diff mbox series

[RFC,v2,1/2] ppc/fpu: Add generic FPU api similar to x86

Message ID 20210721044801.840501-2-Anson.Jacob@amd.com (mailing list archive)
State RFC
Headers show
Series PPC: Add generic FPU api similar to x86 | expand
Related show

Commit Message

Anson Jacob July 21, 2021, 4:48 a.m. UTC 
- Add kernel_fpu_begin & kernel_fpu_end API as x86
- Add logic similar to x86 to ensure fpu
  begin/end call correctness
- Add kernel_fpu_enabled to know if FPU is enabled

v2:
- Added asm/fpu/api.h powerpc variant with kernel_fpu_begin/end()
  and kernel_fpu_enabled() declarations
- Updated kernel_fpu_enabled as EXPORT_SYMBOL_GPL

Signed-off-by: Anson Jacob <Anson.Jacob@amd.com>
CC: Christoph Hellwig <hch@infradead.org>
CC: Rodrigo Siqueira <Rodrigo.Siqueira@amd.com>
CC: Harry Wentland <harry.wentland@amd.com>
CC: Christian König <christian.koenig@amd.com>
---
 arch/powerpc/include/asm/fpu/api.h   |  18 ++++
 arch/powerpc/include/asm/switch_to.h |  25 +-----
 arch/powerpc/kernel/process.c        | 130 +++++++++++++++++++++++++++
 3 files changed, 151 insertions(+), 22 deletions(-)
 create mode 100644 arch/powerpc/include/asm/fpu/api.h

Comments

Christoph Hellwig July 21, 2021, 6:58 a.m. UTC | #1 
> +
> +/*
> + * Use kernel_fpu_begin/end() if you intend to use FPU in kernel context. It
> + * disables preemption so be careful if you intend to use it for long periods
> + * of time.
> + * TODO: If you intend to use the FPU in irq/softirq you need to check first with
> + * irq_fpu_usable() if it is possible.

Please avoid the overly lone lines comments.

> +extern bool kernel_fpu_enabled(void);
> +extern void kernel_fpu_begin(void);
> +extern void kernel_fpu_end(void);

No need for the externs.

> +/*
> + * Track whether the kernel is using the FPU state
> + * currently.

This all fits on a single line.

> +static bool fpu_support(void)
> +{
> +	if (cpu_has_feature(CPU_FTR_VSX_COMP)) {
> +		return true;
> +	} else if (cpu_has_feature(CPU_FTR_ALTIVEC_COMP)) {
> +		return true;
> +	} else if (!cpu_has_feature(CPU_FTR_FPU_UNAVAILABLE)) {
> +		return true;
> +	}

No need for the braces, or else after a return.  In fact this could
be simplified down to:

	return cpu_has_feature(CPU_FTR_VSX_COMP) ||
		cpu_has_feature(CPU_FTR_ALTIVEC_COMP) ||
		cpu_has_feature(CPU_FTR_FPU_UNAVAILABLE));

> +	preempt_disable();
> +
> +#ifdef CONFIG_VSX
> +	if (cpu_has_feature(CPU_FTR_VSX_COMP)) {
> +		enable_kernel_vsx();
> +		return;
> +	}
> +#endif
> +
> +#ifdef CONFIG_ALTIVEC
> +	if (cpu_has_feature(CPU_FTR_ALTIVEC_COMP)) {
> +		enable_kernel_altivec();
> +		return;
> +	}
> +#endif
> +
> +#ifdef CONFIG_PPC_FPU
> +	if (!cpu_has_feature(CPU_FTR_FPU_UNAVAILABLE)) {
> +		enable_kernel_fp();
> +		return;
> +	}
> +#endif

All the features are defined away if not supported (and we already rely
on that in fpu_support()).  So this could become:

	if (cpu_has_feature(CPU_FTR_VSX_COMP))
		enable_kernel_vsx();
	else if (cpu_has_feature(CPU_FTR_ALTIVEC_COMP))
		enable_kernel_altivec();
	else if (!cpu_has_feature(CPU_FTR_FPU_UNAVAILABLE))
		enable_kernel_fp();

Same for the disable path.
diff mbox series

Patch

diff --git a/arch/powerpc/include/asm/fpu/api.h b/arch/powerpc/include/asm/fpu/api.h
new file mode 100644
index 000000000000..57308cdc65c9
--- /dev/null
+++ b/arch/powerpc/include/asm/fpu/api.h
@@ -0,0 +1,18 @@ 
+/* SPDX-License-Identifier: GPL-2.0 */
+
+#ifndef _ASM_POWERPC_FPU_API_H
+#define _ASM_POWERPC_FPU_API_H
+
+/*
+ * Use kernel_fpu_begin/end() if you intend to use FPU in kernel context. It
+ * disables preemption so be careful if you intend to use it for long periods
+ * of time.
+ * TODO: If you intend to use the FPU in irq/softirq you need to check first with
+ * irq_fpu_usable() if it is possible.
+ */
+
+extern bool kernel_fpu_enabled(void);
+extern void kernel_fpu_begin(void);
+extern void kernel_fpu_end(void);
+
+#endif /* _ASM_POWERPC_FPU_API_H */
diff --git a/arch/powerpc/include/asm/switch_to.h b/arch/powerpc/include/asm/switch_to.h
index 9d1fbd8be1c7..a9a919279f48 100644
--- a/arch/powerpc/include/asm/switch_to.h
+++ b/arch/powerpc/include/asm/switch_to.h
@@ -41,10 +41,7 @@  extern void enable_kernel_fp(void);
 extern void flush_fp_to_thread(struct task_struct *);
 extern void giveup_fpu(struct task_struct *);
 extern void save_fpu(struct task_struct *);
-static inline void disable_kernel_fp(void)
-{
-	msr_check_and_clear(MSR_FP);
-}
+extern void disable_kernel_fp(void);
 #else
 static inline void save_fpu(struct task_struct *t) { }
 static inline void flush_fp_to_thread(struct task_struct *t) { }
@@ -55,10 +52,7 @@  extern void enable_kernel_altivec(void);
 extern void flush_altivec_to_thread(struct task_struct *);
 extern void giveup_altivec(struct task_struct *);
 extern void save_altivec(struct task_struct *);
-static inline void disable_kernel_altivec(void)
-{
-	msr_check_and_clear(MSR_VEC);
-}
+extern void disable_kernel_altivec(void);
 #else
 static inline void save_altivec(struct task_struct *t) { }
 static inline void __giveup_altivec(struct task_struct *t) { }
@@ -67,20 +61,7 @@  static inline void __giveup_altivec(struct task_struct *t) { }
 #ifdef CONFIG_VSX
 extern void enable_kernel_vsx(void);
 extern void flush_vsx_to_thread(struct task_struct *);
-static inline void disable_kernel_vsx(void)
-{
-	msr_check_and_clear(MSR_FP|MSR_VEC|MSR_VSX);
-}
-#else
-static inline void enable_kernel_vsx(void)
-{
-	BUILD_BUG();
-}
-
-static inline void disable_kernel_vsx(void)
-{
-	BUILD_BUG();
-}
+extern void disable_kernel_vsx(void);
 #endif
 
 #ifdef CONFIG_SPE
diff --git a/arch/powerpc/kernel/process.c b/arch/powerpc/kernel/process.c
index 185beb290580..969096c0123c 100644
--- a/arch/powerpc/kernel/process.c
+++ b/arch/powerpc/kernel/process.c
@@ -75,6 +75,17 @@ 
 #define TM_DEBUG(x...) do { } while(0)
 #endif
 
+/*
+ * Track whether the kernel is using the FPU state
+ * currently.
+ *
+ * This flag is used:
+ *
+ *   - kernel_fpu_begin()/end() correctness
+ *   - kernel_fpu_enabled info
+ */
+static DEFINE_PER_CPU(bool, in_kernel_fpu);
+
 extern unsigned long _get_SP(void);
 
 #ifdef CONFIG_PPC_TRANSACTIONAL_MEM
@@ -212,6 +223,9 @@  void enable_kernel_fp(void)
 	unsigned long cpumsr;
 
 	WARN_ON(preemptible());
+	WARN_ON_ONCE(this_cpu_read(in_kernel_fpu));
+
+	this_cpu_write(in_kernel_fpu, true);
 
 	cpumsr = msr_check_and_set(MSR_FP);
 
@@ -231,6 +245,15 @@  void enable_kernel_fp(void)
 	}
 }
 EXPORT_SYMBOL(enable_kernel_fp);
+
+void disable_kernel_fp(void)
+{
+	WARN_ON_ONCE(!this_cpu_read(in_kernel_fpu));
+
+	this_cpu_write(in_kernel_fpu, false);
+	msr_check_and_clear(MSR_FP);
+}
+EXPORT_SYMBOL(disable_kernel_fp);
 #else
 static inline void __giveup_fpu(struct task_struct *tsk) { }
 #endif /* CONFIG_PPC_FPU */
@@ -263,6 +286,9 @@  void enable_kernel_altivec(void)
 	unsigned long cpumsr;
 
 	WARN_ON(preemptible());
+	WARN_ON_ONCE(this_cpu_read(in_kernel_fpu));
+
+	this_cpu_write(in_kernel_fpu, true);
 
 	cpumsr = msr_check_and_set(MSR_VEC);
 
@@ -283,6 +309,14 @@  void enable_kernel_altivec(void)
 }
 EXPORT_SYMBOL(enable_kernel_altivec);
 
+void disable_kernel_altivec(void)
+{
+	WARN_ON_ONCE(!this_cpu_read(in_kernel_fpu));
+
+	this_cpu_write(in_kernel_fpu, false);
+	msr_check_and_clear(MSR_VEC);
+}
+EXPORT_SYMBOL(disable_kernel_altivec);
 /*
  * Make sure the VMX/Altivec register state in the
  * the thread_struct is up to date for task tsk.
@@ -333,6 +367,9 @@  void enable_kernel_vsx(void)
 	unsigned long cpumsr;
 
 	WARN_ON(preemptible());
+	WARN_ON_ONCE(this_cpu_read(in_kernel_fpu));
+
+	this_cpu_write(in_kernel_fpu, true);
 
 	cpumsr = msr_check_and_set(MSR_FP|MSR_VEC|MSR_VSX);
 
@@ -354,6 +391,15 @@  void enable_kernel_vsx(void)
 }
 EXPORT_SYMBOL(enable_kernel_vsx);
 
+void disable_kernel_vsx(void)
+{
+	WARN_ON_ONCE(!this_cpu_read(in_kernel_fpu));
+
+	this_cpu_write(in_kernel_fpu, false);
+	msr_check_and_clear(MSR_FP|MSR_VEC|MSR_VSX);
+}
+EXPORT_SYMBOL(disable_kernel_vsx);
+
 void flush_vsx_to_thread(struct task_struct *tsk)
 {
 	if (tsk->thread.regs) {
@@ -406,6 +452,90 @@  void flush_spe_to_thread(struct task_struct *tsk)
 }
 #endif /* CONFIG_SPE */
 
+static bool fpu_support(void)
+{
+	if (cpu_has_feature(CPU_FTR_VSX_COMP)) {
+		return true;
+	} else if (cpu_has_feature(CPU_FTR_ALTIVEC_COMP)) {
+		return true;
+	} else if (!cpu_has_feature(CPU_FTR_FPU_UNAVAILABLE)) {
+		return true;
+	}
+
+	return false;
+}
+
+bool kernel_fpu_enabled(void)
+{
+	return this_cpu_read(in_kernel_fpu);
+}
+EXPORT_SYMBOL_GPL(kernel_fpu_enabled);
+
+void kernel_fpu_begin(void)
+{
+	if (!fpu_support()) {
+		WARN_ON_ONCE(1);
+		return;
+	}
+
+	preempt_disable();
+
+#ifdef CONFIG_VSX
+	if (cpu_has_feature(CPU_FTR_VSX_COMP)) {
+		enable_kernel_vsx();
+		return;
+	}
+#endif
+
+#ifdef CONFIG_ALTIVEC
+	if (cpu_has_feature(CPU_FTR_ALTIVEC_COMP)) {
+		enable_kernel_altivec();
+		return;
+	}
+#endif
+
+#ifdef CONFIG_PPC_FPU
+	if (!cpu_has_feature(CPU_FTR_FPU_UNAVAILABLE)) {
+		enable_kernel_fp();
+		return;
+	}
+#endif
+}
+EXPORT_SYMBOL_GPL(kernel_fpu_begin);
+
+void kernel_fpu_end(void)
+{
+	if (!fpu_support()) {
+		WARN_ON_ONCE(1);
+		return;
+	}
+
+#ifdef CONFIG_VSX
+	if (cpu_has_feature(CPU_FTR_VSX_COMP)) {
+		disable_kernel_vsx();
+		goto done;
+	}
+#endif
+
+#ifdef CONFIG_ALTIVEC
+	if (cpu_has_feature(CPU_FTR_ALTIVEC_COMP)) {
+		disable_kernel_altivec();
+		goto done;
+	}
+#endif
+
+#ifdef CONFIG_PPC_FPU
+	if (!cpu_has_feature(CPU_FTR_FPU_UNAVAILABLE)) {
+		disable_kernel_fp();
+		goto done;
+	}
+#endif
+
+done:
+	preempt_enable();
+}
+EXPORT_SYMBOL_GPL(kernel_fpu_end);
+
 static unsigned long msr_all_available;
 
 static int __init init_msr_all_available(void)