Merge tag 'parisc-for-6.12-rc1' of...

Merge tag 'parisc-for-6.12-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/deller/parisc-linux

Pull parisc architecture updates from Helge Deller:

 - On parisc we now use the generic clockevent framework for timekeeping

 - Although there is no 64-bit glibc/userspace for parisc yet, for
   testing purposes one can run statically linked 64-bit binaries. This
   patchset contains two patches which fix 64-bit userspace which has
   been broken since kernel 4.19

 - Fix the userspace stack position and size when the ADDR_NO_RANDOMIZE
   personality is enabled

 - On other architectures mmap(MAP_GROWSDOWN | MAP_STACK) creates a
   downward-growing stack. On parisc mmap(MAP_STACK) is now sufficient
   to create an upward-growing stack

* tag 'parisc-for-6.12-rc1' of git://git.kernel.org/pub/scm/linux/kernel/git/deller/parisc-linux:
  parisc: Allow mmap(MAP_STACK) memory to automatically expand upwards
  parisc: Use PRIV_USER instead of hardcoded value
  parisc: Fix itlb miss handler for 64-bit programs
  parisc: Fix 64-bit userspace syscall path
  parisc: Fix stack start for ADDR_NO_RANDOMIZE personality
  parisc: Convert to generic clockevents
  parisc: pdc_stable: Constify struct kobj_type

arch/parisc/Kconfig

@@ -72,7 +72,7 @@ config PARISC
 	select GENERIC_SCHED_CLOCK
 	select GENERIC_IRQ_MIGRATION if SMP
 	select HAVE_UNSTABLE_SCHED_CLOCK if SMP
-	select LEGACY_TIMER_TICK
+	select GENERIC_CLOCKEVENTS
 	select CPU_NO_EFFICIENT_FFS
 	select THREAD_INFO_IN_TASK
 	select NEED_DMA_MAP_STATE

arch/parisc/include/asm/mman.h

@@ -11,4 +11,18 @@ static inline bool arch_memory_deny_write_exec_supported(void)
 }
 #define arch_memory_deny_write_exec_supported arch_memory_deny_write_exec_supported
 
+static inline unsigned long arch_calc_vm_flag_bits(unsigned long flags)
+{
+	/*
+	 * The stack on parisc grows upwards, so if userspace requests memory
+	 * for a stack, mark it with VM_GROWSUP so that the stack expansion in
+	 * the fault handler will work.
+	 */
+	if (flags & MAP_STACK)
+		return VM_GROWSUP;
+
+	return 0;
+}
+#define arch_calc_vm_flag_bits(flags) arch_calc_vm_flag_bits(flags)
+
 #endif /* __ASM_MMAN_H__ */

arch/parisc/include/asm/processor.h

@@ -298,7 +298,7 @@ extern unsigned int toc_handler_csum;
 extern void do_cpu_irq_mask(struct pt_regs *);
 extern irqreturn_t timer_interrupt(int, void *);
 extern irqreturn_t ipi_interrupt(int, void *);
-extern void start_cpu_itimer(void);
+extern void parisc_clockevent_init(void);
 extern void handle_interruption(int, struct pt_regs *);
 
 /* called from assembly code: */

arch/parisc/kernel/entry.S

@@ -1051,8 +1051,7 @@ ENTRY_CFI(intr_save)		/* for os_hpmc */
 	STREG           %r16, PT_ISR(%r29)
 	STREG           %r17, PT_IOR(%r29)
 
-#if 0 && defined(CONFIG_64BIT)
-	/* Revisit when we have 64-bit code above 4Gb */
+#if defined(CONFIG_64BIT)
 	b,n		intr_save2
 
 skip_save_ior:
@@ -1060,8 +1059,7 @@ skip_save_ior:
 	 * need to adjust iasq/iaoq here in the same way we adjusted isr/ior
 	 * above.
 	 */
-	extrd,u,*	%r8,PSW_W_BIT,1,%r1
-	cmpib,COND(=),n	1,%r1,intr_save2
+	bb,COND(>=),n	%r8,PSW_W_BIT,intr_save2
 	LDREG		PT_IASQ0(%r29), %r16
 	LDREG		PT_IAOQ0(%r29), %r17
 	/* adjust iasq/iaoq */

arch/parisc/kernel/smp.c

@@ -297,7 +297,7 @@ smp_cpu_init(int cpunum)
 	enter_lazy_tlb(&init_mm, current);
 
 	init_IRQ();   /* make sure no IRQs are enabled or pending */
-	start_cpu_itimer();
+	parisc_clockevent_init();
 }
 
 

arch/parisc/kernel/syscall.S

@@ -243,10 +243,10 @@ linux_gateway_entry:
 
 #ifdef CONFIG_64BIT
 	ldil	L%sys_call_table, %r1
-	or,=	%r2,%r2,%r2
-	addil	L%(sys_call_table64-sys_call_table), %r1
+	or,ev	%r2,%r2,%r2
+	ldil	L%sys_call_table64, %r1
 	ldo	R%sys_call_table(%r1), %r19
-	or,=	%r2,%r2,%r2
+	or,ev	%r2,%r2,%r2
 	ldo	R%sys_call_table64(%r1), %r19
 #else
 	load32	sys_call_table, %r19
@@ -379,10 +379,10 @@ tracesys_next:
 	extrd,u	%r19,63,1,%r2			/* W hidden in bottom bit */
 
 	ldil	L%sys_call_table, %r1
-	or,=	%r2,%r2,%r2
-	addil	L%(sys_call_table64-sys_call_table), %r1
+	or,ev	%r2,%r2,%r2
+	ldil	L%sys_call_table64, %r1
 	ldo	R%sys_call_table(%r1), %r19
-	or,=	%r2,%r2,%r2
+	or,ev	%r2,%r2,%r2
 	ldo	R%sys_call_table64(%r1), %r19
 #else
 	load32	sys_call_table, %r19
@@ -1327,6 +1327,8 @@ ENTRY(sys_call_table)
 END(sys_call_table)
 
 #ifdef CONFIG_64BIT
+#undef __SYSCALL_WITH_COMPAT
+#define __SYSCALL_WITH_COMPAT(nr, native, compat)	__SYSCALL(nr, native)
 	.align 8
 ENTRY(sys_call_table64)
 #include <asm/syscall_table_64.h>    /* 64-bit syscalls */

arch/parisc/kernel/time.c

 // SPDX-License-Identifier: GPL-2.0
 /*
- *  linux/arch/parisc/kernel/time.c
+ * Common time service routines for parisc machines.
+ * based on arch/loongarch/kernel/time.c
  *
- *  Copyright (C) 1991, 1992, 1995  Linus Torvalds
- *  Modifications for ARM (C) 1994, 1995, 1996,1997 Russell King
- *  Copyright (C) 1999 SuSE GmbH, (Philipp Rumpf, prumpf@tux.org)
- *
- * 1994-07-02  Alan Modra
- *             fixed set_rtc_mmss, fixed time.year for >= 2000, new mktime
- * 1998-12-20  Updated NTP code according to technical memorandum Jan '96
- *             "A Kernel Model for Precision Timekeeping" by Dave Mills
+ * Copyright (C) 2024 Helge Deller <deller@gmx.de>
  */
-#include <linux/errno.h>
-#include <linux/module.h>
-#include <linux/rtc.h>
-#include <linux/sched.h>
-#include <linux/sched/clock.h>
-#include <linux/sched_clock.h>
-#include <linux/kernel.h>
-#include <linux/param.h>
-#include <linux/string.h>
-#include <linux/mm.h>
-#include <linux/interrupt.h>
-#include <linux/time.h>
+#include <linux/clockchips.h>
+#include <linux/delay.h>
+#include <linux/export.h>
 #include <linux/init.h>
-#include <linux/smp.h>
-#include <linux/profile.h>
-#include <linux/clocksource.h>
+#include <linux/interrupt.h>
+#include <linux/kernel.h>
+#include <linux/sched_clock.h>
+#include <linux/spinlock.h>
+#include <linux/rtc.h>
 #include <linux/platform_device.h>
-#include <linux/ftrace.h>
+#include <asm/processor.h>
 
-#include <linux/uaccess.h>
-#include <asm/io.h>
-#include <asm/irq.h>
-#include <asm/page.h>
-#include <asm/param.h>
-#include <asm/pdc.h>
-#include <asm/led.h>
+static u64 cr16_clock_freq;
+static unsigned long clocktick;
 
-#include <linux/timex.h>
+int time_keeper_id;	/* CPU used for timekeeping */
 
-int time_keeper_id __read_mostly;	/* CPU used for timekeeping. */
+static DEFINE_PER_CPU(struct clock_event_device, parisc_clockevent_device);
 
-static unsigned long clocktick __ro_after_init;	/* timer cycles per tick */
+static void parisc_event_handler(struct clock_event_device *dev)
+{
+}
 
-/*
- * We keep time on PA-RISC Linux by using the Interval Timer which is
- * a pair of registers; one is read-only and one is write-only; both
- * accessed through CR16.  The read-only register is 32 or 64 bits wide,
- * and increments by 1 every CPU clock tick.  The architecture only
- * guarantees us a rate between 0.5 and 2, but all implementations use a
- * rate of 1.  The write-only register is 32-bits wide.  When the lowest
- * 32 bits of the read-only register compare equal to the write-only
- * register, it raises a maskable external interrupt.  Each processor has
- * an Interval Timer of its own and they are not synchronised.  
- *
- * We want to generate an interrupt every 1/HZ seconds.  So we program
- * CR16 to interrupt every @clocktick cycles.  The it_value in cpu_data
- * is programmed with the intended time of the next tick.  We can be
- * held off for an arbitrarily long period of time by interrupts being
- * disabled, so we may miss one or more ticks.
- */
-irqreturn_t __irq_entry timer_interrupt(int irq, void *dev_id)
+static int parisc_timer_next_event(unsigned long delta, struct clock_event_device *evt)
 {
-	unsigned long now;
-	unsigned long next_tick;
-	unsigned long ticks_elapsed = 0;
-	unsigned int cpu = smp_processor_id();
-	struct cpuinfo_parisc *cpuinfo = &per_cpu(cpu_data, cpu);
-
-	/* gcc can optimize for "read-only" case with a local clocktick */
-	unsigned long cpt = clocktick;
-
-	/* Initialize next_tick to the old expected tick time. */
-	next_tick = cpuinfo->it_value;
-
-	/* Calculate how many ticks have elapsed. */
-	now = mfctl(16);
-	do {
-		++ticks_elapsed;
-		next_tick += cpt;
-	} while (next_tick - now > cpt);
-
-	/* Store (in CR16 cycles) up to when we are accounting right now. */
-	cpuinfo->it_value = next_tick;
-
-	/* Go do system house keeping. */
-	if (IS_ENABLED(CONFIG_SMP) && (cpu != time_keeper_id))
-		ticks_elapsed = 0;
-	legacy_timer_tick(ticks_elapsed);
-
-	/* Skip clockticks on purpose if we know we would miss those.
-	 * The new CR16 must be "later" than current CR16 otherwise
-	 * itimer would not fire until CR16 wrapped - e.g 4 seconds
-	 * later on a 1Ghz processor. We'll account for the missed
-	 * ticks on the next timer interrupt.
-	 * We want IT to fire modulo clocktick even if we miss/skip some.
-	 * But those interrupts don't in fact get delivered that regularly.
-	 *
-	 * "next_tick - now" will always give the difference regardless
-	 * if one or the other wrapped. If "now" is "bigger" we'll end up
-	 * with a very large unsigned number.
-	 */
-	now = mfctl(16);
-	while (next_tick - now > cpt)
-		next_tick += cpt;
-
-	/* Program the IT when to deliver the next interrupt.
-	 * Only bottom 32-bits of next_tick are writable in CR16!
-	 * Timer interrupt will be delivered at least a few hundred cycles
-	 * after the IT fires, so if we are too close (<= 8000 cycles) to the
-	 * next cycle, simply skip it.
-	 */
-	if (next_tick - now <= 8000)
-		next_tick += cpt;
-	mtctl(next_tick, 16);
+	unsigned long new_cr16;
 
-	return IRQ_HANDLED;
-}
+	new_cr16 = mfctl(16) + delta;
+	mtctl(new_cr16, 16);
 
+	return 0;
+}
 
-unsigned long profile_pc(struct pt_regs *regs)
+irqreturn_t timer_interrupt(int irq, void *data)
 {
-	unsigned long pc = instruction_pointer(regs);
+	struct clock_event_device *cd;
+	int cpu = smp_processor_id();
 
-	if (regs->gr[0] & PSW_N)
-		pc -= 4;
+	cd = &per_cpu(parisc_clockevent_device, cpu);
 
-#ifdef CONFIG_SMP
-	if (in_lock_functions(pc))
-		pc = regs->gr[2];
-#endif
+	if (clockevent_state_periodic(cd))
+		parisc_timer_next_event(clocktick, cd);
 
-	return pc;
+	if (clockevent_state_periodic(cd) || clockevent_state_oneshot(cd))
+		cd->event_handler(cd);
+
+	return IRQ_HANDLED;
 }
-EXPORT_SYMBOL(profile_pc);
 
+static int parisc_set_state_oneshot(struct clock_event_device *evt)
+{
+	parisc_timer_next_event(clocktick, evt);
 
-/* clock source code */
+	return 0;
+}
 
-static u64 notrace read_cr16(struct clocksource *cs)
+static int parisc_set_state_periodic(struct clock_event_device *evt)
 {
-	return get_cycles();
+	parisc_timer_next_event(clocktick, evt);
+
+	return 0;
 }
 
-static struct clocksource clocksource_cr16 = {
-	.name			= "cr16",
-	.rating			= 300,
-	.read			= read_cr16,
-	.mask			= CLOCKSOURCE_MASK(BITS_PER_LONG),
-	.flags			= CLOCK_SOURCE_IS_CONTINUOUS,
-};
+static int parisc_set_state_shutdown(struct clock_event_device *evt)
+{
+	return 0;
+}
 
-void start_cpu_itimer(void)
+void parisc_clockevent_init(void)
 {
 	unsigned int cpu = smp_processor_id();
-	unsigned long next_tick = mfctl(16) + clocktick;
+	unsigned long min_delta = 0x600;	/* XXX */
+	unsigned long max_delta = (1UL << (BITS_PER_LONG - 1));
+	struct clock_event_device *cd;
+
+	cd = &per_cpu(parisc_clockevent_device, cpu);
+
+	cd->name = "cr16_clockevent";
+	cd->features = CLOCK_EVT_FEAT_ONESHOT | CLOCK_EVT_FEAT_PERIODIC |
+			CLOCK_EVT_FEAT_PERCPU;
+
+	cd->irq = TIMER_IRQ;
+	cd->rating = 320;
+	cd->cpumask = cpumask_of(cpu);
+	cd->set_state_oneshot = parisc_set_state_oneshot;
+	cd->set_state_oneshot_stopped = parisc_set_state_shutdown;
+	cd->set_state_periodic = parisc_set_state_periodic;
+	cd->set_state_shutdown = parisc_set_state_shutdown;
+	cd->set_next_event = parisc_timer_next_event;
+	cd->event_handler = parisc_event_handler;
+
+	clockevents_config_and_register(cd, cr16_clock_freq, min_delta, max_delta);
+}
+
+unsigned long notrace profile_pc(struct pt_regs *regs)
+{
+	unsigned long pc = instruction_pointer(regs);
 
-	mtctl(next_tick, 16);		/* kick off Interval Timer (CR16) */
+	if (regs->gr[0] & PSW_N)
+		pc -= 4;
+
+#ifdef CONFIG_SMP
+	if (in_lock_functions(pc))
+		pc = regs->gr[2];
+#endif
 
-	per_cpu(cpu_data, cpu).it_value = next_tick;
+	return pc;
 }
+EXPORT_SYMBOL(profile_pc);
 
 #if IS_ENABLED(CONFIG_RTC_DRV_GENERIC)
 static int rtc_generic_get_time(struct device *dev, struct rtc_time *tm)
@@ -224,12 +177,27 @@ void read_persistent_clock64(struct timespec64 *ts)
 	}
 }
 
-
 static u64 notrace read_cr16_sched_clock(void)
 {
 	return get_cycles();
 }
 
+static u64 notrace read_cr16(struct clocksource *cs)
+{
+	return get_cycles();
+}
+
+static struct clocksource clocksource_cr16 = {
+	.name			= "cr16",
+	.rating			= 300,
+	.read			= read_cr16,
+	.mask			= CLOCKSOURCE_MASK(BITS_PER_LONG),
+	.flags			= CLOCK_SOURCE_IS_CONTINUOUS |
+					CLOCK_SOURCE_VALID_FOR_HRES |
+					CLOCK_SOURCE_MUST_VERIFY |
+					CLOCK_SOURCE_VERIFY_PERCPU,
+};
+
 
 /*
  * timer interrupt and sched_clock() initialization
@@ -237,33 +205,14 @@ static u64 notrace read_cr16_sched_clock(void)
 
 void __init time_init(void)
 {
-	unsigned long cr16_hz;
-
-	clocktick = (100 * PAGE0->mem_10msec) / HZ;
-	start_cpu_itimer();	/* get CPU 0 started */
-
-	cr16_hz = 100 * PAGE0->mem_10msec;  /* Hz */
+	cr16_clock_freq = 100 * PAGE0->mem_10msec;  /* Hz */
+	clocktick = cr16_clock_freq / HZ;
 
 	/* register as sched_clock source */
-	sched_clock_register(read_cr16_sched_clock, BITS_PER_LONG, cr16_hz);
-}
+	sched_clock_register(read_cr16_sched_clock, BITS_PER_LONG, cr16_clock_freq);
 
-static int __init init_cr16_clocksource(void)
-{
-	/*
-	 * The cr16 interval timers are not synchronized across CPUs.
-	 */
-	if (num_online_cpus() > 1 && !running_on_qemu) {
-		clocksource_cr16.name = "cr16_unstable";
-		clocksource_cr16.flags = CLOCK_SOURCE_UNSTABLE;
-		clocksource_cr16.rating = 0;
-	}
+	parisc_clockevent_init();
 
 	/* register at clocksource framework */
-	clocksource_register_hz(&clocksource_cr16,
-		100 * PAGE0->mem_10msec);
-
-	return 0;
+	clocksource_register_hz(&clocksource_cr16, cr16_clock_freq);
 }
-
-device_initcall(init_cr16_clocksource);

arch/parisc/kernel/traps.c

@@ -504,7 +504,7 @@ void notrace handle_interruption(int code, struct pt_regs *regs)
 	if (((unsigned long)regs->iaoq[0] & 3) &&
 	    ((unsigned long)regs->iasq[0] != (unsigned long)regs->sr[7])) { 
 		/* Kill the user process later */
-		regs->iaoq[0] = 0 | 3;
+		regs->iaoq[0] = 0 | PRIV_USER;
 		regs->iaoq[1] = regs->iaoq[0] + 4;
 		regs->iasq[0] = regs->iasq[1] = regs->sr[7];
 		regs->gr[0] &= ~PSW_B;

drivers/parisc/pdc_stable.c

@@ -483,7 +483,7 @@ static struct attribute *paths_subsys_attrs[] = {
 ATTRIBUTE_GROUPS(paths_subsys);
 
 /* Specific kobject type for our PDC paths */
-static struct kobj_type ktype_pdcspath = {
+static const struct kobj_type ktype_pdcspath = {
 	.sysfs_ops = &pdcspath_attr_ops,
 	.default_groups = paths_subsys_groups,
 };

fs/exec.c

@@ -811,7 +811,8 @@ int setup_arg_pages(struct linux_binprm *bprm,
 	stack_base = calc_max_stack_size(stack_base);
 
 	/* Add space for stack randomization. */
-	stack_base += (STACK_RND_MASK << PAGE_SHIFT);
+	if (current->flags & PF_RANDOMIZE)
+		stack_base += (STACK_RND_MASK << PAGE_SHIFT);
 
 	/* Make sure we didn't let the argument array grow too large. */
 	if (vma->vm_end - vma->vm_start > stack_base)