Files
Thomas Gleixner 5eede802d4 sched: define TIF_ALLOW_RESCHED
BugLink: https://bugs.launchpad.net/bugs/2060704

On Fri, Sep 22 2023 at 00:55, Thomas Gleixner wrote:
> On Thu, Sep 21 2023 at 09:00, Linus Torvalds wrote:
>> That said - I think as a proof of concept and "look, with this we get
>> the expected scheduling event counts", that patch is perfect. I think
>> you more than proved the concept.
>
> There is certainly quite some analyis work to do to make this a one to
> one replacement.
>
> With a handful of benchmarks the PoC (tweaked with some obvious fixes)
> is pretty much on par with the current mainline variants (NONE/FULL),
> but the memtier benchmark makes a massive dent.
>
> It sports a whopping 10% regression with the LAZY mode versus the mainline
> NONE model. Non-LAZY and FULL behave unsurprisingly in the same way.
>
> That benchmark is really sensitive to the preemption model. With current
> mainline (DYNAMIC_PREEMPT enabled) the preempt=FULL model has ~20%
> performance drop versus preempt=NONE.

That 20% was a tired pilot error. The real number is in the 5% ballpark.

> I have no clue what's going on there yet, but that shows that there is
> obviously quite some work ahead to get this sorted.

It took some head scratching to figure that out. The initial fix broke
the handling of the hog issue, i.e. the problem that Ankur tried to
solve, but I hacked up a "solution" for that too.

With that the memtier benchmark is roughly back to the mainline numbers,
but my throughput benchmark know how is pretty close to zero, so that
should be looked at by people who actually understand these things.

Likewise the hog prevention is just at the PoC level and clearly beyond
my knowledge of scheduler details: It unconditionally forces a
reschedule when the looping task is not responding to a lazy reschedule
request before the next tick. IOW it forces a reschedule on the second
tick, which is obviously different from the cond_resched()/might_sleep()
behaviour.

The changes vs. the original PoC aside of the bug and thinko fixes:

    1) A hack to utilize the TRACE_FLAG_IRQS_NOSUPPORT flag to trace the
       lazy preempt bit as the trace_entry::flags field is full already.

       That obviously breaks the tracer ABI, but if we go there then
       this needs to be fixed. Steven?

    2) debugfs file to validate that loops can be force preempted w/o
       cond_resched()

       The usage is:

       # taskset -c 1 bash
       # echo 1 > /sys/kernel/debug/sched/hog &
       # echo 1 > /sys/kernel/debug/sched/hog &
       # echo 1 > /sys/kernel/debug/sched/hog &

       top shows ~33% CPU for each of the hogs and tracing confirms that
       the crude hack in the scheduler tick works:

            bash-4559    [001] dlh2.  2253.331202: resched_curr <-__update_curr
            bash-4560    [001] dlh2.  2253.340199: resched_curr <-__update_curr
            bash-4561    [001] dlh2.  2253.346199: resched_curr <-__update_curr
            bash-4559    [001] dlh2.  2253.353199: resched_curr <-__update_curr
            bash-4561    [001] dlh2.  2253.358199: resched_curr <-__update_curr
            bash-4560    [001] dlh2.  2253.370202: resched_curr <-__update_curr
            bash-4559    [001] dlh2.  2253.378198: resched_curr <-__update_curr
            bash-4561    [001] dlh2.  2253.389199: resched_curr <-__update_curr

       The 'l' instead of the usual 'N' reflects that the lazy resched
       bit is set. That makes __update_curr() invoke resched_curr()
       instead of the lazy variant. resched_curr() sets TIF_NEED_RESCHED
       and folds it into preempt_count so that preemption happens at the
       next possible point, i.e. either in return from interrupt or at
       the next preempt_enable().

That's as much as I wanted to demonstrate and I'm not going to spend
more cycles on it as I have already too many other things on flight and
the resulting scheduler woes are clearly outside of my expertice.

Though definitely I'm putting a permanent NAK in place for any attempts
to duct tape the preempt=NONE model any further by sprinkling more
cond*() and whatever warts around.

Thanks,

        tglx

[tglx: s@CONFIG_PREEMPT_AUTO@CONFIG_PREEMPT_BUILD_AUTO@ ]

Link: https://lore.kernel.org/all/87jzshhexi.ffs@tglx/
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
Signed-off-by: Sebastian Andrzej Siewior <bigeasy@linutronix.de>
Signed-off-by: Kevin Becker <kevin.becker@canonical.com>
2025-06-25 11:17:16 -04:00

242 lines
7.9 KiB
C

/* SPDX-License-Identifier: GPL-2.0 */
/* thread_info.h: low-level thread information
*
* Copyright (C) 2002 David Howells (dhowells@redhat.com)
* - Incorporating suggestions made by Linus Torvalds and Dave Miller
*/
#ifndef _ASM_X86_THREAD_INFO_H
#define _ASM_X86_THREAD_INFO_H
#include <linux/compiler.h>
#include <asm/page.h>
#include <asm/percpu.h>
#include <asm/types.h>
/*
* TOP_OF_KERNEL_STACK_PADDING is a number of unused bytes that we
* reserve at the top of the kernel stack. We do it because of a nasty
* 32-bit corner case. On x86_32, the hardware stack frame is
* variable-length. Except for vm86 mode, struct pt_regs assumes a
* maximum-length frame. If we enter from CPL 0, the top 8 bytes of
* pt_regs don't actually exist. Ordinarily this doesn't matter, but it
* does in at least one case:
*
* If we take an NMI early enough in SYSENTER, then we can end up with
* pt_regs that extends above sp0. On the way out, in the espfix code,
* we can read the saved SS value, but that value will be above sp0.
* Without this offset, that can result in a page fault. (We are
* careful that, in this case, the value we read doesn't matter.)
*
* In vm86 mode, the hardware frame is much longer still, so add 16
* bytes to make room for the real-mode segments.
*
* x86_64 has a fixed-length stack frame.
*/
#ifdef CONFIG_X86_32
# ifdef CONFIG_VM86
# define TOP_OF_KERNEL_STACK_PADDING 16
# else
# define TOP_OF_KERNEL_STACK_PADDING 8
# endif
#else
# define TOP_OF_KERNEL_STACK_PADDING 0
#endif
/*
* low level task data that entry.S needs immediate access to
* - this struct should fit entirely inside of one cache line
* - this struct shares the supervisor stack pages
*/
#ifndef __ASSEMBLY__
struct task_struct;
#include <asm/cpufeature.h>
#include <linux/atomic.h>
struct thread_info {
unsigned long flags; /* low level flags */
unsigned long syscall_work; /* SYSCALL_WORK_ flags */
u32 status; /* thread synchronous flags */
#ifdef CONFIG_SMP
u32 cpu; /* current CPU */
#endif
};
#define INIT_THREAD_INFO(tsk) \
{ \
.flags = 0, \
}
#else /* !__ASSEMBLY__ */
#include <asm/asm-offsets.h>
#endif
/*
* thread information flags
* - these are process state flags that various assembly files
* may need to access
*/
#define TIF_NOTIFY_RESUME 1 /* callback before returning to user */
#define TIF_SIGPENDING 2 /* signal pending */
#define TIF_NEED_RESCHED 3 /* rescheduling necessary */
#define TIF_ARCH_RESCHED_LAZY 4 /* Lazy rescheduling */
#define TIF_SINGLESTEP 5 /* reenable singlestep on user return*/
#define TIF_SSBD 6 /* Speculative store bypass disable */
#define TIF_SPEC_IB 9 /* Indirect branch speculation mitigation */
#define TIF_SPEC_L1D_FLUSH 10 /* Flush L1D on mm switches (processes) */
#define TIF_USER_RETURN_NOTIFY 11 /* notify kernel of userspace return */
#define TIF_UPROBE 12 /* breakpointed or singlestepping */
#define TIF_PATCH_PENDING 13 /* pending live patching update */
#define TIF_NEED_FPU_LOAD 14 /* load FPU on return to userspace */
#define TIF_NOCPUID 15 /* CPUID is not accessible in userland */
#define TIF_NOTSC 16 /* TSC is not accessible in userland */
#define TIF_NOTIFY_SIGNAL 17 /* signal notifications exist */
#define TIF_MEMDIE 20 /* is terminating due to OOM killer */
#define TIF_POLLING_NRFLAG 21 /* idle is polling for TIF_NEED_RESCHED */
#define TIF_IO_BITMAP 22 /* uses I/O bitmap */
#define TIF_SPEC_FORCE_UPDATE 23 /* Force speculation MSR update in context switch */
#define TIF_FORCED_TF 24 /* true if TF in eflags artificially */
#define TIF_BLOCKSTEP 25 /* set when we want DEBUGCTLMSR_BTF */
#define TIF_LAZY_MMU_UPDATES 27 /* task is updating the mmu lazily */
#define TIF_ADDR32 29 /* 32-bit address space on 64 bits */
#define _TIF_NOTIFY_RESUME (1 << TIF_NOTIFY_RESUME)
#define _TIF_SIGPENDING (1 << TIF_SIGPENDING)
#define _TIF_NEED_RESCHED (1 << TIF_NEED_RESCHED)
#define _TIF_ARCH_RESCHED_LAZY (1 << TIF_ARCH_RESCHED_LAZY)
#define _TIF_SINGLESTEP (1 << TIF_SINGLESTEP)
#define _TIF_SSBD (1 << TIF_SSBD)
#define _TIF_SPEC_IB (1 << TIF_SPEC_IB)
#define _TIF_SPEC_L1D_FLUSH (1 << TIF_SPEC_L1D_FLUSH)
#define _TIF_USER_RETURN_NOTIFY (1 << TIF_USER_RETURN_NOTIFY)
#define _TIF_UPROBE (1 << TIF_UPROBE)
#define _TIF_PATCH_PENDING (1 << TIF_PATCH_PENDING)
#define _TIF_NEED_FPU_LOAD (1 << TIF_NEED_FPU_LOAD)
#define _TIF_NOCPUID (1 << TIF_NOCPUID)
#define _TIF_NOTSC (1 << TIF_NOTSC)
#define _TIF_NOTIFY_SIGNAL (1 << TIF_NOTIFY_SIGNAL)
#define _TIF_POLLING_NRFLAG (1 << TIF_POLLING_NRFLAG)
#define _TIF_IO_BITMAP (1 << TIF_IO_BITMAP)
#define _TIF_SPEC_FORCE_UPDATE (1 << TIF_SPEC_FORCE_UPDATE)
#define _TIF_FORCED_TF (1 << TIF_FORCED_TF)
#define _TIF_BLOCKSTEP (1 << TIF_BLOCKSTEP)
#define _TIF_LAZY_MMU_UPDATES (1 << TIF_LAZY_MMU_UPDATES)
#define _TIF_ADDR32 (1 << TIF_ADDR32)
/* flags to check in __switch_to() */
#define _TIF_WORK_CTXSW_BASE \
(_TIF_NOCPUID | _TIF_NOTSC | _TIF_BLOCKSTEP | \
_TIF_SSBD | _TIF_SPEC_FORCE_UPDATE)
/*
* Avoid calls to __switch_to_xtra() on UP as STIBP is not evaluated.
*/
#ifdef CONFIG_SMP
# define _TIF_WORK_CTXSW (_TIF_WORK_CTXSW_BASE | _TIF_SPEC_IB)
#else
# define _TIF_WORK_CTXSW (_TIF_WORK_CTXSW_BASE)
#endif
#ifdef CONFIG_X86_IOPL_IOPERM
# define _TIF_WORK_CTXSW_PREV (_TIF_WORK_CTXSW| _TIF_USER_RETURN_NOTIFY | \
_TIF_IO_BITMAP)
#else
# define _TIF_WORK_CTXSW_PREV (_TIF_WORK_CTXSW| _TIF_USER_RETURN_NOTIFY)
#endif
#define _TIF_WORK_CTXSW_NEXT (_TIF_WORK_CTXSW)
#define STACK_WARN (THREAD_SIZE/8)
/*
* macros/functions for gaining access to the thread information structure
*
* preempt_count needs to be 1 initially, until the scheduler is functional.
*/
#ifndef __ASSEMBLY__
/*
* Walks up the stack frames to make sure that the specified object is
* entirely contained by a single stack frame.
*
* Returns:
* GOOD_FRAME if within a frame
* BAD_STACK if placed across a frame boundary (or outside stack)
* NOT_STACK unable to determine (no frame pointers, etc)
*
* This function reads pointers from the stack and dereferences them. The
* pointers may not have their KMSAN shadow set up properly, which may result
* in false positive reports. Disable instrumentation to avoid those.
*/
__no_kmsan_checks
static inline int arch_within_stack_frames(const void * const stack,
const void * const stackend,
const void *obj, unsigned long len)
{
#if defined(CONFIG_FRAME_POINTER)
const void *frame = NULL;
const void *oldframe;
oldframe = __builtin_frame_address(1);
if (oldframe)
frame = __builtin_frame_address(2);
/*
* low ----------------------------------------------> high
* [saved bp][saved ip][args][local vars][saved bp][saved ip]
* ^----------------^
* allow copies only within here
*/
while (stack <= frame && frame < stackend) {
/*
* If obj + len extends past the last frame, this
* check won't pass and the next frame will be 0,
* causing us to bail out and correctly report
* the copy as invalid.
*/
if (obj + len <= frame)
return obj >= oldframe + 2 * sizeof(void *) ?
GOOD_FRAME : BAD_STACK;
oldframe = frame;
frame = *(const void * const *)frame;
}
return BAD_STACK;
#else
return NOT_STACK;
#endif
}
#endif /* !__ASSEMBLY__ */
/*
* Thread-synchronous status.
*
* This is different from the flags in that nobody else
* ever touches our thread-synchronous status, so we don't
* have to worry about atomic accesses.
*/
#define TS_COMPAT 0x0002 /* 32bit syscall active (64BIT)*/
#ifndef __ASSEMBLY__
#ifdef CONFIG_COMPAT
#define TS_I386_REGS_POKED 0x0004 /* regs poked by 32-bit ptracer */
#define arch_set_restart_data(restart) \
do { restart->arch_data = current_thread_info()->status; } while (0)
#endif
#ifdef CONFIG_X86_32
#define in_ia32_syscall() true
#else
#define in_ia32_syscall() (IS_ENABLED(CONFIG_IA32_EMULATION) && \
current_thread_info()->status & TS_COMPAT)
#endif
extern void arch_setup_new_exec(void);
#define arch_setup_new_exec arch_setup_new_exec
#endif /* !__ASSEMBLY__ */
#endif /* _ASM_X86_THREAD_INFO_H */