| Message ID | 20230921153204.278427-1-anton.ivanov@cambridgegreys.com |
|---|---|
| State | Superseded |
| Headers | show |
| Series | [v3] um: Enable preemption in UML | expand |
On 21/09/2023 16:32, anton.ivanov@cambridgegreys.com wrote: > From: Anton Ivanov <anton.ivanov@cambridgegreys.com> > > Preemption requires saving/restoring FPU state. This patch > adds support for it using GCC intrinsics. > > Signed-off-by: Anton Ivanov <anton.ivanov@cambridgegreys.com> > --- > arch/um/Kconfig | 1 - > arch/um/include/asm/fpu/api.h | 6 +- > arch/um/include/asm/processor-generic.h | 3 + > arch/um/kernel/Makefile | 4 ++ > arch/um/kernel/fpu.c | 77 +++++++++++++++++++++++++ > arch/um/kernel/irq.c | 2 + > 6 files changed, 91 insertions(+), 2 deletions(-) > create mode 100644 arch/um/kernel/fpu.c Fixes: 1. Defines. I messed them up in v2. 2. Makes build selective on CONFIG_PREEMPT. If preemption is not enabled fpu.o is not build. 3. CFLAGS are altered only for fpu.o Performance change is the same - from 20minutes for find /usr and cat to /dev/null down to 3 and from unusable under stress to fairly reasonable. > > diff --git a/arch/um/Kconfig b/arch/um/Kconfig > index b5e179360534..603f5fd82293 100644 > --- a/arch/um/Kconfig > +++ b/arch/um/Kconfig > @@ -11,7 +11,6 @@ config UML > select ARCH_HAS_KCOV > select ARCH_HAS_STRNCPY_FROM_USER > select ARCH_HAS_STRNLEN_USER > - select ARCH_NO_PREEMPT > select HAVE_ARCH_AUDITSYSCALL > select HAVE_ARCH_KASAN if X86_64 > select HAVE_ARCH_KASAN_VMALLOC if HAVE_ARCH_KASAN > diff --git a/arch/um/include/asm/fpu/api.h b/arch/um/include/asm/fpu/api.h > index 71bfd9ef3938..76ffa9c3c4b9 100644 > --- a/arch/um/include/asm/fpu/api.h > +++ b/arch/um/include/asm/fpu/api.h > @@ -7,9 +7,13 @@ > * A set of "dummy" defines to allow the direct inclusion > * of x86 optimized copy, xor, etc routines into the > * UML code tree. */ > - > +#if defined(CONFIG_PREEMPT) || defined(CONFIG_PREEMPT_VOLUNTRARY) > +extern void kernel_fpu_begin(void); > +extern void kernel_fpu_end(void); > +#else > #define kernel_fpu_begin() (void)0 > #define kernel_fpu_end() (void)0 > +#endif > > static inline bool irq_fpu_usable(void) > { > diff --git a/arch/um/include/asm/processor-generic.h b/arch/um/include/asm/processor-generic.h > index 7414154b8e9a..d9362f5a2212 100644 > --- a/arch/um/include/asm/processor-generic.h > +++ b/arch/um/include/asm/processor-generic.h > @@ -44,6 +44,9 @@ struct thread_struct { > } cb; > } u; > } request; > +#if defined(CONFIG_PREEMPT) || defined(CONFIG_PREEMPT_VOLUNTRARY) > + u8 fpu[2048] __aligned(64); /* Intel docs require xsave/xrestore area to be aligned to 16 bytes */ > +#endif > }; > > #define INIT_THREAD \ > diff --git a/arch/um/kernel/Makefile b/arch/um/kernel/Makefile > index 811188be954c..c616e884a488 100644 > --- a/arch/um/kernel/Makefile > +++ b/arch/um/kernel/Makefile > @@ -26,9 +26,13 @@ obj-$(CONFIG_OF) += dtb.o > obj-$(CONFIG_EARLY_PRINTK) += early_printk.o > obj-$(CONFIG_STACKTRACE) += stacktrace.o > obj-$(CONFIG_GENERIC_PCI_IOMAP) += ioport.o > +obj-$(CONFIG_PREEMPT) += fpu.o > +obj-$(CONFIG_PREEMPT_VOLUNTARY) += fpu.o > > USER_OBJS := config.o > > +CFLAGS_fpu.o += -mxsave -mxsaveopt > + > include $(srctree)/arch/um/scripts/Makefile.rules > > targets := config.c config.tmp capflags.c > diff --git a/arch/um/kernel/fpu.c b/arch/um/kernel/fpu.c > new file mode 100644 > index 000000000000..346c07236185 > --- /dev/null > +++ b/arch/um/kernel/fpu.c > @@ -0,0 +1,77 @@ > +// SPDX-License-Identifier: GPL-2.0-only > +/* > + * Copyright (C) 2023 Cambridge Greys Ltd > + * Copyright (C) 2023 Red Hat Inc > + */ > + > +#include <linux/cpu.h> > +#include <linux/init.h> > +#include <asm/fpu/api.h> > +#include <asm/cpufeature.h> > + > +/* > + * The critical section between kernel_fpu_begin() and kernel_fpu_end() > + * is non-reentrant. It is the caller's responsibility to avoid reentrance. > + */ > + > +static DEFINE_PER_CPU(bool, in_kernel_fpu); > + > +/* UML and driver code it pulls out of the x86 tree knows about 387 features > + * up to and including AVX512. TILE, etc are not yet supported. > + */ > + > +#define KNOWN_387_FEATURES 0xFF > + > + > +void kernel_fpu_begin(void) > +{ > + preempt_disable(); > + > + WARN_ON(this_cpu_read(in_kernel_fpu)); > + > + this_cpu_write(in_kernel_fpu, true); > + > +#ifdef CONFIG_64BIT > + if (likely(cpu_has(&boot_cpu_data, X86_FEATURE_XSAVEOPT))) > + __builtin_ia32_xsaveopt64(¤t->thread.fpu, KNOWN_387_FEATURES); > + else { > + if (likely(cpu_has(&boot_cpu_data, X86_FEATURE_XSAVE))) > + __builtin_ia32_xsave64(¤t->thread.fpu, KNOWN_387_FEATURES); > + else > + __builtin_ia32_fxsave64(¤t->thread.fpu); > + } > +#else > + if (likely(cpu_has(&boot_cpu_data, X86_FEATURE_XSAVEOPT))) > + __builtin_ia32_xsaveopt(¤t->thread.fpu, KNOWN_387_FEATURES); > + else { > + if (likely(cpu_has(&boot_cpu_data, X86_FEATURE_XSAVE))) > + __builtin_ia32_xsave(¤t->thread.fpu, KNOWN_387_FEATURES); > + else > + __builtin_ia32_fxsave(¤t->thread.fpu); > + } > +#endif > +} > + > +EXPORT_SYMBOL_GPL(kernel_fpu_begin); > + > +void kernel_fpu_end(void) > +{ > + WARN_ON(!this_cpu_read(in_kernel_fpu)); > + > +#ifdef CONFIG_64BIT > + if (likely(cpu_has(&boot_cpu_data, X86_FEATURE_XSAVE))) > + __builtin_ia32_xrstor64(¤t->thread.fpu, KNOWN_387_FEATURES); > + else > + __builtin_ia32_fxrstor64(¤t->thread.fpu); > +#else > + if (likely(cpu_has(&boot_cpu_data, X86_FEATURE_XSAVE))) > + __builtin_ia32_xrstor(¤t->thread.fpu, KNOWN_387_FEATURES); > + else > + __builtin_ia32_fxrstor(¤t->thread.fpu); > +#endif > + this_cpu_write(in_kernel_fpu, false); > + > + preempt_enable(); > +} > +EXPORT_SYMBOL_GPL(kernel_fpu_end); > + > diff --git a/arch/um/kernel/irq.c b/arch/um/kernel/irq.c > index 635d44606bfe..c02525da45df 100644 > --- a/arch/um/kernel/irq.c > +++ b/arch/um/kernel/irq.c > @@ -195,7 +195,9 @@ static void _sigio_handler(struct uml_pt_regs *regs, > > void sigio_handler(int sig, struct siginfo *unused_si, struct uml_pt_regs *regs) > { > + preempt_disable(); > _sigio_handler(regs, irqs_suspended); > + preempt_enable(); > } > > static struct irq_entry *get_irq_entry_by_fd(int fd)
On Thu, Sep 21, 2023 at 11:32, anton.ivanov@cambridgegreys.com wrote: > From: Anton Ivanov anton.ivanov@cambridgegreys.com > > > Preemption requires saving/restoring FPU state. This patch > adds support for it using GCC intrinsics. > > Signed-off-by: Anton Ivanov anton.ivanov@cambridgegreys.com > > --- > arch/um/Kconfig | 1 - > arch/um/include/asm/fpu/api.h | 6 +- > arch/um/include/asm/processor-generic.h | 3 + > arch/um/kernel/Makefile | 4 ++ > arch/um/kernel/fpu.c | 77 +++++++++++++++++++++++++ > arch/um/kernel/irq.c | 2 + > 6 files changed, 91 insertions(+), 2 deletions(-) > create mode 100644 arch/um/kernel/fpu.c > > diff --git a/arch/um/Kconfig b/arch/um/Kconfig > index b5e179360534..603f5fd82293 100644 > --- a/arch/um/Kconfig > +++ b/arch/um/Kconfig > @@ -11,7 +11,6 @@ config UML > select ARCH_HAS_KCOV > select ARCH_HAS_STRNCPY_FROM_USER > select ARCH_HAS_STRNLEN_USER > - select ARCH_NO_PREEMPT > select HAVE_ARCH_AUDITSYSCALL > select HAVE_ARCH_KASAN if X86_64 > select HAVE_ARCH_KASAN_VMALLOC if HAVE_ARCH_KASAN > diff --git a/arch/um/include/asm/fpu/api.h b/arch/um/include/asm/fpu/api.h > index 71bfd9ef3938..76ffa9c3c4b9 100644 > --- a/arch/um/include/asm/fpu/api.h > +++ b/arch/um/include/asm/fpu/api.h > @@ -7,9 +7,13 @@ > * A set of "dummy" defines to allow the direct inclusion > * of x86 optimized copy, xor, etc routines into the > * UML code tree. / > - > +#if defined(CONFIG_PREEMPT) || defined(CONFIG_PREEMPT_VOLUNTRARY) Typo, should be CONFIG_PREEMPT_VOLUNTARY > +extern void kernel_fpu_begin(void); > +extern void kernel_fpu_end(void); > +#else > #define kernel_fpu_begin() (void)0 > #define kernel_fpu_end() (void)0 > +#endif > > static inline bool irq_fpu_usable(void) > { > diff --git a/arch/um/include/asm/processor-generic.h b/arch/um/include/asm/processor-generic.h > index 7414154b8e9a..d9362f5a2212 100644 > --- a/arch/um/include/asm/processor-generic.h > +++ b/arch/um/include/asm/processor-generic.h > @@ -44,6 +44,9 @@ struct thread_struct { > } cb; > } u; > } request; > +#if defined(CONFIG_PREEMPT) || defined(CONFIG_PREEMPT_VOLUNTRARY) Same typo again. > + u8 fpu[2048] __aligned(64); / Intel docs require xsave/xrestore area to be aligned to 16 bytes / > +#endif > }; > > #define INIT_THREAD \ > diff --git a/arch/um/kernel/Makefile b/arch/um/kernel/Makefile > index 811188be954c..c616e884a488 100644 > --- a/arch/um/kernel/Makefile > +++ b/arch/um/kernel/Makefile > @@ -26,9 +26,13 @@ obj-$(CONFIG_OF) += dtb.o > obj-$(CONFIG_EARLY_PRINTK) += early_printk.o > obj-$(CONFIG_STACKTRACE) += stacktrace.o > obj-$(CONFIG_GENERIC_PCI_IOMAP) += ioport.o > +obj-$(CONFIG_PREEMPT) += fpu.o > +obj-$(CONFIG_PREEMPT_VOLUNTARY) += fpu.o > > USER_OBJS := config.o > > +CFLAGS_fpu.o += -mxsave -mxsaveopt > + > include $(srctree)/arch/um/scripts/Makefile.rules > > targets := config.c config.tmp capflags.c > diff --git a/arch/um/kernel/fpu.c b/arch/um/kernel/fpu.c > new file mode 100644 > index 000000000000..346c07236185 > --- /dev/null > +++ b/arch/um/kernel/fpu.c > @@ -0,0 +1,77 @@ > +// SPDX-License-Identifier: GPL-2.0-only > +/ > + * Copyright (C) 2023 Cambridge Greys Ltd > + * Copyright (C) 2023 Red Hat Inc > + */ > + > +#include <linux/cpu.h> > > +#include <linux/init.h> > > +#include <asm/fpu/api.h> > > +#include <asm/cpufeature.h> > > + > +/* > + * The critical section between kernel_fpu_begin() and kernel_fpu_end() > + * is non-reentrant. It is the caller's responsibility to avoid reentrance. > + / > + > +static DEFINE_PER_CPU(bool, in_kernel_fpu); > + > +/ UML and driver code it pulls out of the x86 tree knows about 387 features > + * up to and including AVX512. TILE, etc are not yet supported. > + */ > + > +#define KNOWN_387_FEATURES 0xFF > + > + > +void kernel_fpu_begin(void) > +{ > + preempt_disable(); > + > + WARN_ON(this_cpu_read(in_kernel_fpu)); > + > + this_cpu_write(in_kernel_fpu, true); > + > +#ifdef CONFIG_64BIT > + if (likely(cpu_has(&boot_cpu_data, X86_FEATURE_XSAVEOPT))) > + __builtin_ia32_xsaveopt64(¤t->thread.fpu, KNOWN_387_FEATURES); > > + else { > + if (likely(cpu_has(&boot_cpu_data, X86_FEATURE_XSAVE))) > + __builtin_ia32_xsave64(¤t->thread.fpu, KNOWN_387_FEATURES); > > + else > + __builtin_ia32_fxsave64(¤t->thread.fpu); > > + } > +#else > + if (likely(cpu_has(&boot_cpu_data, X86_FEATURE_XSAVEOPT))) > + __builtin_ia32_xsaveopt(¤t->thread.fpu, KNOWN_387_FEATURES); > > + else { > + if (likely(cpu_has(&boot_cpu_data, X86_FEATURE_XSAVE))) > + __builtin_ia32_xsave(¤t->thread.fpu, KNOWN_387_FEATURES); > > + else > + __builtin_ia32_fxsave(¤t->thread.fpu); > > + } > +#endif > +} > + > +EXPORT_SYMBOL_GPL(kernel_fpu_begin); > + > +void kernel_fpu_end(void) > +{ > + WARN_ON(!this_cpu_read(in_kernel_fpu)); > + > +#ifdef CONFIG_64BIT > + if (likely(cpu_has(&boot_cpu_data, X86_FEATURE_XSAVE))) > + __builtin_ia32_xrstor64(¤t->thread.fpu, KNOWN_387_FEATURES); > > + else > + __builtin_ia32_fxrstor64(¤t->thread.fpu); > > +#else > + if (likely(cpu_has(&boot_cpu_data, X86_FEATURE_XSAVE))) > + __builtin_ia32_xrstor(¤t->thread.fpu, KNOWN_387_FEATURES); > > + else > + __builtin_ia32_fxrstor(¤t->thread.fpu); > > +#endif > + this_cpu_write(in_kernel_fpu, false); > + > + preempt_enable(); > +} > +EXPORT_SYMBOL_GPL(kernel_fpu_end); > + > diff --git a/arch/um/kernel/irq.c b/arch/um/kernel/irq.c > index 635d44606bfe..c02525da45df 100644 > --- a/arch/um/kernel/irq.c > +++ b/arch/um/kernel/irq.c > @@ -195,7 +195,9 @@ static void _sigio_handler(struct uml_pt_regs *regs, > > void sigio_handler(int sig, struct siginfo *unused_si, struct uml_pt_regs *regs) > { > + preempt_disable(); > _sigio_handler(regs, irqs_suspended); > + preempt_enable(); > } > > static struct irq_entry *get_irq_entry_by_fd(int fd) Everything other than the typos look good to me. I still think that there should be some limits on reflection, at least when fxsave is being used (when running on a poorly-confgured VM with AVX and no xsave, for example). Cheers, Peter Lafreniere <peter@n8pjl.ca>
On 21/09/2023 16:42, Peter Lafreniere wrote: > On Thu, Sep 21, 2023 at 11:32, anton.ivanov@cambridgegreys.com wrote: > >> From: Anton Ivanov anton.ivanov@cambridgegreys.com >> >> >> Preemption requires saving/restoring FPU state. This patch >> adds support for it using GCC intrinsics. >> >> Signed-off-by: Anton Ivanov anton.ivanov@cambridgegreys.com >> >> --- >> arch/um/Kconfig | 1 - >> arch/um/include/asm/fpu/api.h | 6 +- >> arch/um/include/asm/processor-generic.h | 3 + >> arch/um/kernel/Makefile | 4 ++ >> arch/um/kernel/fpu.c | 77 +++++++++++++++++++++++++ >> arch/um/kernel/irq.c | 2 + >> 6 files changed, 91 insertions(+), 2 deletions(-) >> create mode 100644 arch/um/kernel/fpu.c >> >> diff --git a/arch/um/Kconfig b/arch/um/Kconfig >> index b5e179360534..603f5fd82293 100644 >> --- a/arch/um/Kconfig >> +++ b/arch/um/Kconfig >> @@ -11,7 +11,6 @@ config UML >> select ARCH_HAS_KCOV >> select ARCH_HAS_STRNCPY_FROM_USER >> select ARCH_HAS_STRNLEN_USER >> - select ARCH_NO_PREEMPT >> select HAVE_ARCH_AUDITSYSCALL >> select HAVE_ARCH_KASAN if X86_64 >> select HAVE_ARCH_KASAN_VMALLOC if HAVE_ARCH_KASAN >> diff --git a/arch/um/include/asm/fpu/api.h b/arch/um/include/asm/fpu/api.h >> index 71bfd9ef3938..76ffa9c3c4b9 100644 >> --- a/arch/um/include/asm/fpu/api.h >> +++ b/arch/um/include/asm/fpu/api.h >> @@ -7,9 +7,13 @@ >> * A set of "dummy" defines to allow the direct inclusion >> * of x86 optimized copy, xor, etc routines into the >> * UML code tree. / >> - >> +#if defined(CONFIG_PREEMPT) || defined(CONFIG_PREEMPT_VOLUNTRARY) > Typo, should be CONFIG_PREEMPT_VOLUNTARY Guilty as charged. Fixing it. > >> +extern void kernel_fpu_begin(void); >> +extern void kernel_fpu_end(void); >> +#else >> #define kernel_fpu_begin() (void)0 >> #define kernel_fpu_end() (void)0 >> +#endif >> >> static inline bool irq_fpu_usable(void) >> { >> diff --git a/arch/um/include/asm/processor-generic.h b/arch/um/include/asm/processor-generic.h >> index 7414154b8e9a..d9362f5a2212 100644 >> --- a/arch/um/include/asm/processor-generic.h >> +++ b/arch/um/include/asm/processor-generic.h >> @@ -44,6 +44,9 @@ struct thread_struct { >> } cb; >> } u; >> } request; >> +#if defined(CONFIG_PREEMPT) || defined(CONFIG_PREEMPT_VOLUNTRARY) > Same typo again. > >> + u8 fpu[2048] __aligned(64); / Intel docs require xsave/xrestore area to be aligned to 16 bytes / >> +#endif >> }; >> >> #define INIT_THREAD \ >> diff --git a/arch/um/kernel/Makefile b/arch/um/kernel/Makefile >> index 811188be954c..c616e884a488 100644 >> --- a/arch/um/kernel/Makefile >> +++ b/arch/um/kernel/Makefile >> @@ -26,9 +26,13 @@ obj-$(CONFIG_OF) += dtb.o >> obj-$(CONFIG_EARLY_PRINTK) += early_printk.o >> obj-$(CONFIG_STACKTRACE) += stacktrace.o >> obj-$(CONFIG_GENERIC_PCI_IOMAP) += ioport.o >> +obj-$(CONFIG_PREEMPT) += fpu.o >> +obj-$(CONFIG_PREEMPT_VOLUNTARY) += fpu.o >> >> USER_OBJS := config.o >> >> +CFLAGS_fpu.o += -mxsave -mxsaveopt >> + >> include $(srctree)/arch/um/scripts/Makefile.rules >> >> targets := config.c config.tmp capflags.c >> diff --git a/arch/um/kernel/fpu.c b/arch/um/kernel/fpu.c >> new file mode 100644 >> index 000000000000..346c07236185 >> --- /dev/null >> +++ b/arch/um/kernel/fpu.c >> @@ -0,0 +1,77 @@ >> +// SPDX-License-Identifier: GPL-2.0-only >> +/ >> + * Copyright (C) 2023 Cambridge Greys Ltd >> + * Copyright (C) 2023 Red Hat Inc >> + */ >> + >> +#include <linux/cpu.h> >> >> +#include <linux/init.h> >> >> +#include <asm/fpu/api.h> >> >> +#include <asm/cpufeature.h> >> >> + >> +/* >> + * The critical section between kernel_fpu_begin() and kernel_fpu_end() >> + * is non-reentrant. It is the caller's responsibility to avoid reentrance. >> + / >> + >> +static DEFINE_PER_CPU(bool, in_kernel_fpu); >> + >> +/ UML and driver code it pulls out of the x86 tree knows about 387 features >> + * up to and including AVX512. TILE, etc are not yet supported. >> + */ >> + >> +#define KNOWN_387_FEATURES 0xFF >> + >> + >> +void kernel_fpu_begin(void) >> +{ >> + preempt_disable(); >> + >> + WARN_ON(this_cpu_read(in_kernel_fpu)); >> + >> + this_cpu_write(in_kernel_fpu, true); >> + >> +#ifdef CONFIG_64BIT >> + if (likely(cpu_has(&boot_cpu_data, X86_FEATURE_XSAVEOPT))) >> + __builtin_ia32_xsaveopt64(¤t->thread.fpu, KNOWN_387_FEATURES); >> >> + else { >> + if (likely(cpu_has(&boot_cpu_data, X86_FEATURE_XSAVE))) >> + __builtin_ia32_xsave64(¤t->thread.fpu, KNOWN_387_FEATURES); >> >> + else >> + __builtin_ia32_fxsave64(¤t->thread.fpu); >> >> + } >> +#else >> + if (likely(cpu_has(&boot_cpu_data, X86_FEATURE_XSAVEOPT))) >> + __builtin_ia32_xsaveopt(¤t->thread.fpu, KNOWN_387_FEATURES); >> >> + else { >> + if (likely(cpu_has(&boot_cpu_data, X86_FEATURE_XSAVE))) >> + __builtin_ia32_xsave(¤t->thread.fpu, KNOWN_387_FEATURES); >> >> + else >> + __builtin_ia32_fxsave(¤t->thread.fpu); >> >> + } >> +#endif >> +} >> + >> +EXPORT_SYMBOL_GPL(kernel_fpu_begin); >> + >> +void kernel_fpu_end(void) >> +{ >> + WARN_ON(!this_cpu_read(in_kernel_fpu)); >> + >> +#ifdef CONFIG_64BIT >> + if (likely(cpu_has(&boot_cpu_data, X86_FEATURE_XSAVE))) >> + __builtin_ia32_xrstor64(¤t->thread.fpu, KNOWN_387_FEATURES); >> >> + else >> + __builtin_ia32_fxrstor64(¤t->thread.fpu); >> >> +#else >> + if (likely(cpu_has(&boot_cpu_data, X86_FEATURE_XSAVE))) >> + __builtin_ia32_xrstor(¤t->thread.fpu, KNOWN_387_FEATURES); >> >> + else >> + __builtin_ia32_fxrstor(¤t->thread.fpu); >> >> +#endif >> + this_cpu_write(in_kernel_fpu, false); >> + >> + preempt_enable(); >> +} >> +EXPORT_SYMBOL_GPL(kernel_fpu_end); >> + >> diff --git a/arch/um/kernel/irq.c b/arch/um/kernel/irq.c >> index 635d44606bfe..c02525da45df 100644 >> --- a/arch/um/kernel/irq.c >> +++ b/arch/um/kernel/irq.c >> @@ -195,7 +195,9 @@ static void _sigio_handler(struct uml_pt_regs *regs, >> >> void sigio_handler(int sig, struct siginfo *unused_si, struct uml_pt_regs *regs) >> { >> + preempt_disable(); >> _sigio_handler(regs, irqs_suspended); >> + preempt_enable(); >> } >> >> static struct irq_entry *get_irq_entry_by_fd(int fd) > Everything other than the typos look good to me. I still think that there > should be some limits on reflection, at least when fxsave is being used > (when running on a poorly-confgured VM with AVX and no xsave, for example). > > Cheers, > > Peter Lafreniere > <peter@n8pjl.ca> > >
diff --git a/arch/um/Kconfig b/arch/um/Kconfig index b5e179360534..603f5fd82293 100644 --- a/arch/um/Kconfig +++ b/arch/um/Kconfig @@ -11,7 +11,6 @@ config UML select ARCH_HAS_KCOV select ARCH_HAS_STRNCPY_FROM_USER select ARCH_HAS_STRNLEN_USER - select ARCH_NO_PREEMPT select HAVE_ARCH_AUDITSYSCALL select HAVE_ARCH_KASAN if X86_64 select HAVE_ARCH_KASAN_VMALLOC if HAVE_ARCH_KASAN diff --git a/arch/um/include/asm/fpu/api.h b/arch/um/include/asm/fpu/api.h index 71bfd9ef3938..76ffa9c3c4b9 100644 --- a/arch/um/include/asm/fpu/api.h +++ b/arch/um/include/asm/fpu/api.h @@ -7,9 +7,13 @@ * A set of "dummy" defines to allow the direct inclusion * of x86 optimized copy, xor, etc routines into the * UML code tree. */ - +#if defined(CONFIG_PREEMPT) || defined(CONFIG_PREEMPT_VOLUNTRARY) +extern void kernel_fpu_begin(void); +extern void kernel_fpu_end(void); +#else #define kernel_fpu_begin() (void)0 #define kernel_fpu_end() (void)0 +#endif static inline bool irq_fpu_usable(void) { diff --git a/arch/um/include/asm/processor-generic.h b/arch/um/include/asm/processor-generic.h index 7414154b8e9a..d9362f5a2212 100644 --- a/arch/um/include/asm/processor-generic.h +++ b/arch/um/include/asm/processor-generic.h @@ -44,6 +44,9 @@ struct thread_struct { } cb; } u; } request; +#if defined(CONFIG_PREEMPT) || defined(CONFIG_PREEMPT_VOLUNTRARY) + u8 fpu[2048] __aligned(64); /* Intel docs require xsave/xrestore area to be aligned to 16 bytes */ +#endif }; #define INIT_THREAD \ diff --git a/arch/um/kernel/Makefile b/arch/um/kernel/Makefile index 811188be954c..c616e884a488 100644 --- a/arch/um/kernel/Makefile +++ b/arch/um/kernel/Makefile @@ -26,9 +26,13 @@ obj-$(CONFIG_OF) += dtb.o obj-$(CONFIG_EARLY_PRINTK) += early_printk.o obj-$(CONFIG_STACKTRACE) += stacktrace.o obj-$(CONFIG_GENERIC_PCI_IOMAP) += ioport.o +obj-$(CONFIG_PREEMPT) += fpu.o +obj-$(CONFIG_PREEMPT_VOLUNTARY) += fpu.o USER_OBJS := config.o +CFLAGS_fpu.o += -mxsave -mxsaveopt + include $(srctree)/arch/um/scripts/Makefile.rules targets := config.c config.tmp capflags.c diff --git a/arch/um/kernel/fpu.c b/arch/um/kernel/fpu.c new file mode 100644 index 000000000000..346c07236185 --- /dev/null +++ b/arch/um/kernel/fpu.c @@ -0,0 +1,77 @@ +// SPDX-License-Identifier: GPL-2.0-only +/* + * Copyright (C) 2023 Cambridge Greys Ltd + * Copyright (C) 2023 Red Hat Inc + */ + +#include <linux/cpu.h> +#include <linux/init.h> +#include <asm/fpu/api.h> +#include <asm/cpufeature.h> + +/* + * The critical section between kernel_fpu_begin() and kernel_fpu_end() + * is non-reentrant. It is the caller's responsibility to avoid reentrance. + */ + +static DEFINE_PER_CPU(bool, in_kernel_fpu); + +/* UML and driver code it pulls out of the x86 tree knows about 387 features + * up to and including AVX512. TILE, etc are not yet supported. + */ + +#define KNOWN_387_FEATURES 0xFF + + +void kernel_fpu_begin(void) +{ + preempt_disable(); + + WARN_ON(this_cpu_read(in_kernel_fpu)); + + this_cpu_write(in_kernel_fpu, true); + +#ifdef CONFIG_64BIT + if (likely(cpu_has(&boot_cpu_data, X86_FEATURE_XSAVEOPT))) + __builtin_ia32_xsaveopt64(¤t->thread.fpu, KNOWN_387_FEATURES); + else { + if (likely(cpu_has(&boot_cpu_data, X86_FEATURE_XSAVE))) + __builtin_ia32_xsave64(¤t->thread.fpu, KNOWN_387_FEATURES); + else + __builtin_ia32_fxsave64(¤t->thread.fpu); + } +#else + if (likely(cpu_has(&boot_cpu_data, X86_FEATURE_XSAVEOPT))) + __builtin_ia32_xsaveopt(¤t->thread.fpu, KNOWN_387_FEATURES); + else { + if (likely(cpu_has(&boot_cpu_data, X86_FEATURE_XSAVE))) + __builtin_ia32_xsave(¤t->thread.fpu, KNOWN_387_FEATURES); + else + __builtin_ia32_fxsave(¤t->thread.fpu); + } +#endif +} + +EXPORT_SYMBOL_GPL(kernel_fpu_begin); + +void kernel_fpu_end(void) +{ + WARN_ON(!this_cpu_read(in_kernel_fpu)); + +#ifdef CONFIG_64BIT + if (likely(cpu_has(&boot_cpu_data, X86_FEATURE_XSAVE))) + __builtin_ia32_xrstor64(¤t->thread.fpu, KNOWN_387_FEATURES); + else + __builtin_ia32_fxrstor64(¤t->thread.fpu); +#else + if (likely(cpu_has(&boot_cpu_data, X86_FEATURE_XSAVE))) + __builtin_ia32_xrstor(¤t->thread.fpu, KNOWN_387_FEATURES); + else + __builtin_ia32_fxrstor(¤t->thread.fpu); +#endif + this_cpu_write(in_kernel_fpu, false); + + preempt_enable(); +} +EXPORT_SYMBOL_GPL(kernel_fpu_end); + diff --git a/arch/um/kernel/irq.c b/arch/um/kernel/irq.c index 635d44606bfe..c02525da45df 100644 --- a/arch/um/kernel/irq.c +++ b/arch/um/kernel/irq.c @@ -195,7 +195,9 @@ static void _sigio_handler(struct uml_pt_regs *regs, void sigio_handler(int sig, struct siginfo *unused_si, struct uml_pt_regs *regs) { + preempt_disable(); _sigio_handler(regs, irqs_suspended); + preempt_enable(); } static struct irq_entry *get_irq_entry_by_fd(int fd)