[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]
Message-ID: <20240822073906.2176342-2-xin@zytor.com>
Date: Thu, 22 Aug 2024 00:39:04 -0700
From: "Xin Li (Intel)" <xin@...or.com>
To: linux-kernel@...r.kernel.org
Cc: tglx@...utronix.de, mingo@...hat.com, bp@...en8.de,
dave.hansen@...ux.intel.com, x86@...nel.org, hpa@...or.com,
peterz@...radead.org, andrew.cooper3@...rix.com, seanjc@...gle.com
Subject: [PATCH v2 1/3] x86/entry: Test ti_work for zero before processing individual bits
In most cases, ti_work values passed to arch_exit_to_user_mode_prepare()
are zeros, e.g., 99% in kernel build tests. So an obvious optimization
is to test ti_work for zero before processing individual bits in it.
In addition, Intel 0day tests find no perf regression with this change.
Suggested-by: H. Peter Anvin (Intel) <hpa@...or.com>
Signed-off-by: Xin Li (Intel) <xin@...or.com>
---
Change since v1:
* Leave fpregs_assert_state_consistent() unconditional and independent
of ti_work. (Brian Gerst and Thomas Gleixner)
* Add arch_exit_work() to spare an extra indentation level. (Thomas Gleixner)
---
arch/x86/include/asm/entry-common.h | 10 ++++++++--
1 file changed, 8 insertions(+), 2 deletions(-)
diff --git a/arch/x86/include/asm/entry-common.h b/arch/x86/include/asm/entry-common.h
index fb2809b20b0a..db970828f385 100644
--- a/arch/x86/include/asm/entry-common.h
+++ b/arch/x86/include/asm/entry-common.h
@@ -44,8 +44,7 @@ static __always_inline void arch_enter_from_user_mode(struct pt_regs *regs)
}
#define arch_enter_from_user_mode arch_enter_from_user_mode
-static inline void arch_exit_to_user_mode_prepare(struct pt_regs *regs,
- unsigned long ti_work)
+static inline void arch_exit_work(unsigned long ti_work)
{
if (ti_work & _TIF_USER_RETURN_NOTIFY)
fire_user_return_notifiers();
@@ -56,6 +55,13 @@ static inline void arch_exit_to_user_mode_prepare(struct pt_regs *regs,
fpregs_assert_state_consistent();
if (unlikely(ti_work & _TIF_NEED_FPU_LOAD))
switch_fpu_return();
+}
+
+static inline void arch_exit_to_user_mode_prepare(struct pt_regs *regs,
+ unsigned long ti_work)
+{
+ if (IS_ENABLED(CONFIG_X86_DEBUG_FPU) || unlikely(ti_work))
+ arch_exit_work(ti_work);
#ifdef CONFIG_COMPAT
/*
--
2.46.0
Powered by blists - more mailing lists