From: Anton Blanchard <anton@samba.org>
To: benh@kernel.crashing.org, paulus@samba.org, mpe@ellerman.id.au,
mikey@neuling.org, cyrilbur@gmail.com
Cc: linuxppc-dev@lists.ozlabs.org
Subject: [PATCH 16/19] powerpc: create giveup_all()
Date: Wed, 28 Oct 2015 11:51:04 +1100 [thread overview]
Message-ID: <1445993467-667-16-git-send-email-anton@samba.org> (raw)
In-Reply-To: <1445993467-667-1-git-send-email-anton@samba.org>
Create a single function that gives everything up (FP, VMX, VSX, SPE).
Doing this all at once means we only do one MSR write.
A context switch microbenchmark using yield():
http://ozlabs.org/~anton/junkcode/context_switch2.c
./context_switch2 --test=yield --fp --altivec --vector 0 0
shows an improvement of 3% on POWER8.
Signed-off-by: Anton Blanchard <anton@samba.org>
---
arch/powerpc/include/asm/switch_to.h | 1 +
arch/powerpc/kernel/process.c | 75 ++++++++++++++++++++++++++++--------
arch/powerpc/kvm/book3s_pr.c | 17 +-------
3 files changed, 63 insertions(+), 30 deletions(-)
diff --git a/arch/powerpc/include/asm/switch_to.h b/arch/powerpc/include/asm/switch_to.h
index b6a4951..0b4a339 100644
--- a/arch/powerpc/include/asm/switch_to.h
+++ b/arch/powerpc/include/asm/switch_to.h
@@ -27,6 +27,7 @@ extern void giveup_vsx(struct task_struct *);
extern void enable_kernel_spe(void);
extern void giveup_spe(struct task_struct *);
extern void load_up_spe(struct task_struct *);
+extern void giveup_all(struct task_struct *);
extern void switch_booke_debug_regs(struct debug_reg *new_debug);
#ifdef CONFIG_PPC_FPU
diff --git a/arch/powerpc/kernel/process.c b/arch/powerpc/kernel/process.c
index cf2d811..5ca416c 100644
--- a/arch/powerpc/kernel/process.c
+++ b/arch/powerpc/kernel/process.c
@@ -308,6 +308,64 @@ void flush_spe_to_thread(struct task_struct *tsk)
}
#endif /* CONFIG_SPE */
+static unsigned long msr_all_available;
+
+static int __init init_msr_all_available(void)
+{
+#ifdef CONFIG_PPC_FPU
+ msr_all_available |= MSR_FP;
+#endif
+#ifdef CONFIG_ALTIVEC
+ if (cpu_has_feature(CPU_FTR_ALTIVEC))
+ msr_all_available |= MSR_VEC;
+#endif
+#ifdef CONFIG_VSX
+ if (cpu_has_feature(CPU_FTR_VSX))
+ msr_all_available |= MSR_VSX;
+#endif
+#ifdef CONFIG_SPE
+ if (cpu_has_feature(CPU_FTR_SPE))
+ msr_all_available |= MSR_SPE;
+#endif
+
+ return 0;
+}
+early_initcall(init_msr_all_available);
+
+void giveup_all(struct task_struct *tsk)
+{
+ unsigned long usermsr;
+
+ if (!tsk->thread.regs)
+ return;
+
+ usermsr = tsk->thread.regs->msr;
+
+ if ((usermsr & msr_all_available) == 0)
+ return;
+
+ msr_check_and_set(msr_all_available);
+
+#ifdef CONFIG_PPC_FPU
+ if (usermsr & MSR_FP)
+ __giveup_fpu(tsk);
+#endif
+#ifdef CONFIG_ALTIVEC
+ if (usermsr & MSR_VEC)
+ __giveup_altivec(tsk);
+#endif
+#ifdef CONFIG_VSX
+ if (usermsr & MSR_VSX)
+ __giveup_vsx(tsk);
+#endif
+#ifdef CONFIG_SPE
+ if (usermsr & MSR_SPE)
+ __giveup_spe(tsk);
+#endif
+
+ msr_check_and_clear(msr_all_available);
+}
+
#ifdef CONFIG_PPC_ADV_DEBUG_REGS
void do_send_trap(struct pt_regs *regs, unsigned long address,
unsigned long error_code, int signal_code, int breakpt)
@@ -839,21 +897,8 @@ struct task_struct *__switch_to(struct task_struct *prev,
__switch_to_tm(prev);
- if (prev->thread.regs && (prev->thread.regs->msr & MSR_FP))
- giveup_fpu(prev);
-#ifdef CONFIG_ALTIVEC
- if (prev->thread.regs && (prev->thread.regs->msr & MSR_VEC))
- giveup_altivec(prev);
-#endif /* CONFIG_ALTIVEC */
-#ifdef CONFIG_VSX
- if (prev->thread.regs && (prev->thread.regs->msr & MSR_VSX))
- /* VMX and FPU registers are already save here */
- __giveup_vsx(prev);
-#endif /* CONFIG_VSX */
-#ifdef CONFIG_SPE
- if ((prev->thread.regs && (prev->thread.regs->msr & MSR_SPE)))
- giveup_spe(prev);
-#endif /* CONFIG_SPE */
+ /* Save FPU, Altivec, VSX and SPE state */
+ giveup_all(prev);
#ifdef CONFIG_PPC_ADV_DEBUG_REGS
switch_booke_debug_regs(&new->thread.debug);
diff --git a/arch/powerpc/kvm/book3s_pr.c b/arch/powerpc/kvm/book3s_pr.c
index 49f5dad..a78e0e6 100644
--- a/arch/powerpc/kvm/book3s_pr.c
+++ b/arch/powerpc/kvm/book3s_pr.c
@@ -1490,21 +1490,8 @@ static int kvmppc_vcpu_run_pr(struct kvm_run *kvm_run, struct kvm_vcpu *vcpu)
goto out;
/* interrupts now hard-disabled */
- /* Save FPU state in thread_struct */
- if (current->thread.regs->msr & MSR_FP)
- giveup_fpu(current);
-
-#ifdef CONFIG_ALTIVEC
- /* Save Altivec state in thread_struct */
- if (current->thread.regs->msr & MSR_VEC)
- giveup_altivec(current);
-#endif
-
-#ifdef CONFIG_VSX
- /* Save VSX state in thread_struct */
- if (current->thread.regs->msr & MSR_VSX)
- __giveup_vsx(current);
-#endif
+ /* Save FPU, Altivec and VSX state */
+ giveup_all(current);
/* Preload FPU if it's enabled */
if (kvmppc_get_msr(vcpu) & MSR_FP)
--
2.5.0
next prev parent reply other threads:[~2015-10-28 0:51 UTC|newest]
Thread overview: 24+ messages / expand[flat|nested] mbox.gz Atom feed top
2015-10-28 0:50 [PATCH 01/19] powerpc: Don't disable kernel FP/VMX/VSX MSR bits on context switch Anton Blanchard
2015-10-28 0:50 ` [PATCH 02/19] powerpc: Don't disable MSR bits in do_load_up_transact_*() functions Anton Blanchard
2015-10-28 0:50 ` [PATCH 03/19] powerpc: Create context switch helpers save_sprs() and restore_sprs() Anton Blanchard
2015-10-28 0:50 ` [PATCH 04/19] powerpc: Remove redundant mflr in _switch Anton Blanchard
2015-10-28 0:50 ` [PATCH 05/19] powerpc: Remove UP only lazy floating point and vector optimisations Anton Blanchard
2015-10-28 0:50 ` [PATCH 06/19] powerpc: Simplify TM restore checks Anton Blanchard
2015-10-28 0:50 ` [PATCH 07/19] powerpc: Create mtmsrd_isync() Anton Blanchard
2015-10-28 0:50 ` [PATCH 08/19] powerpc: Remove NULL task struct pointer checks in FP and vector code Anton Blanchard
2015-10-28 0:50 ` [PATCH 09/19] powerpc: Move part of giveup_fpu,altivec,spe into c Anton Blanchard
2015-10-28 3:17 ` [PATCH 09/19] powerpc: Move part of giveup_fpu, altivec, spe " kbuild test robot
2015-10-28 0:50 ` [PATCH 10/19] powerpc: Move part of giveup_vsx " Anton Blanchard
2015-10-28 0:50 ` [PATCH 11/19] crypto: vmx: Only call enable_kernel_vsx() Anton Blanchard
2015-10-28 0:51 ` [PATCH 12/19] powerpc: Create msr_check_and_{set,clear}() Anton Blanchard
2015-10-28 0:51 ` [PATCH 13/19] powerpc: Create disable_kernel_{fp,altivec,vsx,spe}() Anton Blanchard
2015-10-28 0:51 ` [PATCH 14/19] powerpc: Add ppc_strict_facility_enable boot option Anton Blanchard
2015-10-28 0:51 ` [PATCH 15/19] powerpc: Remove fp_enable() and vec_enable(), use msr_check_and_{set, clear}() Anton Blanchard
2015-10-28 0:51 ` Anton Blanchard [this message]
2015-10-28 0:51 ` [PATCH 17/19] powerpc: create flush_all_to_thread() Anton Blanchard
2015-10-28 0:51 ` [PATCH 18/19] powerpc: Rearrange __switch_to() Anton Blanchard
2015-10-28 0:51 ` [PATCH 19/19] powerpc: clean up asm/switch_to.h Anton Blanchard
2015-10-28 3:19 ` kbuild test robot
2015-10-28 4:24 ` kbuild test robot
-- strict thread matches above, loose matches on Subject: below --
2015-10-29 0:43 [PATCH 00/19] Context switch improvements Anton Blanchard
2015-10-29 0:44 ` [PATCH 16/19] powerpc: create giveup_all() Anton Blanchard
2015-12-02 8:24 ` Michael Ellerman
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=1445993467-667-16-git-send-email-anton@samba.org \
--to=anton@samba.org \
--cc=benh@kernel.crashing.org \
--cc=cyrilbur@gmail.com \
--cc=linuxppc-dev@lists.ozlabs.org \
--cc=mikey@neuling.org \
--cc=mpe@ellerman.id.au \
--cc=paulus@samba.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).