AcoSail/kernel_optimize_test
8
0
Fork 0
forked from luck/tmp_suning_uos_patched
Code Pull Requests Activity
a0458b07c1
kernel_optimize_test/arch/sh/include/asm
History
Giuseppe CAVALLARO a0458b07c1 sh: add sleazy FPU optimization 
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: Giuseppe Cavallaro <peppe.cavallaro@st.com>
Signed-off-by: Stuart Menefy <stuart.menefy@st.com>
Signed-off-by: Paul Mundt <lethal@linux-sh.org>
2009-11-24 16:23:38 +09:00
..
.gitignore
adc.h
addrspace.h sh: Support fixed 32-bit PMB mappings from bootloader. 2009-03-10 15:49:54 +09:00
atomic-grb.h
atomic-irq.h sh: Make the atomic functions safe for irqsoff tracing 2009-06-15 00:31:17 +09:00
atomic-llsc.h sh: Fix sh4a llsc-based cmpxchg() 2009-06-11 09:31:55 +03:00
atomic.h sh: Use generic atomic64_t implementation. 2009-06-17 10:43:13 +09:00
auxvec.h
bitops-grb.h sh: Convert to generic bitops for IRQ-toggling implementation. 2008-12-22 18:42:53 +09:00
bitops-llsc.h sh: Relax inline assembly constraints 2009-01-29 11:57:09 +09:00
bitops-op32.h sh: Provide optimized non-atomic bitops for SH-2A. 2008-12-22 18:42:54 +09:00
bitops.h sh: Provide optimized non-atomic bitops for SH-2A. 2008-12-22 18:42:54 +09:00
bitsperlong.h asm-generic: introduce asm/bitsperlong.h 2009-06-11 21:02:14 +02:00
bug.h sh64: dummy unwinder BUG wrappers. 2009-08-22 05:31:45 +09:00
bugs.h sh: Track the CPU family in sh_cpuinfo. 2009-08-15 10:48:13 +09:00
byteorder.h byteorder: make swab.h include asm/swab.h like a regular header 2009-01-14 19:56:50 -08:00
cache.h
cachectl.h sh: fix sys_cacheflush error checking 2009-08-24 18:59:09 +09:00
cacheflush.h sh: Fix up and optimize the kmap_coherent() interface. 2009-09-03 17:21:10 +09:00
checksum_32.h
checksum.h sh: Convert sh64 to use the generic checksum code. 2009-06-15 00:00:42 +09:00
clock.h sh: add enable()/disable()/set_rate() to div6 code 2009-06-11 09:12:58 +03:00
cmpxchg-grb.h
cmpxchg-irq.h
cmpxchg-llsc.h sh: Fix sh4a llsc-based cmpxchg() 2009-06-11 09:31:55 +03:00
cpu-features.h sh: Support for extended ASIDs on PTEAEX-capable SH-X3 cores. 2009-03-17 17:49:49 +09:00
cputime.h
current.h sh: Convert to asm-generic/current.h. 2009-06-14 21:34:26 +09:00
delay.h
device.h sh: Runtime PM for SuperH Mobile platform bus devices 2009-08-23 18:03:17 +09:00
div64.h
dma-mapping.h sh: Fix up more dma-mapping fallout. 2009-06-23 17:36:23 +09:00
dma-sh.h Merge branch 'next' of git://git.kernel.org/pub/scm/linux/kernel/git/djbw/async_tx into for-linus 2009-09-23 18:31:11 +10:00
dma.h sh: Convert to asm-generic/dma.h. 2009-06-14 21:45:06 +09:00
dmabrg.h
dwarf.h sh: unwinder: Fix up uninitialized variable warnings on sh2a build. 2009-08-31 13:47:06 +09:00
elf.h Merge branch 'for-linus' of git://git390.osdl.marist.edu/pub/scm/linux-2.6 2008-12-28 12:33:21 -08:00
emergency-restart.h
entry-macros.S sh: Remove implicit sign extension from assembler immediates 2009-08-24 17:09:53 +09:00
errno.h
fb.h
fcntl.h
fixmap.h
flat.h flat: fix data sections alignment 2009-05-29 08:40:02 -07:00
fpu.h sh: add sleazy FPU optimization 2009-11-24 16:23:38 +09:00
freq.h
ftrace.h sh: Fix an off-by-1 in FTRACE_SYSCALL_MAX. 2009-08-25 18:39:33 +09:00
futex-irq.h
futex.h
gpio.h sh: use gpiolib 2009-01-27 14:49:10 +09:00
hardirq.h sh: convert to asm-generic/hardirq.h 2009-08-05 17:56:07 +09:00
hd64461.h sh: mach-hp6xx: Fix up the hp6xx build for hd64461 changes. 2009-05-20 11:27:13 +09:00
heartbeat.h sh: Prevent heartbeat from scribbling over non-LED bits. 2009-08-18 21:16:29 +09:00
hugetlb.h
hw_irq.h sh: Move the shared INTC code out to drivers/sh/ 2008-10-01 16:13:54 +09:00
hwblk.h sh: Runtime PM for SuperH Mobile platform bus devices 2009-08-23 18:03:17 +09:00
i2c-sh7760.h
ilsel.h
io_generic.h sh: More I/O routine overhauling. 2008-10-04 05:25:52 +09:00
io_trapped.h
io.h sh: Read from CCN_PVR instead of ROM for delay. 2009-08-24 17:13:52 +09:00
ioctl.h
ioctls.h
ipcbuf.h sh: Convert ipc/shm bits to their asm-generic versions. 2009-06-14 23:20:27 +09:00
irq_regs.h
irq.h sh: Tidy up duplication in irq/swab/timex.h. 2009-06-14 23:21:54 +09:00
irqflags_32.h
irqflags_64.h
irqflags.h
Kbuild sh: Add sys_cacheflush() call for SH CPUs. 2009-08-24 18:16:56 +09:00
kdebug.h sh: nmi_debug support. 2009-09-01 17:38:32 +09:00
kexec.h kexec jump: rename KEXEC_CONTROL_CODE_SIZE to KEXEC_CONTROL_PAGE_SIZE 2008-08-15 08:35:42 -07:00
kgdb.h sh: Kill off kgdb's magical NMI debouncing. 2009-09-01 17:45:35 +09:00
kmap_types.h kmap_types: make most arches use generic header file 2009-06-16 19:47:51 -07:00
kprobes.h sh: Rename opcode_t to insn_size_t. 2009-05-09 16:02:08 +09:00
linkage.h
lmb.h sh: Use bootmem ontop of lmb 2009-07-03 16:16:54 +09:00
local.h
machvec.h sh: boot word / mode pin support V2 2009-06-01 15:44:49 +09:00
mc146818rtc.h
mman.h sh: Switch to asm-generic versions for identical headers. 2009-06-14 23:26:48 +09:00
mmu_context_32.h sh: Support for extended ASIDs on PTEAEX-capable SH-X3 cores. 2009-03-17 17:49:49 +09:00
mmu_context_64.h
mmu_context.h sh: Use local TLB flush for get_mmu_context() ASID exhaustion. 2009-08-20 17:24:40 +09:00
mmu.h sh: Sanitize asm/mmu.h for assembly use. 2009-03-31 07:42:37 +09:00
mmzone.h sh: Fix up NUMA build error with se7722_defconfig. 2008-09-12 22:41:30 +09:00
module.h sh: Convert to asm-generic/module.h. 2009-06-14 23:24:14 +09:00
msgbuf.h sh: Convert ipc/shm bits to their asm-generic versions. 2009-06-14 23:20:27 +09:00
mutex-llsc.h sh: Fix up T-bit error handling in SH-4A mutex fastpath. 2009-01-29 11:56:03 +09:00
mutex.h sh: Add SH-4A optimized fastpath mutex implementation. 2008-12-22 18:42:55 +09:00
page.h sh: Convert cache disabled SH-5 over to new cache interface. 2009-08-16 00:53:56 +09:00
param.h sh: Switch to asm-generic versions for identical headers. 2009-06-14 23:26:48 +09:00
parport.h sh: Switch to asm-generic versions for identical headers. 2009-06-14 23:26:48 +09:00
pci.h PCI: remove pcibios_scan_all_fns() 2009-09-09 13:29:18 -07:00
percpu.h
perf_event.h perf: Do the big rename: Performance Counters -> Performance Events 2009-09-21 14:28:04 +02:00
pgalloc.h mm: Remove duplicate definitions in MIPS and SH 2009-07-27 17:26:44 -07:00
pgtable_32.h sh: Fix up and optimize the kmap_coherent() interface. 2009-09-03 17:21:10 +09:00
pgtable_64.h sh: Fix up and optimize the kmap_coherent() interface. 2009-09-03 17:21:10 +09:00
pgtable.h sh: Provide the kmap_coherent() interface generically. 2009-08-15 09:19:19 +09:00
poll.h
posix_types_32.h sh: Convert to asm-generic/posix_types.h. 2009-06-14 23:16:27 +09:00
posix_types_64.h sh: Convert to asm-generic/posix_types.h. 2009-06-14 23:16:27 +09:00
posix_types.h
processor_32.h sh: Fix up DSP context save/restore. 2009-04-04 11:48:11 -04:00
processor_64.h SH: fix start_thread and user_stack_pointer macros 2009-01-29 15:41:15 +09:00
processor.h Merge branch 'master' into sh/smp 2009-09-01 13:54:14 +09:00
ptrace.h sh: Add in some ptrace definitions from GDB. 2009-05-09 00:06:03 +09:00
push-switch.h
resource.h
romimage-macros.h sh: add romimage-macros.h 2009-09-11 13:19:16 +09:00
rtc.h sh: rtc-generic support. 2009-04-27 17:34:39 +09:00
rwsem.h sh: Replace old style lock initializer 2009-11-09 10:47:40 +09:00
scatterlist.h sh: Convert to asm-generic/scatterlist.h. 2009-06-14 23:25:04 +09:00
seccomp.h sh: seccomp support. 2008-08-02 04:39:32 +09:00
sections.h sh: dwarf unwinder support. 2009-08-14 01:58:43 +09:00
segment.h
sembuf.h sh: Convert ipc/shm bits to their asm-generic versions. 2009-06-14 23:20:27 +09:00
serial.h sh: Switch to asm-generic versions for identical headers. 2009-06-14 23:26:48 +09:00
setup.h sh: Derive COMMAND_LINE_SIZE from asm-generic/setup.h. 2009-06-14 23:27:41 +09:00
sfp-machine.h
sh7760fb.h
sh_bios.h sh: Kill off sh_bios_in_gdb_mode(). 2008-12-22 18:44:44 +09:00
sh_eth.h net: sh_eth: add value of ether_link pin in platform_data 2009-08-29 00:19:35 -07:00
sh_keysc.h sh: add kycr2_delay for sh_keysc 2009-09-15 09:30:14 +09:00
shmbuf.h sh: Convert ipc/shm bits to their asm-generic versions. 2009-06-14 23:20:27 +09:00
shmparam.h
sigcontext.h
siginfo.h
signal.h sh: Convert to asm-generic/signal.h. 2009-06-14 23:25:57 +09:00
sizes.h sh: Add a few more definitions to asm/sizes.h. 2008-09-21 10:31:57 +09:00
smc37c93x.h
smp.h cpumask: remove arch_send_call_function_ipi 2009-09-24 09:34:47 +09:30
socket.h sh: Switch to asm-generic versions for identical headers. 2009-06-14 23:26:48 +09:00
sockios.h
sparsemem.h
spi.h
spinlock_types.h
spinlock.h sh: Fix sh4a llsc-based cmpxchg() 2009-06-11 09:31:55 +03:00
stacktrace.h sh: Use the generalized stacktrace ops 2009-08-13 11:50:08 +09:00
stat.h
statfs.h
string_32.h
string_64.h sh: Add SH-5 optimized memcpy()/memset()/strcpy()/strlen(). 2008-12-22 18:44:05 +09:00
string.h
suspend.h sh: cpuidle for SuperH Mobile using hwblk 2009-07-05 00:29:54 +09:00
swab.h sh: Tidy up duplication in irq/swab/timex.h. 2009-06-14 23:21:54 +09:00
syscall_32.h sh: Add ftrace syscall tracing support 2009-07-06 20:16:33 +09:00
syscall_64.h sh: Fix up spurious syscall restarting. 2009-01-29 11:56:03 +09:00
syscall.h sh: Provide the asm/syscall.h interface, needed by tracehook. 2008-09-12 22:13:13 +09:00
syscalls_32.h [CVE-2009-0029] Remove __attribute__((weak)) from sys_pipe/sys_pipe2 2009-01-14 14:15:15 +01:00
syscalls_64.h sh: fixup many sparse errors. 2008-09-08 10:35:04 +09:00
syscalls.h sh: fixup many sparse errors. 2008-09-08 10:35:04 +09:00
system_32.h Merge branch 'master' into sh/smp 2009-09-01 13:54:14 +09:00
system_64.h sh: Fix up the SH-5 build with caches enabled. 2009-08-16 01:50:17 +09:00
system.h sh: nmi_debug support. 2009-09-01 17:38:32 +09:00
termbits.h sh: Convert to asm-generic/termbits.h and termios.h. 2009-06-14 23:17:57 +09:00
termios.h sh: Convert to asm-generic/termbits.h and termios.h. 2009-06-14 23:17:57 +09:00
thread_info.h sh: Wire up HAVE_SYSCALL_TRACEPOINTS. 2009-09-16 14:30:34 +09:00
timex.h sh: Tidy up duplication in irq/swab/timex.h. 2009-06-14 23:21:54 +09:00
tlb_64.h sh: Fix up broken kerneldoc comments. 2008-08-04 12:51:06 +09:00
tlb.h mm: Pass virtual address to [__]p{te,ud,md}_free_tlb() 2009-07-27 12:10:38 -07:00
tlbflush.h
topology.h cpumask: remove obsolete node_to_cpumask now everyone uses cpumask_of_node 2009-09-24 09:34:34 +09:30
types.h sh: Add register alignment helpers for shared flushers. 2009-08-15 01:57:36 +09:00
uaccess_32.h
uaccess_64.h sh: sh_ksyms_64 needs __strncpy_from_user() definition. 2008-09-29 19:45:16 +09:00
uaccess.h
ubc.h sh: Fix UBC setup and registers for SH2A 2009-05-09 00:09:21 +09:00
ucontext.h sh: Switch to asm-generic versions for identical headers. 2009-06-14 23:26:48 +09:00
unaligned-sh4a.h sh: 16-bit get_unaligned() sh4a fix 2009-06-04 20:20:24 +09:00
unaligned.h sh: Convert to asm-generic/unaligned.h. 2009-06-14 23:24:43 +09:00
unistd_32.h perf: Do the big rename: Performance Counters -> Performance Events 2009-09-21 14:28:04 +02:00
unistd_64.h perf: Do the big rename: Performance Counters -> Performance Events 2009-09-21 14:28:04 +02:00
unistd.h
unwinder.h sh: unwinder: Introduce UNWINDER_BUG() and UNWINDER_BUG_ON() 2009-08-21 13:02:44 +01:00
user.h
vga.h
vmlinux.lds.h sh: dwarf unwinder support. 2009-08-14 01:58:43 +09:00
watchdog.h sh: Tidy up watchdog build for SH-2. 2009-08-28 10:19:12 +09:00
xor.h