Merge tag 'for-5.17/parisc-1' of git://git.kernel.org/pub/scm/linux/k…

…ernel/git/deller/parisc-linux

Pull parisc architecture updates from Helge Deller:

 - Fix lpa and lpa_user defines (John David Anglin)

 - Fix symbol lookup of init functions with an __is_kernel() fix (Helge
   Deller)

 - Fix wrong pdc_toc_pim_11 and pdc_toc_pim_20 definitions (Helge
   Deller)

 - Add lws_atomic_xchg and lws_atomic_store syscalls (John David Anglin)

 - Rewrite light-weight syscall and futex code (John David Anglin)

 - Enable TOC (transfer of contents) feature unconditionally (Helge
   Deller)

 - Improve fault handler messages (John David Anglin)

 - Improve build process (Masahiro Yamada)

 - Reduce kernel code footprint of user access functions (Helge Deller)

 - Fix build error due to outX() macros (Bart Van Assche)

 - Ue default_groups in kobj_type in pdc_stable (Greg Kroah-Hartman)

 - Default to 16 CPUs on 32-bit kernel (Helge Deller)

* tag 'for-5.17/parisc-1' of git://git.kernel.org/pub/scm/linux/kernel/git/deller/parisc-linux:
  parisc: Default to 16 CPUs on 32-bit kernel
  sections: Fix __is_kernel() to include init ranges
  parisc: Re-use toc_stack as hpmc_stack
  parisc: Enable TOC (transfer of contents) feature unconditionally
  parisc: io: Improve the outb(), outw() and outl() macros
  parisc: pdc_stable: use default_groups in kobj_type
  parisc: Add kgdb io_module to read chars via PDC
  parisc: Fix pdc_toc_pim_11 and pdc_toc_pim_20 definitions
  parisc: Add lws_atomic_xchg and lws_atomic_store syscalls
  parisc: Rewrite light-weight syscall and futex code
  parisc: Enhance page fault termination message
  parisc: Don't call faulthandler_disabled() in do_page_fault()
  parisc: Switch user access functions to signal errors in r29 instead of r8
  parisc: Avoid calling faulthandler_disabled() twice
  parisc: Fix lpa and lpa_user defines
  parisc: Define depi_safe macro
  parisc: decompressor: do not copy source files while building

arch/parisc/Kconfig

@@ -287,20 +287,6 @@ config SMP
 
 	  If you don't know what to do here, say N.
 
-config TOC
-	bool "Support TOC switch"
-	default y if 64BIT || !SMP
-	help
-	  Most PA-RISC machines have either a switch at the back of the machine
-	  or a command in BMC to trigger a TOC interrupt. If you say Y here a
-	  handler will be installed which will either show a backtrace on all
-	  CPUs, or enter a possible configured debugger like kgdb/kdb.
-
-	  Note that with this option enabled, the kernel will use an additional 16KB
-	  per possible CPU as a special stack for the TOC handler.
-
-	  If you don't want to debug the Kernel, say N.
-
 config PARISC_CPU_TOPOLOGY
 	bool "Support cpu topology definition"
 	depends on SMP
@@ -370,7 +356,8 @@ config NR_CPUS
 	int "Maximum number of CPUs (2-32)"
 	range 2 32
 	depends on SMP
-	default "4"
+	default "4" if 64BIT
+	default "16"
 
 config KEXEC
 	bool "Kexec system call"

arch/parisc/boot/compressed/.gitignore

@@ -1,6 +1,4 @@
 # SPDX-License-Identifier: GPL-2.0-only
-firmware.c
-real2.S
 sizes.h
 vmlinux
 vmlinux.lds

arch/parisc/boot/compressed/Makefile

@@ -13,7 +13,6 @@ OBJECTS := head.o real2.o firmware.o misc.o piggy.o
 targets := vmlinux.lds vmlinux vmlinux.bin vmlinux.bin.gz vmlinux.bin.bz2
 targets += vmlinux.bin.xz vmlinux.bin.lzma vmlinux.bin.lzo vmlinux.bin.lz4
 targets += $(OBJECTS) sizes.h
-targets += real2.S firmware.c
 
 KBUILD_CFLAGS := -D__KERNEL__ -O2 -DBOOTLOADER
 KBUILD_CFLAGS += -DDISABLE_BRANCH_PROFILING
@@ -42,14 +41,7 @@ $(obj)/head.o: $(obj)/sizes.h
 CFLAGS_misc.o += -I$(objtree)/$(obj)
 $(obj)/misc.o: $(obj)/sizes.h
 
-$(obj)/firmware.o: $(obj)/firmware.c
-$(obj)/firmware.c: $(srctree)/arch/$(SRCARCH)/kernel/firmware.c
-	$(call cmd,shipped)
-
 AFLAGS_real2.o += -DBOOTLOADER
-$(obj)/real2.o: $(obj)/real2.S
-$(obj)/real2.S: $(srctree)/arch/$(SRCARCH)/kernel/real2.S
-	$(call cmd,shipped)
 
 CPPFLAGS_vmlinux.lds += -I$(objtree)/$(obj) -DBOOTLOADER
 $(obj)/vmlinux.lds: $(obj)/sizes.h

arch/parisc/boot/compressed/firmware.c

@@ -0,0 +1,2 @@
+// SPDX-License-Identifier: GPL-2.0-only
+#include "../../kernel/firmware.c"

arch/parisc/boot/compressed/real2.S

@@ -0,0 +1,2 @@
+/* SPDX-License-Identifier: GPL-2.0-only */
+#include "../../kernel/real2.S"

arch/parisc/include/asm/assembly.h

@@ -158,6 +158,16 @@
 #endif
 	.endm
 
+	/* The depi instruction leaves the most significant 32 bits of the
+	 * target register in an undefined state on PA 2.0 systems. */
+	.macro depi_safe i, p, len, t
+#ifdef CONFIG_64BIT
+	depdi	\i, 32+(\p), \len, \t
+#else
+	depi	\i, \p, \len, \t
+#endif
+	.endm
+
 	/* load 32-bit 'value' into 'reg' compensating for the ldil
 	 * sign-extension when running in wide mode.
 	 * WARNING!! neither 'value' nor 'reg' can be expressions

arch/parisc/include/asm/futex.h

@@ -8,39 +8,47 @@
 #include <asm/errno.h>
 
 /* The following has to match the LWS code in syscall.S.  We have
-   sixteen four-word locks. */
+ * 256 four-word locks. We use bits 20-27 of the futex virtual
+ * address for the hash index.
+ */
+
+static inline unsigned long _futex_hash_index(unsigned long ua)
+{
+	return (ua >> 2) & 0x3fc;
+}
 
 static inline void
-_futex_spin_lock(u32 __user *uaddr)
+_futex_spin_lock_irqsave(arch_spinlock_t *s, unsigned long *flags)
 {
-	extern u32 lws_lock_start[];
-	long index = ((long)uaddr & 0x7f8) >> 1;
-	arch_spinlock_t *s = (arch_spinlock_t *)&lws_lock_start[index];
-	preempt_disable();
+	local_irq_save(*flags);
 	arch_spin_lock(s);
 }
 
 static inline void
-_futex_spin_unlock(u32 __user *uaddr)
+_futex_spin_unlock_irqrestore(arch_spinlock_t *s, unsigned long *flags)
 {
-	extern u32 lws_lock_start[];
-	long index = ((long)uaddr & 0x7f8) >> 1;
-	arch_spinlock_t *s = (arch_spinlock_t *)&lws_lock_start[index];
 	arch_spin_unlock(s);
-	preempt_enable();
+	local_irq_restore(*flags);
 }
 
 static inline int
 arch_futex_atomic_op_inuser(int op, int oparg, int *oval, u32 __user *uaddr)
 {
+	extern u32 lws_lock_start[];
+	unsigned long ua = (unsigned long)uaddr;
+	arch_spinlock_t *s;
+	unsigned long flags;
 	int oldval, ret;
 	u32 tmp;
 
-	ret = -EFAULT;
+	s = (arch_spinlock_t *)&lws_lock_start[_futex_hash_index(ua)];
+	_futex_spin_lock_irqsave(s, &flags);
 
-	_futex_spin_lock(uaddr);
-	if (unlikely(get_user(oldval, uaddr) != 0))
+	/* Return -EFAULT if we encounter a page fault or COW break */
+	if (unlikely(get_user(oldval, uaddr) != 0)) {
+		ret = -EFAULT;
 		goto out_pagefault_enable;
+	}
 
 	ret = 0;
 	tmp = oldval;
@@ -63,13 +71,14 @@ arch_futex_atomic_op_inuser(int op, int oparg, int *oval, u32 __user *uaddr)
 		break;
 	default:
 		ret = -ENOSYS;
+		goto out_pagefault_enable;
 	}
 
-	if (ret == 0 && unlikely(put_user(tmp, uaddr) != 0))
+	if (unlikely(put_user(tmp, uaddr) != 0))
 		ret = -EFAULT;
 
 out_pagefault_enable:
-	_futex_spin_unlock(uaddr);
+	_futex_spin_unlock_irqrestore(s, &flags);
 
 	if (!ret)
 		*oval = oldval;
@@ -81,7 +90,11 @@ static inline int
 futex_atomic_cmpxchg_inatomic(u32 *uval, u32 __user *uaddr,
 			      u32 oldval, u32 newval)
 {
+	extern u32 lws_lock_start[];
+	unsigned long ua = (unsigned long)uaddr;
+	arch_spinlock_t *s;
 	u32 val;
+	unsigned long flags;
 
 	/* futex.c wants to do a cmpxchg_inatomic on kernel NULL, which is
 	 * our gateway page, and causes no end of trouble...
@@ -94,23 +107,25 @@ futex_atomic_cmpxchg_inatomic(u32 *uval, u32 __user *uaddr,
 
 	/* HPPA has no cmpxchg in hardware and therefore the
 	 * best we can do here is use an array of locks. The
-	 * lock selected is based on a hash of the userspace
-	 * address. This should scale to a couple of CPUs.
+	 * lock selected is based on a hash of the virtual
+	 * address of the futex. This should scale to a couple
+	 * of CPUs.
 	 */
 
-	_futex_spin_lock(uaddr);
+	s = (arch_spinlock_t *)&lws_lock_start[_futex_hash_index(ua)];
+	_futex_spin_lock_irqsave(s, &flags);
 	if (unlikely(get_user(val, uaddr) != 0)) {
-		_futex_spin_unlock(uaddr);
+		_futex_spin_unlock_irqrestore(s, &flags);
 		return -EFAULT;
 	}
 
 	if (val == oldval && unlikely(put_user(newval, uaddr) != 0)) {
-		_futex_spin_unlock(uaddr);
+		_futex_spin_unlock_irqrestore(s, &flags);
 		return -EFAULT;
 	}
 
 	*uval = val;
-	_futex_spin_unlock(uaddr);
+	_futex_spin_unlock_irqrestore(s, &flags);
 
 	return 0;
 }

arch/parisc/include/asm/io.h

@@ -273,9 +273,9 @@ static inline int inl(unsigned long addr)
 	return -1;
 }
 
-#define outb(x, y)	BUG()
-#define outw(x, y)	BUG()
-#define outl(x, y)	BUG()
+#define outb(x, y)	({(void)(x); (void)(y); BUG(); 0;})
+#define outw(x, y)	({(void)(x); (void)(y); BUG(); 0;})
+#define outl(x, y)	({(void)(x); (void)(y); BUG(); 0;})
 #endif
 
 /*

arch/parisc/include/asm/special_insns.h

@@ -2,28 +2,32 @@
 #ifndef __PARISC_SPECIAL_INSNS_H
 #define __PARISC_SPECIAL_INSNS_H
 
-#define lpa(va)	({			\
-	unsigned long pa;		\
-	__asm__ __volatile__(		\
-		"copy %%r0,%0\n\t"	\
-		"lpa %%r0(%1),%0"	\
-		: "=r" (pa)		\
-		: "r" (va)		\
-		: "memory"		\
-	);				\
-	pa;				\
+#define lpa(va)	({					\
+	unsigned long pa;				\
+	__asm__ __volatile__(				\
+		"copy %%r0,%0\n"			\
+		"8:\tlpa %%r0(%1),%0\n"			\
+		"9:\n"					\
+		ASM_EXCEPTIONTABLE_ENTRY(8b, 9b)	\
+		: "=&r" (pa)				\
+		: "r" (va)				\
+		: "memory"				\
+	);						\
+	pa;						\
 })
 
-#define lpa_user(va)	({		\
-	unsigned long pa;		\
-	__asm__ __volatile__(		\
-		"copy %%r0,%0\n\t"	\
-		"lpa %%r0(%%sr3,%1),%0"	\
-		: "=r" (pa)		\
-		: "r" (va)		\
-		: "memory"		\
-	);				\
-	pa;				\
+#define lpa_user(va)	({				\
+	unsigned long pa;				\
+	__asm__ __volatile__(				\
+		"copy %%r0,%0\n"			\
+		"8:\tlpa %%r0(%%sr3,%1),%0\n"		\
+		"9:\n"					\
+		ASM_EXCEPTIONTABLE_ENTRY(8b, 9b)	\
+		: "=&r" (pa)				\
+		: "r" (va)				\
+		: "memory"				\
+	);						\
+	pa;						\
 })
 
 #define mfctl(reg)	({		\

arch/parisc/include/asm/uaccess.h

@@ -53,15 +53,18 @@ struct exception_table_entry {
 /*
  * ASM_EXCEPTIONTABLE_ENTRY_EFAULT() creates a special exception table entry
  * (with lowest bit set) for which the fault handler in fixup_exception() will
- * load -EFAULT into %r8 for a read or write fault, and zeroes the target
+ * load -EFAULT into %r29 for a read or write fault, and zeroes the target
  * register in case of a read fault in get_user().
  */
+#define ASM_EXCEPTIONTABLE_REG	29
+#define ASM_EXCEPTIONTABLE_VAR(__variable)		\
+	register long __variable __asm__ ("r29") = 0
 #define ASM_EXCEPTIONTABLE_ENTRY_EFAULT( fault_addr, except_addr )\
 	ASM_EXCEPTIONTABLE_ENTRY( fault_addr, except_addr + 1)
 
 #define __get_user_internal(sr, val, ptr)		\
 ({							\
-	register long __gu_err __asm__ ("r8") = 0;	\
+	ASM_EXCEPTIONTABLE_VAR(__gu_err);		\
 							\
 	switch (sizeof(*(ptr))) {			\
 	case 1: __get_user_asm(sr, val, "ldb", ptr); break; \
@@ -131,7 +134,7 @@ struct exception_table_entry {
 
 #define __put_user_internal(sr, x, ptr)				\
 ({								\
-	register long __pu_err __asm__ ("r8") = 0;      	\
+	ASM_EXCEPTIONTABLE_VAR(__pu_err);		      	\
         __typeof__(*(ptr)) __x = (__typeof__(*(ptr)))(x);	\
 								\
 	switch (sizeof(*(ptr))) {				\
@@ -168,7 +171,8 @@ struct exception_table_entry {
  * gcc knows about, so there are no aliasing issues. These macros must
  * also be aware that fixups are executed in the context of the fault,
  * and any registers used there must be listed as clobbers.
- * r8 is already listed as err.
+ * The register holding the possible EFAULT error (ASM_EXCEPTIONTABLE_REG)
+ * is already listed as input and output register.
  */
 
 #define __put_user_asm(sr, stx, x, ptr)				\