2015-04-22 15:53:34 +07:00
|
|
|
/*
|
2015-04-26 20:36:46 +07:00
|
|
|
* x86 FPU boot time init code:
|
2015-04-22 15:53:34 +07:00
|
|
|
*/
|
2015-04-24 07:54:44 +07:00
|
|
|
#include <asm/fpu/internal.h>
|
2015-04-22 15:53:34 +07:00
|
|
|
#include <asm/tlbflush.h>
|
x86/fpu: Fix early FPU command-line parsing
The function fpu__init_system() is executed before
parse_early_param(). This causes wrong FPU configuration. This
patch fixes this issue by parsing boot_command_line in the
beginning of fpu__init_system().
With all four patches in this series, each parameter disables
features as the following:
eagerfpu=off: eagerfpu, avx, avx2, avx512, mpx
no387: fpu
nofxsr: fxsr, fxsropt, xmm
noxsave: xsave, xsaveopt, xsaves, xsavec, avx, avx2, avx512,
mpx, xgetbv1 noxsaveopt: xsaveopt
noxsaves: xsaves
Signed-off-by: Yu-cheng Yu <yu-cheng.yu@intel.com>
Cc: Andy Lutomirski <luto@amacapital.net>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Borislav Petkov <bp@suse.de>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: Fenghua Yu <fenghua.yu@intel.com>
Cc: H. Peter Anvin <hpa@zytor.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Quentin Casasnovas <quentin.casasnovas@oracle.com>
Cc: Ravi V. Shankar <ravi.v.shankar@intel.com>
Cc: Sai Praneeth Prakhya <sai.praneeth.prakhya@intel.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: yu-cheng yu <yu-cheng.yu@intel.com>
Link: http://lkml.kernel.org/r/1452119094-7252-2-git-send-email-yu-cheng.yu@intel.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-01-07 05:24:51 +07:00
|
|
|
#include <asm/setup.h>
|
|
|
|
#include <asm/cmdline.h>
|
2015-04-22 15:53:34 +07:00
|
|
|
|
2015-07-17 17:28:12 +07:00
|
|
|
#include <linux/sched.h>
|
x86/fpu: Fix early FPU command-line parsing
The function fpu__init_system() is executed before
parse_early_param(). This causes wrong FPU configuration. This
patch fixes this issue by parsing boot_command_line in the
beginning of fpu__init_system().
With all four patches in this series, each parameter disables
features as the following:
eagerfpu=off: eagerfpu, avx, avx2, avx512, mpx
no387: fpu
nofxsr: fxsr, fxsropt, xmm
noxsave: xsave, xsaveopt, xsaves, xsavec, avx, avx2, avx512,
mpx, xgetbv1 noxsaveopt: xsaveopt
noxsaves: xsaves
Signed-off-by: Yu-cheng Yu <yu-cheng.yu@intel.com>
Cc: Andy Lutomirski <luto@amacapital.net>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Borislav Petkov <bp@suse.de>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: Fenghua Yu <fenghua.yu@intel.com>
Cc: H. Peter Anvin <hpa@zytor.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Quentin Casasnovas <quentin.casasnovas@oracle.com>
Cc: Ravi V. Shankar <ravi.v.shankar@intel.com>
Cc: Sai Praneeth Prakhya <sai.praneeth.prakhya@intel.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: yu-cheng yu <yu-cheng.yu@intel.com>
Link: http://lkml.kernel.org/r/1452119094-7252-2-git-send-email-yu-cheng.yu@intel.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-01-07 05:24:51 +07:00
|
|
|
#include <linux/init.h>
|
2015-07-17 17:28:12 +07:00
|
|
|
|
2015-04-26 20:36:46 +07:00
|
|
|
/*
|
|
|
|
* Initialize the TS bit in CR0 according to the style of context-switches
|
|
|
|
* we are using:
|
|
|
|
*/
|
2015-04-26 20:32:40 +07:00
|
|
|
static void fpu__init_cpu_ctx_switch(void)
|
|
|
|
{
|
2015-12-07 16:39:41 +07:00
|
|
|
if (!boot_cpu_has(X86_FEATURE_EAGER_FPU))
|
2015-04-26 20:32:40 +07:00
|
|
|
stts();
|
|
|
|
else
|
|
|
|
clts();
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Initialize the registers found in all CPUs, CR0 and CR4:
|
|
|
|
*/
|
|
|
|
static void fpu__init_cpu_generic(void)
|
|
|
|
{
|
|
|
|
unsigned long cr0;
|
|
|
|
unsigned long cr4_mask = 0;
|
|
|
|
|
2016-04-05 03:25:01 +07:00
|
|
|
if (boot_cpu_has(X86_FEATURE_FXSR))
|
2015-04-26 20:32:40 +07:00
|
|
|
cr4_mask |= X86_CR4_OSFXSR;
|
2016-04-05 03:24:57 +07:00
|
|
|
if (boot_cpu_has(X86_FEATURE_XMM))
|
2015-04-26 20:32:40 +07:00
|
|
|
cr4_mask |= X86_CR4_OSXMMEXCPT;
|
|
|
|
if (cr4_mask)
|
|
|
|
cr4_set_bits(cr4_mask);
|
|
|
|
|
|
|
|
cr0 = read_cr0();
|
|
|
|
cr0 &= ~(X86_CR0_TS|X86_CR0_EM); /* clear TS and EM */
|
2016-04-05 03:24:58 +07:00
|
|
|
if (!boot_cpu_has(X86_FEATURE_FPU))
|
2015-04-26 20:32:40 +07:00
|
|
|
cr0 |= X86_CR0_EM;
|
|
|
|
write_cr0(cr0);
|
2015-04-29 15:58:03 +07:00
|
|
|
|
|
|
|
/* Flush out any pending x87 state: */
|
2015-08-22 14:52:06 +07:00
|
|
|
#ifdef CONFIG_MATH_EMULATION
|
2016-04-05 03:24:58 +07:00
|
|
|
if (!boot_cpu_has(X86_FEATURE_FPU))
|
2015-08-22 14:52:06 +07:00
|
|
|
fpstate_init_soft(¤t->thread.fpu.state.soft);
|
|
|
|
else
|
|
|
|
#endif
|
|
|
|
asm volatile ("fninit");
|
2015-04-26 20:32:40 +07:00
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
2015-04-26 20:36:46 +07:00
|
|
|
* Enable all supported FPU features. Called when a CPU is brought online:
|
2015-04-26 20:32:40 +07:00
|
|
|
*/
|
|
|
|
void fpu__init_cpu(void)
|
|
|
|
{
|
|
|
|
fpu__init_cpu_generic();
|
|
|
|
fpu__init_cpu_xstate();
|
|
|
|
fpu__init_cpu_ctx_switch();
|
|
|
|
}
|
|
|
|
|
2015-04-26 19:40:54 +07:00
|
|
|
/*
|
2015-04-26 20:07:18 +07:00
|
|
|
* The earliest FPU detection code.
|
|
|
|
*
|
|
|
|
* Set the X86_FEATURE_FPU CPU-capability bit based on
|
|
|
|
* trying to execute an actual sequence of FPU instructions:
|
2015-04-26 19:40:54 +07:00
|
|
|
*/
|
|
|
|
static void fpu__init_system_early_generic(struct cpuinfo_x86 *c)
|
|
|
|
{
|
|
|
|
unsigned long cr0;
|
|
|
|
u16 fsw, fcw;
|
|
|
|
|
|
|
|
fsw = fcw = 0xffff;
|
|
|
|
|
|
|
|
cr0 = read_cr0();
|
|
|
|
cr0 &= ~(X86_CR0_TS | X86_CR0_EM);
|
|
|
|
write_cr0(cr0);
|
|
|
|
|
2016-01-22 06:24:31 +07:00
|
|
|
if (!test_bit(X86_FEATURE_FPU, (unsigned long *)cpu_caps_cleared)) {
|
|
|
|
asm volatile("fninit ; fnstsw %0 ; fnstcw %1"
|
|
|
|
: "+m" (fsw), "+m" (fcw));
|
|
|
|
|
|
|
|
if (fsw == 0 && (fcw & 0x103f) == 0x003f)
|
|
|
|
set_cpu_cap(c, X86_FEATURE_FPU);
|
|
|
|
else
|
|
|
|
clear_cpu_cap(c, X86_FEATURE_FPU);
|
|
|
|
}
|
2015-04-26 19:43:44 +07:00
|
|
|
|
|
|
|
#ifndef CONFIG_MATH_EMULATION
|
2016-04-05 03:24:58 +07:00
|
|
|
if (!boot_cpu_has(X86_FEATURE_FPU)) {
|
2015-04-26 20:36:46 +07:00
|
|
|
pr_emerg("x86/fpu: Giving up, no FPU found and no math emulation present\n");
|
2015-04-26 19:43:44 +07:00
|
|
|
for (;;)
|
|
|
|
asm volatile("hlt");
|
|
|
|
}
|
|
|
|
#endif
|
2015-04-26 19:40:54 +07:00
|
|
|
}
|
|
|
|
|
2015-04-22 18:44:25 +07:00
|
|
|
/*
|
|
|
|
* Boot time FPU feature detection code:
|
|
|
|
*/
|
2015-04-22 15:53:34 +07:00
|
|
|
unsigned int mxcsr_feature_mask __read_mostly = 0xffffffffu;
|
2015-04-24 15:49:11 +07:00
|
|
|
|
2015-05-04 14:52:42 +07:00
|
|
|
static void __init fpu__init_system_mxcsr(void)
|
2015-04-22 15:53:34 +07:00
|
|
|
{
|
2015-04-24 15:49:11 +07:00
|
|
|
unsigned int mask = 0;
|
2015-04-22 15:53:34 +07:00
|
|
|
|
2016-04-05 03:25:01 +07:00
|
|
|
if (boot_cpu_has(X86_FEATURE_FXSR)) {
|
2015-07-04 14:58:19 +07:00
|
|
|
/* Static because GCC does not get 16-byte stack alignment right: */
|
|
|
|
static struct fxregs_state fxregs __initdata;
|
2015-04-24 15:49:11 +07:00
|
|
|
|
2015-07-04 14:58:19 +07:00
|
|
|
asm volatile("fxsave %0" : "+m" (fxregs));
|
2015-04-24 15:49:11 +07:00
|
|
|
|
2015-07-04 14:58:19 +07:00
|
|
|
mask = fxregs.mxcsr_mask;
|
2015-04-24 15:49:11 +07:00
|
|
|
|
|
|
|
/*
|
|
|
|
* If zero then use the default features mask,
|
|
|
|
* which has all features set, except the
|
|
|
|
* denormals-are-zero feature bit:
|
|
|
|
*/
|
2015-04-22 15:53:34 +07:00
|
|
|
if (mask == 0)
|
|
|
|
mask = 0x0000ffbf;
|
|
|
|
}
|
|
|
|
mxcsr_feature_mask &= mask;
|
|
|
|
}
|
|
|
|
|
2015-04-26 19:35:54 +07:00
|
|
|
/*
|
|
|
|
* Once per bootup FPU initialization sequences that will run on most x86 CPUs:
|
|
|
|
*/
|
2015-05-04 14:52:42 +07:00
|
|
|
static void __init fpu__init_system_generic(void)
|
2015-04-26 19:35:54 +07:00
|
|
|
{
|
|
|
|
/*
|
|
|
|
* Set up the legacy init FPU context. (xstate init might overwrite this
|
|
|
|
* with a more modern format, if the CPU supports it.)
|
|
|
|
*/
|
2016-03-11 18:32:06 +07:00
|
|
|
fpstate_init(&init_fpstate);
|
2015-04-26 19:35:54 +07:00
|
|
|
|
|
|
|
fpu__init_system_mxcsr();
|
|
|
|
}
|
|
|
|
|
2015-04-26 20:36:46 +07:00
|
|
|
/*
|
|
|
|
* Size of the FPU context state. All tasks in the system use the
|
|
|
|
* same context size, regardless of what portion they use.
|
|
|
|
* This is inherent to the XSAVE architecture which puts all state
|
|
|
|
* components into a single, continuous memory block:
|
|
|
|
*/
|
2016-05-21 00:47:06 +07:00
|
|
|
unsigned int fpu_kernel_xstate_size;
|
|
|
|
EXPORT_SYMBOL_GPL(fpu_kernel_xstate_size);
|
2015-04-26 20:32:40 +07:00
|
|
|
|
2015-12-21 21:25:30 +07:00
|
|
|
/* Get alignment of the TYPE. */
|
|
|
|
#define TYPE_ALIGN(TYPE) offsetof(struct { char x; TYPE test; }, test)
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Enforce that 'MEMBER' is the last field of 'TYPE'.
|
|
|
|
*
|
|
|
|
* Align the computed size with alignment of the TYPE,
|
|
|
|
* because that's how C aligns structs.
|
|
|
|
*/
|
2015-07-17 17:28:12 +07:00
|
|
|
#define CHECK_MEMBER_AT_END_OF(TYPE, MEMBER) \
|
2015-12-21 21:25:30 +07:00
|
|
|
BUILD_BUG_ON(sizeof(TYPE) != ALIGN(offsetofend(TYPE, MEMBER), \
|
|
|
|
TYPE_ALIGN(TYPE)))
|
2015-07-17 17:28:11 +07:00
|
|
|
|
|
|
|
/*
|
2015-07-17 17:28:12 +07:00
|
|
|
* We append the 'struct fpu' to the task_struct:
|
2015-07-17 17:28:11 +07:00
|
|
|
*/
|
2015-07-17 17:28:12 +07:00
|
|
|
static void __init fpu__init_task_struct_size(void)
|
2015-07-17 17:28:11 +07:00
|
|
|
{
|
|
|
|
int task_size = sizeof(struct task_struct);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Subtract off the static size of the register state.
|
|
|
|
* It potentially has a bunch of padding.
|
|
|
|
*/
|
|
|
|
task_size -= sizeof(((struct task_struct *)0)->thread.fpu.state);
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Add back the dynamically-calculated register state
|
|
|
|
* size.
|
|
|
|
*/
|
2016-05-21 00:47:06 +07:00
|
|
|
task_size += fpu_kernel_xstate_size;
|
2015-07-17 17:28:11 +07:00
|
|
|
|
|
|
|
/*
|
|
|
|
* We dynamically size 'struct fpu', so we require that
|
|
|
|
* it be at the end of 'thread_struct' and that
|
|
|
|
* 'thread_struct' be at the end of 'task_struct'. If
|
|
|
|
* you hit a compile error here, check the structure to
|
|
|
|
* see if something got added to the end.
|
|
|
|
*/
|
|
|
|
CHECK_MEMBER_AT_END_OF(struct fpu, state);
|
|
|
|
CHECK_MEMBER_AT_END_OF(struct thread_struct, fpu);
|
|
|
|
CHECK_MEMBER_AT_END_OF(struct task_struct, thread);
|
|
|
|
|
2015-07-17 17:28:12 +07:00
|
|
|
arch_task_struct_size = task_size;
|
2015-07-17 17:28:11 +07:00
|
|
|
}
|
|
|
|
|
2015-04-26 20:32:40 +07:00
|
|
|
/*
|
2016-05-21 00:47:06 +07:00
|
|
|
* Set up the user and kernel xstate sizes based on the legacy FPU context size.
|
2015-04-26 20:32:40 +07:00
|
|
|
*
|
|
|
|
* We set this up first, and later it will be overwritten by
|
|
|
|
* fpu__init_system_xstate() if the CPU knows about xstates.
|
|
|
|
*/
|
2015-05-04 14:52:42 +07:00
|
|
|
static void __init fpu__init_system_xstate_size_legacy(void)
|
2015-04-22 15:53:34 +07:00
|
|
|
{
|
2015-11-13 21:18:31 +07:00
|
|
|
static int on_boot_cpu __initdata = 1;
|
2015-05-05 16:34:49 +07:00
|
|
|
|
|
|
|
WARN_ON_FPU(!on_boot_cpu);
|
|
|
|
on_boot_cpu = 0;
|
|
|
|
|
2015-04-22 15:53:34 +07:00
|
|
|
/*
|
2016-05-21 00:47:06 +07:00
|
|
|
* Note that xstate sizes might be overwritten later during
|
2015-04-25 11:52:53 +07:00
|
|
|
* fpu__init_system_xstate().
|
2015-04-22 15:53:34 +07:00
|
|
|
*/
|
|
|
|
|
2016-04-05 03:24:58 +07:00
|
|
|
if (!boot_cpu_has(X86_FEATURE_FPU)) {
|
2015-04-22 15:53:34 +07:00
|
|
|
/*
|
|
|
|
* Disable xsave as we do not support it if i387
|
|
|
|
* emulation is enabled.
|
|
|
|
*/
|
|
|
|
setup_clear_cpu_cap(X86_FEATURE_XSAVE);
|
|
|
|
setup_clear_cpu_cap(X86_FEATURE_XSAVEOPT);
|
2016-05-21 00:47:06 +07:00
|
|
|
fpu_kernel_xstate_size = sizeof(struct swregs_state);
|
2015-04-25 09:29:26 +07:00
|
|
|
} else {
|
2016-04-05 03:25:01 +07:00
|
|
|
if (boot_cpu_has(X86_FEATURE_FXSR))
|
2016-05-21 00:47:06 +07:00
|
|
|
fpu_kernel_xstate_size =
|
|
|
|
sizeof(struct fxregs_state);
|
2015-04-25 09:29:26 +07:00
|
|
|
else
|
2016-05-21 00:47:06 +07:00
|
|
|
fpu_kernel_xstate_size =
|
|
|
|
sizeof(struct fregs_state);
|
2015-04-22 15:53:34 +07:00
|
|
|
}
|
2016-05-21 00:47:05 +07:00
|
|
|
|
2016-05-21 00:47:06 +07:00
|
|
|
fpu_user_xstate_size = fpu_kernel_xstate_size;
|
2015-04-22 15:53:34 +07:00
|
|
|
}
|
|
|
|
|
2015-04-26 20:36:46 +07:00
|
|
|
/*
|
|
|
|
* FPU context switching strategies:
|
|
|
|
*
|
|
|
|
* Against popular belief, we don't do lazy FPU saves, due to the
|
|
|
|
* task migration complications it brings on SMP - we only do
|
|
|
|
* lazy FPU restores.
|
|
|
|
*
|
|
|
|
* 'lazy' is the traditional strategy, which is based on setting
|
|
|
|
* CR0::TS to 1 during context-switch (instead of doing a full
|
|
|
|
* restore of the FPU state), which causes the first FPU instruction
|
|
|
|
* after the context switch (whenever it is executed) to fault - at
|
|
|
|
* which point we lazily restore the FPU state into FPU registers.
|
|
|
|
*
|
|
|
|
* Tasks are of course under no obligation to execute FPU instructions,
|
|
|
|
* so it can easily happen that another context-switch occurs without
|
|
|
|
* a single FPU instruction being executed. If we eventually switch
|
|
|
|
* back to the original task (that still owns the FPU) then we have
|
|
|
|
* not only saved the restores along the way, but we also have the
|
|
|
|
* FPU ready to be used for the original task.
|
|
|
|
*
|
x86/fpu: Default eagerfpu=on on all CPUs
We have eager and lazy FPU modes, introduced in:
304bceda6a18 ("x86, fpu: use non-lazy fpu restore for processors supporting xsave")
The result is rather messy. There are two code paths in almost all
of the FPU code, and only one of them (the eager case) is tested
frequently, since most kernel developers have new enough hardware
that we use eagerfpu.
It seems that, on any remotely recent hardware, eagerfpu is a win:
glibc uses SSE2, so laziness is probably overoptimistic, and, in any
case, manipulating TS is far slower that saving and restoring the
full state. (Stores to CR0.TS are serializing and are poorly
optimized.)
To try to shake out any latent issues on old hardware, this changes
the default to eager on all CPUs. If no performance or functionality
problems show up, a subsequent patch could remove lazy mode entirely.
Signed-off-by: Andy Lutomirski <luto@kernel.org>
Cc: Andy Lutomirski <luto@amacapital.net>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: Fenghua Yu <fenghua.yu@intel.com>
Cc: H. Peter Anvin <hpa@zytor.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Quentin Casasnovas <quentin.casasnovas@oracle.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Sai Praneeth Prakhya <sai.praneeth.prakhya@intel.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: yu-cheng yu <yu-cheng.yu@intel.com>
Link: http://lkml.kernel.org/r/ac290de61bf08d9cfc2664a4f5080257ffc1075a.1453675014.git.luto@kernel.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-01-25 05:38:10 +07:00
|
|
|
* 'lazy' is deprecated because it's almost never a performance win
|
|
|
|
* and it's much more complicated than 'eager'.
|
|
|
|
*
|
|
|
|
* 'eager' switching is by default on all CPUs, there we switch the FPU
|
2015-04-26 20:36:46 +07:00
|
|
|
* state during every context switch, regardless of whether the task
|
|
|
|
* has used FPU instructions in that time slice or not. This is done
|
|
|
|
* because modern FPU context saving instructions are able to optimize
|
|
|
|
* state saving and restoration in hardware: they can detect both
|
|
|
|
* unused and untouched FPU state and optimize accordingly.
|
|
|
|
*
|
|
|
|
* [ Note that even in 'lazy' mode we might optimize context switches
|
|
|
|
* to use 'eager' restores, if we detect that a task is using the FPU
|
|
|
|
* frequently. See the fpu->counter logic in fpu/internal.h for that. ]
|
|
|
|
*/
|
x86/fpu: Default eagerfpu=on on all CPUs
We have eager and lazy FPU modes, introduced in:
304bceda6a18 ("x86, fpu: use non-lazy fpu restore for processors supporting xsave")
The result is rather messy. There are two code paths in almost all
of the FPU code, and only one of them (the eager case) is tested
frequently, since most kernel developers have new enough hardware
that we use eagerfpu.
It seems that, on any remotely recent hardware, eagerfpu is a win:
glibc uses SSE2, so laziness is probably overoptimistic, and, in any
case, manipulating TS is far slower that saving and restoring the
full state. (Stores to CR0.TS are serializing and are poorly
optimized.)
To try to shake out any latent issues on old hardware, this changes
the default to eager on all CPUs. If no performance or functionality
problems show up, a subsequent patch could remove lazy mode entirely.
Signed-off-by: Andy Lutomirski <luto@kernel.org>
Cc: Andy Lutomirski <luto@amacapital.net>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: Fenghua Yu <fenghua.yu@intel.com>
Cc: H. Peter Anvin <hpa@zytor.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Quentin Casasnovas <quentin.casasnovas@oracle.com>
Cc: Rik van Riel <riel@redhat.com>
Cc: Sai Praneeth Prakhya <sai.praneeth.prakhya@intel.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: yu-cheng yu <yu-cheng.yu@intel.com>
Link: http://lkml.kernel.org/r/ac290de61bf08d9cfc2664a4f5080257ffc1075a.1453675014.git.luto@kernel.org
Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-01-25 05:38:10 +07:00
|
|
|
static enum { ENABLE, DISABLE } eagerfpu = ENABLE;
|
2015-04-26 01:11:05 +07:00
|
|
|
|
2016-01-07 05:24:53 +07:00
|
|
|
/*
|
|
|
|
* Find supported xfeatures based on cpu features and command-line input.
|
|
|
|
* This must be called after fpu__init_parse_early_param() is called and
|
|
|
|
* xfeatures_mask is enumerated.
|
|
|
|
*/
|
|
|
|
u64 __init fpu__get_supported_xfeatures_mask(void)
|
|
|
|
{
|
|
|
|
/* Support all xfeatures known to us */
|
|
|
|
if (eagerfpu != DISABLE)
|
|
|
|
return XCNTXT_MASK;
|
|
|
|
|
|
|
|
/* Warning of xfeatures being disabled for no eagerfpu mode */
|
|
|
|
if (xfeatures_mask & XFEATURE_MASK_EAGER) {
|
|
|
|
pr_err("x86/fpu: eagerfpu switching disabled, disabling the following xstate features: 0x%llx.\n",
|
|
|
|
xfeatures_mask & XFEATURE_MASK_EAGER);
|
|
|
|
}
|
|
|
|
|
|
|
|
/* Return a mask that masks out all features requiring eagerfpu mode */
|
|
|
|
return ~XFEATURE_MASK_EAGER;
|
|
|
|
}
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Disable features dependent on eagerfpu.
|
|
|
|
*/
|
|
|
|
static void __init fpu__clear_eager_fpu_features(void)
|
|
|
|
{
|
|
|
|
setup_clear_cpu_cap(X86_FEATURE_MPX);
|
|
|
|
}
|
|
|
|
|
2015-04-26 01:11:05 +07:00
|
|
|
/*
|
2015-04-26 20:36:46 +07:00
|
|
|
* Pick the FPU context switching strategy:
|
2016-01-07 05:24:53 +07:00
|
|
|
*
|
|
|
|
* When eagerfpu is AUTO or ENABLE, we ensure it is ENABLE if either of
|
|
|
|
* the following is true:
|
|
|
|
*
|
|
|
|
* (1) the cpu has xsaveopt, as it has the optimization and doing eager
|
|
|
|
* FPU switching has a relatively low cost compared to a plain xsave;
|
|
|
|
* (2) the cpu has xsave features (e.g. MPX) that depend on eager FPU
|
|
|
|
* switching. Should the kernel boot with noxsaveopt, we support MPX
|
|
|
|
* with eager FPU switching at a higher cost.
|
2015-04-26 01:11:05 +07:00
|
|
|
*/
|
2015-05-04 14:52:42 +07:00
|
|
|
static void __init fpu__init_system_ctx_switch(void)
|
2015-04-26 01:11:05 +07:00
|
|
|
{
|
2015-11-13 21:18:31 +07:00
|
|
|
static bool on_boot_cpu __initdata = 1;
|
2015-05-05 16:34:49 +07:00
|
|
|
|
|
|
|
WARN_ON_FPU(!on_boot_cpu);
|
|
|
|
on_boot_cpu = 0;
|
|
|
|
|
|
|
|
WARN_ON_FPU(current->thread.fpu.fpstate_active);
|
2015-04-26 01:11:05 +07:00
|
|
|
|
2015-12-07 16:39:41 +07:00
|
|
|
if (boot_cpu_has(X86_FEATURE_XSAVEOPT) && eagerfpu != DISABLE)
|
2015-04-26 01:11:05 +07:00
|
|
|
eagerfpu = ENABLE;
|
|
|
|
|
2016-01-07 05:24:53 +07:00
|
|
|
if (xfeatures_mask & XFEATURE_MASK_EAGER)
|
|
|
|
eagerfpu = ENABLE;
|
2015-04-26 01:11:05 +07:00
|
|
|
|
|
|
|
if (eagerfpu == ENABLE)
|
|
|
|
setup_force_cpu_cap(X86_FEATURE_EAGER_FPU);
|
|
|
|
|
2015-05-04 14:52:42 +07:00
|
|
|
printk(KERN_INFO "x86/fpu: Using '%s' FPU context switches.\n", eagerfpu == ENABLE ? "eager" : "lazy");
|
2015-04-26 01:11:05 +07:00
|
|
|
}
|
|
|
|
|
x86/fpu: Fix early FPU command-line parsing
The function fpu__init_system() is executed before
parse_early_param(). This causes wrong FPU configuration. This
patch fixes this issue by parsing boot_command_line in the
beginning of fpu__init_system().
With all four patches in this series, each parameter disables
features as the following:
eagerfpu=off: eagerfpu, avx, avx2, avx512, mpx
no387: fpu
nofxsr: fxsr, fxsropt, xmm
noxsave: xsave, xsaveopt, xsaves, xsavec, avx, avx2, avx512,
mpx, xgetbv1 noxsaveopt: xsaveopt
noxsaves: xsaves
Signed-off-by: Yu-cheng Yu <yu-cheng.yu@intel.com>
Cc: Andy Lutomirski <luto@amacapital.net>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Borislav Petkov <bp@suse.de>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: Fenghua Yu <fenghua.yu@intel.com>
Cc: H. Peter Anvin <hpa@zytor.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Quentin Casasnovas <quentin.casasnovas@oracle.com>
Cc: Ravi V. Shankar <ravi.v.shankar@intel.com>
Cc: Sai Praneeth Prakhya <sai.praneeth.prakhya@intel.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: yu-cheng yu <yu-cheng.yu@intel.com>
Link: http://lkml.kernel.org/r/1452119094-7252-2-git-send-email-yu-cheng.yu@intel.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-01-07 05:24:51 +07:00
|
|
|
/*
|
|
|
|
* We parse fpu parameters early because fpu__init_system() is executed
|
|
|
|
* before parse_early_param().
|
|
|
|
*/
|
|
|
|
static void __init fpu__init_parse_early_param(void)
|
|
|
|
{
|
2016-01-07 05:24:53 +07:00
|
|
|
if (cmdline_find_option_bool(boot_command_line, "eagerfpu=off")) {
|
x86/fpu: Fix early FPU command-line parsing
The function fpu__init_system() is executed before
parse_early_param(). This causes wrong FPU configuration. This
patch fixes this issue by parsing boot_command_line in the
beginning of fpu__init_system().
With all four patches in this series, each parameter disables
features as the following:
eagerfpu=off: eagerfpu, avx, avx2, avx512, mpx
no387: fpu
nofxsr: fxsr, fxsropt, xmm
noxsave: xsave, xsaveopt, xsaves, xsavec, avx, avx2, avx512,
mpx, xgetbv1 noxsaveopt: xsaveopt
noxsaves: xsaves
Signed-off-by: Yu-cheng Yu <yu-cheng.yu@intel.com>
Cc: Andy Lutomirski <luto@amacapital.net>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Borislav Petkov <bp@suse.de>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: Fenghua Yu <fenghua.yu@intel.com>
Cc: H. Peter Anvin <hpa@zytor.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Quentin Casasnovas <quentin.casasnovas@oracle.com>
Cc: Ravi V. Shankar <ravi.v.shankar@intel.com>
Cc: Sai Praneeth Prakhya <sai.praneeth.prakhya@intel.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: yu-cheng yu <yu-cheng.yu@intel.com>
Link: http://lkml.kernel.org/r/1452119094-7252-2-git-send-email-yu-cheng.yu@intel.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-01-07 05:24:51 +07:00
|
|
|
eagerfpu = DISABLE;
|
2016-01-07 05:24:53 +07:00
|
|
|
fpu__clear_eager_fpu_features();
|
|
|
|
}
|
x86/fpu: Fix early FPU command-line parsing
The function fpu__init_system() is executed before
parse_early_param(). This causes wrong FPU configuration. This
patch fixes this issue by parsing boot_command_line in the
beginning of fpu__init_system().
With all four patches in this series, each parameter disables
features as the following:
eagerfpu=off: eagerfpu, avx, avx2, avx512, mpx
no387: fpu
nofxsr: fxsr, fxsropt, xmm
noxsave: xsave, xsaveopt, xsaves, xsavec, avx, avx2, avx512,
mpx, xgetbv1 noxsaveopt: xsaveopt
noxsaves: xsaves
Signed-off-by: Yu-cheng Yu <yu-cheng.yu@intel.com>
Cc: Andy Lutomirski <luto@amacapital.net>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Borislav Petkov <bp@suse.de>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: Fenghua Yu <fenghua.yu@intel.com>
Cc: H. Peter Anvin <hpa@zytor.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Quentin Casasnovas <quentin.casasnovas@oracle.com>
Cc: Ravi V. Shankar <ravi.v.shankar@intel.com>
Cc: Sai Praneeth Prakhya <sai.praneeth.prakhya@intel.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: yu-cheng yu <yu-cheng.yu@intel.com>
Link: http://lkml.kernel.org/r/1452119094-7252-2-git-send-email-yu-cheng.yu@intel.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-01-07 05:24:51 +07:00
|
|
|
|
|
|
|
if (cmdline_find_option_bool(boot_command_line, "no387"))
|
|
|
|
setup_clear_cpu_cap(X86_FEATURE_FPU);
|
|
|
|
|
|
|
|
if (cmdline_find_option_bool(boot_command_line, "nofxsr")) {
|
|
|
|
setup_clear_cpu_cap(X86_FEATURE_FXSR);
|
|
|
|
setup_clear_cpu_cap(X86_FEATURE_FXSR_OPT);
|
|
|
|
setup_clear_cpu_cap(X86_FEATURE_XMM);
|
|
|
|
}
|
|
|
|
|
|
|
|
if (cmdline_find_option_bool(boot_command_line, "noxsave"))
|
|
|
|
fpu__xstate_clear_all_cpu_caps();
|
|
|
|
|
|
|
|
if (cmdline_find_option_bool(boot_command_line, "noxsaveopt"))
|
|
|
|
setup_clear_cpu_cap(X86_FEATURE_XSAVEOPT);
|
|
|
|
|
|
|
|
if (cmdline_find_option_bool(boot_command_line, "noxsaves"))
|
|
|
|
setup_clear_cpu_cap(X86_FEATURE_XSAVES);
|
|
|
|
}
|
|
|
|
|
x86/fpu: Split fpu__cpu_init() into early-boot and cpu-boot parts
There are two kinds of FPU initialization sequences necessary to bring FPU
functionality up: once per system bootup activities, such as detection,
feature initialization, etc. of attributes that are shared by all CPUs
in the system - and per cpu initialization sequences run when a CPU is
brought online (either during bootup or during CPU hotplug onlining),
such as CR0/CR4 register setting, etc.
The FPU code is mixing these roles together, with no clear distinction.
Start sorting this out by splitting the main FPU detection routine
(fpu__cpu_init()) into two parts: fpu__init_system() for
one per system init activities, and fpu__init_cpu() for the
per CPU onlining init activities.
Note that xstate_init() is called from both variants for the time being,
because it has a dual nature as well. We'll fix that in upcoming patches.
Just do the split and call it as we used to before, don't introduce any
change in initialization behavior yet, beyond duplicate (and harmless)
fpu__init_cpu() and xstate_init() calls - which we'll fix in later
patches.
Reviewed-by: Borislav Petkov <bp@alien8.de>
Cc: Andy Lutomirski <luto@amacapital.net>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: Fenghua Yu <fenghua.yu@intel.com>
Cc: H. Peter Anvin <hpa@zytor.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
2015-04-25 09:34:48 +07:00
|
|
|
/*
|
2015-04-26 20:36:46 +07:00
|
|
|
* Called on the boot CPU once per system bootup, to set up the initial
|
|
|
|
* FPU state that is later cloned into all processes:
|
x86/fpu: Split fpu__cpu_init() into early-boot and cpu-boot parts
There are two kinds of FPU initialization sequences necessary to bring FPU
functionality up: once per system bootup activities, such as detection,
feature initialization, etc. of attributes that are shared by all CPUs
in the system - and per cpu initialization sequences run when a CPU is
brought online (either during bootup or during CPU hotplug onlining),
such as CR0/CR4 register setting, etc.
The FPU code is mixing these roles together, with no clear distinction.
Start sorting this out by splitting the main FPU detection routine
(fpu__cpu_init()) into two parts: fpu__init_system() for
one per system init activities, and fpu__init_cpu() for the
per CPU onlining init activities.
Note that xstate_init() is called from both variants for the time being,
because it has a dual nature as well. We'll fix that in upcoming patches.
Just do the split and call it as we used to before, don't introduce any
change in initialization behavior yet, beyond duplicate (and harmless)
fpu__init_cpu() and xstate_init() calls - which we'll fix in later
patches.
Reviewed-by: Borislav Petkov <bp@alien8.de>
Cc: Andy Lutomirski <luto@amacapital.net>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: Fenghua Yu <fenghua.yu@intel.com>
Cc: H. Peter Anvin <hpa@zytor.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
2015-04-25 09:34:48 +07:00
|
|
|
*/
|
2015-05-04 14:52:42 +07:00
|
|
|
void __init fpu__init_system(struct cpuinfo_x86 *c)
|
x86/fpu: Split fpu__cpu_init() into early-boot and cpu-boot parts
There are two kinds of FPU initialization sequences necessary to bring FPU
functionality up: once per system bootup activities, such as detection,
feature initialization, etc. of attributes that are shared by all CPUs
in the system - and per cpu initialization sequences run when a CPU is
brought online (either during bootup or during CPU hotplug onlining),
such as CR0/CR4 register setting, etc.
The FPU code is mixing these roles together, with no clear distinction.
Start sorting this out by splitting the main FPU detection routine
(fpu__cpu_init()) into two parts: fpu__init_system() for
one per system init activities, and fpu__init_cpu() for the
per CPU onlining init activities.
Note that xstate_init() is called from both variants for the time being,
because it has a dual nature as well. We'll fix that in upcoming patches.
Just do the split and call it as we used to before, don't introduce any
change in initialization behavior yet, beyond duplicate (and harmless)
fpu__init_cpu() and xstate_init() calls - which we'll fix in later
patches.
Reviewed-by: Borislav Petkov <bp@alien8.de>
Cc: Andy Lutomirski <luto@amacapital.net>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: Fenghua Yu <fenghua.yu@intel.com>
Cc: H. Peter Anvin <hpa@zytor.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
2015-04-25 09:34:48 +07:00
|
|
|
{
|
x86/fpu: Fix early FPU command-line parsing
The function fpu__init_system() is executed before
parse_early_param(). This causes wrong FPU configuration. This
patch fixes this issue by parsing boot_command_line in the
beginning of fpu__init_system().
With all four patches in this series, each parameter disables
features as the following:
eagerfpu=off: eagerfpu, avx, avx2, avx512, mpx
no387: fpu
nofxsr: fxsr, fxsropt, xmm
noxsave: xsave, xsaveopt, xsaves, xsavec, avx, avx2, avx512,
mpx, xgetbv1 noxsaveopt: xsaveopt
noxsaves: xsaves
Signed-off-by: Yu-cheng Yu <yu-cheng.yu@intel.com>
Cc: Andy Lutomirski <luto@amacapital.net>
Cc: Borislav Petkov <bp@alien8.de>
Cc: Borislav Petkov <bp@suse.de>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: Fenghua Yu <fenghua.yu@intel.com>
Cc: H. Peter Anvin <hpa@zytor.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Quentin Casasnovas <quentin.casasnovas@oracle.com>
Cc: Ravi V. Shankar <ravi.v.shankar@intel.com>
Cc: Sai Praneeth Prakhya <sai.praneeth.prakhya@intel.com>
Cc: Thomas Gleixner <tglx@linutronix.de>
Cc: yu-cheng yu <yu-cheng.yu@intel.com>
Link: http://lkml.kernel.org/r/1452119094-7252-2-git-send-email-yu-cheng.yu@intel.com
Signed-off-by: Ingo Molnar <mingo@kernel.org>
2016-01-07 05:24:51 +07:00
|
|
|
fpu__init_parse_early_param();
|
2015-04-26 20:07:18 +07:00
|
|
|
fpu__init_system_early_generic(c);
|
|
|
|
|
2015-04-26 20:36:46 +07:00
|
|
|
/*
|
|
|
|
* The FPU has to be operational for some of the
|
|
|
|
* later FPU init activities:
|
|
|
|
*/
|
x86/fpu: Split fpu__cpu_init() into early-boot and cpu-boot parts
There are two kinds of FPU initialization sequences necessary to bring FPU
functionality up: once per system bootup activities, such as detection,
feature initialization, etc. of attributes that are shared by all CPUs
in the system - and per cpu initialization sequences run when a CPU is
brought online (either during bootup or during CPU hotplug onlining),
such as CR0/CR4 register setting, etc.
The FPU code is mixing these roles together, with no clear distinction.
Start sorting this out by splitting the main FPU detection routine
(fpu__cpu_init()) into two parts: fpu__init_system() for
one per system init activities, and fpu__init_cpu() for the
per CPU onlining init activities.
Note that xstate_init() is called from both variants for the time being,
because it has a dual nature as well. We'll fix that in upcoming patches.
Just do the split and call it as we used to before, don't introduce any
change in initialization behavior yet, beyond duplicate (and harmless)
fpu__init_cpu() and xstate_init() calls - which we'll fix in later
patches.
Reviewed-by: Borislav Petkov <bp@alien8.de>
Cc: Andy Lutomirski <luto@amacapital.net>
Cc: Dave Hansen <dave.hansen@linux.intel.com>
Cc: Fenghua Yu <fenghua.yu@intel.com>
Cc: H. Peter Anvin <hpa@zytor.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Oleg Nesterov <oleg@redhat.com>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Ingo Molnar <mingo@kernel.org>
2015-04-25 09:34:48 +07:00
|
|
|
fpu__init_cpu();
|
2015-04-22 15:53:34 +07:00
|
|
|
|
2015-04-25 13:27:44 +07:00
|
|
|
/*
|
2015-04-26 20:36:46 +07:00
|
|
|
* But don't leave CR0::TS set yet, as some of the FPU setup
|
|
|
|
* methods depend on being able to execute FPU instructions
|
|
|
|
* that will fault on a set TS, such as the FXSAVE in
|
|
|
|
* fpu__init_system_mxcsr().
|
2015-04-25 13:27:44 +07:00
|
|
|
*/
|
|
|
|
clts();
|
|
|
|
|
2015-04-26 19:35:54 +07:00
|
|
|
fpu__init_system_generic();
|
2015-04-26 20:23:37 +07:00
|
|
|
fpu__init_system_xstate_size_legacy();
|
2015-04-25 11:52:53 +07:00
|
|
|
fpu__init_system_xstate();
|
2015-07-17 17:28:12 +07:00
|
|
|
fpu__init_task_struct_size();
|
2015-04-26 15:35:57 +07:00
|
|
|
|
2015-04-26 13:28:31 +07:00
|
|
|
fpu__init_system_ctx_switch();
|
2015-04-22 15:53:34 +07:00
|
|
|
}
|