2022-02-01 10:21:53

by Christophe Leroy

[permalink] [raw]
Subject: [PATCH] powerpc/32s: Make pte_update() non atomic on 603 core

On 603 core, TLB miss handler don't do any change to the
page tables so pte_update() doesn't need to be atomic.

Signed-off-by: Christophe Leroy <[email protected]>
---
arch/powerpc/include/asm/book3s/32/pgtable.h | 37 ++++++++++++--------
1 file changed, 22 insertions(+), 15 deletions(-)

diff --git a/arch/powerpc/include/asm/book3s/32/pgtable.h b/arch/powerpc/include/asm/book3s/32/pgtable.h
index f8b94f78403f..772e00dc4ef1 100644
--- a/arch/powerpc/include/asm/book3s/32/pgtable.h
+++ b/arch/powerpc/include/asm/book3s/32/pgtable.h
@@ -298,28 +298,35 @@ static inline pte_basic_t pte_update(struct mm_struct *mm, unsigned long addr, p
unsigned long clr, unsigned long set, int huge)
{
pte_basic_t old;
- unsigned long tmp;

- __asm__ __volatile__(
+ if (mmu_has_feature(MMU_FTR_HPTE_TABLE)) {
+ unsigned long tmp;
+
+ asm volatile(
#ifndef CONFIG_PTE_64BIT
-"1: lwarx %0, 0, %3\n"
-" andc %1, %0, %4\n"
+ "1: lwarx %0, 0, %3\n"
+ " andc %1, %0, %4\n"
#else
-"1: lwarx %L0, 0, %3\n"
-" lwz %0, -4(%3)\n"
-" andc %1, %L0, %4\n"
+ "1: lwarx %L0, 0, %3\n"
+ " lwz %0, -4(%3)\n"
+ " andc %1, %L0, %4\n"
#endif
-" or %1, %1, %5\n"
-" stwcx. %1, 0, %3\n"
-" bne- 1b"
- : "=&r" (old), "=&r" (tmp), "=m" (*p)
+ " or %1, %1, %5\n"
+ " stwcx. %1, 0, %3\n"
+ " bne- 1b"
+ : "=&r" (old), "=&r" (tmp), "=m" (*p)
#ifndef CONFIG_PTE_64BIT
- : "r" (p),
+ : "r" (p),
#else
- : "b" ((unsigned long)(p) + 4),
+ : "b" ((unsigned long)(p) + 4),
#endif
- "r" (clr), "r" (set), "m" (*p)
- : "cc" );
+ "r" (clr), "r" (set), "m" (*p)
+ : "cc" );
+ } else {
+ old = pte_val(*p);
+
+ *p = __pte((old & ~(pte_basic_t)clr) | set);
+ }

return old;
}
--
2.33.1


2022-02-15 09:11:44

by Michael Ellerman

[permalink] [raw]
Subject: Re: [PATCH] powerpc/32s: Make pte_update() non atomic on 603 core

On Sun, 30 Jan 2022 10:29:34 +0000, Christophe Leroy wrote:
> On 603 core, TLB miss handler don't do any change to the
> page tables so pte_update() doesn't need to be atomic.
>
>

Applied to powerpc/next.

[1/1] powerpc/32s: Make pte_update() non atomic on 603 core
https://git.kernel.org/powerpc/c/4291d085b0b07a78403e845c187428b038c901cd

cheers