, [%g0] ASI_AFSR		! Clear CE sticky bit in AFSR
-	membar	#Sync				! Synchronize ASI stores
-	sllx	%g2, (64 - 41), %g2		! Clear reserved bits
-	srlx	%g2, (64 - 41), %g2		! in latched AFAR
+	.globl		__spitfire_insn_access_exception
+	.globl		__spitfire_insn_access_exception_tl1
+__spitfire_insn_access_exception_tl1:
+	rdpr		%pstate, %g4
+	wrpr		%g4, PSTATE_MG|PSTATE_AG, %pstate
+	mov		TLB_SFSR, %g3
+	ldxa		[%g3] ASI_IMMU, %g4	! Get SFSR
+	rdpr		%tpc, %g5		! IMMU has no SFAR, use TPC
+	stxa		%g0, [%g3] ASI_IMMU	! Clear FaultValid bit
+	membar		#Sync
+	sethi		%hi(109f), %g7
+	ba,pt		%xcc, etraptl1
+109:	 or		%g7, %lo(109b), %g7
+	mov		%l4, %o1
+	mov		%l5, %o2
+	call		spitfire_insn_access_exception_tl1
+	 add		%sp, PTREGS_OFF, %o0
+	ba,pt		%xcc, rtrap
+	 clr		%l6
 
-	andn	%g2, 0x0f, %g2			! Finish resv bit clearing
-	mov	%g1, %g4			! Move AFSR+UDB* into save reg
-	mov	%g2, %g5			! Move AFAR into save reg
-	rdpr	%pil, %g2
-	wrpr	%g0, 15, %pil
-	ba,pt	%xcc, etrap_irq
-	 rd	%pc, %g7
-	mov	%l4, %o0
-
-	mov	%l5, %o1
-	call	cee_log
-	 add	%sp, PTREGS_OFF, %o2
-	ba,a,pt	%xcc, rtrap_irq
+__spitfire_insn_access_exception:
+	rdpr		%pstate, %g4
+	wrpr		%g4, PSTATE_MG|PSTATE_AG, %pstate
+	mov		TLB_SFSR, %g3
+	ldxa		[%g3] ASI_IMMU, %g4	! Get SFSR
+	rdpr		%tpc, %g5		! IMMU has no SFAR, use TPC
+	stxa		%g0, [%g3] ASI_IMMU	! Clear FaultValid bit
+	membar		#Sync
+	sethi		%hi(109f), %g7
+	ba,pt		%xcc, etrap
+109:	 or		%g7, %lo(109b), %g7
+	mov		%l4, %o1
+	mov		%l5, %o2
+	call		spitfire_insn_access_exception
+	 add		%sp, PTREGS_OFF, %o0
+	ba,pt		%xcc, rtrap
+	 clr		%l6
 
 	/* Capture I/D/E-cache state into per-cpu error scoreboard.
 	 *

arch/sparc64/kernel/pci_iommu.c

@@ -466,7 +466,7 @@ do_flush_sync:
 		if (!limit)
 			break;
 		udelay(1);
-		membar("#LoadLoad");
+		rmb();
 	}
 	if (!limit)
 		printk(KERN_WARNING "pci_strbuf_flush: flushflag timeout "

arch/sparc64/kernel/process.c

@@ -103,7 +103,7 @@ void cpu_idle(void)
 		 * other cpus see our increasing idleness for the buddy
 		 * redistribution algorithm.  -DaveM
 		 */
-		membar("#StoreStore | #StoreLoad");
+		membar_storeload_storestore();
 	}
 }
 

arch/sparc64/kernel/sbus.c

@@ -147,7 +147,7 @@ static void sbus_strbuf_flush(struct sbus_iommu *iommu, u32 base, unsigned long
 		if (!limit)
 			break;
 		udelay(1);
-		membar("#LoadLoad");
+		rmb();
 	}
 	if (!limit)
 		printk(KERN_WARNING "sbus_strbuf_flush: flushflag timeout "

arch/sparc64/kernel/setup.c

@@ -33,7 +33,6 @@
 #include <linux/cpu.h>
 #include <linux/initrd.h>
 
-#include <asm/segment.h>
 #include <asm/system.h>
 #include <asm/io.h>
 #include <asm/processor.h>

arch/sparc64/kernel/signal.c

@@ -574,13 +574,12 @@ static inline void handle_signal(unsigned long signr, struct k_sigaction *ka,
 {
 	setup_rt_frame(ka, regs, signr, oldset,
 		       (ka->sa.sa_flags & SA_SIGINFO) ? info : NULL);
-	if (!(ka->sa.sa_flags & SA_NOMASK)) {
-		spin_lock_irq(&current->sighand->siglock);
-		sigorsets(&current->blocked,&current->blocked,&ka->sa.sa_mask);
+	spin_lock_irq(&current->sighand->siglock);
+	sigorsets(&current->blocked,&current->blocked,&ka->sa.sa_mask);
+	if (!(ka->sa.sa_flags & SA_NOMASK))
 		sigaddset(&current->blocked,signr);
-		recalc_sigpending();
-		spin_unlock_irq(&current->sighand->siglock);
-	}
+	recalc_sigpending();
+	spin_unlock_irq(&current->sighand->siglock);
 }
 
 static inline void syscall_restart(unsigned long orig_i0, struct pt_regs *regs,

arch/sparc64/kernel/signal32.c

@@ -877,11 +877,12 @@ static void new_setup_frame32(struct k_sigaction *ka, struct pt_regs *regs,
 			unsigned long page = (unsigned long)
 				page_address(pte_page(*ptep));
 
-			__asm__ __volatile__(
-			"	membar	#StoreStore\n"
-			"	flush	%0 + %1"
-			: : "r" (page), "r" (address & (PAGE_SIZE - 1))
-			: "memory");
+			wmb();
+			__asm__ __volatile__("flush	%0 + %1"
+					     : /* no outputs */
+					     : "r" (page),
+					       "r" (address & (PAGE_SIZE - 1))
+					     : "memory");
 		}
 		pte_unmap(ptep);
 		preempt_enable();
@@ -1292,11 +1293,12 @@ static void setup_rt_frame32(struct k_sigaction *ka, struct pt_regs *regs,
 			unsigned long page = (unsigned long)
 				page_address(pte_page(*ptep));
 
-			__asm__ __volatile__(
-			"	membar	#StoreStore\n"
-			"	flush	%0 + %1"
-			: : "r" (page), "r" (address & (PAGE_SIZE - 1))
-			: "memory");
+			wmb();
+			__asm__ __volatile__("flush	%0 + %1"
+					     : /* no outputs */
+					     : "r" (page),
+					       "r" (address & (PAGE_SIZE - 1))
+					     : "memory");
 		}
 		pte_unmap(ptep);
 		preempt_enable();
@@ -1325,13 +1327,12 @@ static inline void handle_signal32(unsigned long signr, struct k_sigaction *ka,
 		else
 			setup_frame32(&ka->sa, regs, signr, oldset, info);
 	}
-	if (!(ka->sa.sa_flags & SA_NOMASK)) {
-		spin_lock_irq(&current->sighand->siglock);
-		sigorsets(&current->blocked,&current->blocked,&ka->sa.sa_mask);
+	spin_lock_irq(&current->sighand->siglock);
+	sigorsets(&current->blocked,&current->blocked,&ka->sa.sa_mask);
+	if (!(ka->sa.sa_flags & SA_NOMASK))
 		sigaddset(&current->blocked,signr);
-		recalc_sigpending();
-		spin_unlock_irq(&current->sighand->siglock);
-	}
+	recalc_sigpending();
+	spin_unlock_irq(&current->sighand->siglock);
 }
 
 static inline void syscall_restart32(unsigned long orig_i0, struct pt_regs *regs,

arch/sparc64/kernel/smp.c

@@ -144,7 +144,7 @@ void __init smp_callin(void)
 	current->active_mm = &init_mm;
 
 	while (!cpu_isset(cpuid, smp_commenced_mask))
-		membar("#LoadLoad");
+		rmb();
 
 	cpu_set(cpuid, cpu_online_map);
 }
@@ -184,11 +184,11 @@ static inline long get_delta (long *rt, long *master)
 	for (i = 0; i < NUM_ITERS; i++) {
 		t0 = tick_ops->get_tick();
 		go[MASTER] = 1;
-		membar("#StoreLoad");
+		membar_storeload();
 		while (!(tm = go[SLAVE]))
-			membar("#LoadLoad");
+			rmb();
 		go[SLAVE] = 0;
-		membar("#StoreStore");
+		wmb();
 		t1 = tick_ops->get_tick();
 
 		if (t1 - t0 < best_t1 - best_t0)
@@ -221,7 +221,7 @@ void smp_synchronize_tick_client(void)
 	go[MASTER] = 1;
 
 	while (go[MASTER])
-		membar("#LoadLoad");
+		rmb();
 
 	local_irq_save(flags);
 	{
@@ -273,21 +273,21 @@ static void smp_synchronize_one_tick(int cpu)
 
 	/* wait for client to be ready */
 	while (!go[MASTER])
-		membar("#LoadLoad");
+		rmb();
 
 	/* now let the client proceed into his loop */
 	go[MASTER] = 0;
-	membar("#StoreLoad");
+	membar_storeload();
 
 	spin_lock_irqsave(&itc_sync_lock, flags);
 	{
 		for (i = 0; i < NUM_ROUNDS*NUM_ITERS; i++) {
 			while (!go[MASTER])
-				membar("#LoadLoad");
+				rmb();
 			go[MASTER] = 0;
-			membar("#StoreStore");
+			wmb();
 			go[SLAVE] = tick_ops->get_tick();
-			membar("#StoreLoad");
+			membar_storeload();
 		}
 	}
 	spin_unlock_irqrestore(&itc_sync_lock, flags);
@@ -927,11 +927,11 @@ void smp_capture(void)
 		       smp_processor_id());
 #endif
 		penguins_are_doing_time = 1;
-		membar("#StoreStore | #LoadStore");
+		membar_storestore_loadstore();
 		atomic_inc(&smp_capture_registry);
 		smp_cross_call(&xcall_capture, 0, 0, 0);
 		while (atomic_read(&smp_capture_registry) != ncpus)
-			membar("#LoadLoad");
+			rmb();
 #ifdef CAPTURE_DEBUG
 		printk("done\n");
 #endif
@@ -947,7 +947,7 @@ void smp_release(void)
 		       smp_processor_id());
 #endif
 		penguins_are_doing_time = 0;
-		membar("#StoreStore | #StoreLoad");
+		membar_storeload_storestore();
 		atomic_dec(&smp_capture_registry);
 	}
 }
@@ -970,9 +970,9 @@ void smp_penguin_jailcell(int irq, struct pt_regs *regs)
 	save_alternate_globals(global_save);
 	prom_world(1);
 	atomic_inc(&smp_capture_registry);
-	membar("#StoreLoad | #StoreStore");
+	membar_storeload_storestore();
 	while (penguins_are_doing_time)
-		membar("#LoadLoad");
+		rmb();
 	restore_alternate_globals(global_save);
 	atomic_dec(&smp_capture_registry);
 	prom_world(0);

arch/sparc64/kernel/sparc64_ksyms.c

@@ -99,17 +99,6 @@ extern int __ashrdi3(int, int);
 extern void dump_thread(struct pt_regs *, struct user *);
 extern int dump_fpu (struct pt_regs * regs, elf_fpregset_t * fpregs);
 
-#if defined(CONFIG_SMP) && defined(CONFIG_DEBUG_SPINLOCK)
-extern void _do_spin_lock (spinlock_t *lock, char *str);
-extern void _do_spin_unlock (spinlock_t *lock);
-extern int _spin_trylock (spinlock_t *lock);
-extern void _do_read_lock(rwlock_t *rw, char *str);
-extern void _do_read_unlock(rwlock_t *rw, char *str);
-extern void _do_write_lock(rwlock_t *rw, char *str);
-extern void _do_write_unlock(rwlock_t *rw);
-extern int _do_write_trylock(rwlock_t *rw, char *str);
-#endif
-
 extern unsigned long phys_base;
 extern unsigned long pfn_base;
 
@@ -152,18 +141,6 @@ EXPORT_SYMBOL(_mcount);
 EXPORT_SYMBOL(cpu_online_map);
 EXPORT_SYMBOL(phys_cpu_present_map);
 
-/* Spinlock debugging library, optional. */
-#ifdef CONFIG_DEBUG_SPINLOCK
-EXPORT_SYMBOL(_do_spin_lock);
-EXPORT_SYMBOL(_do_spin_unlock);
-EXPORT_SYMBOL(_spin_trylock);
-EXPORT_SYMBOL(_do_read_lock);
-EXPORT_SYMBOL(_do_read_unlock);
-EXPORT_SYMBOL(_do_write_lock);
-EXPORT_SYMBOL(_do_write_unlock);
-EXPORT_SYMBOL(_do_write_trylock);
-#endif
-
 EXPORT_SYMBOL(smp_call_function);
 #endif /* CONFIG_SMP */
 
@@ -429,3 +406,12 @@ EXPORT_SYMBOL(xor_vis_4);
 EXPORT_SYMBOL(xor_vis_5);
 
 EXPORT_SYMBOL(prom_palette);
+
+/* memory barriers */
+EXPORT_SYMBOL(mb);
+EXPORT_SYMBOL(rmb);
+EXPORT_SYMBOL(wmb);
+EXPORT_SYMBOL(membar_storeload);
+EXPORT_SYMBOL(membar_storeload_storestore);
+EXPORT_SYMBOL(membar_storeload_loadload);
+EXPORT_SYMBOL(membar_storestore_loadstore);

arch/sparc64/kernel/traps.c

@@ -33,6 +33,7 @@
 #include <asm/dcu.h>
 #include <asm/estate.h>
 #include <asm/chafsr.h>
+#include <asm/sfafsr.h>
 #include <asm/psrcompat.h>
 #include <asm/processor.h>
 #include <asm/timer.h>
@@ -143,8 +144,7 @@ void do_BUG(const char *file, int line)
 }
 #endif
 
-void instruction_access_exception(struct pt_regs *regs,
-				  unsigned long sfsr, unsigned long sfar)
+void spitfire_insn_access_exception(struct pt_regs *regs, unsigned long sfsr, unsigned long sfar)
 {
 	siginfo_t info;
 
@@ -153,8 +153,8 @@ void instruction_access_exception(struct pt_regs *regs,
 		return;
 
 	if (regs->tstate & TSTATE_PRIV) {
-		printk("instruction_access_exception: SFSR[%016lx] SFAR[%016lx], going.\n",
-		       sfsr, sfar);
+		printk("spitfire_insn_access_exception: SFSR[%016lx] "
+		       "SFAR[%016lx], going.\n", sfsr, sfar);
 		die_if_kernel("Iax", regs);
 	}
 	if (test_thread_flag(TIF_32BIT)) {
@@ -169,19 +169,17 @@ void instruction_access_exception(struct pt_regs *regs,
 	force_sig_info(SIGSEGV, &info, current);
 }
 
-void instruction_access_exception_tl1(struct pt_regs *regs,
-				      unsigned long sfsr, unsigned long sfar)
+void spitfire_insn_access_exception_tl1(struct pt_regs *regs, unsigned long sfsr, unsigned long sfar)
 {
 	if (notify_die(DIE_TRAP_TL1, "instruction access exception tl1", regs,
 		       0, 0x8, SIGTRAP) == NOTIFY_STOP)
 		return;
 
 	dump_tl1_traplog((struct tl1_traplog *)(regs + 1));
-	instruction_access_exception(regs, sfsr, sfar);
+	spitfire_insn_access_exception(regs, sfsr, sfar);
 }
 
-void data_access_exception(struct pt_regs *regs,
-			   unsigned long sfsr, unsigned long sfar)
+void spitfire_data_access_exception(struct pt_regs *regs, unsigned long sfsr, unsigned long sfar)
 {
 	siginfo_t info;
 
@@ -207,8 +205,8 @@ void data_access_exception(struct pt_regs *regs,
 			return;
 		}
 		/* Shit... */
-		printk("data_access_exception: SFSR[%016lx] SFAR[%016lx], going.\n",
-		       sfsr, sfar);
+		printk("spitfire_data_access_exception: SFSR[%016lx] "
+		       "SFAR[%016lx], going.\n", sfsr, sfar);
 		die_if_kernel("Dax", regs);
 	}
 
@@ -220,6 +218,16 @@ void data_access_exception(struct pt_regs *regs,
 	force_sig_info(SIGSEGV, &info, current);
 }
 
+void spitfire_data_access_exception_tl1(struct pt_regs *regs, unsigned long sfsr, unsigned long sfar)
+{
+	if (notify_die(DIE_TRAP_TL1, "data access exception tl1", regs,
+		       0, 0x30, SIGTRAP) == NOTIFY_STOP)
+		return;
+
+	dump_tl1_traplog((struct tl1_traplog *)(regs + 1));
+	spitfire_data_access_exception(regs, sfsr, sfar);
+}
+
 #ifdef CONFIG_PCI
 /* This is really pathetic... */
 extern volatile int pci_poke_in_progress;
@@ -253,54 +261,13 @@ static void spitfire_clean_and_reenable_l1_caches(void)
 			     : "memory");
 }
 
-void do_iae(struct pt_regs *regs)
+static void spitfire_enable_estate_errors(void)
 {
-	siginfo_t info;
-
-	spitfire_clean_and_reenable_l1_caches();
-
-	if (notify_die(DIE_TRAP, "instruction access exception", regs,
-		       0, 0x8, SIGTRAP) == NOTIFY_STOP)
-		return;
-
-	info.si_signo = SIGBUS;
-	info.si_errno = 0;
-	info.si_code = BUS_OBJERR;
-	info.si_addr = (void *)0;
-	info.si_trapno = 0;
-	force_sig_info(SIGBUS, &info, current);
-}
-
-void do_dae(struct pt_regs *regs)
-{
-	siginfo_t info;
-
-#ifdef CONFIG_PCI
-	if (pci_poke_in_progress && pci_poke_cpu == smp_processor_id()) {
-		spitfire_clean_and_reenable_l1_caches();
-
-		pci_poke_faulted = 1;
-
-		/* Why the fuck did they have to change this? */
-		if (tlb_type == cheetah || tlb_type == cheetah_plus)
-			regs->tpc += 4;
-
-		regs->tnpc = regs->tpc + 4;
-		return;
-	}
-#endif
-	spitfire_clean_and_reenable_l1_caches();
-
-	if (notify_die(DIE_TRAP, "data access exception", regs,
-		       0, 0x30, SIGTRAP) == NOTIFY_STOP)
-		return;
-
-	info.si_signo = SIGBUS;
-	info.si_errno = 0;
-	info.si_code = BUS_OBJERR;
-	info.si_addr = (void *)0;
-	info.si_trapno = 0;
-	force_sig_info(SIGBUS, &info, current);
+	__asm__ __volatile__("stxa	%0, [%%g0] %1\n\t"
+			     "membar	#Sync"
+			     : /* no outputs */
+			     : "r" (ESTATE_ERR_ALL),
+			       "i" (ASI_ESTATE_ERROR_EN));
 }
 
 static char ecc_syndrome_table[] = {
@@ -338,65 +305,15 @@ static char ecc_syndrome_table[] = {
 	0x0b, 0x48, 0x48, 0x4b, 0x48, 0x4b, 0x4b, 0x4a
 };
 
-/* cee_trap in entry.S encodes AFSR/UDBH/UDBL error status
- * in the following format.  The AFAR is left as is, with
- * reserved bits cleared, and is a raw 40-bit physical
- * address.
- */
-#define CE_STATUS_UDBH_UE		(1UL << (43 + 9))
-#define CE_STATUS_UDBH_CE		(1UL << (43 + 8))
-#define CE_STATUS_UDBH_ESYNDR		(0xffUL << 43)
-#define CE_STATUS_UDBH_SHIFT		43
-#define CE_STATUS_UDBL_UE		(1UL << (33 + 9))
-#define CE_STATUS_UDBL_CE		(1UL << (33 + 8))
-#define CE_STATUS_UDBL_ESYNDR		(0xffUL << 33)
-#define CE_STATUS_UDBL_SHIFT		33
-#define CE_STATUS_AFSR_MASK		(0x1ffffffffUL)
-#define CE_STATUS_AFSR_ME		(1UL << 32)
-#define CE_STATUS_AFSR_PRIV		(1UL << 31)
-#define CE_STATUS_AFSR_ISAP		(1UL << 30)
-#define CE_STATUS_AFSR_ETP		(1UL << 29)
-#define CE_STATUS_AFSR_IVUE		(1UL << 28)
-#define CE_STATUS_AFSR_TO		(1UL << 27)
-#define CE_STATUS_AFSR_BERR		(1UL << 26)
-#define CE_STATUS_AFSR_LDP		(1UL << 25)
-#define CE_STATUS_AFSR_CP		(1UL << 24)
-#define CE_STATUS_AFSR_WP		(1UL << 23)
-#define CE_STATUS_AFSR_EDP		(1UL << 22)
-#define CE_STATUS_AFSR_UE		(1UL << 21)
-#define CE_STATUS_AFSR_CE		(1UL << 20)
-#define CE_STATUS_AFSR_ETS		(0xfUL << 16)
-#define CE_STATUS_AFSR_ETS_SHIFT	16
-#define CE_STATUS_AFSR_PSYND		(0xffffUL << 0)
-#define CE_STATUS_AFSR_PSYND_SHIFT	0
-
-/* Layout of Ecache TAG Parity Syndrome of AFSR */
-#define AFSR_ETSYNDROME_7_0		0x1UL /* E$-tag bus bits  <7:0> */
-#define AFSR_ETSYNDROME_15_8		0x2UL /* E$-tag bus bits <15:8> */
-#define AFSR_ETSYNDROME_21_16		0x4UL /* E$-tag bus bits <21:16> */
-#define AFSR_ETSYNDROME_24_22		0x8UL /* E$-tag bus bits <24:22> */
-
 static char *syndrome_unknown = "<Unknown>";
 
-asmlinkage void cee_log(unsigned long ce_status,
-			unsigned long afar,
-			struct pt_regs *regs)
+static void spitfire_log_udb_syndrome(unsigned long afar, unsigned long udbh, unsigned long udbl, unsigned long bit)
 {
-	char memmod_str[64];
-	char *p;
-	unsigned short scode, udb_reg;
+	unsigned short scode;
+	char memmod_str[64], *p;
 
-	printk(KERN_WARNING "CPU[%d]: Correctable ECC Error "
-	       "AFSR[%lx] AFAR[%016lx] UDBL[%lx] UDBH[%lx]\n",
-	       smp_processor_id(),
-	       (ce_status & CE_STATUS_AFSR_MASK),
-	       afar,
-	       ((ce_status >> CE_STATUS_UDBL_SHIFT) & 0x3ffUL),
-	       ((ce_status >> CE_STATUS_UDBH_SHIFT) & 0x3ffUL));
-
-	udb_reg = ((ce_status >> CE_STATUS_UDBL_SHIFT) & 0x3ffUL);
-	if (udb_reg & (1 << 8)) {
-		scode = ecc_syndrome_table[udb_reg & 0xff];
+	if (udbl & bit) {
+		scode = ecc_syndrome_table[udbl & 0xff];
 		if (prom_getunumber(scode, afar,
 				    memmod_str, sizeof(memmod_str)) == -1)
 			p = syndrome_unknown;
@@ -407,9 +324,8 @@ asmlinkage void cee_log(unsigned long ce_status,
 		       smp_processor_id(), scode, p);
 	}
 
-	udb_reg = ((ce_status >> CE_STATUS_UDBH_SHIFT) & 0x3ffUL);
-	if (udb_reg & (1 << 8)) {
-		scode = ecc_syndrome_table[udb_reg & 0xff];
+	if (udbh & bit) {
+		scode = ecc_syndrome_table[udbh & 0xff];
 		if (prom_getunumber(scode, afar,
 				    memmod_str, sizeof(memmod_str)) == -1)
 			p = syndrome_unknown;
@@ -419,6 +335,127 @@ asmlinkage void cee_log(unsigned long ce_status,
 		       "Memory Module \"%s\"\n",
 		       smp_processor_id(), scode, p);
 	}
+
+}
+
+static void spitfire_cee_log(unsigned long afsr, unsigned long afar, unsigned long udbh, unsigned long udbl, int tl1, struct pt_regs *regs)
+{
+
+	printk(KERN_WARNING "CPU[%d]: Correctable ECC Error "
+	       "AFSR[%lx] AFAR[%016lx] UDBL[%lx] UDBH[%lx] TL>1[%d]\n",
+	       smp_processor_id(), afsr, afar, udbl, udbh, tl1);
+
+	spitfire_log_udb_syndrome(afar, udbh, udbl, UDBE_CE);
+
+	/* We always log it, even if someone is listening for this
+	 * trap.
+	 */
+	notify_die(DIE_TRAP, "Correctable ECC Error", regs,
+		   0, TRAP_TYPE_CEE, SIGTRAP);
+
+	/* The Correctable ECC Error trap does not disable I/D caches.  So
+	 * we only have to restore the ESTATE Error Enable register.
+	 */
+	spitfire_enable_estate_errors();
+}
+
+static void spitfire_ue_log(unsigned long afsr, unsigned long afar, unsigned long udbh, unsigned long udbl, unsigned long tt, int tl1, struct pt_regs *regs)
+{
+	siginfo_t info;
+
+	printk(KERN_WARNING "CPU[%d]: Uncorrectable Error AFSR[%lx] "
+	       "AFAR[%lx] UDBL[%lx] UDBH[%ld] TT[%lx] TL>1[%d]\n",
+	       smp_processor_id(), afsr, afar, udbl, udbh, tt, tl1);
+
+	/* XXX add more human friendly logging of the error status
+	 * XXX as is implemented for cheetah
+	 */
+
+	spitfire_log_udb_syndrome(afar, udbh, udbl, UDBE_UE);
+
+	/* We always log it, even if someone is listening for this
+	 * trap.
+	 */
+	notify_die(DIE_TRAP, "Uncorrectable Error", regs,
+		   0, tt, SIGTRAP);
+
+	if (regs->tstate & TSTATE_PRIV) {
+		if (tl1)
+			dump_tl1_traplog((struct tl1_traplog *)(regs + 1));
+		die_if_kernel("UE", regs);
+	}
+
+	/* XXX need more intelligent processing here, such as is implemented
+	 * XXX for cheetah errors, in fact if the E-cache still holds the
+	 * XXX line with bad parity this will loop
+	 */
+
+	spitfire_clean_and_reenable_l1_caches();
+	spitfire_enable_estate_errors();
+
+	if (test_thread_flag(TIF_32BIT)) {
+		regs->tpc &= 0xffffffff;
+		regs->tnpc &= 0xffffffff;
+	}
+	info.si_signo = SIGBUS;
+	info.si_errno = 0;
+	info.si_code = BUS_OBJERR;
+	info.si_addr = (void *)0;
+	info.si_trapno = 0;
+	force_sig_info(SIGBUS, &info, current);
+}
+
+void spitfire_access_error(struct pt_regs *regs, unsigned long status_encoded, unsigned long afar)
+{
+	unsigned long afsr, tt, udbh, udbl;
+	int tl1;
+
+	afsr = (status_encoded & SFSTAT_AFSR_MASK) >> SFSTAT_AFSR_SHIFT;
+	tt = (status_encoded & SFSTAT_TRAP_TYPE) >> SFSTAT_TRAP_TYPE_SHIFT;
+	tl1 = (status_encoded & SFSTAT_TL_GT_ONE) ? 1 : 0;
+	udbl = (status_encoded & SFSTAT_UDBL_MASK) >> SFSTAT_UDBL_SHIFT;
+	udbh = (status_encoded & SFSTAT_UDBH_MASK) >> SFSTAT_UDBH_SHIFT;
+
+#ifdef CONFIG_PCI
+	if (tt == TRAP_TYPE_DAE &&
+	    pci_poke_in_progress && pci_poke_cpu == smp_processor_id()) {
+		spitfire_clean_and_reenable_l1_caches();
+		spitfire_enable_estate_errors();
+
+		pci_poke_faulted = 1;
+		regs->tnpc = regs->tpc + 4;
+		return;
+	}
+#endif
+
+	if (afsr & SFAFSR_UE)
+		spitfire_ue_log(afsr, afar, udbh, udbl, tt, tl1, regs);
+
+	if (tt == TRAP_TYPE_CEE) {
+		/* Handle the case where we took a CEE trap, but ACK'd
+		 * only the UE state in the UDB error registers.
+		 */
+		if (afsr & SFAFSR_UE) {
+			if (udbh & UDBE_CE) {
+				__asm__ __volatile__(
+					"stxa	%0, [%1] %2\n\t"
+					"membar	#Sync"
+					: /* no outputs */
+					: "r" (udbh & UDBE_CE),
+					  "r" (0x0), "i" (ASI_UDB_ERROR_W));
+			}
+			if (udbl & UDBE_CE) {
+				__asm__ __volatile__(
+					"stxa	%0, [%1] %2\n\t"
+					"membar	#Sync"
+					: /* no outputs */
+					: "r" (udbl & UDBE_CE),
+					  "r" (0x18), "i" (ASI_UDB_ERROR_W));
+			}
+		}
+
+		spitfire_cee_log(afsr, afar, udbh, udbl, tl1, regs);
+	}
 }
 
 int cheetah_pcache_forced_on;

arch/sparc64/kernel/ttable.S

@@ -18,9 +18,10 @@ sparc64_ttable_tl0:
 tl0_resv000:	BOOT_KERNEL BTRAP(0x1) BTRAP(0x2) BTRAP(0x3)
 tl0_resv004:	BTRAP(0x4)  BTRAP(0x5) BTRAP(0x6) BTRAP(0x7)
 tl0_iax:	membar #Sync
-		TRAP_NOSAVE_7INSNS(__do_instruction_access_exception)
+		TRAP_NOSAVE_7INSNS(__spitfire_insn_access_exception)
 tl0_resv009:	BTRAP(0x9)
-tl0_iae:	TRAP(do_iae)
+tl0_iae:	membar #Sync
+		TRAP_NOSAVE_7INSNS(__spitfire_access_error)
 tl0_resv00b:	BTRAP(0xb) BTRAP(0xc) BTRAP(0xd) BTRAP(0xe) BTRAP(0xf)
 tl0_ill:	membar #Sync
 		TRAP_7INSNS(do_illegal_instruction)
@@ -36,9 +37,10 @@ tl0_cwin:	CLEAN_WINDOW
 tl0_div0:	TRAP(do_div0)
 tl0_resv029:	BTRAP(0x29) BTRAP(0x2a) BTRAP(0x2b) BTRAP(0x2c) BTRAP(0x2d) BTRAP(0x2e)
 tl0_resv02f:	BTRAP(0x2f)
-tl0_dax:	TRAP_NOSAVE(__do_data_access_exception)
+tl0_dax:	TRAP_NOSAVE(__spitfire_data_access_exception)
 tl0_resv031:	BTRAP(0x31)
-tl0_dae:	TRAP(do_dae)
+tl0_dae:	membar #Sync
+		TRAP_NOSAVE_7INSNS(__spitfire_access_error)
 tl0_resv033:	BTRAP(0x33)
 tl0_mna:	TRAP_NOSAVE(do_mna)
 tl0_lddfmna:	TRAP_NOSAVE(do_lddfmna)
@@ -73,7 +75,8 @@ tl0_resv05c:	BTRAP(0x5c) BTRAP(0x5d) BTRAP(0x5e) BTRAP(0x5f)
 tl0_ivec:	TRAP_IVEC
 tl0_paw:	TRAP(do_paw)
 tl0_vaw:	TRAP(do_vaw)
-tl0_cee:	TRAP_NOSAVE(cee_trap)
+tl0_cee:	membar #Sync
+		TRAP_NOSAVE_7INSNS(__spitfire_cee_trap)
 tl0_iamiss:
 #include	"itlb_base.S"
 tl0_damiss:
@@ -175,9 +178,10 @@ tl0_resv1f0:	BTRAPS(0x1f0) BTRAPS(0x1f8)
 sparc64_ttable_tl1:
 tl1_resv000:	BOOT_KERNEL    BTRAPTL1(0x1) BTRAPTL1(0x2) BTRAPTL1(0x3)
 tl1_resv004:	BTRAPTL1(0x4)  BTRAPTL1(0x5) BTRAPTL1(0x6) BTRAPTL1(0x7)
-tl1_iax:	TRAP_NOSAVE(__do_instruction_access_exception_tl1)
+tl1_iax:	TRAP_NOSAVE(__spitfire_insn_access_exception_tl1)
 tl1_resv009:	BTRAPTL1(0x9)
-tl1_iae:	TRAPTL1(do_iae_tl1)
+tl1_iae:	membar #Sync
+		TRAP_NOSAVE_7INSNS(__spitfire_access_error)
 tl1_resv00b:	BTRAPTL1(0xb) BTRAPTL1(0xc) BTRAPTL1(0xd) BTRAPTL1(0xe) BTRAPTL1(0xf)
 tl1_ill:	TRAPTL1(do_ill_tl1)
 tl1_privop:	BTRAPTL1(0x11)
@@ -193,9 +197,10 @@ tl1_cwin:	CLEAN_WINDOW
 tl1_div0:	TRAPTL1(do_div0_tl1)
 tl1_resv029:	BTRAPTL1(0x29) BTRAPTL1(0x2a) BTRAPTL1(0x2b) BTRAPTL1(0x2c)
 tl1_resv02d:	BTRAPTL1(0x2d) BTRAPTL1(0x2e) BTRAPTL1(0x2f)
-tl1_dax:	TRAP_NOSAVE(__do_data_access_exception_tl1)
+tl1_dax:	TRAP_NOSAVE(__spitfire_data_access_exception_tl1)
 tl1_resv031:	BTRAPTL1(0x31)
-tl1_dae:	TRAPTL1(do_dae_tl1)
+tl1_dae:	membar #Sync
+		TRAP_NOSAVE_7INSNS(__spitfire_access_error)
 tl1_resv033:	BTRAPTL1(0x33)
 tl1_mna:	TRAP_NOSAVE(do_mna)
 tl1_lddfmna:	TRAPTL1(do_lddfmna_tl1)
@@ -219,8 +224,8 @@ tl1_paw:	TRAPTL1(do_paw_tl1)
 tl1_vaw:	TRAPTL1(do_vaw_tl1)
 
 		/* The grotty trick to save %g1 into current->thread.cee_stuff
-		 * is because when we take this trap we could be interrupting trap
-		 * code already using the trap alternate global registers.
+		 * is because when we take this trap we could be interrupting
+		 * trap code already using the trap alternate global registers.
 		 *
 		 * We cross our fingers and pray that this store/load does
 		 * not cause yet another CEE trap.

arch/sparc64/kernel/unaligned.c

@@ -349,9 +349,9 @@ int handle_popc(u32 insn, struct pt_regs *regs)
 
 extern void do_fpother(struct pt_regs *regs);
 extern void do_privact(struct pt_regs *regs);
-extern void data_access_exception(struct pt_regs *regs,
-				  unsigned long sfsr,
-				  unsigned long sfar);
+extern void spitfire_data_access_exception(struct pt_regs *regs,
+					   unsigned long sfsr,
+					   unsigned long sfar);
 
 int handle_ldf_stq(u32 insn, struct pt_regs *regs)
 {
@@ -394,14 +394,14 @@ int handle_ldf_stq(u32 insn, struct pt_regs *regs)
 				break;
 			}
 		default:
-			data_access_exception(regs, 0, addr);
+			spitfire_data_access_exception(regs, 0, addr);
 			return 1;
 		}
 		if (put_user (first >> 32, (u32 __user *)addr) ||
 		    __put_user ((u32)first, (u32 __user *)(addr + 4)) ||
 		    __put_user (second >> 32, (u32 __user *)(addr + 8)) ||
 		    __put_user ((u32)second, (u32 __user *)(addr + 12))) {
-		    	data_access_exception(regs, 0, addr);
+		    	spitfire_data_access_exception(regs, 0, addr);
 		    	return 1;
 		}
 	} else {
@@ -414,7 +414,7 @@ int handle_ldf_stq(u32 insn, struct pt_regs *regs)
 			do_privact(regs);
 			return 1;
 		} else if (asi > ASI_SNFL) {
-			data_access_exception(regs, 0, addr);
+			spitfire_data_access_exception(regs, 0, addr);
 			return 1;
 		}
 		switch (insn & 0x180000) {
@@ -431,7 +431,7 @@ int handle_ldf_stq(u32 insn, struct pt_regs *regs)
 				err |= __get_user (data[i], (u32 __user *)(addr + 4*i));
 		}
 		if (err && !(asi & 0x2 /* NF */)) {
-			data_access_exception(regs, 0, addr);
+			spitfire_data_access_exception(regs, 0, addr);
 			return 1;
 		}
 		if (asi & 0x8) /* Little */ {
@@ -534,7 +534,7 @@ void handle_lddfmna(struct pt_regs *regs, unsigned long sfar, unsigned long sfsr
 		*(u64 *)(f->regs + freg) = value;
 		current_thread_info()->fpsaved[0] |= flag;
 	} else {
-daex:		data_access_exception(regs, sfsr, sfar);
+daex:		spitfire_data_access_exception(regs, sfsr, sfar);
 		return;
 	}
 	advance(regs);
@@ -578,7 +578,7 @@ void handle_stdfmna(struct pt_regs *regs, unsigned long sfar, unsigned long sfsr
 		    __put_user ((u32)value, (u32 __user *)(sfar + 4)))
 			goto daex;
 	} else {
-daex:		data_access_exception(regs, sfsr, sfar);
+daex:		spitfire_data_access_exception(regs, sfsr, sfar);
 		return;
 	}
 	advance(regs);

arch/sparc64/kernel/winfixup.S

@@ -318,7 +318,7 @@ fill_fixup_dax:
 	 nop
 	rdpr		%pstate, %l1			! Prepare to change globals.
 	mov		%g4, %o1			! Setup args for
-	mov		%g5, %o2			! final call to data_access_exception.
+	mov		%g5, %o2			! final call to spitfire_data_access_exception.
 	andn		%l1, PSTATE_MM, %l1		! We want to be in RMO
 
 	mov		%g6, %o7			! Stash away current.
@@ -330,7 +330,7 @@ fill_fixup_dax:
 	mov		TSB_REG, %g1
 	ldxa		[%g1] ASI_IMMU, %g5
 #endif
-	call		data_access_exception
+	call		spitfire_data_access_exception
 	 add		%sp, PTREGS_OFF, %o0
 
 	b,pt		%xcc, rtrap
@@ -391,7 +391,7 @@ window_dax_from_user_common:
 109:	 or		%g7, %lo(109b), %g7
 	mov		%l4, %o1
 	mov		%l5, %o2
-	call		data_access_exception
+	call		spitfire_data_access_exception
 	 add		%sp, PTREGS_OFF, %o0
 	ba,pt		%xcc, rtrap
 	 clr		%l6

arch/sparc64/lib/Makefile

@@ -12,7 +12,7 @@ lib-y := PeeCeeI.o copy_page.o clear_page.o strlen.o strncmp.o \
 	 U1memcpy.o U1copy_from_user.o U1copy_to_user.o \
 	 U3memcpy.o U3copy_from_user.o U3copy_to_user.o U3patch.o \
 	 copy_in_user.o user_fixup.o memmove.o \
-	 mcount.o ipcsum.o rwsem.o xor.o find_bit.o delay.o
+	 mcount.o ipcsum.o rwsem.o xor.o find_bit.o delay.o mb.o
 
 lib-$(CONFIG_DEBUG_SPINLOCK) += debuglocks.o
 lib-$(CONFIG_HAVE_DEC_LOCK) += dec_and_lock.o

arch/sparc64/lib/debuglocks.c

@@ -12,8 +12,6 @@
 
 #ifdef CONFIG_SMP
 
-#define GET_CALLER(PC) __asm__ __volatile__("mov %%i7, %0" : "=r" (PC))
-
 static inline void show (char *str, spinlock_t *lock, unsigned long caller)
 {
 	int cpu = smp_processor_id();
@@ -51,20 +49,19 @@ static inline void show_write (char *str, rwlock_t *lock, unsigned long caller)
 #undef INIT_STUCK
 #define INIT_STUCK 100000000
 
-void _do_spin_lock(spinlock_t *lock, char *str)
+void _do_spin_lock(spinlock_t *lock, char *str, unsigned long caller)
 {
-	unsigned long caller, val;
+	unsigned long val;
 	int stuck = INIT_STUCK;
 	int cpu = get_cpu();
 	int shown = 0;
 
-	GET_CALLER(caller);
 again:
 	__asm__ __volatile__("ldstub [%1], %0"
 			     : "=r" (val)
 			     : "r" (&(lock->lock))
 			     : "memory");
-	membar("#StoreLoad | #StoreStore");
+	membar_storeload_storestore();
 	if (val) {
 		while (lock->lock) {
 			if (!--stuck) {
@@ -72,7 +69,7 @@ again:
 					show(str, lock, caller);
 				stuck = INIT_STUCK;
 			}
-			membar("#LoadLoad");
+			rmb();
 		}
 		goto again;
 	}
@@ -84,17 +81,16 @@ again:
 	put_cpu();
 }
 
-int _do_spin_trylock(spinlock_t *lock)
+int _do_spin_trylock(spinlock_t *lock, unsigned long caller)
 {
-	unsigned long val, caller;
+	unsigned long val;
 	int cpu = get_cpu();
 
-	GET_CALLER(caller);
 	__asm__ __volatile__("ldstub [%1], %0"
 			     : "=r" (val)
 			     : "r" (&(lock->lock))
 			     : "memory");
-	membar("#StoreLoad | #StoreStore");
+	membar_storeload_storestore();
 	if (!val) {
 		lock->owner_pc = ((unsigned int)caller);
 		lock->owner_cpu = cpu;
@@ -111,21 +107,20 @@ void _do_spin_unlock(spinlock_t *lock)
 {
 	lock->owner_pc = 0;
 	lock->owner_cpu = NO_PROC_ID;
-	membar("#StoreStore | #LoadStore");
+	membar_storestore_loadstore();
 	lock->lock = 0;
 	current->thread.smp_lock_count--;
 }
 
 /* Keep INIT_STUCK the same... */
 
-void _do_read_lock (rwlock_t *rw, char *str)
+void _do_read_lock(rwlock_t *rw, char *str, unsigned long caller)
 {
-	unsigned long caller, val;
+	unsigned long val;
 	int stuck = INIT_STUCK;
 	int cpu = get_cpu();
 	int shown = 0;
 
-	GET_CALLER(caller);
 wlock_again:
 	/* Wait for any writer to go away.  */
 	while (((long)(rw->lock)) < 0) {
@@ -134,7 +129,7 @@ wlock_again:
 				show_read(str, rw, caller);
 			stuck = INIT_STUCK;
 		}
-		membar("#LoadLoad");
+		rmb();
 	}
 	/* Try once to increment the counter.  */
 	__asm__ __volatile__(
@@ -147,7 +142,7 @@ wlock_again:
 "2:"	: "=r" (val)
 	: "0" (&(rw->lock))
 	: "g1", "g7", "memory");
-	membar("#StoreLoad | #StoreStore");
+	membar_storeload_storestore();
 	if (val)
 		goto wlock_again;
 	rw->reader_pc[cpu] = ((unsigned int)caller);
@@ -157,15 +152,13 @@ wlock_again:
 	put_cpu();
 }
 
-void _do_read_unlock (rwlock_t *rw, char *str)
+void _do_read_unlock(rwlock_t *rw, char *str, unsigned long caller)
 {
-	unsigned long caller, val;
+	unsigned long val;
 	int stuck = INIT_STUCK;
 	int cpu = get_cpu();
 	int shown = 0;
 
-	GET_CALLER(caller);
-
 	/* Drop our identity _first_. */
 	rw->reader_pc[cpu] = 0;
 	current->thread.smp_lock_count--;
@@ -193,14 +186,13 @@ runlock_again:
 	put_cpu();
 }
 
-void _do_write_lock (rwlock_t *rw, char *str)
+void _do_write_lock(rwlock_t *rw, char *str, unsigned long caller)
 {
-	unsigned long caller, val;
+	unsigned long val;
 	int stuck = INIT_STUCK;
 	int cpu = get_cpu();
 	int shown = 0;
 
-	GET_CALLER(caller);
 wlock_again:
 	/* Spin while there is another writer. */
 	while (((long)rw->lock) < 0) {
@@ -209,7 +201,7 @@ wlock_again:
 				show_write(str, rw, caller);
 			stuck = INIT_STUCK;
 		}
-		membar("#LoadLoad");
+		rmb();
 	}
 
 	/* Try to acuire the write bit.  */
@@ -264,7 +256,7 @@ wlock_again:
 					show_write(str, rw, caller);
 				stuck = INIT_STUCK;
 			}
-			membar("#LoadLoad");
+			rmb();
 		}
 		goto wlock_again;
 	}
@@ -278,14 +270,12 @@ wlock_again:
 	put_cpu();
 }
 
-void _do_write_unlock(rwlock_t *rw)
+void _do_write_unlock(rwlock_t *rw, unsigned long caller)
 {
-	unsigned long caller, val;
+	unsigned long val;
 	int stuck = INIT_STUCK;
 	int shown = 0;
 
-	GET_CALLER(caller);
-
 	/* Drop our identity _first_ */
 	rw->writer_pc = 0;
 	rw->writer_cpu = NO_PROC_ID;
@@ -313,13 +303,11 @@ wlock_again:
 	}
 }
 
-int _do_write_trylock (rwlock_t *rw, char *str)
+int _do_write_trylock(rwlock_t *rw, char *str, unsigned long caller)
 {
-	unsigned long caller, val;
+	unsigned long val;
 	int cpu = get_cpu();
 
-	GET_CALLER(caller);
-
 	/* Try to acuire the write bit.  */
 	__asm__ __volatile__(
 "	mov	1, %%g3\n"

arch/sparc64/lib/mb.S

@@ -0,0 +1,73 @@
+/* mb.S: Out of line memory barriers.
+ *
+ * Copyright (C) 2005 David S. Miller (davem@davemloft.net)
+ */
+
+	/* These are here in an effort to more fully work around
+	 * Spitfire Errata #51.  Essentially, if a memory barrier
+	 * occurs soon after a mispredicted branch, the chip can stop
+	 * executing instructions until a trap occurs.  Therefore, if
+	 * interrupts are disabled, the chip can hang forever.
+	 *
+	 * It used to be believed that the memory barrier had to be
+	 * right in the delay slot, but a case has been traced
+	 * recently wherein the memory barrier was one instruction
+	 * after the branch delay slot and the chip still hung.  The
+	 * offending sequence was the following in sym_wakeup_done()
+	 * of the sym53c8xx_2 driver:
+	 *
+	 *	call	sym_ccb_from_dsa, 0
+	 *	 movge	%icc, 0, %l0
+	 *	brz,pn	%o0, .LL1303
+	 *	 mov	%o0, %l2
+	 *	membar	#LoadLoad
+	 *
+	 * The branch has to be mispredicted for the bug to occur.
+	 * Therefore, we put the memory barrier explicitly into a
+	 * "branch always, predicted taken" delay slot to avoid the
+	 * problem case.
+	 */
+
+	.text
+
+99:	retl
+	 nop
+
+	.globl	mb
+mb:	ba,pt	%xcc, 99b
+	 membar	#LoadLoad | #LoadStore | #StoreStore | #StoreLoad
+	.size	mb, .-mb
+
+	.globl	rmb
+rmb:	ba,pt	%xcc, 99b
+	 membar	#LoadLoad
+	.size	rmb, .-rmb
+
+	.globl	wmb
+wmb:	ba,pt	%xcc, 99b
+	 membar	#StoreStore
+	.size	wmb, .-wmb
+
+	.globl	membar_storeload
+membar_storeload:
+	ba,pt	%xcc, 99b
+	 membar	#StoreLoad
+	.size	membar_storeload, .-membar_storeload
+
+	.globl	membar_storeload_storestore
+membar_storeload_storestore:
+	ba,pt	%xcc, 99b
+	 membar	#StoreLoad | #StoreStore
+	.size	membar_storeload_storestore, .-membar_storeload_storestore
+
+	.globl	membar_storeload_loadload
+membar_storeload_loadload:
+	ba,pt	%xcc, 99b
+	 membar	#StoreLoad | #LoadLoad
+	.size	membar_storeload_loadload, .-membar_storeload_loadload
+
+	.globl	membar_storestore_loadstore
+membar_storestore_loadstore:
+	ba,pt	%xcc, 99b
+	 membar	#StoreStore | #LoadStore
+	.size	membar_storestore_loadstore, .-membar_storestore_loadstore

arch/sparc64/solaris/misc.c

@@ -737,7 +737,8 @@ MODULE_LICENSE("GPL");
 extern u32 tl0_solaris[8];
 #define update_ttable(x) 										\
 	tl0_solaris[3] = (((long)(x) - (long)tl0_solaris - 3) >> 2) | 0x40000000;			\
-	__asm__ __volatile__ ("membar #StoreStore; flush %0" : : "r" (&tl0_solaris[3]))
+	wmb();		\
+	__asm__ __volatile__ ("flush %0" : : "r" (&tl0_solaris[3]))
 #else
 #endif	
 
@@ -761,7 +762,8 @@ int init_module(void)
 	entry64_personality_patch |=
 		(offsetof(struct task_struct, personality) +
 		 (sizeof(unsigned long) - 1));
-	__asm__ __volatile__("membar #StoreStore; flush %0"
+	wmb();
+	__asm__ __volatile__("flush %0"
 			     : : "r" (&entry64_personality_patch));
 	return 0;
 }

arch/um/kernel/signal_kern.c

@@ -87,12 +87,12 @@ static int handle_signal(struct pt_regs *regs, unsigned long signr,
 		recalc_sigpending();
 		spin_unlock_irq(&current->sighand->siglock);
 		force_sigsegv(signr, current);
-	}
-	else if(!(ka->sa.sa_flags & SA_NODEFER)){
+	} else {
 		spin_lock_irq(&current->sighand->siglock);
 		sigorsets(&current->blocked, &current->blocked, 
 			  &ka->sa.sa_mask);
-		sigaddset(&current->blocked, signr);
+		 if(!(ka->sa.sa_flags & SA_NODEFER))
+			sigaddset(&current->blocked, signr);
 		recalc_sigpending();
 		spin_unlock_irq(&current->sighand->siglock);
 	}

arch/v850/kernel/signal.c

@@ -462,13 +462,12 @@ handle_signal(unsigned long sig, siginfo_t *info, struct k_sigaction *ka,
 	else
 		setup_frame(sig, ka, oldset, regs);
 
-	if (!(ka->sa.sa_flags & SA_NODEFER)) {
-		spin_lock_irq(&current->sighand->siglock);
-		sigorsets(&current->blocked,&current->blocked,&ka->sa.sa_mask);
+	spin_lock_irq(&current->sighand->siglock);
+	sigorsets(&current->blocked,&current->blocked,&ka->sa.sa_mask);
+	if (!(ka->sa.sa_flags & SA_NODEFER))
 		sigaddset(&current->blocked,sig);
-		recalc_sigpending();
-		spin_unlock_irq(&current->sighand->siglock);
-	}
+	recalc_sigpending();
+	spin_unlock_irq(&current->sighand->siglock);
 }
 
 /*

arch/x86_64/kernel/signal.c

@@ -394,10 +394,11 @@ handle_signal(unsigned long sig, siginfo_t *info, struct k_sigaction *ka,
 #endif
 	ret = setup_rt_frame(sig, ka, info, oldset, regs);
 
-	if (ret && !(ka->sa.sa_flags & SA_NODEFER)) {
+	if (ret) {
 		spin_lock_irq(&current->sighand->siglock);
 		sigorsets(&current->blocked,&current->blocked,&ka->sa.sa_mask);
-		sigaddset(&current->blocked,sig);
+		if (!(ka->sa.sa_flags & SA_NODEFER))
+			sigaddset(&current->blocked,sig);
 		recalc_sigpending();
 		spin_unlock_irq(&current->sighand->siglock);
 	}

arch/xtensa/kernel/signal.c

@@ -702,12 +702,11 @@ int do_signal(struct pt_regs *regs, sigset_t *oldset)
 	if (ka.sa.sa_flags & SA_ONESHOT)
 		ka.sa.sa_handler = SIG_DFL;
 
-	if (!(ka.sa.sa_flags & SA_NODEFER)) {
-		spin_lock_irq(&current->sighand->siglock);
-		sigorsets(&current->blocked, &current->blocked, &ka.sa.sa_mask);
+	spin_lock_irq(&current->sighand->siglock);
+	sigorsets(&current->blocked, &current->blocked, &ka.sa.sa_mask);
+	if (!(ka.sa.sa_flags & SA_NODEFER))
 		sigaddset(&current->blocked, signr);
-		recalc_sigpending();
-		spin_unlock_irq(&current->sighand->siglock);
-	}
+	recalc_sigpending();
+	spin_unlock_irq(&current->sighand->siglock);
 	return 1;
 }

drivers/Kconfig

@@ -48,6 +48,8 @@ source "drivers/hwmon/Kconfig"
 
 source "drivers/misc/Kconfig"
 
+source "drivers/mfd/Kconfig"
+
 source "drivers/media/Kconfig"
 
 source "drivers/video/Kconfig"

drivers/Makefile

@@ -26,7 +26,7 @@ obj-$(CONFIG_FB_INTEL)          += video/intelfb/
 obj-$(CONFIG_SERIO)		+= input/serio/
 obj-y				+= serial/
 obj-$(CONFIG_PARPORT)		+= parport/
-obj-y				+= base/ block/ misc/ net/ media/
+obj-y				+= base/ block/ misc/ mfd/ net/ media/
 obj-$(CONFIG_NUBUS)		+= nubus/
 obj-$(CONFIG_ATM)		+= atm/
 obj-$(CONFIG_PPC_PMAC)		+= macintosh/

drivers/infiniband/core/Makefile

@@ -1,5 +1,3 @@
-EXTRA_CFLAGS += -Idrivers/infiniband/include
-
 obj-$(CONFIG_INFINIBAND) +=		ib_core.o ib_mad.o ib_sa.o \
 					ib_cm.o ib_umad.o ib_ucm.o
 obj-$(CONFIG_INFINIBAND_USER_VERBS) +=	ib_uverbs.o

drivers/infiniband/core/agent.c

@@ -1,9 +1,10 @@
 /*
- * Copyright (c) 2004 Mellanox Technologies Ltd.  All rights reserved.
- * Copyright (c) 2004 Infinicon Corporation.  All rights reserved.
- * Copyright (c) 2004 Intel Corporation.  All rights reserved.
- * Copyright (c) 2004 Topspin Corporation.  All rights reserved.
- * Copyright (c) 2004 Voltaire Corporation.  All rights reserved.
+ * Copyright (c) 2004, 2005 Mellanox Technologies Ltd.  All rights reserved.
+ * Copyright (c) 2004, 2005 Infinicon Corporation.  All rights reserved.
+ * Copyright (c) 2004, 2005 Intel Corporation.  All rights reserved.
+ * Copyright (c) 2004, 2005 Topspin Corporation.  All rights reserved.
+ * Copyright (c) 2004, 2005 Voltaire Corporation.  All rights reserved.
+ * Copyright (c) 2005 Sun Microsystems, Inc. All rights reserved.
  *
  * This software is available to you under a choice of one of two
  * licenses.  You may choose to be licensed under the terms of the GNU
@@ -40,7 +41,7 @@
 
 #include <asm/bug.h>
 
-#include <ib_smi.h>
+#include <rdma/ib_smi.h>
 
 #include "smi.h"
 #include "agent_priv.h"

drivers/infiniband/core/agent_priv.h

@@ -1,9 +1,9 @@
 /*
- * Copyright (c) 2004 Mellanox Technologies Ltd.  All rights reserved.
- * Copyright (c) 2004 Infinicon Corporation.  All rights reserved.
- * Copyright (c) 2004 Intel Corporation.  All rights reserved.
- * Copyright (c) 2004 Topspin Corporation.  All rights reserved.
- * Copyright (c) 2004 Voltaire Corporation.  All rights reserved.
+ * Copyright (c) 2004, 2005 Mellanox Technologies Ltd.  All rights reserved.
+ * Copyright (c) 2004, 2005 Infinicon Corporation.  All rights reserved.
+ * Copyright (c) 2004, 2005 Intel Corporation.  All rights reserved.
+ * Copyright (c) 2004, 2005 Topspin Corporation.  All rights reserved.
+ * Copyright (c) 2004, 2005 Voltaire Corporation.  All rights reserved.
  *
  * This software is available to you under a choice of one of two
  * licenses.  You may choose to be licensed under the terms of the GNU

drivers/infiniband/core/cache.c

@@ -1,5 +1,8 @@
 /*
  * Copyright (c) 2004 Topspin Communications.  All rights reserved.
+ * Copyright (c) 2005 Intel Corporation. All rights reserved.
+ * Copyright (c) 2005 Sun Microsystems, Inc. All rights reserved.
+ * Copyright (c) 2005 Voltaire, Inc. All rights reserved.
  *
  * This software is available to you under a choice of one of two
  * licenses.  You may choose to be licensed under the terms of the GNU
@@ -32,12 +35,11 @@
  * $Id: cache.c 1349 2004-12-16 21:09:43Z roland $
  */
 
-#include <linux/version.h>
 #include <linux/module.h>
 #include <linux/errno.h>
 #include <linux/slab.h>
 
-#include <ib_cache.h>
+#include <rdma/ib_cache.h>
 
 #include "core_priv.h"
 

drivers/infiniband/core/cm.c

@@ -43,8 +43,8 @@
 #include <linux/spinlock.h>
 #include <linux/workqueue.h>
 
-#include <ib_cache.h>
-#include <ib_cm.h>
+#include <rdma/ib_cache.h>
+#include <rdma/ib_cm.h>
 #include "cm_msgs.h"
 
 MODULE_AUTHOR("Sean Hefty");
@@ -83,7 +83,7 @@ struct cm_port {
 struct cm_device {
 	struct list_head list;
 	struct ib_device *device;
-	u64 ca_guid;
+	__be64 ca_guid;
 	struct cm_port port[0];
 };
 
@@ -100,8 +100,8 @@ struct cm_work {
 	struct list_head list;
 	struct cm_port *port;
 	struct ib_mad_recv_wc *mad_recv_wc;	/* Received MADs */
-	u32 local_id;				/* Established / timewait */
-	u32 remote_id;
+	__be32 local_id;			/* Established / timewait */
+	__be32 remote_id;
 	struct ib_cm_event cm_event;
 	struct ib_sa_path_rec path[0];
 };
@@ -110,8 +110,8 @@ struct cm_timewait_info {
 	struct cm_work work;			/* Must be first. */
 	struct rb_node remote_qp_node;
 	struct rb_node remote_id_node;
-	u64 remote_ca_guid;
-	u32 remote_qpn;
+	__be64 remote_ca_guid;
+	__be32 remote_qpn;
 	u8 inserted_remote_qp;
 	u8 inserted_remote_id;
 };
@@ -132,11 +132,11 @@ struct cm_id_private {
 	struct cm_av alt_av;
 
 	void *private_data;
-	u64 tid;
-	u32 local_qpn;
-	u32 remote_qpn;
-	u32 sq_psn;
-	u32 rq_psn;
+	__be64 tid;
+	__be32 local_qpn;
+	__be32 remote_qpn;
+	__be32 sq_psn;
+	__be32 rq_psn;
 	int timeout_ms;
 	enum ib_mtu path_mtu;
 	u8 private_data_len;
@@ -253,7 +253,7 @@ static void cm_set_ah_attr(struct ib_ah_attr *ah_attr, u8 port_num,
 			   u16 dlid, u8 sl, u16 src_path_bits)
 {
 	memset(ah_attr, 0, sizeof ah_attr);
-	ah_attr->dlid = be16_to_cpu(dlid);
+	ah_attr->dlid = dlid;
 	ah_attr->sl = sl;
 	ah_attr->src_path_bits = src_path_bits;
 	ah_attr->port_num = port_num;
@@ -264,7 +264,7 @@ static void cm_init_av_for_response(struct cm_port *port,
 {
 	av->port = port;
 	av->pkey_index = wc->pkey_index;
-	cm_set_ah_attr(&av->ah_attr, port->port_num, cpu_to_be16(wc->slid),
+	cm_set_ah_attr(&av->ah_attr, port->port_num, wc->slid,
 		       wc->sl, wc->dlid_path_bits);
 }
 
@@ -295,8 +295,9 @@ static int cm_init_av_by_path(struct ib_sa_path_rec *path, struct cm_av *av)
 		return ret;
 
 	av->port = port;
-	cm_set_ah_attr(&av->ah_attr, av->port->port_num, path->dlid,
-		       path->sl, path->slid & 0x7F);
+	cm_set_ah_attr(&av->ah_attr, av->port->port_num,
+		       be16_to_cpu(path->dlid), path->sl,
+		       be16_to_cpu(path->slid) & 0x7F);
 	av->packet_life_time = path->packet_life_time;
 	return 0;
 }
@@ -309,26 +310,26 @@ static int cm_alloc_id(struct cm_id_private *cm_id_priv)
 	do {
 		spin_lock_irqsave(&cm.lock, flags);
 		ret = idr_get_new_above(&cm.local_id_table, cm_id_priv, 1,
-					(int *) &cm_id_priv->id.local_id);
+					(__force int *) &cm_id_priv->id.local_id);
 		spin_unlock_irqrestore(&cm.lock, flags);
 	} while( (ret == -EAGAIN) && idr_pre_get(&cm.local_id_table, GFP_KERNEL) );
 	return ret;
 }
 
-static void cm_free_id(u32 local_id)
+static void cm_free_id(__be32 local_id)
 {
 	unsigned long flags;
 
 	spin_lock_irqsave(&cm.lock, flags);
-	idr_remove(&cm.local_id_table, (int) local_id);
+	idr_remove(&cm.local_id_table, (__force int) local_id);
 	spin_unlock_irqrestore(&cm.lock, flags);
 }
 
-static struct cm_id_private * cm_get_id(u32 local_id, u32 remote_id)
+static struct cm_id_private * cm_get_id(__be32 local_id, __be32 remote_id)
 {
 	struct cm_id_private *cm_id_priv;
 
-	cm_id_priv = idr_find(&cm.local_id_table, (int) local_id);
+	cm_id_priv = idr_find(&cm.local_id_table, (__force int) local_id);
 	if (cm_id_priv) {
 		if (cm_id_priv->id.remote_id == remote_id)
 			atomic_inc(&cm_id_priv->refcount);
@@ -339,7 +340,7 @@ static struct cm_id_private * cm_get_id(u32 local_id, u32 remote_id)
 	return cm_id_priv;
 }
 
-static struct cm_id_private * cm_acquire_id(u32 local_id, u32 remote_id)
+static struct cm_id_private * cm_acquire_id(__be32 local_id, __be32 remote_id)
 {
 	struct cm_id_private *cm_id_priv;
 	unsigned long flags;
@@ -356,8 +357,8 @@ static struct cm_id_private * cm_insert_listen(struct cm_id_private *cm_id_priv)
 	struct rb_node **link = &cm.listen_service_table.rb_node;
 	struct rb_node *parent = NULL;
 	struct cm_id_private *cur_cm_id_priv;
-	u64 service_id = cm_id_priv->id.service_id;
-	u64 service_mask = cm_id_priv->id.service_mask;
+	__be64 service_id = cm_id_priv->id.service_id;
+	__be64 service_mask = cm_id_priv->id.service_mask;
 
 	while (*link) {
 		parent = *link;
@@ -376,7 +377,7 @@ static struct cm_id_private * cm_insert_listen(struct cm_id_private *cm_id_priv)
 	return NULL;
 }
 
-static struct cm_id_private * cm_find_listen(u64 service_id)
+static struct cm_id_private * cm_find_listen(__be64 service_id)
 {
 	struct rb_node *node = cm.listen_service_table.rb_node;
 	struct cm_id_private *cm_id_priv;
@@ -400,8 +401,8 @@ static struct cm_timewait_info * cm_insert_remote_id(struct cm_timewait_info
 	struct rb_node **link = &cm.remote_id_table.rb_node;
 	struct rb_node *parent = NULL;
 	struct cm_timewait_info *cur_timewait_info;
-	u64 remote_ca_guid = timewait_info->remote_ca_guid;
-	u32 remote_id = timewait_info->work.remote_id;
+	__be64 remote_ca_guid = timewait_info->remote_ca_guid;
+	__be32 remote_id = timewait_info->work.remote_id;
 
 	while (*link) {
 		parent = *link;
@@ -424,8 +425,8 @@ static struct cm_timewait_info * cm_insert_remote_id(struct cm_timewait_info
 	return NULL;
 }
 
-static struct cm_timewait_info * cm_find_remote_id(u64 remote_ca_guid,
-						   u32 remote_id)
+static struct cm_timewait_info * cm_find_remote_id(__be64 remote_ca_guid,
+						   __be32 remote_id)
 {
 	struct rb_node *node = cm.remote_id_table.rb_node;
 	struct cm_timewait_info *timewait_info;
@@ -453,8 +454,8 @@ static struct cm_timewait_info * cm_insert_remote_qpn(struct cm_timewait_info
 	struct rb_node **link = &cm.remote_qp_table.rb_node;
 	struct rb_node *parent = NULL;
 	struct cm_timewait_info *cur_timewait_info;
-	u64 remote_ca_guid = timewait_info->remote_ca_guid;
-	u32 remote_qpn = timewait_info->remote_qpn;
+	__be64 remote_ca_guid = timewait_info->remote_ca_guid;
+	__be32 remote_qpn = timewait_info->remote_qpn;
 
 	while (*link) {
 		parent = *link;
@@ -484,7 +485,7 @@ static struct cm_id_private * cm_insert_remote_sidr(struct cm_id_private
 	struct rb_node *parent = NULL;
 	struct cm_id_private *cur_cm_id_priv;
 	union ib_gid *port_gid = &cm_id_priv->av.dgid;
-	u32 remote_id = cm_id_priv->id.remote_id;
+	__be32 remote_id = cm_id_priv->id.remote_id;
 
 	while (*link) {
 		parent = *link;
@@ -598,7 +599,7 @@ static void cm_cleanup_timewait(struct cm_timewait_info *timewait_info)
 	spin_unlock_irqrestore(&cm.lock, flags);
 }
 
-static struct cm_timewait_info * cm_create_timewait_info(u32 local_id)
+static struct cm_timewait_info * cm_create_timewait_info(__be32 local_id)
 {
 	struct cm_timewait_info *timewait_info;
 
@@ -715,14 +716,15 @@ retest:
 EXPORT_SYMBOL(ib_destroy_cm_id);
 
 int ib_cm_listen(struct ib_cm_id *cm_id,
-		 u64 service_id,
-		 u64 service_mask)
+		 __be64 service_id,
+		 __be64 service_mask)
 {
 	struct cm_id_private *cm_id_priv, *cur_cm_id_priv;
 	unsigned long flags;
 	int ret = 0;
 
-	service_mask = service_mask ? service_mask : ~0ULL;
+	service_mask = service_mask ? service_mask :
+		       __constant_cpu_to_be64(~0ULL);
 	service_id &= service_mask;
 	if ((service_id & IB_SERVICE_ID_AGN_MASK) == IB_CM_ASSIGN_SERVICE_ID &&
 	    (service_id != IB_CM_ASSIGN_SERVICE_ID))
@@ -735,8 +737,8 @@ int ib_cm_listen(struct ib_cm_id *cm_id,
 
 	spin_lock_irqsave(&cm.lock, flags);
 	if (service_id == IB_CM_ASSIGN_SERVICE_ID) {
-		cm_id->service_id = __cpu_to_be64(cm.listen_service_id++);
-		cm_id->service_mask = ~0ULL;
+		cm_id->service_id = cpu_to_be64(cm.listen_service_id++);
+		cm_id->service_mask = __constant_cpu_to_be64(~0ULL);
 	} else {
 		cm_id->service_id = service_id;
 		cm_id->service_mask = service_mask;
@@ -752,18 +754,19 @@ int ib_cm_listen(struct ib_cm_id *cm_id,
 }
 EXPORT_SYMBOL(ib_cm_listen);
 
-static u64 cm_form_tid(struct cm_id_private *cm_id_priv,
-		       enum cm_msg_sequence msg_seq)
+static __be64 cm_form_tid(struct cm_id_private *cm_id_priv,
+			  enum cm_msg_sequence msg_seq)
 {
 	u64 hi_tid, low_tid;
 
 	hi_tid   = ((u64) cm_id_priv->av.port->mad_agent->hi_tid) << 32;
-	low_tid  = (u64) (cm_id_priv->id.local_id | (msg_seq << 30));
+	low_tid  = (u64) ((__force u32)cm_id_priv->id.local_id |
+			  (msg_seq << 30));
 	return cpu_to_be64(hi_tid | low_tid);
 }
 
 static void cm_format_mad_hdr(struct ib_mad_hdr *hdr,
-			      enum cm_msg_attr_id attr_id, u64 tid)
+			      __be16 attr_id, __be64 tid)
 {
 	hdr->base_version  = IB_MGMT_BASE_VERSION;
 	hdr->mgmt_class	   = IB_MGMT_CLASS_CM;
@@ -896,7 +899,7 @@ int ib_send_cm_req(struct ib_cm_id *cm_id,
 			goto error1;
 	}
 	cm_id->service_id = param->service_id;
-	cm_id->service_mask = ~0ULL;
+	cm_id->service_mask = __constant_cpu_to_be64(~0ULL);
 	cm_id_priv->timeout_ms = cm_convert_to_ms(
 				    param->primary_path->packet_life_time) * 2 +
 				 cm_convert_to_ms(
@@ -963,7 +966,7 @@ static int cm_issue_rej(struct cm_port *port,
 	rej_msg->remote_comm_id = rcv_msg->local_comm_id;
 	rej_msg->local_comm_id = rcv_msg->remote_comm_id;
 	cm_rej_set_msg_rejected(rej_msg, msg_rejected);
-	rej_msg->reason = reason;
+	rej_msg->reason = cpu_to_be16(reason);
 
 	if (ari && ari_length) {
 		cm_rej_set_reject_info_len(rej_msg, ari_length);
@@ -977,8 +980,8 @@ static int cm_issue_rej(struct cm_port *port,
 	return ret;
 }
 
-static inline int cm_is_active_peer(u64 local_ca_guid, u64 remote_ca_guid,
-				    u32 local_qpn, u32 remote_qpn)
+static inline int cm_is_active_peer(__be64 local_ca_guid, __be64 remote_ca_guid,
+				    __be32 local_qpn, __be32 remote_qpn)
 {
 	return (be64_to_cpu(local_ca_guid) > be64_to_cpu(remote_ca_guid) ||
 		((local_ca_guid == remote_ca_guid) &&
@@ -1137,7 +1140,7 @@ static void cm_format_rej(struct cm_rej_msg *rej_msg,
 		break;
 	}
 
-	rej_msg->reason = reason;
+	rej_msg->reason = cpu_to_be16(reason);
 	if (ari && ari_length) {
 		cm_rej_set_reject_info_len(rej_msg, ari_length);
 		memcpy(rej_msg->ari, ari, ari_length);
@@ -1276,7 +1279,7 @@ static int cm_req_handler(struct cm_work *work)
 	cm_id_priv->id.cm_handler = listen_cm_id_priv->id.cm_handler;
 	cm_id_priv->id.context = listen_cm_id_priv->id.context;
 	cm_id_priv->id.service_id = req_msg->service_id;
-	cm_id_priv->id.service_mask = ~0ULL;
+	cm_id_priv->id.service_mask = __constant_cpu_to_be64(~0ULL);
 
 	cm_format_paths_from_req(req_msg, &work->path[0], &work->path[1]);
 	ret = cm_init_av_by_path(&work->path[0], &cm_id_priv->av);
@@ -1969,7 +1972,7 @@ static void cm_format_rej_event(struct cm_work *work)
 	param = &work->cm_event.param.rej_rcvd;
 	param->ari = rej_msg->ari;
 	param->ari_length = cm_rej_get_reject_info_len(rej_msg);
-	param->reason = rej_msg->reason;
+	param->reason = __be16_to_cpu(rej_msg->reason);
 	work->cm_event.private_data = &rej_msg->private_data;
 }
 
@@ -1978,20 +1981,20 @@ static struct cm_id_private * cm_acquire_rejected_id(struct cm_rej_msg *rej_msg)
 	struct cm_timewait_info *timewait_info;
 	struct cm_id_private *cm_id_priv;
 	unsigned long flags;
-	u32 remote_id;
+	__be32 remote_id;
 
 	remote_id = rej_msg->local_comm_id;
 
-	if (rej_msg->reason == IB_CM_REJ_TIMEOUT) {
+	if (__be16_to_cpu(rej_msg->reason) == IB_CM_REJ_TIMEOUT) {
 		spin_lock_irqsave(&cm.lock, flags);
-		timewait_info = cm_find_remote_id( *((u64 *) rej_msg->ari),
+		timewait_info = cm_find_remote_id( *((__be64 *) rej_msg->ari),
 						  remote_id);
 		if (!timewait_info) {
 			spin_unlock_irqrestore(&cm.lock, flags);
 			return NULL;
 		}
 		cm_id_priv = idr_find(&cm.local_id_table,
-				      (int) timewait_info->work.local_id);
+				      (__force int) timewait_info->work.local_id);
 		if (cm_id_priv) {
 			if (cm_id_priv->id.remote_id == remote_id)
 				atomic_inc(&cm_id_priv->refcount);
@@ -2032,7 +2035,7 @@ static int cm_rej_handler(struct cm_work *work)
 		/* fall through */
 	case IB_CM_REQ_RCVD:
 	case IB_CM_MRA_REQ_SENT:
-		if (rej_msg->reason == IB_CM_REJ_STALE_CONN)
+		if (__be16_to_cpu(rej_msg->reason) == IB_CM_REJ_STALE_CONN)
 			cm_enter_timewait(cm_id_priv);
 		else
 			cm_reset_to_idle(cm_id_priv);
@@ -2553,7 +2556,7 @@ static void cm_format_sidr_req(struct cm_sidr_req_msg *sidr_req_msg,
 	cm_format_mad_hdr(&sidr_req_msg->hdr, CM_SIDR_REQ_ATTR_ID,
 			  cm_form_tid(cm_id_priv, CM_MSG_SEQUENCE_SIDR));
 	sidr_req_msg->request_id = cm_id_priv->id.local_id;
-	sidr_req_msg->pkey = param->pkey;
+	sidr_req_msg->pkey = cpu_to_be16(param->pkey);
 	sidr_req_msg->service_id = param->service_id;
 
 	if (param->private_data && param->private_data_len)
@@ -2580,7 +2583,7 @@ int ib_send_cm_sidr_req(struct ib_cm_id *cm_id,
 		goto out;
 
 	cm_id->service_id = param->service_id;
-	cm_id->service_mask = ~0ULL;
+	cm_id->service_mask = __constant_cpu_to_be64(~0ULL);
 	cm_id_priv->timeout_ms = param->timeout_ms;
 	cm_id_priv->max_cm_retries = param->max_cm_retries;
 	ret = cm_alloc_msg(cm_id_priv, &msg);
@@ -2621,7 +2624,7 @@ static void cm_format_sidr_req_event(struct cm_work *work,
 	sidr_req_msg = (struct cm_sidr_req_msg *)
 				work->mad_recv_wc->recv_buf.mad;
 	param = &work->cm_event.param.sidr_req_rcvd;
-	param->pkey = sidr_req_msg->pkey;
+	param->pkey = __be16_to_cpu(sidr_req_msg->pkey);
 	param->listen_id = listen_id;
 	param->device = work->port->mad_agent->device;
 	param->port = work->port->port_num;
@@ -2645,7 +2648,7 @@ static int cm_sidr_req_handler(struct cm_work *work)
 	sidr_req_msg = (struct cm_sidr_req_msg *)
 				work->mad_recv_wc->recv_buf.mad;
 	wc = work->mad_recv_wc->wc;
-	cm_id_priv->av.dgid.global.subnet_prefix = wc->slid;
+	cm_id_priv->av.dgid.global.subnet_prefix = cpu_to_be64(wc->slid);
 	cm_id_priv->av.dgid.global.interface_id = 0;
 	cm_init_av_for_response(work->port, work->mad_recv_wc->wc,
 				&cm_id_priv->av);
@@ -2673,7 +2676,7 @@ static int cm_sidr_req_handler(struct cm_work *work)
 	cm_id_priv->id.cm_handler = cur_cm_id_priv->id.cm_handler;
 	cm_id_priv->id.context = cur_cm_id_priv->id.context;
 	cm_id_priv->id.service_id = sidr_req_msg->service_id;
-	cm_id_priv->id.service_mask = ~0ULL;
+	cm_id_priv->id.service_mask = __constant_cpu_to_be64(~0ULL);
 
 	cm_format_sidr_req_event(work, &cur_cm_id_priv->id);
 	cm_process_work(cm_id_priv, work);
@@ -3175,10 +3178,10 @@ int ib_cm_init_qp_attr(struct ib_cm_id *cm_id,
 }
 EXPORT_SYMBOL(ib_cm_init_qp_attr);
 
-static u64 cm_get_ca_guid(struct ib_device *device)
+static __be64 cm_get_ca_guid(struct ib_device *device)
 {
 	struct ib_device_attr *device_attr;
-	u64 guid;
+	__be64 guid;
 	int ret;
 
 	device_attr = kmalloc(sizeof *device_attr, GFP_KERNEL);

drivers/infiniband/core/cm_msgs.h

@@ -34,7 +34,7 @@
 #if !defined(CM_MSGS_H)
 #define CM_MSGS_H
 
-#include <ib_mad.h>
+#include <rdma/ib_mad.h>
 
 /*
  * Parameters to routines below should be in network-byte order, and values
@@ -43,19 +43,17 @@
 
 #define IB_CM_CLASS_VERSION	2 /* IB specification 1.2 */
 
-enum cm_msg_attr_id {
-	CM_REQ_ATTR_ID	    = __constant_htons(0x0010),
-	CM_MRA_ATTR_ID	    = __constant_htons(0x0011),
-	CM_REJ_ATTR_ID	    = __constant_htons(0x0012),
-	CM_REP_ATTR_ID	    = __constant_htons(0x0013),
-	CM_RTU_ATTR_ID	    = __constant_htons(0x0014),
-	CM_DREQ_ATTR_ID	    = __constant_htons(0x0015),
-	CM_DREP_ATTR_ID	    = __constant_htons(0x0016),
-	CM_SIDR_REQ_ATTR_ID = __constant_htons(0x0017),
-	CM_SIDR_REP_ATTR_ID = __constant_htons(0x0018),
-	CM_LAP_ATTR_ID      = __constant_htons(0x0019),
-	CM_APR_ATTR_ID      = __constant_htons(0x001A)
-};
+#define CM_REQ_ATTR_ID	    __constant_htons(0x0010)
+#define CM_MRA_ATTR_ID	    __constant_htons(0x0011)
+#define CM_REJ_ATTR_ID	    __constant_htons(0x0012)
+#define CM_REP_ATTR_ID	    __constant_htons(0x0013)
+#define CM_RTU_ATTR_ID	    __constant_htons(0x0014)
+#define CM_DREQ_ATTR_ID	    __constant_htons(0x0015)
+#define CM_DREP_ATTR_ID	    __constant_htons(0x0016)
+#define CM_SIDR_REQ_ATTR_ID __constant_htons(0x0017)
+#define CM_SIDR_REP_ATTR_ID __constant_htons(0x0018)
+#define CM_LAP_ATTR_ID      __constant_htons(0x0019)
+#define CM_APR_ATTR_ID      __constant_htons(0x001A)
 
 enum cm_msg_sequence {
 	CM_MSG_SEQUENCE_REQ,
@@ -67,35 +65,35 @@ enum cm_msg_sequence {
 struct cm_req_msg {
 	struct ib_mad_hdr hdr;
 
-	u32 local_comm_id;
-	u32 rsvd4;
-	u64 service_id;
-	u64 local_ca_guid;
-	u32 rsvd24;
-	u32 local_qkey;
+	__be32 local_comm_id;
+	__be32 rsvd4;
+	__be64 service_id;
+	__be64 local_ca_guid;
+	__be32 rsvd24;
+	__be32 local_qkey;
 	/* local QPN:24, responder resources:8 */
-	u32 offset32;
+	__be32 offset32;
 	/* local EECN:24, initiator depth:8 */
-	u32 offset36;
+	__be32 offset36;
 	/*
 	 * remote EECN:24, remote CM response timeout:5,
 	 * transport service type:2, end-to-end flow control:1
 	 */
-	u32 offset40;
+	__be32 offset40;
 	/* starting PSN:24, local CM response timeout:5, retry count:3 */
-	u32 offset44;
-	u16 pkey;
+	__be32 offset44;
+	__be16 pkey;
 	/* path MTU:4, RDC exists:1, RNR retry count:3. */
 	u8 offset50;
 	/* max CM Retries:4, SRQ:1, rsvd:3 */
 	u8 offset51;
 
-	u16 primary_local_lid;
-	u16 primary_remote_lid;
+	__be16 primary_local_lid;
+	__be16 primary_remote_lid;
 	union ib_gid primary_local_gid;
 	union ib_gid primary_remote_gid;
 	/* flow label:20, rsvd:6, packet rate:6 */
-	u32 primary_offset88;
+	__be32 primary_offset88;
 	u8 primary_traffic_class;
 	u8 primary_hop_limit;
 	/* SL:4, subnet local:1, rsvd:3 */
@@ -103,12 +101,12 @@ struct cm_req_msg {
 	/* local ACK timeout:5, rsvd:3 */
 	u8 primary_offset95;
 
-	u16 alt_local_lid;
-	u16 alt_remote_lid;
+	__be16 alt_local_lid;
+	__be16 alt_remote_lid;
 	union ib_gid alt_local_gid;
 	union ib_gid alt_remote_gid;
 	/* flow label:20, rsvd:6, packet rate:6 */
-	u32 alt_offset132;
+	__be32 alt_offset132;
 	u8 alt_traffic_class;
 	u8 alt_hop_limit;
 	/* SL:4, subnet local:1, rsvd:3 */
@@ -120,12 +118,12 @@ struct cm_req_msg {
 
 } __attribute__ ((packed));
 
-static inline u32 cm_req_get_local_qpn(struct cm_req_msg *req_msg)
+static inline __be32 cm_req_get_local_qpn(struct cm_req_msg *req_msg)
 {
 	return cpu_to_be32(be32_to_cpu(req_msg->offset32) >> 8);
 }
 
-static inline void cm_req_set_local_qpn(struct cm_req_msg *req_msg, u32 qpn)
+static inline void cm_req_set_local_qpn(struct cm_req_msg *req_msg, __be32 qpn)
 {
 	req_msg->offset32 = cpu_to_be32((be32_to_cpu(qpn) << 8) |
 					 (be32_to_cpu(req_msg->offset32) &
@@ -208,13 +206,13 @@ static inline void cm_req_set_flow_ctrl(struct cm_req_msg *req_msg,
 					  0xFFFFFFFE));
 }
 
-static inline u32 cm_req_get_starting_psn(struct cm_req_msg *req_msg)
+static inline __be32 cm_req_get_starting_psn(struct cm_req_msg *req_msg)
 {
 	return cpu_to_be32(be32_to_cpu(req_msg->offset44) >> 8);
 }
 
 static inline void cm_req_set_starting_psn(struct cm_req_msg *req_msg,
-					   u32 starting_psn)
+					   __be32 starting_psn)
 {
 	req_msg->offset44 = cpu_to_be32((be32_to_cpu(starting_psn) << 8) |
 			    (be32_to_cpu(req_msg->offset44) & 0x000000FF));
@@ -288,13 +286,13 @@ static inline void cm_req_set_srq(struct cm_req_msg *req_msg, u8 srq)
 				  ((srq & 0x1) << 3));
 }
 
-static inline u32 cm_req_get_primary_flow_label(struct cm_req_msg *req_msg)
+static inline __be32 cm_req_get_primary_flow_label(struct cm_req_msg *req_msg)
 {
-	return cpu_to_be32((be32_to_cpu(req_msg->primary_offset88) >> 12));
+	return cpu_to_be32(be32_to_cpu(req_msg->primary_offset88) >> 12);
 }
 
 static inline void cm_req_set_primary_flow_label(struct cm_req_msg *req_msg,
-						 u32 flow_label)
+						 __be32 flow_label)
 {
 	req_msg->primary_offset88 = cpu_to_be32(
 				    (be32_to_cpu(req_msg->primary_offset88) &
@@ -350,13 +348,13 @@ static inline void cm_req_set_primary_local_ack_timeout(struct cm_req_msg *req_m
 					  (local_ack_timeout << 3));
 }
 
-static inline u32 cm_req_get_alt_flow_label(struct cm_req_msg *req_msg)
+static inline __be32 cm_req_get_alt_flow_label(struct cm_req_msg *req_msg)
 {
-	return cpu_to_be32((be32_to_cpu(req_msg->alt_offset132) >> 12));
+	return cpu_to_be32(be32_to_cpu(req_msg->alt_offset132) >> 12);
 }
 
 static inline void cm_req_set_alt_flow_label(struct cm_req_msg *req_msg,
-					     u32 flow_label)
+					     __be32 flow_label)
 {
 	req_msg->alt_offset132 = cpu_to_be32(
 				 (be32_to_cpu(req_msg->alt_offset132) &
@@ -422,8 +420,8 @@ enum cm_msg_response {
  struct cm_mra_msg {
 	struct ib_mad_hdr hdr;
 
-	u32 local_comm_id;
-	u32 remote_comm_id;
+	__be32 local_comm_id;
+	__be32 remote_comm_id;
 	/* message MRAed:2, rsvd:6 */
 	u8 offset8;
 	/* service timeout:5, rsvd:3 */
@@ -458,13 +456,13 @@ static inline void cm_mra_set_service_timeout(struct cm_mra_msg *mra_msg,
 struct cm_rej_msg {
 	struct ib_mad_hdr hdr;
 
-	u32 local_comm_id;
-	u32 remote_comm_id;
+	__be32 local_comm_id;
+	__be32 remote_comm_id;
 	/* message REJected:2, rsvd:6 */
 	u8 offset8;
 	/* reject info length:7, rsvd:1. */
 	u8 offset9;
-	u16 reason;
+	__be16 reason;
 	u8 ari[IB_CM_REJ_ARI_LENGTH];
 
 	u8 private_data[IB_CM_REJ_PRIVATE_DATA_SIZE];
@@ -495,45 +493,45 @@ static inline void cm_rej_set_reject_info_len(struct cm_rej_msg *rej_msg,
 struct cm_rep_msg {
 	struct ib_mad_hdr hdr;
 
-	u32 local_comm_id;
-	u32 remote_comm_id;
-	u32 local_qkey;
+	__be32 local_comm_id;
+	__be32 remote_comm_id;
+	__be32 local_qkey;
 	/* local QPN:24, rsvd:8 */
-	u32 offset12;
+	__be32 offset12;
 	/* local EECN:24, rsvd:8 */
-	u32 offset16;
+	__be32 offset16;
 	/* starting PSN:24 rsvd:8 */
-	u32 offset20;
+	__be32 offset20;
 	u8 resp_resources;
 	u8 initiator_depth;
 	/* target ACK delay:5, failover accepted:2, end-to-end flow control:1 */
 	u8 offset26;
 	/* RNR retry count:3, SRQ:1, rsvd:5 */
 	u8 offset27;
-	u64 local_ca_guid;
+	__be64 local_ca_guid;
 
 	u8 private_data[IB_CM_REP_PRIVATE_DATA_SIZE];
 
 } __attribute__ ((packed));
 
-static inline u32 cm_rep_get_local_qpn(struct cm_rep_msg *rep_msg)
+static inline __be32 cm_rep_get_local_qpn(struct cm_rep_msg *rep_msg)
 {
 	return cpu_to_be32(be32_to_cpu(rep_msg->offset12) >> 8);
 }
 
-static inline void cm_rep_set_local_qpn(struct cm_rep_msg *rep_msg, u32 qpn)
+static inline void cm_rep_set_local_qpn(struct cm_rep_msg *rep_msg, __be32 qpn)
 {
 	rep_msg->offset12 = cpu_to_be32((be32_to_cpu(qpn) << 8) |
 			    (be32_to_cpu(rep_msg->offset12) & 0x000000FF));
 }
 
-static inline u32 cm_rep_get_starting_psn(struct cm_rep_msg *rep_msg)
+static inline __be32 cm_rep_get_starting_psn(struct cm_rep_msg *rep_msg)
 {
 	return cpu_to_be32(be32_to_cpu(rep_msg->offset20) >> 8);
 }
 
 static inline void cm_rep_set_starting_psn(struct cm_rep_msg *rep_msg,
-					   u32 starting_psn)
+					   __be32 starting_psn)
 {
 	rep_msg->offset20 = cpu_to_be32((be32_to_cpu(starting_psn) << 8) |
 			    (be32_to_cpu(rep_msg->offset20) & 0x000000FF));
@@ -600,8 +598,8 @@ static inline void cm_rep_set_srq(struct cm_rep_msg *rep_msg, u8 srq)
 struct cm_rtu_msg {
 	struct ib_mad_hdr hdr;
 
-	u32 local_comm_id;
-	u32 remote_comm_id;
+	__be32 local_comm_id;
+	__be32 remote_comm_id;
 
 	u8 private_data[IB_CM_RTU_PRIVATE_DATA_SIZE];
 
@@ -610,21 +608,21 @@ struct cm_rtu_msg {
 struct cm_dreq_msg {
 	struct ib_mad_hdr hdr;
 
-	u32 local_comm_id;
-	u32 remote_comm_id;
+	__be32 local_comm_id;
+	__be32 remote_comm_id;
 	/* remote QPN/EECN:24, rsvd:8 */
-	u32 offset8;
+	__be32 offset8;
 
 	u8 private_data[IB_CM_DREQ_PRIVATE_DATA_SIZE];
 
 } __attribute__ ((packed));
 
-static inline u32 cm_dreq_get_remote_qpn(struct cm_dreq_msg *dreq_msg)
+static inline __be32 cm_dreq_get_remote_qpn(struct cm_dreq_msg *dreq_msg)
 {
 	return cpu_to_be32(be32_to_cpu(dreq_msg->offset8) >> 8);
 }
 
-static inline void cm_dreq_set_remote_qpn(struct cm_dreq_msg *dreq_msg, u32 qpn)
+static inline void cm_dreq_set_remote_qpn(struct cm_dreq_msg *dreq_msg, __be32 qpn)
 {
 	dreq_msg->offset8 = cpu_to_be32((be32_to_cpu(qpn) << 8) |
 			    (be32_to_cpu(dreq_msg->offset8) & 0x000000FF));
@@ -633,8 +631,8 @@ static inline void cm_dreq_set_remote_qpn(struct cm_dreq_msg *dreq_msg, u32 qpn)
 struct cm_drep_msg {
 	struct ib_mad_hdr hdr;
 
-	u32 local_comm_id;
-	u32 remote_comm_id;
+	__be32 local_comm_id;
+	__be32 remote_comm_id;
 
 	u8 private_data[IB_CM_DREP_PRIVATE_DATA_SIZE];
 
@@ -643,37 +641,37 @@ struct cm_drep_msg {
 struct cm_lap_msg {
 	struct ib_mad_hdr hdr;
 
-	u32 local_comm_id;
-	u32 remote_comm_id;
+	__be32 local_comm_id;
+	__be32 remote_comm_id;
 
-	u32 rsvd8;
+	__be32 rsvd8;
 	/* remote QPN/EECN:24, remote CM response timeout:5, rsvd:3 */
-	u32 offset12;
-	u32 rsvd16;
+	__be32 offset12;
+	__be32 rsvd16;
 
-	u16 alt_local_lid;
-	u16 alt_remote_lid;
+	__be16 alt_local_lid;
+	__be16 alt_remote_lid;
 	union ib_gid alt_local_gid;
 	union ib_gid alt_remote_gid;
 	/* flow label:20, rsvd:4, traffic class:8 */
-	u32 offset56;
+	__be32 offset56;
 	u8 alt_hop_limit;
 	/* rsvd:2, packet rate:6 */
-	uint8_t offset61;
+	u8 offset61;
 	/* SL:4, subnet local:1, rsvd:3 */
-	uint8_t offset62;
+	u8 offset62;
 	/* local ACK timeout:5, rsvd:3 */
-	uint8_t offset63;
+	u8 offset63;
 
 	u8 private_data[IB_CM_LAP_PRIVATE_DATA_SIZE];
 } __attribute__  ((packed));
 
-static inline u32 cm_lap_get_remote_qpn(struct cm_lap_msg *lap_msg)
+static inline __be32 cm_lap_get_remote_qpn(struct cm_lap_msg *lap_msg)
 {
 	return cpu_to_be32(be32_to_cpu(lap_msg->offset12) >> 8);
 }
 
-static inline void cm_lap_set_remote_qpn(struct cm_lap_msg *lap_msg, u32 qpn)
+static inline void cm_lap_set_remote_qpn(struct cm_lap_msg *lap_msg, __be32 qpn)
 {
 	lap_msg->offset12 = cpu_to_be32((be32_to_cpu(qpn) << 8) |
 					 (be32_to_cpu(lap_msg->offset12) &
@@ -693,17 +691,17 @@ static inline void cm_lap_set_remote_resp_timeout(struct cm_lap_msg *lap_msg,
 					  0xFFFFFF07));
 }
 
-static inline u32 cm_lap_get_flow_label(struct cm_lap_msg *lap_msg)
+static inline __be32 cm_lap_get_flow_label(struct cm_lap_msg *lap_msg)
 {
-	return be32_to_cpu(lap_msg->offset56) >> 12;
+	return cpu_to_be32(be32_to_cpu(lap_msg->offset56) >> 12);
 }
 
 static inline void cm_lap_set_flow_label(struct cm_lap_msg *lap_msg,
-					 u32 flow_label)
+					 __be32 flow_label)
 {
-	lap_msg->offset56 = cpu_to_be32((flow_label << 12) |
-					 (be32_to_cpu(lap_msg->offset56) &
-					  0x00000FFF));
+	lap_msg->offset56 = cpu_to_be32(
+				 (be32_to_cpu(lap_msg->offset56) & 0x00000FFF) |
+				 (be32_to_cpu(flow_label) << 12));
 }
 
 static inline u8 cm_lap_get_traffic_class(struct cm_lap_msg *lap_msg)
@@ -766,8 +764,8 @@ static inline void cm_lap_set_local_ack_timeout(struct cm_lap_msg *lap_msg,
 struct cm_apr_msg {
 	struct ib_mad_hdr hdr;
 
-	u32 local_comm_id;
-	u32 remote_comm_id;
+	__be32 local_comm_id;
+	__be32 remote_comm_id;
 
 	u8 info_length;
 	u8 ap_status;
@@ -779,10 +777,10 @@ struct cm_apr_msg {
 struct cm_sidr_req_msg {
 	struct ib_mad_hdr hdr;
 
-	u32 request_id;
-	u16 pkey;
-	u16 rsvd;
-	u64 service_id;
+	__be32 request_id;
+	__be16 pkey;
+	__be16 rsvd;
+	__be64 service_id;
 
 	u8 private_data[IB_CM_SIDR_REQ_PRIVATE_DATA_SIZE];
 } __attribute__ ((packed));
@@ -790,26 +788,26 @@ struct cm_sidr_req_msg {
 struct cm_sidr_rep_msg {
 	struct ib_mad_hdr hdr;
 
-	u32 request_id;
+	__be32 request_id;
 	u8 status;
 	u8 info_length;
-	u16 rsvd;
+	__be16 rsvd;
 	/* QPN:24, rsvd:8 */
-	u32 offset8;
-	u64 service_id;
-	u32 qkey;
+	__be32 offset8;
+	__be64 service_id;
+	__be32 qkey;
 	u8 info[IB_CM_SIDR_REP_INFO_LENGTH];
 
 	u8 private_data[IB_CM_SIDR_REP_PRIVATE_DATA_SIZE];
 } __attribute__ ((packed));
 
-static inline u32 cm_sidr_rep_get_qpn(struct cm_sidr_rep_msg *sidr_rep_msg)
+static inline __be32 cm_sidr_rep_get_qpn(struct cm_sidr_rep_msg *sidr_rep_msg)
 {
 	return cpu_to_be32(be32_to_cpu(sidr_rep_msg->offset8) >> 8);
 }
 
 static inline void cm_sidr_rep_set_qpn(struct cm_sidr_rep_msg *sidr_rep_msg,
-				       u32 qpn)
+				       __be32 qpn)
 {
 	sidr_rep_msg->offset8 = cpu_to_be32((be32_to_cpu(qpn) << 8) |
 					(be32_to_cpu(sidr_rep_msg->offset8) &

drivers/infiniband/core/core_priv.h

@@ -38,7 +38,7 @@
 #include <linux/list.h>
 #include <linux/spinlock.h>
 
-#include <ib_verbs.h>
+#include <rdma/ib_verbs.h>
 
 int  ib_device_register_sysfs(struct ib_device *device);
 void ib_device_unregister_sysfs(struct ib_device *device);

drivers/infiniband/core/device.c

@@ -1,5 +1,6 @@
 /*
  * Copyright (c) 2004 Topspin Communications.  All rights reserved.
+ * Copyright (c) 2005 Sun Microsystems, Inc. All rights reserved.
  *
  * This software is available to you under a choice of one of two
  * licenses.  You may choose to be licensed under the terms of the GNU

drivers/infiniband/core/fmr_pool.c

@@ -39,7 +39,7 @@
 #include <linux/jhash.h>
 #include <linux/kthread.h>
 
-#include <ib_fmr_pool.h>
+#include <rdma/ib_fmr_pool.h>
 
 #include "core_priv.h"
 
@@ -334,6 +334,7 @@ void ib_destroy_fmr_pool(struct ib_fmr_pool *pool)
 {
 	struct ib_pool_fmr *fmr;
 	struct ib_pool_fmr *tmp;
+	LIST_HEAD(fmr_list);
 	int                 i;
 
 	kthread_stop(pool->thread);
@@ -341,6 +342,11 @@ void ib_destroy_fmr_pool(struct ib_fmr_pool *pool)
 
 	i = 0;
 	list_for_each_entry_safe(fmr, tmp, &pool->free_list, list) {
+		if (fmr->remap_count) {
+			INIT_LIST_HEAD(&fmr_list);
+			list_add_tail(&fmr->fmr->list, &fmr_list);
+			ib_unmap_fmr(&fmr_list);
+		}
 		ib_dealloc_fmr(fmr->fmr);
 		list_del(&fmr->list);
 		kfree(fmr);

drivers/infiniband/core/mad.c

@@ -693,7 +693,8 @@ static int handle_outgoing_dr_smp(struct ib_mad_agent_private *mad_agent_priv,
 		goto out;
 	}
 
-	build_smp_wc(send_wr->wr_id, smp->dr_slid, send_wr->wr.ud.pkey_index,
+	build_smp_wc(send_wr->wr_id, be16_to_cpu(smp->dr_slid),
+		     send_wr->wr.ud.pkey_index,
 		     send_wr->wr.ud.port_num, &mad_wc);
 
 	/* No GRH for DR SMP */
@@ -1554,7 +1555,7 @@ static int is_data_mad(struct ib_mad_agent_private *mad_agent_priv,
 }
 
 struct ib_mad_send_wr_private*
-ib_find_send_mad(struct ib_mad_agent_private *mad_agent_priv, u64 tid)
+ib_find_send_mad(struct ib_mad_agent_private *mad_agent_priv, __be64 tid)
 {
 	struct ib_mad_send_wr_private *mad_send_wr;
 
@@ -1597,7 +1598,7 @@ static void ib_mad_complete_recv(struct ib_mad_agent_private *mad_agent_priv,
 	struct ib_mad_send_wr_private *mad_send_wr;
 	struct ib_mad_send_wc mad_send_wc;
 	unsigned long flags;
-	u64 tid;
+	__be64 tid;
 
 	INIT_LIST_HEAD(&mad_recv_wc->rmpp_list);
 	list_add(&mad_recv_wc->recv_buf.list, &mad_recv_wc->rmpp_list);
@@ -2165,7 +2166,8 @@ static void local_completions(void *data)
 			 * Defined behavior is to complete response
 			 * before request
 			 */
-			build_smp_wc(local->wr_id, IB_LID_PERMISSIVE,
+			build_smp_wc(local->wr_id,
+				     be16_to_cpu(IB_LID_PERMISSIVE),
 				     0 /* pkey index */,
 				     recv_mad_agent->agent.port_num, &wc);
 
@@ -2294,7 +2296,7 @@ static void timeout_sends(void *data)
 	spin_unlock_irqrestore(&mad_agent_priv->lock, flags);
 }
 
-static void ib_mad_thread_completion_handler(struct ib_cq *cq)
+static void ib_mad_thread_completion_handler(struct ib_cq *cq, void *arg)
 {
 	struct ib_mad_port_private *port_priv = cq->cq_context;
 
@@ -2574,8 +2576,7 @@ static int ib_mad_port_open(struct ib_device *device,
 
 	cq_size = (IB_MAD_QP_SEND_SIZE + IB_MAD_QP_RECV_SIZE) * 2;
 	port_priv->cq = ib_create_cq(port_priv->device,
-				     (ib_comp_handler)
-					ib_mad_thread_completion_handler,
+				     ib_mad_thread_completion_handler,
 				     NULL, port_priv, cq_size);
 	if (IS_ERR(port_priv->cq)) {
 		printk(KERN_ERR PFX "Couldn't create ib_mad CQ\n");

drivers/infiniband/core/mad_priv.h

@@ -40,8 +40,8 @@
 #include <linux/pci.h>
 #include <linux/kthread.h>
 #include <linux/workqueue.h>
-#include <ib_mad.h>
-#include <ib_smi.h>
+#include <rdma/ib_mad.h>
+#include <rdma/ib_smi.h>
 
 
 #define PFX "ib_mad: "
@@ -121,7 +121,7 @@ struct ib_mad_send_wr_private {
 	struct ib_send_wr send_wr;
 	struct ib_sge sg_list[IB_MAD_SEND_REQ_MAX_SG];
 	u64 wr_id;			/* client WR ID */
-	u64 tid;
+	__be64 tid;
 	unsigned long timeout;
 	int retries;
 	int retry;
@@ -144,7 +144,7 @@ struct ib_mad_local_private {
 	struct ib_send_wr send_wr;
 	struct ib_sge sg_list[IB_MAD_SEND_REQ_MAX_SG];
 	u64 wr_id;			/* client WR ID */
-	u64 tid;
+	__be64 tid;
 };
 
 struct ib_mad_mgmt_method_table {
@@ -210,7 +210,7 @@ extern kmem_cache_t *ib_mad_cache;
 int ib_send_mad(struct ib_mad_send_wr_private *mad_send_wr);
 
 struct ib_mad_send_wr_private *
-ib_find_send_mad(struct ib_mad_agent_private *mad_agent_priv, u64 tid);
+ib_find_send_mad(struct ib_mad_agent_private *mad_agent_priv, __be64 tid);
 
 void ib_mad_complete_send_wr(struct ib_mad_send_wr_private *mad_send_wr,
 			     struct ib_mad_send_wc *mad_send_wc);

drivers/infiniband/core/mad_rmpp.c

@@ -61,7 +61,7 @@ struct mad_rmpp_recv {
 	int seg_num;
 	int newwin;
 
-	u64 tid;
+	__be64 tid;
 	u32 src_qp;
 	u16 slid;
 	u8 mgmt_class;
@@ -100,6 +100,121 @@ void ib_cancel_rmpp_recvs(struct ib_mad_agent_private *agent)
 	}
 }
 
+static int data_offset(u8 mgmt_class)
+{
+	if (mgmt_class == IB_MGMT_CLASS_SUBN_ADM)
+		return offsetof(struct ib_sa_mad, data);
+	else if ((mgmt_class >= IB_MGMT_CLASS_VENDOR_RANGE2_START) &&
+		 (mgmt_class <= IB_MGMT_CLASS_VENDOR_RANGE2_END))
+		return offsetof(struct ib_vendor_mad, data);
+	else
+		return offsetof(struct ib_rmpp_mad, data);
+}
+
+static void format_ack(struct ib_rmpp_mad *ack,
+		       struct ib_rmpp_mad *data,
+		       struct mad_rmpp_recv *rmpp_recv)
+{
+	unsigned long flags;
+
+	memcpy(&ack->mad_hdr, &data->mad_hdr,
+	       data_offset(data->mad_hdr.mgmt_class));
+
+	ack->mad_hdr.method ^= IB_MGMT_METHOD_RESP;
+	ack->rmpp_hdr.rmpp_type = IB_MGMT_RMPP_TYPE_ACK;
+	ib_set_rmpp_flags(&ack->rmpp_hdr, IB_MGMT_RMPP_FLAG_ACTIVE);
+
+	spin_lock_irqsave(&rmpp_recv->lock, flags);
+	rmpp_recv->last_ack = rmpp_recv->seg_num;
+	ack->rmpp_hdr.seg_num = cpu_to_be32(rmpp_recv->seg_num);
+	ack->rmpp_hdr.paylen_newwin = cpu_to_be32(rmpp_recv->newwin);
+	spin_unlock_irqrestore(&rmpp_recv->lock, flags);
+}
+
+static void ack_recv(struct mad_rmpp_recv *rmpp_recv,
+		     struct ib_mad_recv_wc *recv_wc)
+{
+	struct ib_mad_send_buf *msg;
+	struct ib_send_wr *bad_send_wr;
+	int hdr_len, ret;
+
+	hdr_len = sizeof(struct ib_mad_hdr) + sizeof(struct ib_rmpp_hdr);
+	msg = ib_create_send_mad(&rmpp_recv->agent->agent, recv_wc->wc->src_qp,
+				 recv_wc->wc->pkey_index, rmpp_recv->ah, 1,
+				 hdr_len, sizeof(struct ib_rmpp_mad) - hdr_len,
+				 GFP_KERNEL);
+	if (!msg)
+		return;
+
+	format_ack((struct ib_rmpp_mad *) msg->mad,
+		   (struct ib_rmpp_mad *) recv_wc->recv_buf.mad, rmpp_recv);
+	ret = ib_post_send_mad(&rmpp_recv->agent->agent, &msg->send_wr,
+			       &bad_send_wr);
+	if (ret)
+		ib_free_send_mad(msg);
+}
+
+static int alloc_response_msg(struct ib_mad_agent *agent,
+			      struct ib_mad_recv_wc *recv_wc,
+			      struct ib_mad_send_buf **msg)
+{
+	struct ib_mad_send_buf *m;
+	struct ib_ah *ah;
+	int hdr_len;
+
+	ah = ib_create_ah_from_wc(agent->qp->pd, recv_wc->wc,
+				  recv_wc->recv_buf.grh, agent->port_num);
+	if (IS_ERR(ah))
+		return PTR_ERR(ah);
+
+	hdr_len = sizeof(struct ib_mad_hdr) + sizeof(struct ib_rmpp_hdr);
+	m = ib_create_send_mad(agent, recv_wc->wc->src_qp,
+			       recv_wc->wc->pkey_index, ah, 1, hdr_len,
+			       sizeof(struct ib_rmpp_mad) - hdr_len,
+			       GFP_KERNEL);
+	if (IS_ERR(m)) {
+		ib_destroy_ah(ah);
+		return PTR_ERR(m);
+	}
+	*msg = m;
+	return 0;
+}
+
+static void free_msg(struct ib_mad_send_buf *msg)
+{
+	ib_destroy_ah(msg->send_wr.wr.ud.ah);
+	ib_free_send_mad(msg);
+}
+
+static void nack_recv(struct ib_mad_agent_private *agent,
+		      struct ib_mad_recv_wc *recv_wc, u8 rmpp_status)
+{
+	struct ib_mad_send_buf *msg;
+	struct ib_rmpp_mad *rmpp_mad;
+	struct ib_send_wr *bad_send_wr;
+	int ret;
+
+	ret = alloc_response_msg(&agent->agent, recv_wc, &msg);
+	if (ret)
+		return;
+
+	rmpp_mad = (struct ib_rmpp_mad *) msg->mad;
+	memcpy(rmpp_mad, recv_wc->recv_buf.mad,
+	       data_offset(recv_wc->recv_buf.mad->mad_hdr.mgmt_class));
+
+	rmpp_mad->mad_hdr.method ^= IB_MGMT_METHOD_RESP;
+	rmpp_mad->rmpp_hdr.rmpp_version = IB_MGMT_RMPP_VERSION;
+	rmpp_mad->rmpp_hdr.rmpp_type = IB_MGMT_RMPP_TYPE_ABORT;
+	ib_set_rmpp_flags(&rmpp_mad->rmpp_hdr, IB_MGMT_RMPP_FLAG_ACTIVE);
+	rmpp_mad->rmpp_hdr.rmpp_status = rmpp_status;
+	rmpp_mad->rmpp_hdr.seg_num = 0;
+	rmpp_mad->rmpp_hdr.paylen_newwin = 0;
+
+	ret = ib_post_send_mad(&agent->agent, &msg->send_wr, &bad_send_wr);
+	if (ret)
+		free_msg(msg);
+}
+
 static void recv_timeout_handler(void *data)
 {
 	struct mad_rmpp_recv *rmpp_recv = data;
@@ -115,8 +230,8 @@ static void recv_timeout_handler(void *data)
 	list_del(&rmpp_recv->list);
 	spin_unlock_irqrestore(&rmpp_recv->agent->lock, flags);
 
-	/* TODO: send abort. */
 	rmpp_wc = rmpp_recv->rmpp_wc;
+	nack_recv(rmpp_recv->agent, rmpp_wc, IB_MGMT_RMPP_STATUS_T2L);
 	destroy_rmpp_recv(rmpp_recv);
 	ib_free_recv_mad(rmpp_wc);
 }
@@ -230,60 +345,6 @@ insert_rmpp_recv(struct ib_mad_agent_private *agent,
 	return cur_rmpp_recv;
 }
 
-static int data_offset(u8 mgmt_class)
-{
-	if (mgmt_class == IB_MGMT_CLASS_SUBN_ADM)
-		return offsetof(struct ib_sa_mad, data);
-	else if ((mgmt_class >= IB_MGMT_CLASS_VENDOR_RANGE2_START) &&
-		 (mgmt_class <= IB_MGMT_CLASS_VENDOR_RANGE2_END))
-		return offsetof(struct ib_vendor_mad, data);
-	else
-		return offsetof(struct ib_rmpp_mad, data);
-}
-
-static void format_ack(struct ib_rmpp_mad *ack,
-		       struct ib_rmpp_mad *data,
-		       struct mad_rmpp_recv *rmpp_recv)
-{
-	unsigned long flags;
-
-	memcpy(&ack->mad_hdr, &data->mad_hdr,
-	       data_offset(data->mad_hdr.mgmt_class));
-
-	ack->mad_hdr.method ^= IB_MGMT_METHOD_RESP;
-	ack->rmpp_hdr.rmpp_type = IB_MGMT_RMPP_TYPE_ACK;
-	ib_set_rmpp_flags(&ack->rmpp_hdr, IB_MGMT_RMPP_FLAG_ACTIVE);
-
-	spin_lock_irqsave(&rmpp_recv->lock, flags);
-	rmpp_recv->last_ack = rmpp_recv->seg_num;
-	ack->rmpp_hdr.seg_num = cpu_to_be32(rmpp_recv->seg_num);
-	ack->rmpp_hdr.paylen_newwin = cpu_to_be32(rmpp_recv->newwin);
-	spin_unlock_irqrestore(&rmpp_recv->lock, flags);
-}
-
-static void ack_recv(struct mad_rmpp_recv *rmpp_recv,
-		     struct ib_mad_recv_wc *recv_wc)
-{
-	struct ib_mad_send_buf *msg;
-	struct ib_send_wr *bad_send_wr;
-	int hdr_len, ret;
-
-	hdr_len = sizeof(struct ib_mad_hdr) + sizeof(struct ib_rmpp_hdr);
-	msg = ib_create_send_mad(&rmpp_recv->agent->agent, recv_wc->wc->src_qp,
-				 recv_wc->wc->pkey_index, rmpp_recv->ah, 1,
-				 hdr_len, sizeof(struct ib_rmpp_mad) - hdr_len,
-				 GFP_KERNEL);
-	if (!msg)
-		return;
-
-	format_ack((struct ib_rmpp_mad *) msg->mad,
-		   (struct ib_rmpp_mad *) recv_wc->recv_buf.mad, rmpp_recv);
-	ret = ib_post_send_mad(&rmpp_recv->agent->agent, &msg->send_wr,
-			       &bad_send_wr);
-	if (ret)
-		ib_free_send_mad(msg);
-}
-
 static inline int get_last_flag(struct ib_mad_recv_buf *seg)
 {
 	struct ib_rmpp_mad *rmpp_mad;
@@ -559,6 +620,34 @@ static int send_next_seg(struct ib_mad_send_wr_private *mad_send_wr)
 	return ib_send_mad(mad_send_wr);
 }
 
+static void abort_send(struct ib_mad_agent_private *agent, __be64 tid,
+		       u8 rmpp_status)
+{
+	struct ib_mad_send_wr_private *mad_send_wr;
+	struct ib_mad_send_wc wc;
+	unsigned long flags;
+
+	spin_lock_irqsave(&agent->lock, flags);
+	mad_send_wr = ib_find_send_mad(agent, tid);
+	if (!mad_send_wr)
+		goto out;	/* Unmatched send */
+
+	if ((mad_send_wr->last_ack == mad_send_wr->total_seg) ||
+	    (!mad_send_wr->timeout) || (mad_send_wr->status != IB_WC_SUCCESS))
+		goto out;	/* Send is already done */
+
+	ib_mark_mad_done(mad_send_wr);
+	spin_unlock_irqrestore(&agent->lock, flags);
+
+	wc.status = IB_WC_REM_ABORT_ERR;
+	wc.vendor_err = rmpp_status;
+	wc.wr_id = mad_send_wr->wr_id;
+	ib_mad_complete_send_wr(mad_send_wr, &wc);
+	return;
+out:
+	spin_unlock_irqrestore(&agent->lock, flags);
+}
+
 static void process_rmpp_ack(struct ib_mad_agent_private *agent,
 			     struct ib_mad_recv_wc *mad_recv_wc)
 {
@@ -568,11 +657,21 @@ static void process_rmpp_ack(struct ib_mad_agent_private *agent,
 	int seg_num, newwin, ret;
 
 	rmpp_mad = (struct ib_rmpp_mad *)mad_recv_wc->recv_buf.mad;
-	if (rmpp_mad->rmpp_hdr.rmpp_status)
+	if (rmpp_mad->rmpp_hdr.rmpp_status) {
+		abort_send(agent, rmpp_mad->mad_hdr.tid,
+			   IB_MGMT_RMPP_STATUS_BAD_STATUS);
+		nack_recv(agent, mad_recv_wc, IB_MGMT_RMPP_STATUS_BAD_STATUS);
 		return;
+	}
 
 	seg_num = be32_to_cpu(rmpp_mad->rmpp_hdr.seg_num);
 	newwin = be32_to_cpu(rmpp_mad->rmpp_hdr.paylen_newwin);
+	if (newwin < seg_num) {
+		abort_send(agent, rmpp_mad->mad_hdr.tid,
+			   IB_MGMT_RMPP_STATUS_W2S);
+		nack_recv(agent, mad_recv_wc, IB_MGMT_RMPP_STATUS_W2S);
+		return;
+	}
 
 	spin_lock_irqsave(&agent->lock, flags);
 	mad_send_wr = ib_find_send_mad(agent, rmpp_mad->mad_hdr.tid);
@@ -583,8 +682,13 @@ static void process_rmpp_ack(struct ib_mad_agent_private *agent,
 	    (!mad_send_wr->timeout) || (mad_send_wr->status != IB_WC_SUCCESS))
 		goto out;	/* Send is already done */
 
-	if (seg_num > mad_send_wr->total_seg)
-		goto out;	/* Bad ACK */
+	if (seg_num > mad_send_wr->total_seg || seg_num > mad_send_wr->newwin) {
+		spin_unlock_irqrestore(&agent->lock, flags);
+		abort_send(agent, rmpp_mad->mad_hdr.tid,
+			   IB_MGMT_RMPP_STATUS_S2B);
+		nack_recv(agent, mad_recv_wc, IB_MGMT_RMPP_STATUS_S2B);
+		return;
+	}
 
 	if (newwin < mad_send_wr->newwin || seg_num < mad_send_wr->last_ack)
 		goto out;	/* Old ACK */
@@ -628,6 +732,72 @@ out:
 	spin_unlock_irqrestore(&agent->lock, flags);
 }
 
+static struct ib_mad_recv_wc *
+process_rmpp_data(struct ib_mad_agent_private *agent,
+		  struct ib_mad_recv_wc *mad_recv_wc)
+{
+	struct ib_rmpp_hdr *rmpp_hdr;
+	u8 rmpp_status;
+
+	rmpp_hdr = &((struct ib_rmpp_mad *)mad_recv_wc->recv_buf.mad)->rmpp_hdr;
+
+	if (rmpp_hdr->rmpp_status) {
+		rmpp_status = IB_MGMT_RMPP_STATUS_BAD_STATUS;
+		goto bad;
+	}
+
+	if (rmpp_hdr->seg_num == __constant_htonl(1)) {
+		if (!(ib_get_rmpp_flags(rmpp_hdr) & IB_MGMT_RMPP_FLAG_FIRST)) {
+			rmpp_status = IB_MGMT_RMPP_STATUS_BAD_SEG;
+			goto bad;
+		}
+		return start_rmpp(agent, mad_recv_wc);
+	} else {
+		if (ib_get_rmpp_flags(rmpp_hdr) & IB_MGMT_RMPP_FLAG_FIRST) {
+			rmpp_status = IB_MGMT_RMPP_STATUS_BAD_SEG;
+			goto bad;
+		}
+		return continue_rmpp(agent, mad_recv_wc);
+	}
+bad:
+	nack_recv(agent, mad_recv_wc, rmpp_status);
+	ib_free_recv_mad(mad_recv_wc);
+	return NULL;
+}
+
+static void process_rmpp_stop(struct ib_mad_agent_private *agent,
+			      struct ib_mad_recv_wc *mad_recv_wc)
+{
+	struct ib_rmpp_mad *rmpp_mad;
+
+	rmpp_mad = (struct ib_rmpp_mad *)mad_recv_wc->recv_buf.mad;
+
+	if (rmpp_mad->rmpp_hdr.rmpp_status != IB_MGMT_RMPP_STATUS_RESX) {
+		abort_send(agent, rmpp_mad->mad_hdr.tid,
+			   IB_MGMT_RMPP_STATUS_BAD_STATUS);
+		nack_recv(agent, mad_recv_wc, IB_MGMT_RMPP_STATUS_BAD_STATUS);
+	} else
+		abort_send(agent, rmpp_mad->mad_hdr.tid,
+			   rmpp_mad->rmpp_hdr.rmpp_status);
+}
+
+static void process_rmpp_abort(struct ib_mad_agent_private *agent,
+			       struct ib_mad_recv_wc *mad_recv_wc)
+{
+	struct ib_rmpp_mad *rmpp_mad;
+
+	rmpp_mad = (struct ib_rmpp_mad *)mad_recv_wc->recv_buf.mad;
+
+	if (rmpp_mad->rmpp_hdr.rmpp_status < IB_MGMT_RMPP_STATUS_ABORT_MIN ||
+	    rmpp_mad->rmpp_hdr.rmpp_status > IB_MGMT_RMPP_STATUS_ABORT_MAX) {
+		abort_send(agent, rmpp_mad->mad_hdr.tid,
+			   IB_MGMT_RMPP_STATUS_BAD_STATUS);
+		nack_recv(agent, mad_recv_wc, IB_MGMT_RMPP_STATUS_BAD_STATUS);
+	} else
+		abort_send(agent, rmpp_mad->mad_hdr.tid,
+			   rmpp_mad->rmpp_hdr.rmpp_status);
+}
+
 struct ib_mad_recv_wc *
 ib_process_rmpp_recv_wc(struct ib_mad_agent_private *agent,
 			struct ib_mad_recv_wc *mad_recv_wc)
@@ -638,23 +808,29 @@ ib_process_rmpp_recv_wc(struct ib_mad_agent_private *agent,
 	if (!(rmpp_mad->rmpp_hdr.rmpp_rtime_flags & IB_MGMT_RMPP_FLAG_ACTIVE))
 		return mad_recv_wc;
 
-	if (rmpp_mad->rmpp_hdr.rmpp_version != IB_MGMT_RMPP_VERSION)
+	if (rmpp_mad->rmpp_hdr.rmpp_version != IB_MGMT_RMPP_VERSION) {
+		abort_send(agent, rmpp_mad->mad_hdr.tid,
+			   IB_MGMT_RMPP_STATUS_UNV);
+		nack_recv(agent, mad_recv_wc, IB_MGMT_RMPP_STATUS_UNV);
 		goto out;
+	}
 
 	switch (rmpp_mad->rmpp_hdr.rmpp_type) {
 	case IB_MGMT_RMPP_TYPE_DATA:
-		if (rmpp_mad->rmpp_hdr.seg_num == __constant_htonl(1))
-			return start_rmpp(agent, mad_recv_wc);
-		else
-			return continue_rmpp(agent, mad_recv_wc);
+		return process_rmpp_data(agent, mad_recv_wc);
 	case IB_MGMT_RMPP_TYPE_ACK:
 		process_rmpp_ack(agent, mad_recv_wc);
 		break;
 	case IB_MGMT_RMPP_TYPE_STOP:
+		process_rmpp_stop(agent, mad_recv_wc);
+		break;
 	case IB_MGMT_RMPP_TYPE_ABORT:
-		/* TODO: process_rmpp_nack(agent, mad_recv_wc); */
+		process_rmpp_abort(agent, mad_recv_wc);
 		break;
 	default:
+		abort_send(agent, rmpp_mad->mad_hdr.tid,
+			   IB_MGMT_RMPP_STATUS_BADT);
+		nack_recv(agent, mad_recv_wc, IB_MGMT_RMPP_STATUS_BADT);
 		break;
 	}
 out:
@@ -714,7 +890,10 @@ int ib_process_rmpp_send_wc(struct ib_mad_send_wr_private *mad_send_wr,
 	if (rmpp_mad->rmpp_hdr.rmpp_type != IB_MGMT_RMPP_TYPE_DATA) {
 		msg = (struct ib_mad_send_buf *) (unsigned long)
 		      mad_send_wc->wr_id;
-		ib_free_send_mad(msg);
+		if (rmpp_mad->rmpp_hdr.rmpp_type == IB_MGMT_RMPP_TYPE_ACK)
+			ib_free_send_mad(msg);
+		else
+			free_msg(msg);
 		return IB_RMPP_RESULT_INTERNAL;	 /* ACK, STOP, or ABORT */
 	}
 

drivers/infiniband/core/packer.c

@@ -1,5 +1,6 @@
 /*
  * Copyright (c) 2004 Topspin Corporation.  All rights reserved.
+ * Copyright (c) 2005 Sun Microsystems, Inc. All rights reserved.
  *
  * This software is available to you under a choice of one of two
  * licenses.  You may choose to be licensed under the terms of the GNU
@@ -32,7 +33,7 @@
  * $Id: packer.c 1349 2004-12-16 21:09:43Z roland $
  */
 
-#include <ib_pack.h>
+#include <rdma/ib_pack.h>
 
 static u64 value_read(int offset, int size, void *structure)
 {

drivers/infiniband/core/sa_query.c

@@ -1,6 +1,6 @@
 /*
  * Copyright (c) 2004 Topspin Communications.  All rights reserved.
- * Copyright (c) 2005 Voltaire, Inc.  All rights reserved.
+ * Copyright (c) 2005 Voltaire, Inc.  All rights reserved.
  *
  * This software is available to you under a choice of one of two
  * licenses.  You may choose to be licensed under the terms of the GNU
@@ -44,8 +44,8 @@
 #include <linux/kref.h>
 #include <linux/idr.h>
 
-#include <ib_pack.h>
-#include <ib_sa.h>
+#include <rdma/ib_pack.h>
+#include <rdma/ib_sa.h>
 
 MODULE_AUTHOR("Roland Dreier");
 MODULE_DESCRIPTION("InfiniBand subnet administration query support");

drivers/infiniband/core/smi.c

@@ -1,9 +1,10 @@
 /*
- * Copyright (c) 2004 Mellanox Technologies Ltd.  All rights reserved.
- * Copyright (c) 2004 Infinicon Corporation.  All rights reserved.
- * Copyright (c) 2004 Intel Corporation.  All rights reserved.
- * Copyright (c) 2004 Topspin Corporation.  All rights reserved.
- * Copyright (c) 2004 Voltaire Corporation.  All rights reserved.
+ * Copyright (c) 2004, 2005 Mellanox Technologies Ltd.  All rights reserved.
+ * Copyright (c) 2004, 2005 Infinicon Corporation.  All rights reserved.
+ * Copyright (c) 2004, 2005 Intel Corporation.  All rights reserved.
+ * Copyright (c) 2004, 2005 Topspin Corporation.  All rights reserved.
+ * Copyright (c) 2004, 2005 Voltaire Corporation.  All rights reserved.
+ * Copyright (c) 2005 Sun Microsystems, Inc. All rights reserved.
  *
  * This software is available to you under a choice of one of two
  * licenses.  You may choose to be licensed under the terms of the GNU
@@ -36,7 +37,7 @@
  * $Id: smi.c 1389 2004-12-27 22:56:47Z roland $
  */
 
-#include <ib_smi.h>
+#include <rdma/ib_smi.h>
 #include "smi.h"
 
 /*

drivers/infiniband/core/sysfs.c

@@ -1,5 +1,7 @@
 /*
  * Copyright (c) 2004, 2005 Topspin Communications.  All rights reserved.
+ * Copyright (c) 2005 Mellanox Technologies Ltd.  All rights reserved.
+ * Copyright (c) 2005 Sun Microsystems, Inc. All rights reserved.
  *
  * This software is available to you under a choice of one of two
  * licenses.  You may choose to be licensed under the terms of the GNU
@@ -34,7 +36,7 @@
 
 #include "core_priv.h"
 
-#include <ib_mad.h>
+#include <rdma/ib_mad.h>
 
 struct ib_port {
 	struct kobject         kobj;
@@ -253,14 +255,14 @@ static ssize_t show_port_gid(struct ib_port *p, struct port_attribute *attr,
 		return ret;
 
 	return sprintf(buf, "%04x:%04x:%04x:%04x:%04x:%04x:%04x:%04x\n",
-		       be16_to_cpu(((u16 *) gid.raw)[0]),
-		       be16_to_cpu(((u16 *) gid.raw)[1]),
-		       be16_to_cpu(((u16 *) gid.raw)[2]),
-		       be16_to_cpu(((u16 *) gid.raw)[3]),
-		       be16_to_cpu(((u16 *) gid.raw)[4]),
-		       be16_to_cpu(((u16 *) gid.raw)[5]),
-		       be16_to_cpu(((u16 *) gid.raw)[6]),
-		       be16_to_cpu(((u16 *) gid.raw)[7]));
+		       be16_to_cpu(((__be16 *) gid.raw)[0]),
+		       be16_to_cpu(((__be16 *) gid.raw)[1]),
+		       be16_to_cpu(((__be16 *) gid.raw)[2]),
+		       be16_to_cpu(((__be16 *) gid.raw)[3]),
+		       be16_to_cpu(((__be16 *) gid.raw)[4]),
+		       be16_to_cpu(((__be16 *) gid.raw)[5]),
+		       be16_to_cpu(((__be16 *) gid.raw)[6]),
+		       be16_to_cpu(((__be16 *) gid.raw)[7]));
 }
 
 static ssize_t show_port_pkey(struct ib_port *p, struct port_attribute *attr,
@@ -332,11 +334,11 @@ static ssize_t show_pma_counter(struct ib_port *p, struct port_attribute *attr,
 		break;
 	case 16:
 		ret = sprintf(buf, "%u\n",
-			      be16_to_cpup((u16 *)(out_mad->data + 40 + offset / 8)));
+			      be16_to_cpup((__be16 *)(out_mad->data + 40 + offset / 8)));
 		break;
 	case 32:
 		ret = sprintf(buf, "%u\n",
-			      be32_to_cpup((u32 *)(out_mad->data + 40 + offset / 8)));
+			      be32_to_cpup((__be32 *)(out_mad->data + 40 + offset / 8)));
 		break;
 	default:
 		ret = 0;
@@ -598,10 +600,10 @@ static ssize_t show_sys_image_guid(struct class_device *cdev, char *buf)
 		return ret;
 
 	return sprintf(buf, "%04x:%04x:%04x:%04x\n",
-		       be16_to_cpu(((u16 *) &attr.sys_image_guid)[0]),
-		       be16_to_cpu(((u16 *) &attr.sys_image_guid)[1]),
-		       be16_to_cpu(((u16 *) &attr.sys_image_guid)[2]),
-		       be16_to_cpu(((u16 *) &attr.sys_image_guid)[3]));
+		       be16_to_cpu(((__be16 *) &attr.sys_image_guid)[0]),
+		       be16_to_cpu(((__be16 *) &attr.sys_image_guid)[1]),
+		       be16_to_cpu(((__be16 *) &attr.sys_image_guid)[2]),
+		       be16_to_cpu(((__be16 *) &attr.sys_image_guid)[3]));
 }
 
 static ssize_t show_node_guid(struct class_device *cdev, char *buf)
@@ -615,10 +617,10 @@ static ssize_t show_node_guid(struct class_device *cdev, char *buf)
 		return ret;
 
 	return sprintf(buf, "%04x:%04x:%04x:%04x\n",
-		       be16_to_cpu(((u16 *) &attr.node_guid)[0]),
-		       be16_to_cpu(((u16 *) &attr.node_guid)[1]),
-		       be16_to_cpu(((u16 *) &attr.node_guid)[2]),
-		       be16_to_cpu(((u16 *) &attr.node_guid)[3]));
+		       be16_to_cpu(((__be16 *) &attr.node_guid)[0]),
+		       be16_to_cpu(((__be16 *) &attr.node_guid)[1]),
+		       be16_to_cpu(((__be16 *) &attr.node_guid)[2]),
+		       be16_to_cpu(((__be16 *) &attr.node_guid)[3]));
 }
 
 static CLASS_DEVICE_ATTR(node_type, S_IRUGO, show_node_type, NULL);

drivers/infiniband/core/ucm.c

@@ -1,5 +1,6 @@
 /*
  * Copyright (c) 2005 Topspin Communications.  All rights reserved.
+ * Copyright (c) 2005 Intel Corporation.  All rights reserved.
  *
  * This software is available to you under a choice of one of two
  * licenses.  You may choose to be licensed under the terms of the GNU
@@ -73,14 +74,18 @@ static struct semaphore ctx_id_mutex;
 static struct idr       ctx_id_table;
 static int              ctx_id_rover = 0;
 
-static struct ib_ucm_context *ib_ucm_ctx_get(int id)
+static struct ib_ucm_context *ib_ucm_ctx_get(struct ib_ucm_file *file, int id)
 {
 	struct ib_ucm_context *ctx;
 
 	down(&ctx_id_mutex);
 	ctx = idr_find(&ctx_id_table, id);
-	if (ctx)
-		ctx->ref++;
+	if (!ctx)
+		ctx = ERR_PTR(-ENOENT);
+	else if (ctx->file != file)
+		ctx = ERR_PTR(-EINVAL);
+	else
+		atomic_inc(&ctx->ref);
 	up(&ctx_id_mutex);
 
 	return ctx;
@@ -88,21 +93,37 @@ static struct ib_ucm_context *ib_ucm_ctx_get(int id)
 
 static void ib_ucm_ctx_put(struct ib_ucm_context *ctx)
 {
+	if (atomic_dec_and_test(&ctx->ref))
+		wake_up(&ctx->wait);
+}
+
+static ssize_t ib_ucm_destroy_ctx(struct ib_ucm_file *file, int id)
+{
+	struct ib_ucm_context *ctx;
 	struct ib_ucm_event *uevent;
 
 	down(&ctx_id_mutex);
-
-	ctx->ref--;
-	if (!ctx->ref)
+	ctx = idr_find(&ctx_id_table, id);
+	if (!ctx)
+		ctx = ERR_PTR(-ENOENT);
+	else if (ctx->file != file)
+		ctx = ERR_PTR(-EINVAL);
+	else
 		idr_remove(&ctx_id_table, ctx->id);
-
 	up(&ctx_id_mutex);
 
-	if (ctx->ref)
-		return;
+	if (IS_ERR(ctx))
+		return PTR_ERR(ctx);
 
-	down(&ctx->file->mutex);
+	atomic_dec(&ctx->ref);
+	wait_event(ctx->wait, !atomic_read(&ctx->ref));
 
+	/* No new events will be generated after destroying the cm_id. */
+	if (!IS_ERR(ctx->cm_id))
+		ib_destroy_cm_id(ctx->cm_id);
+
+	/* Cleanup events not yet reported to the user. */
+	down(&file->mutex);
 	list_del(&ctx->file_list);
 	while (!list_empty(&ctx->events)) {
 
@@ -117,13 +138,10 @@ static void ib_ucm_ctx_put(struct ib_ucm_context *ctx)
 
 		kfree(uevent);
 	}
+	up(&file->mutex);
 
-	up(&ctx->file->mutex);
-
-	ucm_dbg("Destroyed CM ID <%d>\n", ctx->id);
-
-	ib_destroy_cm_id(ctx->cm_id);
 	kfree(ctx);
+	return 0;
 }
 
 static struct ib_ucm_context *ib_ucm_ctx_alloc(struct ib_ucm_file *file)
@@ -135,11 +153,11 @@ static struct ib_ucm_context *ib_ucm_ctx_alloc(struct ib_ucm_file *file)
 	if (!ctx)
 		return NULL;
 
-	ctx->ref  = 1; /* user reference */
+	atomic_set(&ctx->ref, 1);
+	init_waitqueue_head(&ctx->wait);
 	ctx->file = file;
 
 	INIT_LIST_HEAD(&ctx->events);
-	init_MUTEX(&ctx->mutex);
 
 	list_add_tail(&ctx->file_list, &file->ctxs);
 
@@ -177,8 +195,8 @@ static void ib_ucm_event_path_get(struct ib_ucm_path_rec *upath,
 	if (!kpath || !upath)
 		return;
 
-	memcpy(upath->dgid, kpath->dgid.raw, sizeof(union ib_gid));
-	memcpy(upath->sgid, kpath->sgid.raw, sizeof(union ib_gid));
+	memcpy(upath->dgid, kpath->dgid.raw, sizeof *upath->dgid);
+	memcpy(upath->sgid, kpath->sgid.raw, sizeof *upath->sgid);
 
 	upath->dlid             = kpath->dlid;
 	upath->slid             = kpath->slid;
@@ -201,10 +219,11 @@ static void ib_ucm_event_path_get(struct ib_ucm_path_rec *upath,
 		kpath->packet_life_time_selector;
 }
 
-static void ib_ucm_event_req_get(struct ib_ucm_req_event_resp *ureq,
+static void ib_ucm_event_req_get(struct ib_ucm_context *ctx,
+				 struct ib_ucm_req_event_resp *ureq,
 				 struct ib_cm_req_event_param *kreq)
 {
-	ureq->listen_id = (long)kreq->listen_id->context;
+	ureq->listen_id = ctx->id;
 
 	ureq->remote_ca_guid             = kreq->remote_ca_guid;
 	ureq->remote_qkey                = kreq->remote_qkey;
@@ -240,34 +259,11 @@ static void ib_ucm_event_rep_get(struct ib_ucm_rep_event_resp *urep,
 	urep->srq                 = krep->srq;
 }
 
-static void ib_ucm_event_rej_get(struct ib_ucm_rej_event_resp *urej,
-				 struct ib_cm_rej_event_param *krej)
-{
-	urej->reason = krej->reason;
-}
-
-static void ib_ucm_event_mra_get(struct ib_ucm_mra_event_resp *umra,
-				 struct ib_cm_mra_event_param *kmra)
-{
-	umra->timeout = kmra->service_timeout;
-}
-
-static void ib_ucm_event_lap_get(struct ib_ucm_lap_event_resp *ulap,
-				 struct ib_cm_lap_event_param *klap)
-{
-	ib_ucm_event_path_get(&ulap->path, klap->alternate_path);
-}
-
-static void ib_ucm_event_apr_get(struct ib_ucm_apr_event_resp *uapr,
-				 struct ib_cm_apr_event_param *kapr)
-{
-	uapr->status = kapr->ap_status;
-}
-
-static void ib_ucm_event_sidr_req_get(struct ib_ucm_sidr_req_event_resp *ureq,
+static void ib_ucm_event_sidr_req_get(struct ib_ucm_context *ctx,
+				      struct ib_ucm_sidr_req_event_resp *ureq,
 				      struct ib_cm_sidr_req_event_param *kreq)
 {
-	ureq->listen_id = (long)kreq->listen_id->context;
+	ureq->listen_id = ctx->id;
 	ureq->pkey      = kreq->pkey;
 }
 
@@ -279,19 +275,18 @@ static void ib_ucm_event_sidr_rep_get(struct ib_ucm_sidr_rep_event_resp *urep,
 	urep->qpn    = krep->qpn;
 };
 
-static int ib_ucm_event_process(struct ib_cm_event *evt,
+static int ib_ucm_event_process(struct ib_ucm_context *ctx,
+				struct ib_cm_event *evt,
 				struct ib_ucm_event *uvt)
 {
 	void *info = NULL;
-	int result;
 
 	switch (evt->event) {
 	case IB_CM_REQ_RECEIVED:
-		ib_ucm_event_req_get(&uvt->resp.u.req_resp,
+		ib_ucm_event_req_get(ctx, &uvt->resp.u.req_resp,
 				     &evt->param.req_rcvd);
 		uvt->data_len      = IB_CM_REQ_PRIVATE_DATA_SIZE;
-		uvt->resp.present |= (evt->param.req_rcvd.primary_path ?
-				      IB_UCM_PRES_PRIMARY : 0);
+		uvt->resp.present  = IB_UCM_PRES_PRIMARY;
 		uvt->resp.present |= (evt->param.req_rcvd.alternate_path ?
 				      IB_UCM_PRES_ALTERNATE : 0);
 		break;
@@ -299,57 +294,46 @@ static int ib_ucm_event_process(struct ib_cm_event *evt,
 		ib_ucm_event_rep_get(&uvt->resp.u.rep_resp,
 				     &evt->param.rep_rcvd);
 		uvt->data_len = IB_CM_REP_PRIVATE_DATA_SIZE;
-
 		break;
 	case IB_CM_RTU_RECEIVED:
 		uvt->data_len = IB_CM_RTU_PRIVATE_DATA_SIZE;
 		uvt->resp.u.send_status = evt->param.send_status;
-
 		break;
 	case IB_CM_DREQ_RECEIVED:
 		uvt->data_len = IB_CM_DREQ_PRIVATE_DATA_SIZE;
 		uvt->resp.u.send_status = evt->param.send_status;
-
 		break;
 	case IB_CM_DREP_RECEIVED:
 		uvt->data_len = IB_CM_DREP_PRIVATE_DATA_SIZE;
 		uvt->resp.u.send_status = evt->param.send_status;
-
 		break;
 	case IB_CM_MRA_RECEIVED:
-		ib_ucm_event_mra_get(&uvt->resp.u.mra_resp,
-				     &evt->param.mra_rcvd);
+		uvt->resp.u.mra_resp.timeout =
+					evt->param.mra_rcvd.service_timeout;
 		uvt->data_len = IB_CM_MRA_PRIVATE_DATA_SIZE;
-
 		break;
 	case IB_CM_REJ_RECEIVED:
-		ib_ucm_event_rej_get(&uvt->resp.u.rej_resp,
-				     &evt->param.rej_rcvd);
+		uvt->resp.u.rej_resp.reason = evt->param.rej_rcvd.reason;
 		uvt->data_len = IB_CM_REJ_PRIVATE_DATA_SIZE;
 		uvt->info_len = evt->param.rej_rcvd.ari_length;
 		info	      = evt->param.rej_rcvd.ari;
-
 		break;
 	case IB_CM_LAP_RECEIVED:
-		ib_ucm_event_lap_get(&uvt->resp.u.lap_resp,
-				     &evt->param.lap_rcvd);
+		ib_ucm_event_path_get(&uvt->resp.u.lap_resp.path,
+				      evt->param.lap_rcvd.alternate_path);
 		uvt->data_len = IB_CM_LAP_PRIVATE_DATA_SIZE;
-		uvt->resp.present |= (evt->param.lap_rcvd.alternate_path ?
-				      IB_UCM_PRES_ALTERNATE : 0);
+		uvt->resp.present = IB_UCM_PRES_ALTERNATE;
 		break;
 	case IB_CM_APR_RECEIVED:
-		ib_ucm_event_apr_get(&uvt->resp.u.apr_resp,
-				     &evt->param.apr_rcvd);
+		uvt->resp.u.apr_resp.status = evt->param.apr_rcvd.ap_status;
 		uvt->data_len = IB_CM_APR_PRIVATE_DATA_SIZE;
 		uvt->info_len = evt->param.apr_rcvd.info_len;
 		info	      = evt->param.apr_rcvd.apr_info;
-
 		break;
 	case IB_CM_SIDR_REQ_RECEIVED:
-		ib_ucm_event_sidr_req_get(&uvt->resp.u.sidr_req_resp,
+		ib_ucm_event_sidr_req_get(ctx, &uvt->resp.u.sidr_req_resp,
 					  &evt->param.sidr_req_rcvd);
 		uvt->data_len = IB_CM_SIDR_REQ_PRIVATE_DATA_SIZE;
-
 		break;
 	case IB_CM_SIDR_REP_RECEIVED:
 		ib_ucm_event_sidr_rep_get(&uvt->resp.u.sidr_rep_resp,
@@ -357,43 +341,35 @@ static int ib_ucm_event_process(struct ib_cm_event *evt,
 		uvt->data_len = IB_CM_SIDR_REP_PRIVATE_DATA_SIZE;
 		uvt->info_len = evt->param.sidr_rep_rcvd.info_len;
 		info	      = evt->param.sidr_rep_rcvd.info;
-
 		break;
 	default:
 		uvt->resp.u.send_status = evt->param.send_status;
-
 		break;
 	}
 
-	if (uvt->data_len && evt->private_data) {
-
+	if (uvt->data_len) {
 		uvt->data = kmalloc(uvt->data_len, GFP_KERNEL);
-		if (!uvt->data) {
-			result = -ENOMEM;
-			goto error;
-		}
+		if (!uvt->data)
+			goto err1;
 
 		memcpy(uvt->data, evt->private_data, uvt->data_len);
 		uvt->resp.present |= IB_UCM_PRES_DATA;
 	}
 
-	if (uvt->info_len && info) {
-
+	if (uvt->info_len) {
 		uvt->info = kmalloc(uvt->info_len, GFP_KERNEL);
-		if (!uvt->info) {
-			result = -ENOMEM;
-			goto error;
-		}
+		if (!uvt->info)
+			goto err2;
 
 		memcpy(uvt->info, info, uvt->info_len);
 		uvt->resp.present |= IB_UCM_PRES_INFO;
 	}
-
 	return 0;
-error:
-	kfree(uvt->info);
+
+err2:
 	kfree(uvt->data);
-	return result;
+err1:
+	return -ENOMEM;
 }
 
 static int ib_ucm_event_handler(struct ib_cm_id *cm_id,
@@ -403,63 +379,42 @@ static int ib_ucm_event_handler(struct ib_cm_id *cm_id,
 	struct ib_ucm_context *ctx;
 	int result = 0;
 	int id;
-	/*
-	 * lookup correct context based on event type.
-	 */
-	switch (event->event) {
-	case IB_CM_REQ_RECEIVED:
-		id = (long)event->param.req_rcvd.listen_id->context;
-		break;
-	case IB_CM_SIDR_REQ_RECEIVED:
-		id = (long)event->param.sidr_req_rcvd.listen_id->context;
-		break;
-	default:
-		id = (long)cm_id->context;
-		break;
-	}
 
-	ucm_dbg("Event. CM ID <%d> event <%d>\n", id, event->event);
-
-	ctx = ib_ucm_ctx_get(id);
-	if (!ctx)
-		return -ENOENT;
+	ctx = cm_id->context;
 
 	if (event->event == IB_CM_REQ_RECEIVED ||
 	    event->event == IB_CM_SIDR_REQ_RECEIVED)
 		id = IB_UCM_CM_ID_INVALID;
+	else
+		id = ctx->id;
 
 	uevent = kmalloc(sizeof(*uevent), GFP_KERNEL);
-	if (!uevent) {
-		result = -ENOMEM;
-		goto done;
-	}
+	if (!uevent)
+		goto err1;
 
 	memset(uevent, 0, sizeof(*uevent));
-
 	uevent->resp.id    = id;
 	uevent->resp.event = event->event;
 
-	result = ib_ucm_event_process(event, uevent);
+	result = ib_ucm_event_process(ctx, event, uevent);
 	if (result)
-		goto done;
+		goto err2;
 
 	uevent->ctx   = ctx;
-	uevent->cm_id = ((event->event == IB_CM_REQ_RECEIVED ||
-			  event->event == IB_CM_SIDR_REQ_RECEIVED ) ?
-			 cm_id : NULL);
+	uevent->cm_id = (id == IB_UCM_CM_ID_INVALID) ? cm_id : NULL;
 
 	down(&ctx->file->mutex);
-
 	list_add_tail(&uevent->file_list, &ctx->file->events);
 	list_add_tail(&uevent->ctx_list, &ctx->events);
-
 	wake_up_interruptible(&ctx->file->poll_wait);
-
 	up(&ctx->file->mutex);
-done:
-	ctx->error = result;
-	ib_ucm_ctx_put(ctx); /* func reference */
-	return result;
+	return 0;
+
+err2:
+	kfree(uevent);
+err1:
+	/* Destroy new cm_id's */
+	return (id == IB_UCM_CM_ID_INVALID);
 }
 
 static ssize_t ib_ucm_event(struct ib_ucm_file *file,
@@ -517,9 +472,8 @@ static ssize_t ib_ucm_event(struct ib_ucm_file *file,
 		goto done;
 	}
 
-	ctx->cm_id             = uevent->cm_id;
-	ctx->cm_id->cm_handler = ib_ucm_event_handler;
-	ctx->cm_id->context    = (void *)(unsigned long)ctx->id;
+	ctx->cm_id          = uevent->cm_id;
+	ctx->cm_id->context = ctx;
 
 	uevent->resp.id = ctx->id;
 
@@ -585,30 +539,29 @@ static ssize_t ib_ucm_create_id(struct ib_ucm_file *file,
 	if (copy_from_user(&cmd, inbuf, sizeof(cmd)))
 		return -EFAULT;
 
+	down(&file->mutex);
 	ctx = ib_ucm_ctx_alloc(file);
+	up(&file->mutex);
 	if (!ctx)
 		return -ENOMEM;
 
-	ctx->cm_id = ib_create_cm_id(ib_ucm_event_handler,
-				     (void *)(unsigned long)ctx->id);
-	if (!ctx->cm_id) {
-		result = -ENOMEM;
-		goto err_cm;
+	ctx->cm_id = ib_create_cm_id(ib_ucm_event_handler, ctx);
+	if (IS_ERR(ctx->cm_id)) {
+		result = PTR_ERR(ctx->cm_id);
+		goto err;
 	}
 
 	resp.id = ctx->id;
 	if (copy_to_user((void __user *)(unsigned long)cmd.response,
 			 &resp, sizeof(resp))) {
 		result = -EFAULT;
-		goto err_ret;
+		goto err;
 	}
 
 	return 0;
-err_ret:
-	ib_destroy_cm_id(ctx->cm_id);
-err_cm:
-	ib_ucm_ctx_put(ctx); /* user reference */
 
+err:
+	ib_ucm_destroy_ctx(file, ctx->id);
 	return result;
 }
 
@@ -617,19 +570,11 @@ static ssize_t ib_ucm_destroy_id(struct ib_ucm_file *file,
 				 int in_len, int out_len)
 {
 	struct ib_ucm_destroy_id cmd;
-	struct ib_ucm_context *ctx;
 
 	if (copy_from_user(&cmd, inbuf, sizeof(cmd)))
 		return -EFAULT;
 
-	ctx = ib_ucm_ctx_get(cmd.id);
-	if (!ctx)
-		return -ENOENT;
-
-	ib_ucm_ctx_put(ctx); /* user reference */
-	ib_ucm_ctx_put(ctx); /* func reference */
-
-	return 0;
+	return ib_ucm_destroy_ctx(file, cmd.id);
 }
 
 static ssize_t ib_ucm_attr_id(struct ib_ucm_file *file,
@@ -647,15 +592,9 @@ static ssize_t ib_ucm_attr_id(struct ib_ucm_file *file,
 	if (copy_from_user(&cmd, inbuf, sizeof(cmd)))
 		return -EFAULT;
 
-	ctx = ib_ucm_ctx_get(cmd.id);
-	if (!ctx)
-		return -ENOENT;
-
- 	down(&ctx->file->mutex);
-	if (ctx->file != file) {
-		result = -EINVAL;
-		goto done;
-	}
+	ctx = ib_ucm_ctx_get(file, cmd.id);
+	if (IS_ERR(ctx))
+		return PTR_ERR(ctx);
 
 	resp.service_id   = ctx->cm_id->service_id;
 	resp.service_mask = ctx->cm_id->service_mask;
@@ -666,9 +605,7 @@ static ssize_t ib_ucm_attr_id(struct ib_ucm_file *file,
 			 &resp, sizeof(resp)))
 		result = -EFAULT;
 
-done:
-	up(&ctx->file->mutex);
-	ib_ucm_ctx_put(ctx); /* func reference */
+	ib_ucm_ctx_put(ctx);
 	return result;
 }
 
@@ -683,19 +620,12 @@ static ssize_t ib_ucm_listen(struct ib_ucm_file *file,
 	if (copy_from_user(&cmd, inbuf, sizeof(cmd)))
 		return -EFAULT;
 
-	ctx = ib_ucm_ctx_get(cmd.id);
-	if (!ctx)
-		return -ENOENT;
+	ctx = ib_ucm_ctx_get(file, cmd.id);
+	if (IS_ERR(ctx))
+		return PTR_ERR(ctx);
 
- 	down(&ctx->file->mutex);
-	if (ctx->file != file)
-		result = -EINVAL;
-	else
-		result = ib_cm_listen(ctx->cm_id, cmd.service_id,
-				      cmd.service_mask);
-
-	up(&ctx->file->mutex);
-	ib_ucm_ctx_put(ctx); /* func reference */
+	result = ib_cm_listen(ctx->cm_id, cmd.service_id, cmd.service_mask);
+	ib_ucm_ctx_put(ctx);
 	return result;
 }
 
@@ -710,18 +640,12 @@ static ssize_t ib_ucm_establish(struct ib_ucm_file *file,
 	if (copy_from_user(&cmd, inbuf, sizeof(cmd)))
 		return -EFAULT;
 
-	ctx = ib_ucm_ctx_get(cmd.id);
-	if (!ctx)
-		return -ENOENT;
+	ctx = ib_ucm_ctx_get(file, cmd.id);
+	if (IS_ERR(ctx))
+		return PTR_ERR(ctx);
 
- 	down(&ctx->file->mutex);
-	if (ctx->file != file)
-		result = -EINVAL;
-	else
-		result = ib_cm_establish(ctx->cm_id);
-
-	up(&ctx->file->mutex);
-	ib_ucm_ctx_put(ctx); /* func reference */
+	result = ib_cm_establish(ctx->cm_id);
+	ib_ucm_ctx_put(ctx);
 	return result;
 }
 
@@ -768,8 +692,8 @@ static int ib_ucm_path_get(struct ib_sa_path_rec **path, u64 src)
 		return -EFAULT;
 	}
 
-	memcpy(sa_path->dgid.raw, ucm_path.dgid, sizeof(union ib_gid));
-	memcpy(sa_path->sgid.raw, ucm_path.sgid, sizeof(union ib_gid));
+	memcpy(sa_path->dgid.raw, ucm_path.dgid, sizeof sa_path->dgid);
+	memcpy(sa_path->sgid.raw, ucm_path.sgid, sizeof sa_path->sgid);
 
 	sa_path->dlid	          = ucm_path.dlid;
 	sa_path->slid	          = ucm_path.slid;
@@ -839,25 +763,17 @@ static ssize_t ib_ucm_send_req(struct ib_ucm_file *file,
 	param.max_cm_retries             = cmd.max_cm_retries;
 	param.srq                        = cmd.srq;
 
-	ctx = ib_ucm_ctx_get(cmd.id);
-	if (!ctx) {
-		result = -ENOENT;
-		goto done;
-	}
-
- 	down(&ctx->file->mutex);
-	if (ctx->file != file)
-		result = -EINVAL;
-	else
+	ctx = ib_ucm_ctx_get(file, cmd.id);
+	if (!IS_ERR(ctx)) {
 		result = ib_send_cm_req(ctx->cm_id, &param);
+		ib_ucm_ctx_put(ctx);
+	} else
+		result = PTR_ERR(ctx);
 
-	up(&ctx->file->mutex);
-	ib_ucm_ctx_put(ctx); /* func reference */
 done:
 	kfree(param.private_data);
 	kfree(param.primary_path);
 	kfree(param.alternate_path);
-
 	return result;
 }
 
@@ -890,23 +806,14 @@ static ssize_t ib_ucm_send_rep(struct ib_ucm_file *file,
 	param.rnr_retry_count     = cmd.rnr_retry_count;
 	param.srq                 = cmd.srq;
 
-	ctx = ib_ucm_ctx_get(cmd.id);
-	if (!ctx) {
-		result = -ENOENT;
-		goto done;
-	}
-
- 	down(&ctx->file->mutex);
-	if (ctx->file != file)
-		result = -EINVAL;
-	else
+	ctx = ib_ucm_ctx_get(file, cmd.id);
+	if (!IS_ERR(ctx)) {
 		result = ib_send_cm_rep(ctx->cm_id, &param);
+		ib_ucm_ctx_put(ctx);
+	} else
+		result = PTR_ERR(ctx);
 
-	up(&ctx->file->mutex);
-	ib_ucm_ctx_put(ctx); /* func reference */
-done:
 	kfree(param.private_data);
-
 	return result;
 }
 
@@ -928,23 +835,14 @@ static ssize_t ib_ucm_send_private_data(struct ib_ucm_file *file,
 	if (result)
 		return result;
 
-	ctx = ib_ucm_ctx_get(cmd.id);
-	if (!ctx) {
-		result = -ENOENT;
-		goto done;
-	}
-
- 	down(&ctx->file->mutex);
-	if (ctx->file != file)
-		result = -EINVAL;
-	else
+	ctx = ib_ucm_ctx_get(file, cmd.id);
+	if (!IS_ERR(ctx)) {
 		result = func(ctx->cm_id, private_data, cmd.len);
+		ib_ucm_ctx_put(ctx);
+	} else
+		result = PTR_ERR(ctx);
 
-	up(&ctx->file->mutex);
-	ib_ucm_ctx_put(ctx); /* func reference */
-done:
 	kfree(private_data);
-
 	return result;
 }
 
@@ -995,26 +893,17 @@ static ssize_t ib_ucm_send_info(struct ib_ucm_file *file,
 	if (result)
 		goto done;
 
-	ctx = ib_ucm_ctx_get(cmd.id);
-	if (!ctx) {
-		result = -ENOENT;
-		goto done;
-	}
-
- 	down(&ctx->file->mutex);
-	if (ctx->file != file)
-		result = -EINVAL;
-	else
-		result = func(ctx->cm_id, cmd.status,
-			      info, cmd.info_len,
+	ctx = ib_ucm_ctx_get(file, cmd.id);
+	if (!IS_ERR(ctx)) {
+		result = func(ctx->cm_id, cmd.status, info, cmd.info_len,
 			      data, cmd.data_len);
+		ib_ucm_ctx_put(ctx);
+	} else
+		result = PTR_ERR(ctx);
 
-	up(&ctx->file->mutex);
-	ib_ucm_ctx_put(ctx); /* func reference */
 done:
 	kfree(data);
 	kfree(info);
-
 	return result;
 }
 
@@ -1048,24 +937,14 @@ static ssize_t ib_ucm_send_mra(struct ib_ucm_file *file,
 	if (result)
 		return result;
 
-	ctx = ib_ucm_ctx_get(cmd.id);
-	if (!ctx) {
-		result = -ENOENT;
-		goto done;
-	}
+	ctx = ib_ucm_ctx_get(file, cmd.id);
+	if (!IS_ERR(ctx)) {
+		result = ib_send_cm_mra(ctx->cm_id, cmd.timeout, data, cmd.len);
+		ib_ucm_ctx_put(ctx);
+	} else
+		result = PTR_ERR(ctx);
 
- 	down(&ctx->file->mutex);
-	if (ctx->file != file)
-		result = -EINVAL;
-	else
-		result = ib_send_cm_mra(ctx->cm_id, cmd.timeout,
-					data, cmd.len);
-
-	up(&ctx->file->mutex);
-	ib_ucm_ctx_put(ctx); /* func reference */
-done:
 	kfree(data);
-
 	return result;
 }
 
@@ -1090,24 +969,16 @@ static ssize_t ib_ucm_send_lap(struct ib_ucm_file *file,
 	if (result)
 		goto done;
 
-	ctx = ib_ucm_ctx_get(cmd.id);
-	if (!ctx) {
-		result = -ENOENT;
-		goto done;
-	}
-
- 	down(&ctx->file->mutex);
-	if (ctx->file != file)
-		result = -EINVAL;
-	else
+	ctx = ib_ucm_ctx_get(file, cmd.id);
+	if (!IS_ERR(ctx)) {
 		result = ib_send_cm_lap(ctx->cm_id, path, data, cmd.len);
+		ib_ucm_ctx_put(ctx);
+	} else
+		result = PTR_ERR(ctx);
 
-	up(&ctx->file->mutex);
-	ib_ucm_ctx_put(ctx); /* func reference */
 done:
 	kfree(data);
 	kfree(path);
-
 	return result;
 }
 
@@ -1140,24 +1011,16 @@ static ssize_t ib_ucm_send_sidr_req(struct ib_ucm_file *file,
 	param.max_cm_retries   = cmd.max_cm_retries;
 	param.pkey             = cmd.pkey;
 
-	ctx = ib_ucm_ctx_get(cmd.id);
-	if (!ctx) {
-		result = -ENOENT;
-		goto done;
-	}
-
- 	down(&ctx->file->mutex);
-	if (ctx->file != file)
-		result = -EINVAL;
-	else
+	ctx = ib_ucm_ctx_get(file, cmd.id);
+	if (!IS_ERR(ctx)) {
 		result = ib_send_cm_sidr_req(ctx->cm_id, &param);
+		ib_ucm_ctx_put(ctx);
+	} else
+		result = PTR_ERR(ctx);
 
-	up(&ctx->file->mutex);
-	ib_ucm_ctx_put(ctx); /* func reference */
 done:
 	kfree(param.private_data);
 	kfree(param.path);
-
 	return result;
 }
 
@@ -1184,30 +1047,22 @@ static ssize_t ib_ucm_send_sidr_rep(struct ib_ucm_file *file,
 	if (result)
 		goto done;
 
-	param.qp_num	   = cmd.qpn;
-	param.qkey	     = cmd.qkey;
-	param.status	   = cmd.status;
-	param.info_length      = cmd.info_len;
-	param.private_data_len = cmd.data_len;
+	param.qp_num		= cmd.qpn;
+	param.qkey		= cmd.qkey;
+	param.status		= cmd.status;
+	param.info_length	= cmd.info_len;
+	param.private_data_len	= cmd.data_len;
 
-	ctx = ib_ucm_ctx_get(cmd.id);
-	if (!ctx) {
-		result = -ENOENT;
-		goto done;
-	}
-
- 	down(&ctx->file->mutex);
-	if (ctx->file != file)
-		result = -EINVAL;
-	else
+	ctx = ib_ucm_ctx_get(file, cmd.id);
+	if (!IS_ERR(ctx)) {
 		result = ib_send_cm_sidr_rep(ctx->cm_id, &param);
+		ib_ucm_ctx_put(ctx);
+	} else
+		result = PTR_ERR(ctx);
 
-	up(&ctx->file->mutex);
-	ib_ucm_ctx_put(ctx); /* func reference */
 done:
 	kfree(param.private_data);
 	kfree(param.info);
-
 	return result;
 }
 
@@ -1305,22 +1160,17 @@ static int ib_ucm_close(struct inode *inode, struct file *filp)
 	struct ib_ucm_context *ctx;
 
 	down(&file->mutex);
-
 	while (!list_empty(&file->ctxs)) {
 
 		ctx = list_entry(file->ctxs.next,
 				 struct ib_ucm_context, file_list);
 
-		up(&ctx->file->mutex);
-		ib_ucm_ctx_put(ctx); /* user reference */
+		up(&file->mutex);
+		ib_ucm_destroy_ctx(file, ctx->id);
 		down(&file->mutex);
 	}
-
 	up(&file->mutex);
-
 	kfree(file);
-
-	ucm_dbg("Deleted struct\n");
 	return 0;
 }
 

drivers/infiniband/core/ucm.h

@@ -40,17 +40,15 @@
 #include <linux/cdev.h>
 #include <linux/idr.h>
 
-#include <ib_cm.h>
-#include <ib_user_cm.h>
+#include <rdma/ib_cm.h>
+#include <rdma/ib_user_cm.h>
 
 #define IB_UCM_CM_ID_INVALID 0xffffffff
 
 struct ib_ucm_file {
 	struct semaphore mutex;
 	struct file *filp;
-	/*
-	 * list of pending events
-	 */
+
 	struct list_head  ctxs;   /* list of active connections */
 	struct list_head  events; /* list of pending events */
 	wait_queue_head_t poll_wait;
@@ -58,12 +56,11 @@ struct ib_ucm_file {
 
 struct ib_ucm_context {
 	int                 id;
-	int                 ref;
-	int                 error;
+	wait_queue_head_t   wait;
+	atomic_t            ref;
 
 	struct ib_ucm_file *file;
 	struct ib_cm_id    *cm_id;
-	struct semaphore    mutex;
 
 	struct list_head    events;    /* list of pending events. */
 	struct list_head    file_list; /* member in file ctx list */

drivers/infiniband/core/ud_header.c

@@ -1,5 +1,6 @@
 /*
  * Copyright (c) 2004 Topspin Corporation.  All rights reserved.
+ * Copyright (c) 2005 Sun Microsystems, Inc. All rights reserved.
  *
  * This software is available to you under a choice of one of two
  * licenses.  You may choose to be licensed under the terms of the GNU
@@ -34,7 +35,7 @@
 
 #include <linux/errno.h>
 
-#include <ib_pack.h>
+#include <rdma/ib_pack.h>
 
 #define STRUCT_FIELD(header, field) \
 	.struct_offset_bytes = offsetof(struct ib_unpacked_ ## header, field),      \
@@ -194,6 +195,7 @@ void ib_ud_header_init(int     		    payload_bytes,
 		       struct ib_ud_header *header)
 {
 	int header_len;
+	u16 packet_length;
 
 	memset(header, 0, sizeof *header);
 
@@ -208,7 +210,7 @@ void ib_ud_header_init(int     		    payload_bytes,
 	header->lrh.link_version     = 0;
 	header->lrh.link_next_header =
 		grh_present ? IB_LNH_IBA_GLOBAL : IB_LNH_IBA_LOCAL;
-	header->lrh.packet_length    = (IB_LRH_BYTES     +
+	packet_length		     = (IB_LRH_BYTES     +
 					IB_BTH_BYTES     +
 					IB_DETH_BYTES    +
 					payload_bytes    +
@@ -217,8 +219,7 @@ void ib_ud_header_init(int     		    payload_bytes,
 
 	header->grh_present          = grh_present;
 	if (grh_present) {
-		header->lrh.packet_length  += IB_GRH_BYTES / 4;
-
+		packet_length		   += IB_GRH_BYTES / 4;
 		header->grh.ip_version      = 6;
 		header->grh.payload_length  =
 			cpu_to_be16((IB_BTH_BYTES     +
@@ -229,7 +230,7 @@ void ib_ud_header_init(int     		    payload_bytes,
 		header->grh.next_header     = 0x1b;
 	}
 
-	cpu_to_be16s(&header->lrh.packet_length);
+	header->lrh.packet_length = cpu_to_be16(packet_length);
 
 	if (header->immediate_present)
 		header->bth.opcode           = IB_OPCODE_UD_SEND_ONLY_WITH_IMMEDIATE;

drivers/infiniband/core/user_mad.c

@@ -1,6 +1,6 @@
 /*
  * Copyright (c) 2004 Topspin Communications.  All rights reserved.
- * Copyright (c) 2005 Voltaire, Inc. All rights reserved.
+ * Copyright (c) 2005 Voltaire, Inc. All rights reserved. 
  * Copyright (c) 2005 Sun Microsystems, Inc. All rights reserved.
  *
  * This software is available to you under a choice of one of two
@@ -49,8 +49,8 @@
 #include <asm/uaccess.h>
 #include <asm/semaphore.h>
 
-#include <ib_mad.h>
-#include <ib_user_mad.h>
+#include <rdma/ib_mad.h>
+#include <rdma/ib_user_mad.h>
 
 MODULE_AUTHOR("Roland Dreier");
 MODULE_DESCRIPTION("InfiniBand userspace MAD packet access");
@@ -271,7 +271,7 @@ static ssize_t ib_umad_write(struct file *filp, const char __user *buf,
 	struct ib_send_wr *bad_wr;
 	struct ib_rmpp_mad *rmpp_mad;
 	u8 method;
-	u64 *tid;
+	__be64 *tid;
 	int ret, length, hdr_len, data_len, rmpp_hdr_size;
 	int rmpp_active = 0;
 
@@ -316,7 +316,7 @@ static ssize_t ib_umad_write(struct file *filp, const char __user *buf,
 	if (packet->mad.hdr.grh_present) {
 		ah_attr.ah_flags = IB_AH_GRH;
 		memcpy(ah_attr.grh.dgid.raw, packet->mad.hdr.gid, 16);
-		ah_attr.grh.flow_label 	   = packet->mad.hdr.flow_label;
+		ah_attr.grh.flow_label 	   = be32_to_cpu(packet->mad.hdr.flow_label);
 		ah_attr.grh.hop_limit  	   = packet->mad.hdr.hop_limit;
 		ah_attr.grh.traffic_class  = packet->mad.hdr.traffic_class;
 	}

drivers/infiniband/core/uverbs.h

@@ -1,6 +1,8 @@
 /*
  * Copyright (c) 2005 Topspin Communications.  All rights reserved.
  * Copyright (c) 2005 Cisco Systems.  All rights reserved.
+ * Copyright (c) 2005 Mellanox Technologies. All rights reserved.
+ * Copyright (c) 2005 Voltaire, Inc. All rights reserved.
  *
  * This software is available to you under a choice of one of two
  * licenses.  You may choose to be licensed under the terms of the GNU
@@ -43,8 +45,8 @@
 #include <linux/kref.h>
 #include <linux/idr.h>
 
-#include <ib_verbs.h>
-#include <ib_user_verbs.h>
+#include <rdma/ib_verbs.h>
+#include <rdma/ib_user_verbs.h>
 
 struct ib_uverbs_device {
 	int					devnum;
@@ -97,10 +99,12 @@ extern struct idr ib_uverbs_mw_idr;
 extern struct idr ib_uverbs_ah_idr;
 extern struct idr ib_uverbs_cq_idr;
 extern struct idr ib_uverbs_qp_idr;
+extern struct idr ib_uverbs_srq_idr;
 
 void ib_uverbs_comp_handler(struct ib_cq *cq, void *cq_context);
 void ib_uverbs_cq_event_handler(struct ib_event *event, void *context_ptr);
 void ib_uverbs_qp_event_handler(struct ib_event *event, void *context_ptr);
+void ib_uverbs_srq_event_handler(struct ib_event *event, void *context_ptr);
 
 int ib_umem_get(struct ib_device *dev, struct ib_umem *mem,
 		void *addr, size_t size, int write);
@@ -129,5 +133,8 @@ IB_UVERBS_DECLARE_CMD(modify_qp);
 IB_UVERBS_DECLARE_CMD(destroy_qp);
 IB_UVERBS_DECLARE_CMD(attach_mcast);
 IB_UVERBS_DECLARE_CMD(detach_mcast);
+IB_UVERBS_DECLARE_CMD(create_srq);
+IB_UVERBS_DECLARE_CMD(modify_srq);
+IB_UVERBS_DECLARE_CMD(destroy_srq);
 
 #endif /* UVERBS_H */

drivers/infiniband/core/uverbs_cmd.c

@@ -724,6 +724,7 @@ ssize_t ib_uverbs_create_qp(struct ib_uverbs_file *file,
 	struct ib_uobject              *uobj;
 	struct ib_pd                   *pd;
 	struct ib_cq                   *scq, *rcq;
+	struct ib_srq                  *srq;
 	struct ib_qp                   *qp;
 	struct ib_qp_init_attr          attr;
 	int ret;
@@ -747,10 +748,12 @@ ssize_t ib_uverbs_create_qp(struct ib_uverbs_file *file,
 	pd  = idr_find(&ib_uverbs_pd_idr, cmd.pd_handle);
 	scq = idr_find(&ib_uverbs_cq_idr, cmd.send_cq_handle);
 	rcq = idr_find(&ib_uverbs_cq_idr, cmd.recv_cq_handle);
+	srq = cmd.is_srq ? idr_find(&ib_uverbs_srq_idr, cmd.srq_handle) : NULL;
 
 	if (!pd  || pd->uobject->context  != file->ucontext ||
 	    !scq || scq->uobject->context != file->ucontext ||
-	    !rcq || rcq->uobject->context != file->ucontext) {
+	    !rcq || rcq->uobject->context != file->ucontext ||
+	    (cmd.is_srq && (!srq || srq->uobject->context != file->ucontext))) {
 		ret = -EINVAL;
 		goto err_up;
 	}
@@ -759,7 +762,7 @@ ssize_t ib_uverbs_create_qp(struct ib_uverbs_file *file,
 	attr.qp_context    = file;
 	attr.send_cq       = scq;
 	attr.recv_cq       = rcq;
-	attr.srq           = NULL;
+	attr.srq           = srq;
 	attr.sq_sig_type   = cmd.sq_sig_all ? IB_SIGNAL_ALL_WR : IB_SIGNAL_REQ_WR;
 	attr.qp_type       = cmd.qp_type;
 
@@ -1004,3 +1007,178 @@ ssize_t ib_uverbs_detach_mcast(struct ib_uverbs_file *file,
 
 	return ret ? ret : in_len;
 }
+
+ssize_t ib_uverbs_create_srq(struct ib_uverbs_file *file,
+			     const char __user *buf, int in_len,
+			     int out_len)
+{
+	struct ib_uverbs_create_srq      cmd;
+	struct ib_uverbs_create_srq_resp resp;
+	struct ib_udata                  udata;
+	struct ib_uobject               *uobj;
+	struct ib_pd                    *pd;
+	struct ib_srq                   *srq;
+	struct ib_srq_init_attr          attr;
+	int ret;
+
+	if (out_len < sizeof resp)
+		return -ENOSPC;
+
+	if (copy_from_user(&cmd, buf, sizeof cmd))
+		return -EFAULT;
+
+	INIT_UDATA(&udata, buf + sizeof cmd,
+		   (unsigned long) cmd.response + sizeof resp,
+		   in_len - sizeof cmd, out_len - sizeof resp);
+
+	uobj = kmalloc(sizeof *uobj, GFP_KERNEL);
+	if (!uobj)
+		return -ENOMEM;
+
+	down(&ib_uverbs_idr_mutex);
+
+	pd  = idr_find(&ib_uverbs_pd_idr, cmd.pd_handle);
+
+	if (!pd || pd->uobject->context != file->ucontext) {
+		ret = -EINVAL;
+		goto err_up;
+	}
+
+	attr.event_handler  = ib_uverbs_srq_event_handler;
+	attr.srq_context    = file;
+	attr.attr.max_wr    = cmd.max_wr;
+	attr.attr.max_sge   = cmd.max_sge;
+	attr.attr.srq_limit = cmd.srq_limit;
+
+	uobj->user_handle = cmd.user_handle;
+	uobj->context     = file->ucontext;
+
+	srq = pd->device->create_srq(pd, &attr, &udata);
+	if (IS_ERR(srq)) {
+		ret = PTR_ERR(srq);
+		goto err_up;
+	}
+
+	srq->device    	   = pd->device;
+	srq->pd        	   = pd;
+	srq->uobject       = uobj;
+	srq->event_handler = attr.event_handler;
+	srq->srq_context   = attr.srq_context;
+	atomic_inc(&pd->usecnt);
+	atomic_set(&srq->usecnt, 0);
+
+	memset(&resp, 0, sizeof resp);
+
+retry:
+	if (!idr_pre_get(&ib_uverbs_srq_idr, GFP_KERNEL)) {
+		ret = -ENOMEM;
+		goto err_destroy;
+	}
+
+	ret = idr_get_new(&ib_uverbs_srq_idr, srq, &uobj->id);
+
+	if (ret == -EAGAIN)
+		goto retry;
+	if (ret)
+		goto err_destroy;
+
+	resp.srq_handle = uobj->id;
+
+	spin_lock_irq(&file->ucontext->lock);
+	list_add_tail(&uobj->list, &file->ucontext->srq_list);
+	spin_unlock_irq(&file->ucontext->lock);
+
+	if (copy_to_user((void __user *) (unsigned long) cmd.response,
+			 &resp, sizeof resp)) {
+		ret = -EFAULT;
+		goto err_list;
+	}
+
+	up(&ib_uverbs_idr_mutex);
+
+	return in_len;
+
+err_list:
+	spin_lock_irq(&file->ucontext->lock);
+	list_del(&uobj->list);
+	spin_unlock_irq(&file->ucontext->lock);
+
+err_destroy:
+	ib_destroy_srq(srq);
+
+err_up:
+	up(&ib_uverbs_idr_mutex);
+
+	kfree(uobj);
+	return ret;
+}
+
+ssize_t ib_uverbs_modify_srq(struct ib_uverbs_file *file,
+			     const char __user *buf, int in_len,
+			     int out_len)
+{
+	struct ib_uverbs_modify_srq cmd;
+	struct ib_srq              *srq;
+	struct ib_srq_attr          attr;
+	int                         ret;
+
+	if (copy_from_user(&cmd, buf, sizeof cmd))
+		return -EFAULT;
+
+	down(&ib_uverbs_idr_mutex);
+
+	srq = idr_find(&ib_uverbs_srq_idr, cmd.srq_handle);
+	if (!srq || srq->uobject->context != file->ucontext) {
+		ret = -EINVAL;
+		goto out;
+	}
+
+	attr.max_wr    = cmd.max_wr;
+	attr.max_sge   = cmd.max_sge;
+	attr.srq_limit = cmd.srq_limit;
+
+	ret = ib_modify_srq(srq, &attr, cmd.attr_mask);
+
+out:
+	up(&ib_uverbs_idr_mutex);
+
+	return ret ? ret : in_len;
+}
+
+ssize_t ib_uverbs_destroy_srq(struct ib_uverbs_file *file,
+			      const char __user *buf, int in_len,
+			      int out_len)
+{
+	struct ib_uverbs_destroy_srq cmd;
+	struct ib_srq               *srq;
+	struct ib_uobject           *uobj;
+	int                          ret = -EINVAL;
+
+	if (copy_from_user(&cmd, buf, sizeof cmd))
+		return -EFAULT;
+
+	down(&ib_uverbs_idr_mutex);
+
+	srq = idr_find(&ib_uverbs_srq_idr, cmd.srq_handle);
+	if (!srq || srq->uobject->context != file->ucontext)
+		goto out;
+
+	uobj = srq->uobject;
+
+	ret = ib_destroy_srq(srq);
+	if (ret)
+		goto out;
+
+	idr_remove(&ib_uverbs_srq_idr, cmd.srq_handle);
+
+	spin_lock_irq(&file->ucontext->lock);
+	list_del(&uobj->list);
+	spin_unlock_irq(&file->ucontext->lock);
+
+	kfree(uobj);
+
+out:
+	up(&ib_uverbs_idr_mutex);
+
+	return ret ? ret : in_len;
+}

drivers/infiniband/core/uverbs_main.c

@@ -1,6 +1,8 @@
 /*
  * Copyright (c) 2005 Topspin Communications.  All rights reserved.
  * Copyright (c) 2005 Cisco Systems.  All rights reserved.
+ * Copyright (c) 2005 Mellanox Technologies. All rights reserved.
+ * Copyright (c) 2005 Voltaire, Inc. All rights reserved.
  *
  * This software is available to you under a choice of one of two
  * licenses.  You may choose to be licensed under the terms of the GNU
@@ -67,6 +69,7 @@ DEFINE_IDR(ib_uverbs_mw_idr);
 DEFINE_IDR(ib_uverbs_ah_idr);
 DEFINE_IDR(ib_uverbs_cq_idr);
 DEFINE_IDR(ib_uverbs_qp_idr);
+DEFINE_IDR(ib_uverbs_srq_idr);
 
 static spinlock_t map_lock;
 static DECLARE_BITMAP(dev_map, IB_UVERBS_MAX_DEVICES);
@@ -91,6 +94,9 @@ static ssize_t (*uverbs_cmd_table[])(struct ib_uverbs_file *file,
 	[IB_USER_VERBS_CMD_DESTROY_QP]    = ib_uverbs_destroy_qp,
 	[IB_USER_VERBS_CMD_ATTACH_MCAST]  = ib_uverbs_attach_mcast,
 	[IB_USER_VERBS_CMD_DETACH_MCAST]  = ib_uverbs_detach_mcast,
+	[IB_USER_VERBS_CMD_CREATE_SRQ]    = ib_uverbs_create_srq,
+	[IB_USER_VERBS_CMD_MODIFY_SRQ]    = ib_uverbs_modify_srq,
+	[IB_USER_VERBS_CMD_DESTROY_SRQ]   = ib_uverbs_destroy_srq,
 };
 
 static struct vfsmount *uverbs_event_mnt;
@@ -125,7 +131,14 @@ static int ib_dealloc_ucontext(struct ib_ucontext *context)
 		kfree(uobj);
 	}
 
-	/* XXX Free SRQs */
+	list_for_each_entry_safe(uobj, tmp, &context->srq_list, list) {
+		struct ib_srq *srq = idr_find(&ib_uverbs_srq_idr, uobj->id);
+		idr_remove(&ib_uverbs_srq_idr, uobj->id);
+		ib_destroy_srq(srq);
+		list_del(&uobj->list);
+		kfree(uobj);
+	}
+
 	/* XXX Free MWs */
 
 	list_for_each_entry_safe(uobj, tmp, &context->mr_list, list) {
@@ -344,6 +357,13 @@ void ib_uverbs_qp_event_handler(struct ib_event *event, void *context_ptr)
 				event->event);
 }
 
+void ib_uverbs_srq_event_handler(struct ib_event *event, void *context_ptr)
+{
+	ib_uverbs_async_handler(context_ptr,
+				event->element.srq->uobject->user_handle,
+				event->event);
+}
+
 static void ib_uverbs_event_handler(struct ib_event_handler *handler,
 				    struct ib_event *event)
 {

drivers/infiniband/core/uverbs_mem.c

@@ -1,6 +1,7 @@
 /*
  * Copyright (c) 2005 Topspin Communications.  All rights reserved.
  * Copyright (c) 2005 Cisco Systems.  All rights reserved.
+ * Copyright (c) 2005 Mellanox Technologies. All rights reserved.
  *
  * This software is available to you under a choice of one of two
  * licenses.  You may choose to be licensed under the terms of the GNU

drivers/infiniband/core/verbs.c

@@ -4,6 +4,7 @@
  * Copyright (c) 2004 Intel Corporation.  All rights reserved.
  * Copyright (c) 2004 Topspin Corporation.  All rights reserved.
  * Copyright (c) 2004 Voltaire Corporation.  All rights reserved.
+ * Copyright (c) 2005 Sun Microsystems, Inc. All rights reserved.
  * Copyright (c) 2005 Cisco Systems.  All rights reserved.
  *
  * This software is available to you under a choice of one of two
@@ -40,8 +41,8 @@
 #include <linux/errno.h>
 #include <linux/err.h>
 
-#include <ib_verbs.h>
-#include <ib_cache.h>
+#include <rdma/ib_verbs.h>
+#include <rdma/ib_cache.h>
 
 /* Protection domains */
 
@@ -153,6 +154,66 @@ int ib_destroy_ah(struct ib_ah *ah)
 }
 EXPORT_SYMBOL(ib_destroy_ah);
 
+/* Shared receive queues */
+
+struct ib_srq *ib_create_srq(struct ib_pd *pd,
+			     struct ib_srq_init_attr *srq_init_attr)
+{
+	struct ib_srq *srq;
+
+	if (!pd->device->create_srq)
+		return ERR_PTR(-ENOSYS);
+
+	srq = pd->device->create_srq(pd, srq_init_attr, NULL);
+
+	if (!IS_ERR(srq)) {
+		srq->device    	   = pd->device;
+		srq->pd        	   = pd;
+		srq->uobject       = NULL;
+		srq->event_handler = srq_init_attr->event_handler;
+		srq->srq_context   = srq_init_attr->srq_context;
+		atomic_inc(&pd->usecnt);
+		atomic_set(&srq->usecnt, 0);
+	}
+
+	return srq;
+}
+EXPORT_SYMBOL(ib_create_srq);
+
+int ib_modify_srq(struct ib_srq *srq,
+		  struct ib_srq_attr *srq_attr,
+		  enum ib_srq_attr_mask srq_attr_mask)
+{
+	return srq->device->modify_srq(srq, srq_attr, srq_attr_mask);
+}
+EXPORT_SYMBOL(ib_modify_srq);
+
+int ib_query_srq(struct ib_srq *srq,
+		 struct ib_srq_attr *srq_attr)
+{
+	return srq->device->query_srq ?
+		srq->device->query_srq(srq, srq_attr) : -ENOSYS;
+}
+EXPORT_SYMBOL(ib_query_srq);
+
+int ib_destroy_srq(struct ib_srq *srq)
+{
+	struct ib_pd *pd;
+	int ret;
+
+	if (atomic_read(&srq->usecnt))
+		return -EBUSY;
+
+	pd = srq->pd;
+
+	ret = srq->device->destroy_srq(srq);
+	if (!ret)
+		atomic_dec(&pd->usecnt);
+
+	return ret;
+}
+EXPORT_SYMBOL(ib_destroy_srq);
+
 /* Queue pairs */
 
 struct ib_qp *ib_create_qp(struct ib_pd *pd,

drivers/infiniband/hw/mthca/Makefile

@@ -1,5 +1,3 @@
-EXTRA_CFLAGS += -Idrivers/infiniband/include
-
 ifdef CONFIG_INFINIBAND_MTHCA_DEBUG
 EXTRA_CFLAGS += -DDEBUG
 endif
@@ -9,4 +7,4 @@ obj-$(CONFIG_INFINIBAND_MTHCA) += ib_mthca.o
 ib_mthca-y :=	mthca_main.o mthca_cmd.o mthca_profile.o mthca_reset.o \
 		mthca_allocator.o mthca_eq.o mthca_pd.o mthca_cq.o \
 		mthca_mr.o mthca_qp.o mthca_av.o mthca_mcg.o mthca_mad.o \
-		mthca_provider.o mthca_memfree.o mthca_uar.o
+		mthca_provider.o mthca_memfree.o mthca_uar.o mthca_srq.o

drivers/infiniband/hw/mthca/mthca_allocator.c

@@ -177,3 +177,119 @@ void mthca_array_cleanup(struct mthca_array *array, int nent)
 
 	kfree(array->page_list);
 }
+
+/*
+ * Handling for queue buffers -- we allocate a bunch of memory and
+ * register it in a memory region at HCA virtual address 0.  If the
+ * requested size is > max_direct, we split the allocation into
+ * multiple pages, so we don't require too much contiguous memory.
+ */
+
+int mthca_buf_alloc(struct mthca_dev *dev, int size, int max_direct,
+		    union mthca_buf *buf, int *is_direct, struct mthca_pd *pd,
+		    int hca_write, struct mthca_mr *mr)
+{
+	int err = -ENOMEM;
+	int npages, shift;
+	u64 *dma_list = NULL;
+	dma_addr_t t;
+	int i;
+
+	if (size <= max_direct) {
+		*is_direct = 1;
+		npages     = 1;
+		shift      = get_order(size) + PAGE_SHIFT;
+
+		buf->direct.buf = dma_alloc_coherent(&dev->pdev->dev,
+						     size, &t, GFP_KERNEL);
+		if (!buf->direct.buf)
+			return -ENOMEM;
+
+		pci_unmap_addr_set(&buf->direct, mapping, t);
+
+		memset(buf->direct.buf, 0, size);
+
+		while (t & ((1 << shift) - 1)) {
+			--shift;
+			npages *= 2;
+		}
+
+		dma_list = kmalloc(npages * sizeof *dma_list, GFP_KERNEL);
+		if (!dma_list)
+			goto err_free;
+
+		for (i = 0; i < npages; ++i)
+			dma_list[i] = t + i * (1 << shift);
+	} else {
+		*is_direct = 0;
+		npages     = (size + PAGE_SIZE - 1) / PAGE_SIZE;
+		shift      = PAGE_SHIFT;
+
+		dma_list = kmalloc(npages * sizeof *dma_list, GFP_KERNEL);
+		if (!dma_list)
+			return -ENOMEM;
+
+		buf->page_list = kmalloc(npages * sizeof *buf->page_list,
+					 GFP_KERNEL);
+		if (!buf->page_list)
+			goto err_out;
+
+		for (i = 0; i < npages; ++i)
+			buf->page_list[i].buf = NULL;
+
+		for (i = 0; i < npages; ++i) {
+			buf->page_list[i].buf =
+				dma_alloc_coherent(&dev->pdev->dev, PAGE_SIZE,
+						   &t, GFP_KERNEL);
+			if (!buf->page_list[i].buf)
+				goto err_free;
+
+			dma_list[i] = t;
+			pci_unmap_addr_set(&buf->page_list[i], mapping, t);
+
+			memset(buf->page_list[i].buf, 0, PAGE_SIZE);
+		}
+	}
+
+	err = mthca_mr_alloc_phys(dev, pd->pd_num,
+				  dma_list, shift, npages,
+				  0, size,
+				  MTHCA_MPT_FLAG_LOCAL_READ |
+				  (hca_write ? MTHCA_MPT_FLAG_LOCAL_WRITE : 0),
+				  mr);
+	if (err)
+		goto err_free;
+
+	kfree(dma_list);
+
+	return 0;
+
+err_free:
+	mthca_buf_free(dev, size, buf, *is_direct, NULL);
+
+err_out:
+	kfree(dma_list);
+
+	return err;
+}
+
+void mthca_buf_free(struct mthca_dev *dev, int size, union mthca_buf *buf,
+		    int is_direct, struct mthca_mr *mr)
+{
+	int i;
+
+	if (mr)
+		mthca_free_mr(dev, mr);
+
+	if (is_direct)
+		dma_free_coherent(&dev->pdev->dev, size, buf->direct.buf,
+				  pci_unmap_addr(&buf->direct, mapping));
+	else {
+		for (i = 0; i < (size + PAGE_SIZE - 1) / PAGE_SIZE; ++i)
+			dma_free_coherent(&dev->pdev->dev, PAGE_SIZE,
+					  buf->page_list[i].buf,
+					  pci_unmap_addr(&buf->page_list[i],
+							 mapping));
+		kfree(buf->page_list);
+	}
+}

drivers/infiniband/hw/mthca/mthca_av.c

@@ -35,22 +35,22 @@
 
 #include <linux/init.h>
 
-#include <ib_verbs.h>
-#include <ib_cache.h>
+#include <rdma/ib_verbs.h>
+#include <rdma/ib_cache.h>
 
 #include "mthca_dev.h"
 
 struct mthca_av {
-	u32 port_pd;
-	u8  reserved1;
-	u8  g_slid;
-	u16 dlid;
-	u8  reserved2;
-	u8  gid_index;
-	u8  msg_sr;
-	u8  hop_limit;
-	u32 sl_tclass_flowlabel;
-	u32 dgid[4];
+	__be32 port_pd;
+	u8     reserved1;
+	u8     g_slid;
+	__be16 dlid;
+	u8     reserved2;
+	u8     gid_index;
+	u8     msg_sr;
+	u8     hop_limit;
+	__be32 sl_tclass_flowlabel;
+	__be32 dgid[4];
 };
 
 int mthca_create_ah(struct mthca_dev *dev,
@@ -128,7 +128,7 @@ on_hca_fail:
 			  av, (unsigned long) ah->avdma);
 		for (j = 0; j < 8; ++j)
 			printk(KERN_DEBUG "  [%2x] %08x\n",
-			       j * 4, be32_to_cpu(((u32 *) av)[j]));
+			       j * 4, be32_to_cpu(((__be32 *) av)[j]));
 	}
 
 	if (ah->type == MTHCA_AH_ON_HCA) {
@@ -169,7 +169,7 @@ int mthca_read_ah(struct mthca_dev *dev, struct mthca_ah *ah,
 
 	header->lrh.service_level   = be32_to_cpu(ah->av->sl_tclass_flowlabel) >> 28;
 	header->lrh.destination_lid = ah->av->dlid;
-	header->lrh.source_lid      = ah->av->g_slid & 0x7f;
+	header->lrh.source_lid      = cpu_to_be16(ah->av->g_slid & 0x7f);
 	if (ah->av->g_slid & 0x80) {
 		header->grh_present = 1;
 		header->grh.traffic_class =

drivers/infiniband/hw/mthca/mthca_cmd.c

@@ -1,5 +1,6 @@
 /*
  * Copyright (c) 2004, 2005 Topspin Communications.  All rights reserved.
+ * Copyright (c) 2005 Mellanox Technologies. All rights reserved.
  *
  * This software is available to you under a choice of one of two
  * licenses.  You may choose to be licensed under the terms of the GNU
@@ -36,7 +37,7 @@
 #include <linux/pci.h>
 #include <linux/errno.h>
 #include <asm/io.h>
-#include <ib_mad.h>
+#include <rdma/ib_mad.h>
 
 #include "mthca_dev.h"
 #include "mthca_config_reg.h"
@@ -108,6 +109,7 @@ enum {
 	CMD_SW2HW_SRQ 	    = 0x35,
 	CMD_HW2SW_SRQ 	    = 0x36,
 	CMD_QUERY_SRQ       = 0x37,
+	CMD_ARM_SRQ         = 0x40,
 
 	/* QP/EE commands */
 	CMD_RST2INIT_QPEE   = 0x19,
@@ -219,20 +221,20 @@ static int mthca_cmd_post(struct mthca_dev *dev,
 	 * (and some architectures such as ia64 implement memcpy_toio
 	 * in terms of writeb).
 	 */
-	__raw_writel(cpu_to_be32(in_param >> 32),           dev->hcr + 0 * 4);
-	__raw_writel(cpu_to_be32(in_param & 0xfffffffful),  dev->hcr + 1 * 4);
-	__raw_writel(cpu_to_be32(in_modifier),              dev->hcr + 2 * 4);
-	__raw_writel(cpu_to_be32(out_param >> 32),          dev->hcr + 3 * 4);
-	__raw_writel(cpu_to_be32(out_param & 0xfffffffful), dev->hcr + 4 * 4);
-	__raw_writel(cpu_to_be32(token << 16),              dev->hcr + 5 * 4);
+	__raw_writel((__force u32) cpu_to_be32(in_param >> 32),           dev->hcr + 0 * 4);
+	__raw_writel((__force u32) cpu_to_be32(in_param & 0xfffffffful),  dev->hcr + 1 * 4);
+	__raw_writel((__force u32) cpu_to_be32(in_modifier),              dev->hcr + 2 * 4);
+	__raw_writel((__force u32) cpu_to_be32(out_param >> 32),          dev->hcr + 3 * 4);
+	__raw_writel((__force u32) cpu_to_be32(out_param & 0xfffffffful), dev->hcr + 4 * 4);
+	__raw_writel((__force u32) cpu_to_be32(token << 16),              dev->hcr + 5 * 4);
 
 	/* __raw_writel may not order writes. */
 	wmb();
 
-	__raw_writel(cpu_to_be32((1 << HCR_GO_BIT)                |
-				 (event ? (1 << HCA_E_BIT) : 0)   |
-				 (op_modifier << HCR_OPMOD_SHIFT) |
-				 op),                       dev->hcr + 6 * 4);
+	__raw_writel((__force u32) cpu_to_be32((1 << HCR_GO_BIT)                |
+					       (event ? (1 << HCA_E_BIT) : 0)   |
+					       (op_modifier << HCR_OPMOD_SHIFT) |
+					       op),                       dev->hcr + 6 * 4);
 
 out:
 	up(&dev->cmd.hcr_sem);
@@ -273,12 +275,14 @@ static int mthca_cmd_poll(struct mthca_dev *dev,
 		goto out;
 	}
 
-	if (out_is_imm) {
-		memcpy_fromio(out_param, dev->hcr + HCR_OUT_PARAM_OFFSET, sizeof (u64));
-		be64_to_cpus(out_param);
-	}
+	if (out_is_imm)
+		*out_param = 
+			(u64) be32_to_cpu((__force __be32)
+					  __raw_readl(dev->hcr + HCR_OUT_PARAM_OFFSET)) << 32 |
+			(u64) be32_to_cpu((__force __be32)
+					  __raw_readl(dev->hcr + HCR_OUT_PARAM_OFFSET + 4));
 
-	*status = be32_to_cpu(__raw_readl(dev->hcr + HCR_STATUS_OFFSET)) >> 24;
+	*status = be32_to_cpu((__force __be32) __raw_readl(dev->hcr + HCR_STATUS_OFFSET)) >> 24;
 
 out:
 	up(&dev->cmd.poll_sem);
@@ -1029,6 +1033,8 @@ int mthca_QUERY_DEV_LIM(struct mthca_dev *dev,
 
 	mthca_dbg(dev, "Max QPs: %d, reserved QPs: %d, entry size: %d\n",
 		  dev_lim->max_qps, dev_lim->reserved_qps, dev_lim->qpc_entry_sz);
+	mthca_dbg(dev, "Max SRQs: %d, reserved SRQs: %d, entry size: %d\n",
+		  dev_lim->max_srqs, dev_lim->reserved_srqs, dev_lim->srq_entry_sz);
 	mthca_dbg(dev, "Max CQs: %d, reserved CQs: %d, entry size: %d\n",
 		  dev_lim->max_cqs, dev_lim->reserved_cqs, dev_lim->cqc_entry_sz);
 	mthca_dbg(dev, "Max EQs: %d, reserved EQs: %d, entry size: %d\n",
@@ -1082,6 +1088,34 @@ out:
 	return err;
 }
 
+static void get_board_id(void *vsd, char *board_id)
+{
+	int i;
+
+#define VSD_OFFSET_SIG1		0x00
+#define VSD_OFFSET_SIG2		0xde
+#define VSD_OFFSET_MLX_BOARD_ID	0xd0
+#define VSD_OFFSET_TS_BOARD_ID	0x20
+
+#define VSD_SIGNATURE_TOPSPIN	0x5ad
+
+	memset(board_id, 0, MTHCA_BOARD_ID_LEN);
+
+	if (be16_to_cpup(vsd + VSD_OFFSET_SIG1) == VSD_SIGNATURE_TOPSPIN &&
+	    be16_to_cpup(vsd + VSD_OFFSET_SIG2) == VSD_SIGNATURE_TOPSPIN) {
+		strlcpy(board_id, vsd + VSD_OFFSET_TS_BOARD_ID, MTHCA_BOARD_ID_LEN);
+	} else {
+		/*
+		 * The board ID is a string but the firmware byte
+		 * swaps each 4-byte word before passing it back to
+		 * us.  Therefore we need to swab it before printing.
+		 */
+		for (i = 0; i < 4; ++i)
+			((u32 *) board_id)[i] =
+				swab32(*(u32 *) (vsd + VSD_OFFSET_MLX_BOARD_ID + i * 4));
+	}
+}
+
 int mthca_QUERY_ADAPTER(struct mthca_dev *dev,
 			struct mthca_adapter *adapter, u8 *status)
 {
@@ -1094,6 +1128,7 @@ int mthca_QUERY_ADAPTER(struct mthca_dev *dev,
 #define QUERY_ADAPTER_DEVICE_ID_OFFSET     0x04
 #define QUERY_ADAPTER_REVISION_ID_OFFSET   0x08
 #define QUERY_ADAPTER_INTA_PIN_OFFSET      0x10
+#define QUERY_ADAPTER_VSD_OFFSET           0x20
 
 	mailbox = mthca_alloc_mailbox(dev, GFP_KERNEL);
 	if (IS_ERR(mailbox))
@@ -1111,6 +1146,9 @@ int mthca_QUERY_ADAPTER(struct mthca_dev *dev,
 	MTHCA_GET(adapter->revision_id, outbox, QUERY_ADAPTER_REVISION_ID_OFFSET);
 	MTHCA_GET(adapter->inta_pin, outbox,    QUERY_ADAPTER_INTA_PIN_OFFSET);
 
+	get_board_id(outbox + QUERY_ADAPTER_VSD_OFFSET / 4,
+		     adapter->board_id);
+
 out:
 	mthca_free_mailbox(dev, mailbox);
 	return err;
@@ -1121,7 +1159,7 @@ int mthca_INIT_HCA(struct mthca_dev *dev,
 		   u8 *status)
 {
 	struct mthca_mailbox *mailbox;
-	u32 *inbox;
+	__be32 *inbox;
 	int err;
 
 #define INIT_HCA_IN_SIZE             	 0x200
@@ -1247,10 +1285,8 @@ int mthca_INIT_IB(struct mthca_dev *dev,
 #define INIT_IB_FLAG_SIG         (1 << 18)
 #define INIT_IB_FLAG_NG          (1 << 17)
 #define INIT_IB_FLAG_G0          (1 << 16)
-#define INIT_IB_FLAG_1X          (1 << 8)
-#define INIT_IB_FLAG_4X          (1 << 9)
-#define INIT_IB_FLAG_12X         (1 << 11)
 #define INIT_IB_VL_SHIFT         4
+#define INIT_IB_PORT_WIDTH_SHIFT 8
 #define INIT_IB_MTU_SHIFT        12
 #define INIT_IB_MAX_GID_OFFSET   0x06
 #define INIT_IB_MAX_PKEY_OFFSET  0x0a
@@ -1266,12 +1302,11 @@ int mthca_INIT_IB(struct mthca_dev *dev,
 	memset(inbox, 0, INIT_IB_IN_SIZE);
 
 	flags = 0;
-	flags |= param->enable_1x     ? INIT_IB_FLAG_1X  : 0;
-	flags |= param->enable_4x     ? INIT_IB_FLAG_4X  : 0;
 	flags |= param->set_guid0     ? INIT_IB_FLAG_G0  : 0;
 	flags |= param->set_node_guid ? INIT_IB_FLAG_NG  : 0;
 	flags |= param->set_si_guid   ? INIT_IB_FLAG_SIG : 0;
 	flags |= param->vl_cap << INIT_IB_VL_SHIFT;
+	flags |= param->port_width << INIT_IB_PORT_WIDTH_SHIFT;
 	flags |= param->mtu_cap << INIT_IB_MTU_SHIFT;
 	MTHCA_PUT(inbox, flags, INIT_IB_FLAGS_OFFSET);
 
@@ -1342,7 +1377,7 @@ int mthca_MAP_ICM(struct mthca_dev *dev, struct mthca_icm *icm, u64 virt, u8 *st
 int mthca_MAP_ICM_page(struct mthca_dev *dev, u64 dma_addr, u64 virt, u8 *status)
 {
 	struct mthca_mailbox *mailbox;
-	u64 *inbox;
+	__be64 *inbox;
 	int err;
 
 	mailbox = mthca_alloc_mailbox(dev, GFP_KERNEL);
@@ -1468,6 +1503,27 @@ int mthca_HW2SW_CQ(struct mthca_dev *dev, struct mthca_mailbox *mailbox,
 			     CMD_TIME_CLASS_A, status);
 }
 
+int mthca_SW2HW_SRQ(struct mthca_dev *dev, struct mthca_mailbox *mailbox,
+		    int srq_num, u8 *status)
+{
+	return mthca_cmd(dev, mailbox->dma, srq_num, 0, CMD_SW2HW_SRQ,
+			CMD_TIME_CLASS_A, status);
+}
+
+int mthca_HW2SW_SRQ(struct mthca_dev *dev, struct mthca_mailbox *mailbox,
+		    int srq_num, u8 *status)
+{
+	return mthca_cmd_box(dev, 0, mailbox->dma, srq_num, 0,
+			     CMD_HW2SW_SRQ,
+			     CMD_TIME_CLASS_A, status);
+}
+
+int mthca_ARM_SRQ(struct mthca_dev *dev, int srq_num, int limit, u8 *status)
+{
+	return mthca_cmd(dev, limit, srq_num, 0, CMD_ARM_SRQ,
+			 CMD_TIME_CLASS_B, status);
+}
+
 int mthca_MODIFY_QP(struct mthca_dev *dev, int trans, u32 num,
 		    int is_ee, struct mthca_mailbox *mailbox, u32 optmask,
 		    u8 *status)
@@ -1513,7 +1569,7 @@ int mthca_MODIFY_QP(struct mthca_dev *dev, int trans, u32 num,
 				if (i % 8 == 0)
 					printk("  [%02x] ", i * 4);
 				printk(" %08x",
-				       be32_to_cpu(((u32 *) mailbox->buf)[i + 2]));
+				       be32_to_cpu(((__be32 *) mailbox->buf)[i + 2]));
 				if ((i + 1) % 8 == 0)
 					printk("\n");
 			}
@@ -1533,7 +1589,7 @@ int mthca_MODIFY_QP(struct mthca_dev *dev, int trans, u32 num,
 				if (i % 8 == 0)
 					printk("[%02x] ", i * 4);
 				printk(" %08x",
-				       be32_to_cpu(((u32 *) mailbox->buf)[i + 2]));
+				       be32_to_cpu(((__be32 *) mailbox->buf)[i + 2]));
 				if ((i + 1) % 8 == 0)
 					printk("\n");
 			}

drivers/infiniband/hw/mthca/mthca_cmd.h

@@ -1,5 +1,6 @@
 /*
  * Copyright (c) 2004, 2005 Topspin Communications.  All rights reserved.
+ * Copyright (c) 2005 Mellanox Technologies. All rights reserved.
  *
  * This software is available to you under a choice of one of two
  * licenses.  You may choose to be licensed under the terms of the GNU
@@ -35,7 +36,7 @@
 #ifndef MTHCA_CMD_H
 #define MTHCA_CMD_H
 
-#include <ib_verbs.h>
+#include <rdma/ib_verbs.h>
 
 #define MTHCA_MAILBOX_SIZE 4096
 
@@ -183,10 +184,11 @@ struct mthca_dev_lim {
 };
 
 struct mthca_adapter {
-	u32 vendor_id;
-	u32 device_id;
-	u32 revision_id;
-	u8  inta_pin;
+	u32  vendor_id;
+	u32  device_id;
+	u32  revision_id;
+	char board_id[MTHCA_BOARD_ID_LEN];
+	u8   inta_pin;
 };
 
 struct mthca_init_hca_param {
@@ -218,8 +220,7 @@ struct mthca_init_hca_param {
 };
 
 struct mthca_init_ib_param {
-	int enable_1x;
-	int enable_4x;
+	int port_width;
 	int vl_cap;
 	int mtu_cap;
 	u16 gid_cap;
@@ -297,6 +298,11 @@ int mthca_SW2HW_CQ(struct mthca_dev *dev, struct mthca_mailbox *mailbox,
 		   int cq_num, u8 *status);
 int mthca_HW2SW_CQ(struct mthca_dev *dev, struct mthca_mailbox *mailbox,
 		   int cq_num, u8 *status);
+int mthca_SW2HW_SRQ(struct mthca_dev *dev, struct mthca_mailbox *mailbox,
+		    int srq_num, u8 *status);
+int mthca_HW2SW_SRQ(struct mthca_dev *dev, struct mthca_mailbox *mailbox,
+		    int srq_num, u8 *status);
+int mthca_ARM_SRQ(struct mthca_dev *dev, int srq_num, int limit, u8 *status);
 int mthca_MODIFY_QP(struct mthca_dev *dev, int trans, u32 num,
 		    int is_ee, struct mthca_mailbox *mailbox, u32 optmask,
 		    u8 *status);

drivers/infiniband/hw/mthca/mthca_config_reg.h

@@ -1,5 +1,6 @@
 /*
  * Copyright (c) 2004 Topspin Communications.  All rights reserved.
+ * Copyright (c) 2005 Mellanox Technologies. All rights reserved.
  *
  * This software is available to you under a choice of one of two
  * licenses.  You may choose to be licensed under the terms of the GNU

drivers/infiniband/hw/mthca/mthca_cq.c

@@ -2,6 +2,8 @@
  * Copyright (c) 2004, 2005 Topspin Communications.  All rights reserved.
  * Copyright (c) 2005 Sun Microsystems, Inc. All rights reserved.
  * Copyright (c) 2005 Cisco Systems, Inc. All rights reserved.
+ * Copyright (c) 2005 Mellanox Technologies. All rights reserved.
+ * Copyright (c) 2004 Voltaire, Inc. All rights reserved.
  *
  * This software is available to you under a choice of one of two
  * licenses.  You may choose to be licensed under the terms of the GNU
@@ -37,7 +39,7 @@
 #include <linux/init.h>
 #include <linux/hardirq.h>
 
-#include <ib_pack.h>
+#include <rdma/ib_pack.h>
 
 #include "mthca_dev.h"
 #include "mthca_cmd.h"
@@ -55,21 +57,21 @@ enum {
  * Must be packed because start is 64 bits but only aligned to 32 bits.
  */
 struct mthca_cq_context {
-	u32 flags;
-	u64 start;
-	u32 logsize_usrpage;
-	u32 error_eqn;		/* Tavor only */
-	u32 comp_eqn;
-	u32 pd;
-	u32 lkey;
-	u32 last_notified_index;
-	u32 solicit_producer_index;
-	u32 consumer_index;
-	u32 producer_index;
-	u32 cqn;
-	u32 ci_db;		/* Arbel only */
-	u32 state_db;		/* Arbel only */
-	u32 reserved;
+	__be32 flags;
+	__be64 start;
+	__be32 logsize_usrpage;
+	__be32 error_eqn;	/* Tavor only */
+	__be32 comp_eqn;
+	__be32 pd;
+	__be32 lkey;
+	__be32 last_notified_index;
+	__be32 solicit_producer_index;
+	__be32 consumer_index;
+	__be32 producer_index;
+	__be32 cqn;
+	__be32 ci_db;		/* Arbel only */
+	__be32 state_db;	/* Arbel only */
+	u32    reserved;
 } __attribute__((packed));
 
 #define MTHCA_CQ_STATUS_OK          ( 0 << 28)
@@ -108,31 +110,31 @@ enum {
 };
 
 struct mthca_cqe {
-	u32 my_qpn;
-	u32 my_ee;
-	u32 rqpn;
-	u16 sl_g_mlpath;
-	u16 rlid;
-	u32 imm_etype_pkey_eec;
-	u32 byte_cnt;
-	u32 wqe;
-	u8  opcode;
-	u8  is_send;
-	u8  reserved;
-	u8  owner;
+	__be32 my_qpn;
+	__be32 my_ee;
+	__be32 rqpn;
+	__be16 sl_g_mlpath;
+	__be16 rlid;
+	__be32 imm_etype_pkey_eec;
+	__be32 byte_cnt;
+	__be32 wqe;
+	u8     opcode;
+	u8     is_send;
+	u8     reserved;
+	u8     owner;
 };
 
 struct mthca_err_cqe {
-	u32 my_qpn;
-	u32 reserved1[3];
-	u8  syndrome;
-	u8  reserved2;
-	u16 db_cnt;
-	u32 reserved3;
-	u32 wqe;
-	u8  opcode;
-	u8  reserved4[2];
-	u8  owner;
+	__be32 my_qpn;
+	u32    reserved1[3];
+	u8     syndrome;
+	u8     reserved2;
+	__be16 db_cnt;
+	u32    reserved3;
+	__be32 wqe;
+	u8     opcode;
+	u8     reserved4[2];
+	u8     owner;
 };
 
 #define MTHCA_CQ_ENTRY_OWNER_SW      (0 << 7)
@@ -191,7 +193,7 @@ static void dump_cqe(struct mthca_dev *dev, void *cqe_ptr)
 static inline void update_cons_index(struct mthca_dev *dev, struct mthca_cq *cq,
 				     int incr)
 {
-	u32 doorbell[2];
+	__be32 doorbell[2];
 
 	if (mthca_is_memfree(dev)) {
 		*cq->set_ci_db = cpu_to_be32(cq->cons_index);
@@ -222,7 +224,8 @@ void mthca_cq_event(struct mthca_dev *dev, u32 cqn)
 	cq->ibcq.comp_handler(&cq->ibcq, cq->ibcq.cq_context);
 }
 
-void mthca_cq_clean(struct mthca_dev *dev, u32 cqn, u32 qpn)
+void mthca_cq_clean(struct mthca_dev *dev, u32 cqn, u32 qpn,
+		    struct mthca_srq *srq)
 {
 	struct mthca_cq *cq;
 	struct mthca_cqe *cqe;
@@ -263,8 +266,11 @@ void mthca_cq_clean(struct mthca_dev *dev, u32 cqn, u32 qpn)
 	 */
 	while (prod_index > cq->cons_index) {
 		cqe = get_cqe(cq, (prod_index - 1) & cq->ibcq.cqe);
-		if (cqe->my_qpn == cpu_to_be32(qpn))
+		if (cqe->my_qpn == cpu_to_be32(qpn)) {
+			if (srq)
+				mthca_free_srq_wqe(srq, be32_to_cpu(cqe->wqe));
 			++nfreed;
+		}
 		else if (nfreed)
 			memcpy(get_cqe(cq, (prod_index - 1 + nfreed) &
 				       cq->ibcq.cqe),
@@ -291,7 +297,7 @@ static int handle_error_cqe(struct mthca_dev *dev, struct mthca_cq *cq,
 {
 	int err;
 	int dbd;
-	u32 new_wqe;
+	__be32 new_wqe;
 
 	if (cqe->syndrome == SYNDROME_LOCAL_QP_OP_ERR) {
 		mthca_dbg(dev, "local QP operation err "
@@ -365,6 +371,13 @@ static int handle_error_cqe(struct mthca_dev *dev, struct mthca_cq *cq,
 		break;
 	}
 
+	/*
+	 * Mem-free HCAs always generate one CQE per WQE, even in the
+	 * error case, so we don't have to check the doorbell count, etc.
+	 */
+	if (mthca_is_memfree(dev))
+		return 0;
+
 	err = mthca_free_err_wqe(dev, qp, is_send, wqe_index, &dbd, &new_wqe);
 	if (err)
 		return err;
@@ -373,12 +386,8 @@ static int handle_error_cqe(struct mthca_dev *dev, struct mthca_cq *cq,
 	 * If we're at the end of the WQE chain, or we've used up our
 	 * doorbell count, free the CQE.  Otherwise just update it for
 	 * the next poll operation.
-	 *
-	 * This does not apply to mem-free HCAs: they don't use the
-	 * doorbell count field, and so we should always free the CQE.
 	 */
-	if (mthca_is_memfree(dev) ||
-	    !(new_wqe & cpu_to_be32(0x3f)) || (!cqe->db_cnt && dbd))
+	if (!(new_wqe & cpu_to_be32(0x3f)) || (!cqe->db_cnt && dbd))
 		return 0;
 
 	cqe->db_cnt   = cpu_to_be16(be16_to_cpu(cqe->db_cnt) - dbd);
@@ -450,23 +459,27 @@ static inline int mthca_poll_one(struct mthca_dev *dev,
 			     >> wq->wqe_shift);
 		entry->wr_id = (*cur_qp)->wrid[wqe_index +
 					       (*cur_qp)->rq.max];
+	} else if ((*cur_qp)->ibqp.srq) {
+		struct mthca_srq *srq = to_msrq((*cur_qp)->ibqp.srq);
+		u32 wqe = be32_to_cpu(cqe->wqe);
+		wq = NULL;
+		wqe_index = wqe >> srq->wqe_shift;
+		entry->wr_id = srq->wrid[wqe_index];
+		mthca_free_srq_wqe(srq, wqe);
 	} else {
 		wq = &(*cur_qp)->rq;
 		wqe_index = be32_to_cpu(cqe->wqe) >> wq->wqe_shift;
 		entry->wr_id = (*cur_qp)->wrid[wqe_index];
 	}
 
-	if (wq->last_comp < wqe_index)
-		wq->tail += wqe_index - wq->last_comp;
-	else
-		wq->tail += wqe_index + wq->max - wq->last_comp;
+	if (wq) {
+		if (wq->last_comp < wqe_index)
+			wq->tail += wqe_index - wq->last_comp;
+		else
+			wq->tail += wqe_index + wq->max - wq->last_comp;
 
-	wq->last_comp = wqe_index;
-
-	if (0)
-		mthca_dbg(dev, "%s completion for QP %06x, index %d (nr %d)\n",
-			  is_send ? "Send" : "Receive",
-			  (*cur_qp)->qpn, wqe_index, wq->max);
+		wq->last_comp = wqe_index;
+	}
 
 	if (is_error) {
 		err = handle_error_cqe(dev, cq, *cur_qp, wqe_index, is_send,
@@ -584,13 +597,13 @@ int mthca_poll_cq(struct ib_cq *ibcq, int num_entries,
 
 int mthca_tavor_arm_cq(struct ib_cq *cq, enum ib_cq_notify notify)
 {
-	u32 doorbell[2];
+	__be32 doorbell[2];
 
 	doorbell[0] = cpu_to_be32((notify == IB_CQ_SOLICITED ?
 				   MTHCA_TAVOR_CQ_DB_REQ_NOT_SOL :
 				   MTHCA_TAVOR_CQ_DB_REQ_NOT)      |
 				  to_mcq(cq)->cqn);
-	doorbell[1] = 0xffffffff;
+	doorbell[1] = (__force __be32) 0xffffffff;
 
 	mthca_write64(doorbell,
 		      to_mdev(cq->device)->kar + MTHCA_CQ_DOORBELL,
@@ -602,9 +615,9 @@ int mthca_tavor_arm_cq(struct ib_cq *cq, enum ib_cq_notify notify)
 int mthca_arbel_arm_cq(struct ib_cq *ibcq, enum ib_cq_notify notify)
 {
 	struct mthca_cq *cq = to_mcq(ibcq);
-	u32 doorbell[2];
+	__be32 doorbell[2];
 	u32 sn;
-	u32 ci;
+	__be32 ci;
 
 	sn = cq->arm_sn & 3;
 	ci = cpu_to_be32(cq->cons_index);
@@ -637,113 +650,8 @@ int mthca_arbel_arm_cq(struct ib_cq *ibcq, enum ib_cq_notify notify)
 
 static void mthca_free_cq_buf(struct mthca_dev *dev, struct mthca_cq *cq)
 {
-	int i;
-	int size;
-
-	if (cq->is_direct)
-		dma_free_coherent(&dev->pdev->dev,
-				  (cq->ibcq.cqe + 1) * MTHCA_CQ_ENTRY_SIZE,
-				  cq->queue.direct.buf,
-				  pci_unmap_addr(&cq->queue.direct,
-						 mapping));
-	else {
-		size = (cq->ibcq.cqe + 1) * MTHCA_CQ_ENTRY_SIZE;
-		for (i = 0; i < (size + PAGE_SIZE - 1) / PAGE_SIZE; ++i)
-			if (cq->queue.page_list[i].buf)
-				dma_free_coherent(&dev->pdev->dev, PAGE_SIZE,
-						  cq->queue.page_list[i].buf,
-						  pci_unmap_addr(&cq->queue.page_list[i],
-								 mapping));
-
-		kfree(cq->queue.page_list);
-	}
-}
-
-static int mthca_alloc_cq_buf(struct mthca_dev *dev, int size,
-			      struct mthca_cq *cq)
-{
-	int err = -ENOMEM;
-	int npages, shift;
-	u64 *dma_list = NULL;
-	dma_addr_t t;
-	int i;
-
-	if (size <= MTHCA_MAX_DIRECT_CQ_SIZE) {
-		cq->is_direct = 1;
-		npages        = 1;
-		shift         = get_order(size) + PAGE_SHIFT;
-
-		cq->queue.direct.buf = dma_alloc_coherent(&dev->pdev->dev,
-							  size, &t, GFP_KERNEL);
-		if (!cq->queue.direct.buf)
-			return -ENOMEM;
-
-		pci_unmap_addr_set(&cq->queue.direct, mapping, t);
-
-		memset(cq->queue.direct.buf, 0, size);
-
-		while (t & ((1 << shift) - 1)) {
-			--shift;
-			npages *= 2;
-		}
-
-		dma_list = kmalloc(npages * sizeof *dma_list, GFP_KERNEL);
-		if (!dma_list)
-			goto err_free;
-
-		for (i = 0; i < npages; ++i)
-			dma_list[i] = t + i * (1 << shift);
-	} else {
-		cq->is_direct = 0;
-		npages        = (size + PAGE_SIZE - 1) / PAGE_SIZE;
-		shift         = PAGE_SHIFT;
-
-		dma_list = kmalloc(npages * sizeof *dma_list, GFP_KERNEL);
-		if (!dma_list)
-			return -ENOMEM;
-
-		cq->queue.page_list = kmalloc(npages * sizeof *cq->queue.page_list,
-					      GFP_KERNEL);
-		if (!cq->queue.page_list)
-			goto err_out;
-
-		for (i = 0; i < npages; ++i)
-			cq->queue.page_list[i].buf = NULL;
-
-		for (i = 0; i < npages; ++i) {
-			cq->queue.page_list[i].buf =
-				dma_alloc_coherent(&dev->pdev->dev, PAGE_SIZE,
-						   &t, GFP_KERNEL);
-			if (!cq->queue.page_list[i].buf)
-				goto err_free;
-
-			dma_list[i] = t;
-			pci_unmap_addr_set(&cq->queue.page_list[i], mapping, t);
-
-			memset(cq->queue.page_list[i].buf, 0, PAGE_SIZE);
-		}
-	}
-
-	err = mthca_mr_alloc_phys(dev, dev->driver_pd.pd_num,
-				  dma_list, shift, npages,
-				  0, size,
-				  MTHCA_MPT_FLAG_LOCAL_WRITE |
-				  MTHCA_MPT_FLAG_LOCAL_READ,
-				  &cq->mr);
-	if (err)
-		goto err_free;
-
-	kfree(dma_list);
-
-	return 0;
-
-err_free:
-	mthca_free_cq_buf(dev, cq);
-
-err_out:
-	kfree(dma_list);
-
-	return err;
+	mthca_buf_free(dev, (cq->ibcq.cqe + 1) * MTHCA_CQ_ENTRY_SIZE,
+		       &cq->queue, cq->is_direct, &cq->mr);
 }
 
 int mthca_init_cq(struct mthca_dev *dev, int nent,
@@ -795,7 +703,9 @@ int mthca_init_cq(struct mthca_dev *dev, int nent,
 	cq_context = mailbox->buf;
 
 	if (cq->is_kernel) {
-		err = mthca_alloc_cq_buf(dev, size, cq);
+		err = mthca_buf_alloc(dev, size, MTHCA_MAX_DIRECT_CQ_SIZE,
+				      &cq->queue, &cq->is_direct,
+				      &dev->driver_pd, 1, &cq->mr);
 		if (err)
 			goto err_out_mailbox;
 
@@ -811,7 +721,6 @@ int mthca_init_cq(struct mthca_dev *dev, int nent,
 	cq_context->flags           = cpu_to_be32(MTHCA_CQ_STATUS_OK      |
 						  MTHCA_CQ_STATE_DISARMED |
 						  MTHCA_CQ_FLAG_TR);
-	cq_context->start           = cpu_to_be64(0);
 	cq_context->logsize_usrpage = cpu_to_be32((ffs(nent) - 1) << 24);
 	if (ctx)
 		cq_context->logsize_usrpage |= cpu_to_be32(ctx->uar.index);
@@ -857,10 +766,8 @@ int mthca_init_cq(struct mthca_dev *dev, int nent,
 	return 0;
 
 err_out_free_mr:
-	if (cq->is_kernel) {
-		mthca_free_mr(dev, &cq->mr);
+	if (cq->is_kernel)
 		mthca_free_cq_buf(dev, cq);
-	}
 
 err_out_mailbox:
 	mthca_free_mailbox(dev, mailbox);
@@ -904,7 +811,7 @@ void mthca_free_cq(struct mthca_dev *dev,
 		mthca_warn(dev, "HW2SW_CQ returned status 0x%02x\n", status);
 
 	if (0) {
-		u32 *ctx = mailbox->buf;
+		__be32 *ctx = mailbox->buf;
 		int j;
 
 		printk(KERN_ERR "context for CQN %x (cons index %x, next sw %d)\n",
@@ -928,7 +835,6 @@ void mthca_free_cq(struct mthca_dev *dev,
 	wait_event(cq->wait, !atomic_read(&cq->refcount));
 
 	if (cq->is_kernel) {
-		mthca_free_mr(dev, &cq->mr);
 		mthca_free_cq_buf(dev, cq);
 		if (mthca_is_memfree(dev)) {
 			mthca_free_db(dev, MTHCA_DB_TYPE_CQ_ARM,    cq->arm_db_index);

drivers/infiniband/hw/mthca/mthca_dev.h

@@ -2,6 +2,8 @@
  * Copyright (c) 2004, 2005 Topspin Communications.  All rights reserved.
  * Copyright (c) 2005 Sun Microsystems, Inc. All rights reserved.
  * Copyright (c) 2005 Cisco Systems.  All rights reserved.
+ * Copyright (c) 2005 Mellanox Technologies. All rights reserved.
+ * Copyright (c) 2004 Voltaire, Inc. All rights reserved.
  *
  * This software is available to you under a choice of one of two
  * licenses.  You may choose to be licensed under the terms of the GNU
@@ -66,6 +68,10 @@ enum {
 	MTHCA_MAX_PORTS = 2
 };
 
+enum {
+	MTHCA_BOARD_ID_LEN = 64
+};
+
 enum {
 	MTHCA_EQ_CONTEXT_SIZE =  0x40,
 	MTHCA_CQ_CONTEXT_SIZE =  0x40,
@@ -142,6 +148,7 @@ struct mthca_limits {
 	int      reserved_mcgs;
 	int      num_pds;
 	int      reserved_pds;
+	u8       port_width_cap;
 };
 
 struct mthca_alloc {
@@ -211,6 +218,13 @@ struct mthca_cq_table {
 	struct mthca_icm_table *table;
 };
 
+struct mthca_srq_table {
+	struct mthca_alloc 	alloc;
+	spinlock_t         	lock;
+	struct mthca_array      srq;
+	struct mthca_icm_table *table;
+};
+
 struct mthca_qp_table {
 	struct mthca_alloc     	alloc;
 	u32                    	rdb_base;
@@ -246,6 +260,7 @@ struct mthca_dev {
 	unsigned long    device_cap_flags;
 
 	u32              rev_id;
+	char             board_id[MTHCA_BOARD_ID_LEN];
 
 	/* firmware info */
 	u64              fw_ver;
@@ -291,6 +306,7 @@ struct mthca_dev {
 	struct mthca_mr_table  mr_table;
 	struct mthca_eq_table  eq_table;
 	struct mthca_cq_table  cq_table;
+	struct mthca_srq_table srq_table;
 	struct mthca_qp_table  qp_table;
 	struct mthca_av_table  av_table;
 	struct mthca_mcg_table mcg_table;
@@ -331,14 +347,13 @@ extern void __buggy_use_of_MTHCA_PUT(void);
 
 #define MTHCA_PUT(dest, source, offset)                               \
 	do {                                                          \
-		__typeof__(source) *__p =                             \
-			(__typeof__(source) *) ((char *) (dest) + (offset)); \
+		void *__d = ((char *) (dest) + (offset));	      \
 		switch (sizeof(source)) {                             \
-			case 1: *__p = (source);            break;    \
-			case 2: *__p = cpu_to_be16(source); break;    \
-			case 4: *__p = cpu_to_be32(source); break;    \
-			case 8: *__p = cpu_to_be64(source); break;    \
-			default: __buggy_use_of_MTHCA_PUT();          \
+		case 1: *(u8 *) __d = (source);                break; \
+		case 2:	*(__be16 *) __d = cpu_to_be16(source); break; \
+		case 4:	*(__be32 *) __d = cpu_to_be32(source); break; \
+		case 8:	*(__be64 *) __d = cpu_to_be64(source); break; \
+		default: __buggy_use_of_MTHCA_PUT();		      \
 		}                                                     \
 	} while (0)
 
@@ -354,12 +369,18 @@ int mthca_array_set(struct mthca_array *array, int index, void *value);
 void mthca_array_clear(struct mthca_array *array, int index);
 int mthca_array_init(struct mthca_array *array, int nent);
 void mthca_array_cleanup(struct mthca_array *array, int nent);
+int mthca_buf_alloc(struct mthca_dev *dev, int size, int max_direct,
+		    union mthca_buf *buf, int *is_direct, struct mthca_pd *pd,
+		    int hca_write, struct mthca_mr *mr);
+void mthca_buf_free(struct mthca_dev *dev, int size, union mthca_buf *buf,
+		    int is_direct, struct mthca_mr *mr);
 
 int mthca_init_uar_table(struct mthca_dev *dev);
 int mthca_init_pd_table(struct mthca_dev *dev);
 int mthca_init_mr_table(struct mthca_dev *dev);
 int mthca_init_eq_table(struct mthca_dev *dev);
 int mthca_init_cq_table(struct mthca_dev *dev);
+int mthca_init_srq_table(struct mthca_dev *dev);
 int mthca_init_qp_table(struct mthca_dev *dev);
 int mthca_init_av_table(struct mthca_dev *dev);
 int mthca_init_mcg_table(struct mthca_dev *dev);
@@ -369,6 +390,7 @@ void mthca_cleanup_pd_table(struct mthca_dev *dev);
 void mthca_cleanup_mr_table(struct mthca_dev *dev);
 void mthca_cleanup_eq_table(struct mthca_dev *dev);
 void mthca_cleanup_cq_table(struct mthca_dev *dev);
+void mthca_cleanup_srq_table(struct mthca_dev *dev);
 void mthca_cleanup_qp_table(struct mthca_dev *dev);
 void mthca_cleanup_av_table(struct mthca_dev *dev);
 void mthca_cleanup_mcg_table(struct mthca_dev *dev);
@@ -419,7 +441,19 @@ int mthca_init_cq(struct mthca_dev *dev, int nent,
 void mthca_free_cq(struct mthca_dev *dev,
 		   struct mthca_cq *cq);
 void mthca_cq_event(struct mthca_dev *dev, u32 cqn);
-void mthca_cq_clean(struct mthca_dev *dev, u32 cqn, u32 qpn);
+void mthca_cq_clean(struct mthca_dev *dev, u32 cqn, u32 qpn,
+		    struct mthca_srq *srq);
+
+int mthca_alloc_srq(struct mthca_dev *dev, struct mthca_pd *pd,
+		    struct ib_srq_attr *attr, struct mthca_srq *srq);
+void mthca_free_srq(struct mthca_dev *dev, struct mthca_srq *srq);
+void mthca_srq_event(struct mthca_dev *dev, u32 srqn,
+		     enum ib_event_type event_type);
+void mthca_free_srq_wqe(struct mthca_srq *srq, u32 wqe_addr);
+int mthca_tavor_post_srq_recv(struct ib_srq *srq, struct ib_recv_wr *wr,
+			      struct ib_recv_wr **bad_wr);
+int mthca_arbel_post_srq_recv(struct ib_srq *srq, struct ib_recv_wr *wr,
+			      struct ib_recv_wr **bad_wr);
 
 void mthca_qp_event(struct mthca_dev *dev, u32 qpn,
 		    enum ib_event_type event_type);
@@ -433,7 +467,7 @@ int mthca_arbel_post_send(struct ib_qp *ibqp, struct ib_send_wr *wr,
 int mthca_arbel_post_receive(struct ib_qp *ibqp, struct ib_recv_wr *wr,
 			     struct ib_recv_wr **bad_wr);
 int mthca_free_err_wqe(struct mthca_dev *dev, struct mthca_qp *qp, int is_send,
-		       int index, int *dbd, u32 *new_wqe);
+		       int index, int *dbd, __be32 *new_wqe);
 int mthca_alloc_qp(struct mthca_dev *dev,
 		   struct mthca_pd *pd,
 		   struct mthca_cq *send_cq,

drivers/infiniband/hw/mthca/mthca_doorbell.h

@@ -1,6 +1,7 @@
 /*
  * Copyright (c) 2004 Topspin Communications.  All rights reserved.
  * Copyright (c) 2005 Sun Microsystems, Inc. All rights reserved.
+ * Copyright (c) 2005 Mellanox Technologies. All rights reserved.
  *
  * This software is available to you under a choice of one of two
  * licenses.  You may choose to be licensed under the terms of the GNU
@@ -57,13 +58,13 @@ static inline void mthca_write64_raw(__be64 val, void __iomem *dest)
 	__raw_writeq((__force u64) val, dest);
 }
 
-static inline void mthca_write64(u32 val[2], void __iomem *dest,
+static inline void mthca_write64(__be32 val[2], void __iomem *dest,
 				 spinlock_t *doorbell_lock)
 {
 	__raw_writeq(*(u64 *) val, dest);
 }
 
-static inline void mthca_write_db_rec(u32 val[2], u32 *db)
+static inline void mthca_write_db_rec(__be32 val[2], __be32 *db)
 {
 	*(u64 *) db = *(u64 *) val;
 }
@@ -86,18 +87,18 @@ static inline void mthca_write64_raw(__be64 val, void __iomem *dest)
 	__raw_writel(((__force u32 *) &val)[1], dest + 4);
 }
 
-static inline void mthca_write64(u32 val[2], void __iomem *dest,
+static inline void mthca_write64(__be32 val[2], void __iomem *dest,
 				 spinlock_t *doorbell_lock)
 {
 	unsigned long flags;
 
 	spin_lock_irqsave(doorbell_lock, flags);
-	__raw_writel(val[0], dest);
-	__raw_writel(val[1], dest + 4);
+	__raw_writel((__force u32) val[0], dest);
+	__raw_writel((__force u32) val[1], dest + 4);
 	spin_unlock_irqrestore(doorbell_lock, flags);
 }
 
-static inline void mthca_write_db_rec(u32 val[2], u32 *db)
+static inline void mthca_write_db_rec(__be32 val[2], __be32 *db)
 {
 	db[0] = val[0];
 	wmb();

drivers/infiniband/hw/mthca/mthca_eq.c

@@ -1,5 +1,6 @@
 /*
  * Copyright (c) 2004, 2005 Topspin Communications.  All rights reserved.
+ * Copyright (c) 2005 Mellanox Technologies. All rights reserved.
  *
  * This software is available to you under a choice of one of two
  * licenses.  You may choose to be licensed under the terms of the GNU
@@ -51,18 +52,18 @@ enum {
  * Must be packed because start is 64 bits but only aligned to 32 bits.
  */
 struct mthca_eq_context {
-	u32 flags;
-	u64 start;
-	u32 logsize_usrpage;
-	u32 tavor_pd;		/* reserved for Arbel */
-	u8  reserved1[3];
-	u8  intr;
-	u32 arbel_pd;		/* lost_count for Tavor */
-	u32 lkey;
-	u32 reserved2[2];
-	u32 consumer_index;
-	u32 producer_index;
-	u32 reserved3[4];
+	__be32 flags;
+	__be64 start;
+	__be32 logsize_usrpage;
+	__be32 tavor_pd;	/* reserved for Arbel */
+	u8     reserved1[3];
+	u8     intr;
+	__be32 arbel_pd;	/* lost_count for Tavor */
+	__be32 lkey;
+	u32    reserved2[2];
+	__be32 consumer_index;
+	__be32 producer_index;
+	u32    reserved3[4];
 } __attribute__((packed));
 
 #define MTHCA_EQ_STATUS_OK          ( 0 << 28)
@@ -127,28 +128,28 @@ struct mthca_eqe {
 	union {
 		u32 raw[6];
 		struct {
-			u32 cqn;
+			__be32 cqn;
 		} __attribute__((packed)) comp;
 		struct {
-			u16 reserved1;
-			u16 token;
-			u32 reserved2;
-			u8  reserved3[3];
-			u8  status;
-			u64 out_param;
+			u16    reserved1;
+			__be16 token;
+			u32    reserved2;
+			u8     reserved3[3];
+			u8     status;
+			__be64 out_param;
 		} __attribute__((packed)) cmd;
 		struct {
-			u32 qpn;
+			__be32 qpn;
 		} __attribute__((packed)) qp;
 		struct {
-			u32 cqn;
-			u32 reserved1;
-			u8  reserved2[3];
-			u8  syndrome;
+			__be32 cqn;
+			u32    reserved1;
+			u8     reserved2[3];
+			u8     syndrome;
 		} __attribute__((packed)) cq_err;
 		struct {
-			u32 reserved1[2];
-			u32 port;
+			u32    reserved1[2];
+			__be32 port;
 		} __attribute__((packed)) port_change;
 	} event;
 	u8 reserved3[3];
@@ -167,7 +168,7 @@ static inline u64 async_mask(struct mthca_dev *dev)
 
 static inline void tavor_set_eq_ci(struct mthca_dev *dev, struct mthca_eq *eq, u32 ci)
 {
-	u32 doorbell[2];
+	__be32 doorbell[2];
 
 	doorbell[0] = cpu_to_be32(MTHCA_EQ_DB_SET_CI | eq->eqn);
 	doorbell[1] = cpu_to_be32(ci & (eq->nent - 1));
@@ -190,8 +191,8 @@ static inline void arbel_set_eq_ci(struct mthca_dev *dev, struct mthca_eq *eq, u
 {
 	/* See comment in tavor_set_eq_ci() above. */
 	wmb();
-	__raw_writel(cpu_to_be32(ci), dev->eq_regs.arbel.eq_set_ci_base +
-		     eq->eqn * 8);
+	__raw_writel((__force u32) cpu_to_be32(ci),
+		     dev->eq_regs.arbel.eq_set_ci_base + eq->eqn * 8);
 	/* We still want ordering, just not swabbing, so add a barrier */
 	mb();
 }
@@ -206,7 +207,7 @@ static inline void set_eq_ci(struct mthca_dev *dev, struct mthca_eq *eq, u32 ci)
 
 static inline void tavor_eq_req_not(struct mthca_dev *dev, int eqn)
 {
-	u32 doorbell[2];
+	__be32 doorbell[2];
 
 	doorbell[0] = cpu_to_be32(MTHCA_EQ_DB_REQ_NOT | eqn);
 	doorbell[1] = 0;
@@ -224,7 +225,7 @@ static inline void arbel_eq_req_not(struct mthca_dev *dev, u32 eqn_mask)
 static inline void disarm_cq(struct mthca_dev *dev, int eqn, int cqn)
 {
 	if (!mthca_is_memfree(dev)) {
-		u32 doorbell[2];
+		__be32 doorbell[2];
 
 		doorbell[0] = cpu_to_be32(MTHCA_EQ_DB_DISARM_CQ | eqn);
 		doorbell[1] = cpu_to_be32(cqn);

drivers/infiniband/hw/mthca/mthca_mad.c

@@ -1,5 +1,7 @@
 /*
  * Copyright (c) 2004 Topspin Communications.  All rights reserved.
+ * Copyright (c) 2005 Mellanox Technologies. All rights reserved.
+ * Copyright (c) 2004 Voltaire, Inc. All rights reserved.
  *
  * This software is available to you under a choice of one of two
  * licenses.  You may choose to be licensed under the terms of the GNU
@@ -32,9 +34,9 @@
  * $Id: mthca_mad.c 1349 2004-12-16 21:09:43Z roland $
  */
 
-#include <ib_verbs.h>
-#include <ib_mad.h>
-#include <ib_smi.h>
+#include <rdma/ib_verbs.h>
+#include <rdma/ib_mad.h>
+#include <rdma/ib_smi.h>
 
 #include "mthca_dev.h"
 #include "mthca_cmd.h"
@@ -192,7 +194,7 @@ int mthca_process_mad(struct ib_device *ibdev,
 {
 	int err;
 	u8 status;
-	u16 slid = in_wc ? in_wc->slid : IB_LID_PERMISSIVE;
+	u16 slid = in_wc ? in_wc->slid : be16_to_cpu(IB_LID_PERMISSIVE);
 
 	/* Forward locally generated traps to the SM */
 	if (in_mad->mad_hdr.method == IB_MGMT_METHOD_TRAP &&