[v5, 1/6] riscv: Add support for kernel mode vector
Andy Chiu
andy.chiu at sifive.com
Mon Dec 18 22:04:55 PST 2023
On Sat, Dec 16, 2023 at 2:41 AM Charlie Jenkins <charlie at rivosinc.com> wrote:
>
> On Sat, Dec 16, 2023 at 12:01:53AM +0800, Andy Chiu wrote:
> > On Fri, Dec 15, 2023 at 2:24 PM Charlie Jenkins <charlie at rivosinc.com> wrote:
> > >
> > > On Thu, Dec 14, 2023 at 03:57:16PM +0000, Andy Chiu wrote:
> > > > From: Greentime Hu <greentime.hu at sifive.com>
> > > >
> > > > Add kernel_vector_begin() and kernel_vector_end() function declarations
> > > > and corresponding definitions in kernel_mode_vector.c
> > > >
> > > > These are needed to wrap uses of vector in kernel mode.
> > > >
> > > > Co-developed-by: Vincent Chen <vincent.chen at sifive.com>
> > > > Signed-off-by: Vincent Chen <vincent.chen at sifive.com>
> > > > Signed-off-by: Greentime Hu <greentime.hu at sifive.com>
> > > > Signed-off-by: Andy Chiu <andy.chiu at sifive.com>
> > > > ---
> > > > Changelog v4:
> > > > - Use kernel_v_flags and helpers to track vector context.
> > > > Changelog v3:
> > > > - Reorder patch 1 to patch 3 to make use of
> > > > {get,put}_cpu_vector_context later.
> > > > - Export {get,put}_cpu_vector_context.
> > > > - Save V context after disabling preemption. (Guo)
> > > > - Fix a build fail. (Conor)
> > > > - Remove irqs_disabled() check as it is not needed, fix styling. (Björn)
> > > > Changelog v2:
> > > > - 's/kernel_rvv/kernel_vector' and return void in kernel_vector_begin
> > > > (Conor)
> > > > - export may_use_simd to include/asm/simd.h
> > > > ---
> > > > arch/riscv/include/asm/processor.h | 15 +++-
> > > > arch/riscv/include/asm/simd.h | 42 ++++++++++++
> > > > arch/riscv/include/asm/vector.h | 21 ++++++
> > > > arch/riscv/kernel/Makefile | 1 +
> > > > arch/riscv/kernel/kernel_mode_vector.c | 95 ++++++++++++++++++++++++++
> > > > arch/riscv/kernel/process.c | 2 +-
> > > > 6 files changed, 174 insertions(+), 2 deletions(-)
> > > > create mode 100644 arch/riscv/include/asm/simd.h
> > > > create mode 100644 arch/riscv/kernel/kernel_mode_vector.c
> > > >
> > > > diff --git a/arch/riscv/include/asm/processor.h b/arch/riscv/include/asm/processor.h
> > > > index f19f861cda54..a47763c262e1 100644
> > > > --- a/arch/riscv/include/asm/processor.h
> > > > +++ b/arch/riscv/include/asm/processor.h
> > > > @@ -73,6 +73,18 @@
> > > > struct task_struct;
> > > > struct pt_regs;
> > > >
> > > > +/*
> > > > + * We use a flag to track in-kernel Vector context. Currently the flag has the
> > > > + * following meaning:
> > > > + *
> > > > + * - bit 0 indicates whether the in-kernel Vector context is active. The
> > > > + * activation of this state disables the preemption.
> > > > + */
> > > > +
> > > > +#define RISCV_KERNEL_MODE_V_MASK 0x1
> > > > +
> > > > +#define RISCV_KERNEL_MODE_V 0x1
> > > > +
> > > > /* CPU-specific state of a task */
> > > > struct thread_struct {
> > > > /* Callee-saved registers */
> > > > @@ -81,7 +93,8 @@ struct thread_struct {
> > > > unsigned long s[12]; /* s[0]: frame pointer */
> > > > struct __riscv_d_ext_state fstate;
> > > > unsigned long bad_cause;
> > > > - unsigned long vstate_ctrl;
> > > > + u32 riscv_v_flags;
> > > > + u32 vstate_ctrl;
> > > > struct __riscv_v_ext_state vstate;
> > > > unsigned long align_ctl;
> > > > };
> > > > diff --git a/arch/riscv/include/asm/simd.h b/arch/riscv/include/asm/simd.h
> > > > new file mode 100644
> > > > index 000000000000..269752bfa2cc
> > > > --- /dev/null
> > > > +++ b/arch/riscv/include/asm/simd.h
> > > > @@ -0,0 +1,42 @@
> > > > +/* SPDX-License-Identifier: GPL-2.0-only */
> > > > +/*
> > > > + * Copyright (C) 2017 Linaro Ltd. <ard.biesheuvel at linaro.org>
> > > > + * Copyright (C) 2023 SiFive
> > > > + */
> > > > +
> > > > +#ifndef __ASM_SIMD_H
> > > > +#define __ASM_SIMD_H
> > > > +
> > > > +#include <linux/compiler.h>
> > > > +#include <linux/irqflags.h>
> > > > +#include <linux/percpu.h>
> > > > +#include <linux/preempt.h>
> > > > +#include <linux/types.h>
> > > > +
> > > > +#ifdef CONFIG_RISCV_ISA_V
> > > > +/*
> > > > + * may_use_simd - whether it is allowable at this time to issue vector
> > > > + * instructions or access the vector register file
> > > > + *
> > > > + * Callers must not assume that the result remains true beyond the next
> > > > + * preempt_enable() or return from softirq context.
> > > > + */
> > > > +static __must_check inline bool may_use_simd(void)
> > > > +{
> > > > + /*
> > > > + * RISCV_KERNEL_MODE_V is only set while preemption is disabled,
> > > > + * and is clear whenever preemption is enabled.
> > > > + */
> > > > + return !in_hardirq() && !in_nmi() && !(riscv_v_ctx_cnt() & RISCV_KERNEL_MODE_V_MASK);
> > > > +}
> > > > +
> > > > +#else /* ! CONFIG_RISCV_ISA_V */
> > > > +
> > > > +static __must_check inline bool may_use_simd(void)
> > > > +{
> > > > + return false;
> > > > +}
> > > > +
> > > > +#endif /* ! CONFIG_RISCV_ISA_V */
> > > > +
> > > > +#endif
> > > > diff --git a/arch/riscv/include/asm/vector.h b/arch/riscv/include/asm/vector.h
> > > > index 87aaef656257..6254830c0668 100644
> > > > --- a/arch/riscv/include/asm/vector.h
> > > > +++ b/arch/riscv/include/asm/vector.h
> > > > @@ -22,6 +22,27 @@
> > > > extern unsigned long riscv_v_vsize;
> > > > int riscv_v_setup_vsize(void);
> > > > bool riscv_v_first_use_handler(struct pt_regs *regs);
> > > > +void kernel_vector_begin(void);
> > > > +void kernel_vector_end(void);
> > > > +void get_cpu_vector_context(void);
> > > > +void put_cpu_vector_context(void);
> > > > +
> > > > +static inline void riscv_v_ctx_cnt_add(u32 offset)
> > > > +{
> > > > + current->thread.riscv_v_flags += offset;
> > > > + barrier();
> > > > +}
> > > > +
> > > > +static inline void riscv_v_ctx_cnt_sub(u32 offset)
> > > > +{
> > > > + barrier();
> > > > + current->thread.riscv_v_flags -= offset;
> > > > +}
> > > > +
> > > > +static inline u32 riscv_v_ctx_cnt(void)
> > > > +{
> > > > + return READ_ONCE(current->thread.riscv_v_flags);
> > > > +}
> > > >
> > > > static __always_inline bool has_vector(void)
> > > > {
> > > > diff --git a/arch/riscv/kernel/Makefile b/arch/riscv/kernel/Makefile
> > > > index fee22a3d1b53..8c58595696b3 100644
> > > > --- a/arch/riscv/kernel/Makefile
> > > > +++ b/arch/riscv/kernel/Makefile
> > > > @@ -63,6 +63,7 @@ obj-$(CONFIG_MMU) += vdso.o vdso/
> > > > obj-$(CONFIG_RISCV_MISALIGNED) += traps_misaligned.o
> > > > obj-$(CONFIG_FPU) += fpu.o
> > > > obj-$(CONFIG_RISCV_ISA_V) += vector.o
> > > > +obj-$(CONFIG_RISCV_ISA_V) += kernel_mode_vector.o
> > > > obj-$(CONFIG_SMP) += smpboot.o
> > > > obj-$(CONFIG_SMP) += smp.o
> > > > obj-$(CONFIG_SMP) += cpu_ops.o
> > > > diff --git a/arch/riscv/kernel/kernel_mode_vector.c b/arch/riscv/kernel/kernel_mode_vector.c
> > > > new file mode 100644
> > > > index 000000000000..c9ccf21dd16c
> > > > --- /dev/null
> > > > +++ b/arch/riscv/kernel/kernel_mode_vector.c
> > > > @@ -0,0 +1,95 @@
> > > > +// SPDX-License-Identifier: GPL-2.0-or-later
> > > > +/*
> > > > + * Copyright (C) 2012 ARM Ltd.
> > > > + * Author: Catalin Marinas <catalin.marinas at arm.com>
> > > > + * Copyright (C) 2017 Linaro Ltd. <ard.biesheuvel at linaro.org>
> > > > + * Copyright (C) 2021 SiFive
> > > > + */
> > > > +#include <linux/compiler.h>
> > > > +#include <linux/irqflags.h>
> > > > +#include <linux/percpu.h>
> > > > +#include <linux/preempt.h>
> > > > +#include <linux/types.h>
> > > > +
> > > > +#include <asm/vector.h>
> > > > +#include <asm/switch_to.h>
> > > > +#include <asm/simd.h>
> > > > +
> > > > +/*
> > > > + * Claim ownership of the CPU vector context for use by the calling context.
> > > > + *
> > > > + * The caller may freely manipulate the vector context metadata until
> > > > + * put_cpu_vector_context() is called.
> > > > + */
> > > > +void get_cpu_vector_context(void)
> > > > +{
> > > > + preempt_disable();
> > > > +
> > > > + WARN_ON(riscv_v_ctx_cnt() & RISCV_KERNEL_MODE_V_MASK);
> > >
> > > This is a bigger issue than a warn. Calling riscv_v_ctx_cnt_add with
> > > the same flag an even number of times will cause (riscv_v_ctx_cnt() &
> > > RISCV_KERNEL_MODE_V_MASK) to return 0, even though vector is being used.
> > > This could be solved by using a bitwise or instead of addition when
> > > setting the flag.
> >
> > Yes, we should use bitwise to operate it. At the same time, I am
> > thinking if we should allow calling kernel_vector_begin() multiple
> > times on a call chain. Or provide a fast check (in_kernel_vector()) in
> > a vectorized function to reduce the cost of calling multiple
> > kernel_vector_begin(). For example,
> >
> > kernel_vector_begin()
> > memcpy()<- allow calling kernel_vector_begin/end() again
> > - or prevent calling it again with in_kernel_vector()
> > do_something_with_vector()
> > kernel_vector_end()
> >
>
> Yes that seems like that could be useful.
>
> > >
> > > > + riscv_v_ctx_cnt_add(RISCV_KERNEL_MODE_V);
> > > > +}
> > > > +
> > > > +/*
> > > > + * Release the CPU vector context.
> > > > + *
> > > > + * Must be called from a context in which get_cpu_vector_context() was
> > > > + * previously called, with no call to put_cpu_vector_context() in the
> > > > + * meantime.
> > > > + */
> > > > +void put_cpu_vector_context(void)
> > > > +{
> > > > + WARN_ON(!(riscv_v_ctx_cnt() & RISCV_KERNEL_MODE_V_MASK));
> > > > + riscv_v_ctx_cnt_sub(RISCV_KERNEL_MODE_V);
> > > > +
> > > > + preempt_enable();
> > > > +}
> > > > +
> > > > +/*
> > > > + * kernel_vector_begin(): obtain the CPU vector registers for use by the calling
> > > > + * context
> > > > + *
> > > > + * Must not be called unless may_use_simd() returns true.
> > > > + * Task context in the vector registers is saved back to memory as necessary.
> > > > + *
> > > > + * A matching call to kernel_vector_end() must be made before returning from the
> > > > + * calling context.
> > > > + *
> > > > + * The caller may freely use the vector registers until kernel_vector_end() is
> > > > + * called.
> > > > + */
> > > > +void kernel_vector_begin(void)
> > > > +{
> > > > + if (WARN_ON(!has_vector()))
> > >
> > > Should this be WARN_ONCE? If somebody runs a kernel compiled with vector
> > > on hardware without vector, this warning has the potential to be thrown
> > > an excessive amount of times.
> >
> > Callers of this function should check with may_use_simd() and only
> > proceed to call this function if it returns true.
> >
>
> Yes it is a bug if they don't call may_use_simd() first, but I was more
> concerned about the number of logs that are generated with WARN_ON.
> A single log seems like it would be sufficient.
In fact this will just panic the kernel if a user proceeds
kernel_vector_begin() regardless of may_use_simd(). To demonstrate,
has_vector() returns false under these conditions:
1. The kernel is compiled without CONFIG_RISCV_ISA_V.
2. The kernel has CONFIG_RISCV_ISA_V but the hardware doesn't support V
3. The kernel has CONFIG_RISCV_ISA_V. The hardware supports V, but in
a way where it is not recognized by the kernel, e.g uneven VLEN.
For case 1, kernel_vector_begin() won't even exist so the user will
experience a compile failure.
For case 2 and 3, the kernel will panic, hitting an illegal
instruction error when it starts executing V because VS will be off by
then. Unless the user goes very off and manually enables V by calling
riscv_v_enable() on case 3.
Please tell me if you still have any concerns about this.
>
> - Charlie
>
> > >
> > > > + return;
> > > > +
> > > > + BUG_ON(!may_use_simd());
> > > > +
> > > > + get_cpu_vector_context();
> > > > +
> > > > + riscv_v_vstate_save(current, task_pt_regs(current));
> > > > +
> > > > + riscv_v_enable();
> > > > +}
> > > > +EXPORT_SYMBOL_GPL(kernel_vector_begin);
> > > > +
> > > > +/*
> > > > + * kernel_vector_end(): give the CPU vector registers back to the current task
> > > > + *
> > > > + * Must be called from a context in which kernel_vector_begin() was previously
> > > > + * called, with no call to kernel_vector_end() in the meantime.
> > > > + *
> > > > + * The caller must not use the vector registers after this function is called,
> > > > + * unless kernel_vector_begin() is called again in the meantime.
> > > > + */
> > > > +void kernel_vector_end(void)
> > > > +{
> > > > + if (WARN_ON(!has_vector()))
> > >
> > > Same as above.
> > >
> > > - Charlie
> > >
> > > >+ return;
> > > > +
> > > > + riscv_v_vstate_restore(current, task_pt_regs(current));
> > > > +
> > > > + riscv_v_disable();
> > > > +
> > > > + put_cpu_vector_context();
> > > > +}
> > > > +EXPORT_SYMBOL_GPL(kernel_vector_end);
> > > > diff --git a/arch/riscv/kernel/process.c b/arch/riscv/kernel/process.c
> > > > index 4f21d970a129..5c4dcf518684 100644
> > > > --- a/arch/riscv/kernel/process.c
> > > > +++ b/arch/riscv/kernel/process.c
> > > > @@ -187,7 +187,6 @@ int arch_dup_task_struct(struct task_struct *dst, struct task_struct *src)
> > > > *dst = *src;
> > > > /* clear entire V context, including datap for a new task */
> > > > memset(&dst->thread.vstate, 0, sizeof(struct __riscv_v_ext_state));
> > > > -
> > > > return 0;
> > > > }
> > > >
> > > > @@ -221,6 +220,7 @@ int copy_thread(struct task_struct *p, const struct kernel_clone_args *args)
> > > > childregs->a0 = 0; /* Return value of fork() */
> > > > p->thread.s[0] = 0;
> > > > }
> > > > + p->thread.riscv_v_flags = 0;
> > > > p->thread.ra = (unsigned long)ret_from_fork;
> > > > p->thread.sp = (unsigned long)childregs; /* kernel sp */
> > > > return 0;
> > > > --
> > > > 2.17.1
> > > >
> >
> > Regards,
> > Andy
More information about the linux-riscv
mailing list