Re: [RFC PATCH 13/22] x86/fpu/xstate: Expand dynamic user state area on first use

From: Andy Lutomirski
Date: Thu Oct 01 2020 - 19:39:26 EST


On Thu, Oct 1, 2020 at 1:43 PM Chang S. Bae <chang.seok.bae@xxxxxxxxx> wrote:
>
> Intel's Extended Feature Disable (XFD) feature is an extension of the XSAVE
> architecture. XFD allows the kernel to enable a feature state in XCR0 and
> to receive a #NM trap when a task uses instructions accessing that state.
> In this way, Linux can allocate the large task->fpu buffer only for tasks
> that use it.
>
> XFD introduces two MSRs: IA32_XFD to enable/disable the feature and
> IA32_XFD_ERR to assist the #NM trap handler. Both use the same
> state-component bitmap format, used by XCR0.
>
> Use this hardware capability to find the right time to expand xstate area.
> Introduce two sets of helper functions for that:
>
> 1. The first set is primarily for interacting with the XFD hardware
> feature. Helpers for configuring disablement, e.g. in context switching,
> are:
> xdisable_setbits()
> xdisable_getbits()
> xdisable_switch()
>
> 2. The second set is for managing the first-use status and handling #NM
> trap:
> xfirstuse_enabled()
> xfirstuse_not_detected()
> xfirstuse_event_handler()
>
> The #NM handler induces the xstate area expansion to save the first-used
> states.
>
> No functional change until the kernel enables dynamic user states and XFD.
>
> Signed-off-by: Chang S. Bae <chang.seok.bae@xxxxxxxxx>
> Reviewed-by: Len Brown <len.brown@xxxxxxxxx>
> Cc: x86@xxxxxxxxxx
> Cc: linux-kernel@xxxxxxxxxxxxxxx
> ---
> arch/x86/include/asm/cpufeatures.h | 1 +
> arch/x86/include/asm/fpu/internal.h | 53 ++++++++++++++++++++++++++++-
> arch/x86/include/asm/msr-index.h | 2 ++
> arch/x86/kernel/fpu/core.c | 37 ++++++++++++++++++++
> arch/x86/kernel/fpu/xstate.c | 34 ++++++++++++++++--
> arch/x86/kernel/process.c | 5 +++
> arch/x86/kernel/process_32.c | 2 +-
> arch/x86/kernel/process_64.c | 2 +-
> arch/x86/kernel/traps.c | 3 ++
> 9 files changed, 133 insertions(+), 6 deletions(-)
>
> diff --git a/arch/x86/include/asm/cpufeatures.h b/arch/x86/include/asm/cpufeatures.h
> index 2901d5df4366..7d7fe1d82966 100644
> --- a/arch/x86/include/asm/cpufeatures.h
> +++ b/arch/x86/include/asm/cpufeatures.h
> @@ -274,6 +274,7 @@
> #define X86_FEATURE_XSAVEC (10*32+ 1) /* XSAVEC instruction */
> #define X86_FEATURE_XGETBV1 (10*32+ 2) /* XGETBV with ECX = 1 instruction */
> #define X86_FEATURE_XSAVES (10*32+ 3) /* XSAVES/XRSTORS instructions */
> +#define X86_FEATURE_XFD (10*32+ 4) /* eXtended Feature Disabling */
>
> /*
> * Extended auxiliary flags: Linux defined - for features scattered in various
> diff --git a/arch/x86/include/asm/fpu/internal.h b/arch/x86/include/asm/fpu/internal.h
> index 3b03ead87a46..f5dbbaa060fb 100644
> --- a/arch/x86/include/asm/fpu/internal.h
> +++ b/arch/x86/include/asm/fpu/internal.h
> @@ -572,11 +572,60 @@ static inline void switch_fpu_prepare(struct fpu *old_fpu, int cpu)
> * Misc helper functions:
> */
>
> +/* The first-use detection helpers: */
> +
> +static inline void xdisable_setbits(u64 value)
> +{
> + wrmsrl_safe(MSR_IA32_XFD, value);
> +}
> +
> +static inline u64 xdisable_getbits(void)
> +{
> + u64 value;
> +
> + rdmsrl_safe(MSR_IA32_XFD, &value);
> + return value;
> +}
> +
> +static inline u64 xfirstuse_enabled(void)
> +{
> + /* All the dynamic user components are first-use enabled. */
> + return xfeatures_mask_user_dynamic;
> +}
> +
> +/*
> + * Convert fpu->firstuse_bv to xdisable configuration in MSR IA32_XFD.
> + * xdisable_setbits() only uses this.
> + */
> +static inline u64 xfirstuse_not_detected(struct fpu *fpu)
> +{
> + u64 firstuse_bv = (fpu->state_mask & xfirstuse_enabled());
> +
> + /*
> + * If first-use is not detected, set the bit. If the detection is
> + * not enabled, the bit is always zero in firstuse_bv. So, make
> + * following conversion:
> + */
> + return (xfirstuse_enabled() ^ firstuse_bv);
> +}
> +
> +/* Update MSR IA32_XFD based on fpu->firstuse_bv */
> +static inline void xdisable_switch(struct fpu *prev, struct fpu *next)
> +{
> + if (!static_cpu_has(X86_FEATURE_XFD) || !xfirstuse_enabled())
> + return;
> +
> + if (unlikely(prev->state_mask != next->state_mask))
> + xdisable_setbits(xfirstuse_not_detected(next));
> +}
> +
> +bool xfirstuse_event_handler(struct fpu *fpu);
> +
> /*
> * Load PKRU from the FPU context if available. Delay loading of the
> * complete FPU state until the return to userland.
> */
> -static inline void switch_fpu_finish(struct fpu *new_fpu)
> +static inline void switch_fpu_finish(struct fpu *old_fpu, struct fpu *new_fpu)
> {
> u32 pkru_val = init_pkru_value;
> struct pkru_state *pk;
> @@ -586,6 +635,8 @@ static inline void switch_fpu_finish(struct fpu *new_fpu)
>
> set_thread_flag(TIF_NEED_FPU_LOAD);
>
> + xdisable_switch(old_fpu, new_fpu);
> +
> if (!cpu_feature_enabled(X86_FEATURE_OSPKE))
> return;
>
> diff --git a/arch/x86/include/asm/msr-index.h b/arch/x86/include/asm/msr-index.h
> index 2859ee4f39a8..0ccbe8cc99ad 100644
> --- a/arch/x86/include/asm/msr-index.h
> +++ b/arch/x86/include/asm/msr-index.h
> @@ -610,6 +610,8 @@
> #define MSR_IA32_BNDCFGS_RSVD 0x00000ffc
>
> #define MSR_IA32_XSS 0x00000da0
> +#define MSR_IA32_XFD 0x000001c4
> +#define MSR_IA32_XFD_ERR 0x000001c5
>
> #define MSR_IA32_APICBASE 0x0000001b
> #define MSR_IA32_APICBASE_BSP (1<<8)
> diff --git a/arch/x86/kernel/fpu/core.c b/arch/x86/kernel/fpu/core.c
> index ece6428ba85b..2e07bfcd54b3 100644
> --- a/arch/x86/kernel/fpu/core.c
> +++ b/arch/x86/kernel/fpu/core.c
> @@ -518,3 +518,40 @@ int fpu__exception_code(struct fpu *fpu, int trap_nr)
> */
> return 0;
> }
> +
> +bool xfirstuse_event_handler(struct fpu *fpu)
> +{
> + bool handled = false;
> + u64 event_mask;
> +
> + /* Check whether the first-use detection is running. */
> + if (!static_cpu_has(X86_FEATURE_XFD) || !xfirstuse_enabled())
> + return handled;
> +
> + rdmsrl_safe(MSR_IA32_XFD_ERR, &event_mask);

NAK.

MSR_IA32_XFD_ERR needs to be wired up in the exception handler, not in
some helper called farther down the stack

But this raises an interesting point -- what happens if allocation
fails? I think that, from kernel code, we simply cannot support this
exception mechanism. If kernel code wants to use AMX (and that would
be very strange indeed), it should call x86_i_am_crazy_amx_begin() and
handle errors, not rely on exceptions. From user code, I assume we
send a signal if allocation fails.