Merge tag 'arc-v3.14-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/vgupta/arc

Pull ARC changes from Vineet Gupta:
 - IPI optimization and cleanups
 - Support for bootloader provided external Device Tree blobs

* tag 'arc-v3.14-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/vgupta/arc:
  ARC: [cmdline] support External Device Trees from u-boot
  ARC: [cmdline] uboot cmdline handling rework
  ARC: [SMP] optimize IPI send and receive
  ARC: [SMP] simplify IPI code
  ARC: [SMP] cpu halt interface doesn't need "self" cpu-id
  ARC: [SMP] IPI ACK interface doesn't need "self" cpu-id
  ARC: [SMP] cpumask not needed in IPI send path

6 files changed, 121 insertions, 108 deletions

diff --git a/arch/arc/Kconfig b/arch/arc/Kconfig
index 5438cabbc45d..9be30c8cb0c2 100644
--- a/arch/arc/Kconfig
+++ b/arch/arc/Kconfig
@@ -409,17 +409,6 @@ config ARC_DBG_TLB_MISS_COUNT
           Counts number of I and D TLB Misses and exports them via Debugfs
           The counters can be cleared via Debugfs as well
 
-config CMDLINE_UBOOT
-        bool "Support U-boot kernel command line passing"
-        default n
-        help
-          If you are using U-boot (www.denx.de) and wish to pass the kernel
-          command line from the U-boot environment to the Linux kernel then
-          switch this option on.
-          ARC U-boot will setup the cmdline in RAM/flash and set r2 to point
-          to it. kernel startup code will append this to DeviceTree
-          /bootargs provided cmdline args.
-
 config ARC_BUILTIN_DTB_NAME
         string "Built in DTB"
         help
diff --git a/arch/arc/include/asm/smp.h b/arch/arc/include/asm/smp.h
index eefc29f08cdb..5d06eee43ea9 100644
--- a/arch/arc/include/asm/smp.h
+++ b/arch/arc/include/asm/smp.h
@@ -46,14 +46,14 @@ extern int smp_ipi_irq_setup(int cpu, int irq);
  *
  * @info:               SoC SMP specific info for /proc/cpuinfo etc
  * @cpu_kick:           For Master to kickstart a cpu (optionally at a PC)
- * @ipi_send:           To send IPI to a @cpumask
- * @ips_clear:          To clear IPI received by @cpu at @irq
+ * @ipi_send:           To send IPI to a @cpu
+ * @ips_clear:          To clear IPI received at @irq
  */
 struct plat_smp_ops {
         const char      *info;
         void            (*cpu_kick)(int cpu, unsigned long pc);
-        void            (*ipi_send)(void *callmap);
-        void            (*ipi_clear)(int cpu, int irq);
+        void            (*ipi_send)(int cpu);
+        void            (*ipi_clear)(int irq);
 };
 
 /* TBD: stop exporting it for direct population by platform */
diff --git a/arch/arc/kernel/head.S b/arch/arc/kernel/head.S
index 2c878e964a64..991997269d02 100644
--- a/arch/arc/kernel/head.S
+++ b/arch/arc/kernel/head.S
@@ -49,25 +49,13 @@ stext:
         st.ab   0, [r5,4]
         brlt    r5, r6, 1b
 
-#ifdef CONFIG_CMDLINE_UBOOT
-        ; support for bootloader provided cmdline
-        ;    If cmdline passed by u-boot, then
-        ;    r0 = 1  (because ATAGS parsing, now retired, used to use 0)
-        ;    r1 = magic number (board identity)
-        ;    r2 = addr of cmdline string (somewhere in memory/flash)
-
-        brne    r0, 1, .Lother_bootup_chores    ; u-boot didn't pass cmdline
-        breq    r2, 0, .Lother_bootup_chores    ; or cmdline is NULL
-
-        mov     r5, @command_line
-1:
-        ldb.ab  r6, [r2, 1]
-        breq    r6, 0, .Lother_bootup_chores
-        b.d     1b
-        stb.ab  r6, [r5, 1]
-#endif
-
-.Lother_bootup_chores:
+        ; Uboot - kernel ABI
+        ;    r0 = [0] No uboot interaction, [1] cmdline in r2, [2] DTB in r2
+        ;    r1 = magic number (board identity, unused as of now
+        ;    r2 = pointer to uboot provided cmdline or external DTB in mem
+        ; These are handled later in setup_arch()
+        st      r0, [@uboot_tag]
+        st      r2, [@uboot_arg]
 
         ; Identify if running on ISS vs Silicon
         ;       IDENTITY Reg [ 3  2  1  0 ]
diff --git a/arch/arc/kernel/setup.c b/arch/arc/kernel/setup.c
index 643eae4436e0..119dddb752b2 100644
--- a/arch/arc/kernel/setup.c
+++ b/arch/arc/kernel/setup.c
@@ -29,7 +29,10 @@
 
 int running_on_hw = 1;  /* vs. on ISS */
 
-char __initdata command_line[COMMAND_LINE_SIZE];
+/* Part of U-boot ABI: see head.S */
+int __initdata uboot_tag;
+char __initdata *uboot_arg;
+
 const struct machine_desc *machine_desc;
 
 struct task_struct *_current_task[NR_CPUS];     /* For stack switching */
@@ -311,19 +314,40 @@ void setup_processor(void)
         arc_chk_fpu();
 }
 
+static inline int is_kernel(unsigned long addr)
+{
+        if (addr >= (unsigned long)_stext && addr <= (unsigned long)_end)
+                return 1;
+        return 0;
+}
+
 void __init setup_arch(char **cmdline_p)
 {
-        /* This also populates @boot_command_line from /bootargs */
-        machine_desc = setup_machine_fdt(__dtb_start);
-        if (!machine_desc)
-                panic("Embedded DT invalid\n");
-
-        /* Append any u-boot provided cmdline */
-#ifdef CONFIG_CMDLINE_UBOOT
-        /* Add a whitespace seperator between the 2 cmdlines */
-        strlcat(boot_command_line, " ", COMMAND_LINE_SIZE);
-        strlcat(boot_command_line, command_line, COMMAND_LINE_SIZE);
-#endif
+        /* make sure that uboot passed pointer to cmdline/dtb is valid */
+        if (uboot_tag && is_kernel((unsigned long)uboot_arg))
+                panic("Invalid uboot arg\n");
+
+        /* See if u-boot passed an external Device Tree blob */
+        machine_desc = setup_machine_fdt(uboot_arg);    /* uboot_tag == 2 */
+        if (!machine_desc) {
+                /* No, so try the embedded one */
+                machine_desc = setup_machine_fdt(__dtb_start);
+                if (!machine_desc)
+                        panic("Embedded DT invalid\n");
+
+                /*
+                 * If we are here, it is established that @uboot_arg didn't
+                 * point to DT blob. Instead if u-boot says it is cmdline,
+                 * Appent to embedded DT cmdline.
+                 * setup_machine_fdt() would have populated @boot_command_line
+                 */
+                if (uboot_tag == 1) {
+                        /* Ensure a whitespace between the 2 cmdlines */
+                        strlcat(boot_command_line, " ", COMMAND_LINE_SIZE);
+                        strlcat(boot_command_line, uboot_arg,
+                                COMMAND_LINE_SIZE);
+                }
+        }
 
         /* Save unparsed command line copy for /proc/cmdline */
         *cmdline_p = boot_command_line;
diff --git a/arch/arc/kernel/smp.c b/arch/arc/kernel/smp.c
index c2f9ebbc38f6..40859e5619f9 100644
--- a/arch/arc/kernel/smp.c
+++ b/arch/arc/kernel/smp.c
@@ -197,51 +197,65 @@ int __init setup_profiling_timer(unsigned int multiplier)
 /*              Inter Processor Interrupt Handling                           */
 /*****************************************************************************/
 
-/*
- * structures for inter-processor calls
- * A Collection of single bit ipi messages
- *
- */
-
-/*
- * TODO_rajesh investigate tlb message types.
- * IPI Timer not needed because each ARC has an individual Interrupting Timer
- */
 enum ipi_msg_type {
-        IPI_NOP = 0,
+        IPI_EMPTY = 0,
         IPI_RESCHEDULE = 1,
         IPI_CALL_FUNC,
-        IPI_CPU_STOP
+        IPI_CPU_STOP,
 };
 
-struct ipi_data {
-        unsigned long bits;
-};
+/*
+ * In arches with IRQ for each msg type (above), receiver can use IRQ-id  to
+ * figure out what msg was sent. For those which don't (ARC has dedicated IPI
+ * IRQ), the msg-type needs to be conveyed via per-cpu data
+ */
 
-static DEFINE_PER_CPU(struct ipi_data, ipi_data);
+static DEFINE_PER_CPU(unsigned long, ipi_data);
 
-static void ipi_send_msg(const struct cpumask *callmap, enum ipi_msg_type msg)
+static void ipi_send_msg_one(int cpu, enum ipi_msg_type msg)
 {
+        unsigned long __percpu *ipi_data_ptr = per_cpu_ptr(&ipi_data, cpu);
+        unsigned long old, new;
         unsigned long flags;
-        unsigned int cpu;
+
+        pr_debug("%d Sending msg [%d] to %d\n", smp_processor_id(), msg, cpu);
 
         local_irq_save(flags);
 
-        for_each_cpu(cpu, callmap) {
-                struct ipi_data *ipi = &per_cpu(ipi_data, cpu);
-                set_bit(msg, &ipi->bits);
-        }
+        /*
+         * Atomically write new msg bit (in case others are writing too),
+         * and read back old value
+         */
+        do {
+                new = old = *ipi_data_ptr;
+                new |= 1U << msg;
+        } while (cmpxchg(ipi_data_ptr, old, new) != old);
 
-        /* Call the platform specific cross-CPU call function  */
-        if (plat_smp_ops.ipi_send)
-                plat_smp_ops.ipi_send((void *)callmap);
+        /*
+         * Call the platform specific IPI kick function, but avoid if possible:
+         * Only do so if there's no pending msg from other concurrent sender(s).
+         * Otherwise, recevier will see this msg as well when it takes the
+         * IPI corresponding to that msg. This is true, even if it is already in
+         * IPI handler, because !@old means it has not yet dequeued the msg(s)
+         * so @new msg can be a free-loader
+         */
+        if (plat_smp_ops.ipi_send && !old)
+                plat_smp_ops.ipi_send(cpu);
 
         local_irq_restore(flags);
 }
 
+static void ipi_send_msg(const struct cpumask *callmap, enum ipi_msg_type msg)
+{
+        unsigned int cpu;
+
+        for_each_cpu(cpu, callmap)
+                ipi_send_msg_one(cpu, msg);
+}
+
 void smp_send_reschedule(int cpu)
 {
-        ipi_send_msg(cpumask_of(cpu), IPI_RESCHEDULE);
+        ipi_send_msg_one(cpu, IPI_RESCHEDULE);
 }
 
 void smp_send_stop(void)
@@ -254,7 +268,7 @@ void smp_send_stop(void)
 
 void arch_send_call_function_single_ipi(int cpu)
 {
-        ipi_send_msg(cpumask_of(cpu), IPI_CALL_FUNC);
+        ipi_send_msg_one(cpu, IPI_CALL_FUNC);
 }
 
 void arch_send_call_function_ipi_mask(const struct cpumask *mask)
@@ -265,33 +279,29 @@ void arch_send_call_function_ipi_mask(const struct cpumask *mask)
 /*
  * ipi_cpu_stop - handle IPI from smp_send_stop()
  */
-static void ipi_cpu_stop(unsigned int cpu)
+static void ipi_cpu_stop(void)
 {
         machine_halt();
 }
 
-static inline void __do_IPI(unsigned long *ops, struct ipi_data *ipi, int cpu)
+static inline void __do_IPI(unsigned long msg)
 {
-        unsigned long msg = 0;
+        switch (msg) {
+        case IPI_RESCHEDULE:
+                scheduler_ipi();
+                break;
 
-        do {
-                msg = find_next_bit(ops, BITS_PER_LONG, msg+1);
+        case IPI_CALL_FUNC:
+                generic_smp_call_function_interrupt();
+                break;
 
-                switch (msg) {
-                case IPI_RESCHEDULE:
-                        scheduler_ipi();
-                        break;
-
-                case IPI_CALL_FUNC:
-                        generic_smp_call_function_interrupt();
-                        break;
-
-                case IPI_CPU_STOP:
-                        ipi_cpu_stop(cpu);
-                        break;
-                }
-        } while (msg < BITS_PER_LONG);
+        case IPI_CPU_STOP:
+                ipi_cpu_stop();
+                break;
 
+        default:
+                pr_warn("IPI with unexpected msg %ld\n", msg);
+        }
 }
 
 /*
@@ -300,19 +310,25 @@ static inline void __do_IPI(unsigned long *ops, struct ipi_data *ipi, int cpu)
  */
 irqreturn_t do_IPI(int irq, void *dev_id)
 {
-        int cpu = smp_processor_id();
-        struct ipi_data *ipi = &per_cpu(ipi_data, cpu);
-        unsigned long ops;
+        unsigned long pending;
+
+        pr_debug("IPI [%ld] received on cpu %d\n",
+                 *this_cpu_ptr(&ipi_data), smp_processor_id());
 
         if (plat_smp_ops.ipi_clear)
-                plat_smp_ops.ipi_clear(cpu, irq);
+                plat_smp_ops.ipi_clear(irq);
 
         /*
-         * XXX: is this loop really needed
-         * And do we need to move ipi_clean inside
+         * "dequeue" the msg corresponding to this IPI (and possibly other
+         * piggybacked msg from elided IPIs: see ipi_send_msg_one() above)
          */
-        while ((ops = xchg(&ipi->bits, 0)) != 0)
-                __do_IPI(&ops, ipi, cpu);
+        pending = xchg(this_cpu_ptr(&ipi_data), 0);
+
+        do {
+                unsigned long msg = __ffs(pending);
+                __do_IPI(msg);
+                pending &= ~(1U << msg);
+        } while (pending);
 
         return IRQ_HANDLED;
 }
diff --git a/arch/arc/plat-arcfpga/smp.c b/arch/arc/plat-arcfpga/smp.c
index 91b55349a5f8..8a12741f5f7a 100644
--- a/arch/arc/plat-arcfpga/smp.c
+++ b/arch/arc/plat-arcfpga/smp.c
@@ -88,18 +88,14 @@ void iss_model_init_smp(unsigned int cpu)
         smp_ipi_irq_setup(cpu, IDU_INTERRUPT_0 + cpu);
 }
 
-static void iss_model_ipi_send(void *arg)
+static void iss_model_ipi_send(int cpu)
 {
-        struct cpumask *callmap = arg;
-        unsigned int cpu;
-
-        for_each_cpu(cpu, callmap)
-                idu_irq_assert(cpu);
+        idu_irq_assert(cpu);
 }
 
-static void iss_model_ipi_clear(int cpu, int irq)
+static void iss_model_ipi_clear(int irq)
 {
-        idu_irq_clear(IDU_INTERRUPT_0 + cpu);
+        idu_irq_clear(IDU_INTERRUPT_0 + smp_processor_id());
 }
 
 void iss_model_init_early_smp(void)