Skip to content

R
raspberrypi-kernel

openeuler / raspberrypi-kernel

通知 13
Star 1
Fork 0
R
raspberrypi-kernel
切换分支/标签
  1. 12 1月, 2010 1 次提交
  3. 25 11月, 2009 1 次提交
  5. 24 11月, 2009 2 次提交
    • S
      sh: Minor optimisations to FPU handling · d3ea9fa0
      由 Stuart Menefy 提交于 
      A number of small optimisations to FPU handling, in particular:

 - move the task USEDFPU flag from the thread_info flags field (which
   is accessed asynchronously to the thread) to a new status field,
   which is only accessed by the thread itself. This allows locking to
   be removed in most cases, or can be reduced to a preempt_lock().
   This mimics the i386 behaviour.

 - move the modification of regs->sr and thread_info->status flags out
   of save_fpu() to __unlazy_fpu(). This gives the compiler a better
   chance to optimise things, as well as making save_fpu() symmetrical
   with restore_fpu() and init_fpu().

 - implement prepare_to_copy(), so that when creating a thread, we can
   unlazy the FPU prior to copying the thread data structures.

Also make sure that the FPU is disabled while in the kernel, in
particular while booting, and for newly created kernel threads,

In a very artificial benchmark, the execution time for 2500000
context switches was reduced from 50 to 45 seconds.
Signed-off-by: NStuart Menefy <stuart.menefy@st.com>
Signed-off-by: NPaul Mundt <lethal@linux-sh.org>
      d3ea9fa0
    • G
      sh: add sleazy FPU optimization · a0458b07
      由 Giuseppe CAVALLARO 提交于 
      sh port of the sLeAZY-fpu feature currently implemented for some architectures
such us i386.

Right now the SH kernel has a 100% lazy fpu behaviour.
This is of course great for applications that have very sporadic or no FPU use.
However for very frequent FPU users...  you take an extra trap every context
switch.
The patch below adds a simple heuristic to this code: after 5 consecutive
context switches of FPU use, the lazy behavior is disabled and the context
gets restored every context switch.
After 256 switches, this is reset and the 100% lazy behavior is returned.

Tests with LMbench showed no regression.
I saw a little improvement due to the prefetching (~2%).

The tests below also show that, with this sLeazy patch, indeed,
the number of FPU exceptions is reduced.
To test this. I hacked the lat_ctx LMBench to use the FPU a little more.

   sLeasy implementation
   ===========================================
   switch_to calls            |  79326
   sleasy   calls             |  42577
   do_fpu_state_restore  calls|  59232
   restore_fpu   calls        |  59032

   Exceptions:  0x800 (FPU disabled  ): 16604

   100% Leazy (default implementation)
   ===========================================
   switch_to  calls            |  79690
   do_fpu_state_restore calls  |  53299
   restore_fpu  calls          |   53101

   Exceptions: 0x800 (FPU disabled  ):  53273
Signed-off-by: NGiuseppe Cavallaro <peppe.cavallaro@st.com>
Signed-off-by: NStuart Menefy <stuart.menefy@st.com>
Signed-off-by: NPaul Mundt <lethal@linux-sh.org>
      a0458b07
  7. 27 10月, 2009 1 次提交
  9. 24 8月, 2009 1 次提交
  11. 11 7月, 2009 1 次提交
    • M
      sh: Mark __switch_to() as __notrace_funcgraph · 7816fecd
      由 Matt Fleming 提交于 
      Annotate __switch_to() so that the function graph tracer does not try to
trace it. Use __notrace_funcgraph, as opposed to notrace, so that other
tracers can continue to trace __switch_to().

The reason that we don't want to trace __switch_to() with the function
graph tracer is because of how the return address stack in task_struct
is implemented. When we enter __switch_to we store the real return
address on prev's ret_stack. When we return from __switch_to() we've
patched the return address on the kernel stack to be
return_to_handler. Calling return_to_handler we do,

       -> ftrace_return_to_handler()
       	  -> ftrace_pop_return_ftrace()

Which tries to pop the real return address from current->ret_stack. The
problem being that we stored the return address on prev->ret_stack, but
current now points to next, and next->ret_stack doesn't contain the
correct return address (and is possibly even empty).
Signed-off-by: NMatt Fleming <matt@console-pimps.org>
Signed-off-by: NPaul Mundt <lethal@linux-sh.org>
      7816fecd
  13. 19 6月, 2009 1 次提交
  15. 18 6月, 2009 1 次提交
  17. 08 5月, 2009 1 次提交
  19. 04 4月, 2009 1 次提交
    • M
      sh: Fix up DSP context save/restore. · 01ab1039
      由 Michael Trimarchi 提交于 
      There were a number of issues with the DSP context save/restore code,
mostly left-over relics from when it was introduced on SH3-DSP with
little follow-up testing, resulting in things like task_pt_dspregs()
referencing incorrect state on the stack.

This follows the MIPS convention of tracking the DSP state in the
thread_struct and handling the state save/restore in switch_to() and
finish_arch_switch() respectively. The regset interface is also updated,
which allows us to finally be rid of task_pt_dspregs() and the special
cased task_pt_regs().
Signed-off-by: NMichael Trimarchi <michael@evidence.eu.com>
Signed-off-by: NPaul Mundt <lethal@linux-sh.org>
      01ab1039
  21. 03 4月, 2009 1 次提交
  23. 22 12月, 2008 5 次提交
  25. 21 9月, 2008 2 次提交
    • P
      sh: Add FPU registers to regset interface. · e7ab3cd2
      由 Paul Mundt 提交于 
      Signed-off-by: NPaul Mundt <lethal@linux-sh.org>
      e7ab3cd2
    • P
      sh: Trivial trace_mark() instrumentation for core events. · 3d58695e
      由 Paul Mundt 提交于 
      This implements a few trace points across events that are deemed
interesting. This implements a number of trace points:

	- The page fault handler / TLB miss
	- IPC calls
	- Kernel thread creation

The original LTTng patch had the slow-path instrumented, which
fails to account for the vast majority of events. In general
placing this in the fast-path is not a huge performance hit, as
we don't take page faults for kernel addresses.

The other bits of interest are some of the other trap handlers, as
well as the syscall entry/exit (which is better off being handled
through the tracehook API). Most of the other trap handlers are corner
cases where alternate means of notification exist, so there is little
value in placing extra trace points in these locations.

Based on top of the points provided both by the LTTng instrumentation
patch as well as the patch shipping in the ST-Linux tree, albeit in a
stripped down form.
Signed-off-by: NPaul Mundt <lethal@linux-sh.org>
      3d58695e
  27. 08 9月, 2008 2 次提交
  29. 28 7月, 2008 1 次提交
    • A
      sh/kernel/ cleanups · 4c1cfab1
      由 Adrian Bunk 提交于 
      This patch contains the following cleanups:
- make the following needlessly global code static:
  - cf-enabler.c: cf_init()
  - cpu/clock.c: __clk_enable()
  - cpu/clock.c: __clk_disable()
  - process_32.c: default_idle()
  - time_32.c: struct clocksource_sh
  - timers/timer-tmu.c: struct tmu_timer_ops
- remove the following unused functions (no CONFIG_BLK_DEV_FD on sh):
  - process_{32,64}.c: disable_hlt()
  - process_{32,64}.c: enable_hlt()
Signed-off-by: NAdrian Bunk <bunk@kernel.org>
Signed-off-by: NPaul Mundt <lethal@linux-sh.org>
      4c1cfab1
  31. 19 7月, 2008 1 次提交
    • T
      nohz: prevent tick stop outside of the idle loop · b8f8c3cf
      由 Thomas Gleixner 提交于 
      Jack Ren and Eric Miao tracked down the following long standing
problem in the NOHZ code:

	scheduler switch to idle task
	enable interrupts

Window starts here

	----> interrupt happens (does not set NEED_RESCHED)
	      	irq_exit() stops the tick

	----> interrupt happens (does set NEED_RESCHED)

	return from schedule()
	
	cpu_idle(): preempt_disable();

Window ends here

The interrupts can happen at any point inside the race window. The
first interrupt stops the tick, the second one causes the scheduler to
rerun and switch away from idle again and we end up with the tick
disabled.

The fact that it needs two interrupts where the first one does not set
NEED_RESCHED and the second one does made the bug obscure and extremly
hard to reproduce and analyse. Kudos to Jack and Eric.

Solution: Limit the NOHZ functionality to the idle loop to make sure
that we can not run into such a situation ever again.

cpu_idle()
{
	preempt_disable();

	while(1) {
		 tick_nohz_stop_sched_tick(1); <- tell NOHZ code that we
		 			          are in the idle loop

		 while (!need_resched())
		       halt();

		 tick_nohz_restart_sched_tick(); <- disables NOHZ mode
		 preempt_enable_no_resched();
		 schedule();
		 preempt_disable();
	}
}

In hindsight we should have done this forever, but ... 

/me grabs a large brown paperbag.

Debugged-by: Jack Ren <jack.ren@marvell.com>, 
Debugged-by: Neric miao <eric.y.miao@gmail.com>
Signed-off-by: NThomas Gleixner <tglx@linutronix.de>
      b8f8c3cf
  33. 26 3月, 2008 1 次提交
  35. 28 1月, 2008 4 次提交
  37. 20 10月, 2007 1 次提交
  39. 28 9月, 2007 2 次提交
  41. 31 7月, 2007 1 次提交
  43. 26 7月, 2007 1 次提交
  45. 11 6月, 2007 1 次提交
  47. 08 6月, 2007 2 次提交
  49. 21 5月, 2007 1 次提交
    • P
      sh: sr.bl toggling around idle sleep. · f3a9022f
      由 Paul Mundt 提交于 
      As pointed out by Saito-san, without the sr.bl manipulation we can
occasionally hit delays in the idle loop due to interrupt handling, so
ensure that interrupts are blocked before going to sleep.

At the same time, we throw in TIF_POLLING_NRFLAG for the !hlt_counter
case (primarily used by the ST-40 parts).
Signed-off-by: NPaul Mundt <lethal@linux-sh.org>
      f3a9022f
  51. 09 5月, 2007 3 次提交
    • P
      sh: clockevent/clocksource/hrtimers/nohz TMU support. · 57be2b48
      由 Paul Mundt 提交于 
      This adds basic support for clockevents and clocksources,
presently only implemented for TMU-based systems (which
are the majority of SH-3 and SH-4 systems).

The old NO_IDLE_HZ implementation is also dropped completely,
the only users of this were on TMU-based systems anyways.

More work needs to be done to generalize the TMU handling,
in that the current implementation is rather tied to the
notion of TMU0 and TMU1 utilization.

Additionally, as more SH timers switch over to this scheme,
we'll be able to gut most of the remaining system timer
infrastructure that existed before.
Signed-off-by: NPaul Mundt <lethal@linux-sh.org>
      57be2b48
    • P
      sh: Convert to common die chain. · b118ca57
      由 Paul Mundt 提交于 
      This went in immediately after SH added the die chain notifiers,
so move over to that instead..
Signed-off-by: NPaul Mundt <lethal@linux-sh.org>
      b118ca57
    • P
      sh: Fix PC adjustments for varying opcode length. · 53f983a9
      由 Paul Mundt 提交于 
      There are a few different cases for figuring out how to
size the instruction. We read in the instruction located
at regs->pc - 4 when rewinding the opcode to figure out if
there's a 32-bit opcode before the faulting instruction, with
a default of a - 2 adjustment on a mismatch. In practice this
works for the cases where pc - 4 is just another 16-bit opcode,
or we happen to have a 32-bit and a 16-bit immediately
preceeding the pc value.

In the cases where we aren't rewinding, this is much less ugly..

We also don't bother fixing up the places where we're explicitly
dealing with 16-bit instructions, since this might lead to
confusion regarding the encoding size possibilities on other
CPU variants.
Signed-off-by: NPaul Mundt <lethal@linux-sh.org>
      53f983a9