* 'x86-xen-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip: (42 commits)
xen: cache cr0 value to avoid trap'n'emulate for read_cr0
xen/x86-64: clean up warnings about IST-using traps
xen/x86-64: fix breakpoints and hardware watchpoints
xen: reserve Xen start_info rather than e820 reserving
xen: add FIX_TEXT_POKE to fixmap
lguest: update lazy mmu changes to match lguest's use of kvm hypercalls
xen: honour VCPU availability on boot
xen: add "capabilities" file
xen: drop kexec bits from /sys/hypervisor since kexec isn't implemented yet
xen/sys/hypervisor: change writable_pt to features
xen: add /sys/hypervisor support
xen/xenbus: export xenbus_dev_changed
xen: use device model for suspending xenbus devices
xen: remove suspend_cancel hook
xen/dev-evtchn: clean up locking in evtchn
xen: export ioctl headers to userspace
xen: add /dev/xen/evtchn driver
xen: add irq_from_evtchn
xen: clean up gate trap/interrupt constants
xen: set _PAGE_NX in __supported_pte_mask before pagetable construction
...
Xiaohui Xin and some other folks at Intel have been looking into what's
behind the performance hit of paravirt_ops when running native.
It appears that the hit is entirely due to the paravirtualized
spinlocks introduced by:
| commit 8efcbab674
| Date: Mon Jul 7 12:07:51 2008 -0700
|
| paravirt: introduce a "lock-byte" spinlock implementation
The extra call/return in the spinlock path is somehow
causing an increase in the cycles/instruction of somewhere around 2-7%
(seems to vary quite a lot from test to test). The working theory is
that the CPU's pipeline is getting upset about the
call->call->locked-op->return->return, and seems to be failing to
speculate (though I haven't seen anything definitive about the precise
reasons). This doesn't entirely make sense, because the performance
hit is also visible on unlock and other operations which don't involve
locked instructions. But spinlock operations clearly swamp all the
other pvops operations, even though I can't imagine that they're
nearly as common (there's only a .05% increase in instructions
executed).
If I disable just the pv-spinlock calls, my tests show that pvops is
identical to non-pvops performance on native (my measurements show that
it is actually about .1% faster, but Xiaohui shows a .05% slowdown).
Summary of results, averaging 10 runs of the "mmperf" test, using a
no-pvops build as baseline:
nopv Pv-nospin Pv-spin
CPU cycles 100.00% 99.89% 102.18%
instructions 100.00% 100.10% 100.15%
CPI 100.00% 99.79% 102.03%
cache ref 100.00% 100.84% 100.28%
cache miss 100.00% 90.47% 88.56%
cache miss rate 100.00% 89.72% 88.31%
branches 100.00% 99.93% 100.04%
branch miss 100.00% 103.66% 107.72%
branch miss rt 100.00% 103.73% 107.67%
wallclock 100.00% 99.90% 102.20%
The clear effect here is that the 2% increase in CPI is
directly reflected in the final wallclock time.
(The other interesting effect is that the more ops are
out of line calls via pvops, the lower the cache access
and miss rates. Not too surprising, but it suggests that
the non-pvops kernel is over-inlined. On the flipside,
the branch misses go up correspondingly...)
So, what's the fix?
Paravirt patching turns all the pvops calls into direct calls, so
_spin_lock etc do end up having direct calls. For example, the compiler
generated code for paravirtualized _spin_lock is:
<_spin_lock+0>: mov %gs:0xb4c8,%rax
<_spin_lock+9>: incl 0xffffffffffffe044(%rax)
<_spin_lock+15>: callq *0xffffffff805a5b30
<_spin_lock+22>: retq
The indirect call will get patched to:
<_spin_lock+0>: mov %gs:0xb4c8,%rax
<_spin_lock+9>: incl 0xffffffffffffe044(%rax)
<_spin_lock+15>: callq <__ticket_spin_lock>
<_spin_lock+20>: nop; nop /* or whatever 2-byte nop */
<_spin_lock+22>: retq
One possibility is to inline _spin_lock, etc, when building an
optimised kernel (ie, when there's no spinlock/preempt
instrumentation/debugging enabled). That will remove the outer
call/return pair, returning the instruction stream to a single
call/return, which will presumably execute the same as the non-pvops
case. The downsides arel 1) it will replicate the
preempt_disable/enable code at eack lock/unlock callsite; this code is
fairly small, but not nothing; and 2) the spinlock definitions are
already a very heavily tangled mass of #ifdefs and other preprocessor
magic, and making any changes will be non-trivial.
The other obvious answer is to disable pv-spinlocks. Making them a
separate config option is fairly easy, and it would be trivial to
enable them only when Xen is enabled (as the only non-default user).
But it doesn't really address the common case of a distro build which
is going to have Xen support enabled, and leaves the open question of
whether the native performance cost of pv-spinlocks is worth the
performance improvement on a loaded Xen system (10% saving of overall
system CPU when guests block rather than spin). Still it is a
reasonable short-term workaround.
[ Impact: fix pvops performance regression when running native ]
Analysed-by: "Xin Xiaohui" <xiaohui.xin@intel.com>
Analysed-by: "Li Xin" <xin.li@intel.com>
Analysed-by: "Nakajima Jun" <jun.nakajima@intel.com>
Signed-off-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com>
Acked-by: H. Peter Anvin <hpa@zytor.com>
Cc: Nick Piggin <npiggin@suse.de>
Cc: Xen-devel <xen-devel@lists.xensource.com>
LKML-Reference: <4A0B62F7.5030802@goop.org>
[ fixed the help text ]
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Conflicts:
arch/frv/include/asm/pgtable.h
arch/x86/include/asm/required-features.h
arch/x86/xen/mmu.c
Merge reason: x86/xen was on a .29 base still, move it to a fresher
branch and pick up Xen fixes as well, plus resolve
conflicts
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Use phys_addr_t for receiving a physical address argument instead of
unsigned long. This allows fixmap to handle pages higher than 4GB on
x86-32.
Signed-off-by: Masami Hiramatsu <mhiramat@redhat.com>
Cc: Ingo Molnar <mingo@elte.hu>
Acked-by: Mathieu Desnoyers <mathieu.desnoyers@polymtl.ca>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
* commit 'origin/master': (4825 commits)
Fix build errors due to CONFIG_BRANCH_TRACER=y
parport: Use the PCI IRQ if offered
tty: jsm cleanups
Adjust path to gpio headers
KGDB_SERIAL_CONSOLE check for module
Change KCONFIG name
tty: Blackin CTS/RTS
Change hardware flow control from poll to interrupt driven
Add support for the MAX3100 SPI UART.
lanana: assign a device name and numbering for MAX3100
serqt: initial clean up pass for tty side
tty: Use the generic RS485 ioctl on CRIS
tty: Correct inline types for tty_driver_kref_get()
splice: fix deadlock in splicing to file
nilfs2: support nanosecond timestamp
nilfs2: introduce secondary super block
nilfs2: simplify handling of active state of segments
nilfs2: mark minor flag for checkpoint created by internal operation
nilfs2: clean up sketch file
nilfs2: super block operations fix endian bug
...
Conflicts:
arch/x86/include/asm/thread_info.h
arch/x86/lguest/boot.c
drivers/xen/manage.c
Impact: fix lazy context switch API
Pass the previous and next tasks into the context switch start
end calls, so that the called functions can properly access the
task state (esp in end_context_switch, in which the next task
is not yet completely current).
Signed-off-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Impact: allow preemption during lazy mmu updates
If we're in lazy mmu mode when context switching, leave
lazy mmu mode, but remember the task's state in
TIF_LAZY_MMU_UPDATES. When we resume the task, check this
flag and re-enter lazy mmu mode if its set.
This sets things up for allowing lazy mmu mode while preemptible,
though that won't actually be active until the next change.
Signed-off-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Impact: simplification, prepare for later changes
Make lazy cpu mode more specific to context switching, so that
it makes sense to do more context-switch specific things in
the callbacks.
Signed-off-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com>
Acked-by: Peter Zijlstra <a.p.zijlstra@chello.nl>
Impact: fix crash on VMI (VMware)
When we generate a call sequence for calling a paravirtualized
function, we presume that the generated code is "call *0xXXXXX",
which is a 6 byte opcode; this is larger than a normal
direct call, and so we can patch a direct call over it.
At the moment, however we give gcc enough rope to hang us by
putting the address in a register and generating a two byte
indirect-via-register call. Prevent this by explicitly
dereferencing the function pointer and passing it into the
asm as a constant.
This prevents crashes in VMI, as it cannot handle unpatchable
callsites.
Signed-off-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com>
Cc: Alok Kataria <akataria@vmware.com>
LKML-Reference: <49BEEDC2.2070809@goop.org>
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
* 'x86-fixes-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/tip/linux-2.6-tip:
x86, vm86: fix preemption bug
x86, olpc: fix model detection without OFW
x86, hpet: fix for LS21 + HPET = boot hang
x86: CPA avoid repeated lazy mmu flush
x86: warn if arch_flush_lazy_mmu_cpu is called in preemptible context
x86/paravirt: make arch_flush_lazy_mmu/cpu disable preemption
x86, pat: fix warn_on_once() while mapping 0-1MB range with /dev/mem
x86/cpa: make sure cpa is safe to call in lazy mmu mode
x86, ptrace, mm: fix double-free on race
Impact: avoid access to percpu vars in preempible context
They are intended to be used whenever there's the possibility
that there's some stale state which is going to be overwritten
with a queued update, or to force a state change when we may be
in lazy mode. Either way, we could end up calling it with
preemption enabled, so wrap the functions in their own little
preempt-disable section so they can be safely called in any
context (though preemption should never be enabled if we're actually
in a lazy state).
(Move out of line to avoid #include dependencies.)
Signed-off-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com>
Signed-off-by: Thomas Gleixner <tglx@linutronix.de>
pgtable*.h is intended for definitions relating to actual pagetables
and their entries, so move all the definitions for
(pte|pmd|pud|pgd)(val)?_t to the appropriate pgtable*.h headers.
Signed-off-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com>
Architectures other than mips and x86 are not using ticket spinlocks.
Therefore, the contention on the lock is meaningless, since there is
nobody known to be waiting on it (arguably /fairly/ unfair locks).
Dummy it out to return 0 on other architectures.
Signed-off-by: Kyle McMartin <kyle@redhat.com>
Acked-by: Ralf Baechle <ralf@linux-mips.org>
Acked-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Impact: Bug fix
A hunk went missing in the original patch, and callee-save callsites were
not marked as returning the upper 32-bit of result, causing Badness.
Signed-off-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com>
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Impact: bugfix
In the 32-bit calling convention, %eax:%edx is used to return 64-bit
values. Don't save and restore %edx around wrapped functions, or they
can't return a full 64-bit result.
Signed-off-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com>
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Impact: Fix build when CONFIG_PARAVIRT_DEBUG is enabled
Fix missed convertion to using callee-saved calls for pud_val, which
causes a compile error when CONFIG_PARAVIRT_DEBUG is enabled.
Signed-off-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com>
Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>
Impact: Optimization
In the native case, pte_val, make_pte, etc are all just identity
functions, so there's no need to clobber a lot of registers over them.
(This changes the 32-bit callee-save calling convention to return both
EAX and EDX so functions can return 64-bit values.)
Signed-off-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com>
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Impact: Optimization
Functions with the callee save calling convention clobber many fewer
registers than the normal C calling convention. Implement variants of
PVOP_V?CALL* accordingly. This only bothers with functions up to 3
args, since functions with more args may as well use the normal
calling convention.
Signed-off-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com>
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Impact: Optimization
One of the problems with inserting a pile of C calls where previously
there were none is that the register pressure is greatly increased.
The C calling convention says that the caller must expect a certain
set of registers may be trashed by the callee, and that the callee can
use those registers without restriction. This includes the function
argument registers, and several others.
This patch seeks to alleviate this pressure by introducing wrapper
thunks that will do the register saving/restoring, so that the
callsite doesn't need to worry about it, but the callee function can
be conventional compiler-generated code. In many cases (particularly
performance-sensitive cases) the callee will be in assembler anyway,
and need not use the compiler's calling convention.
Standard calling convention is:
arguments return scratch
x86-32 eax edx ecx eax ?
x86-64 rdi rsi rdx rcx rax r8 r9 r10 r11
The thunk preserves all argument and scratch registers. The return
register is not preserved, and is available as a scratch register for
unwrapped callee code (and of course the return value).
Wrapped function pointers are themselves wrapped in a struct
paravirt_callee_save structure, in order to get some warning from the
compiler when functions with mismatched calling conventions are used.
The most common paravirt ops, both statically and dynamically, are
interrupt enable/disable/save/restore, so handle them first. This is
particularly easy since their calls are handled specially anyway.
XXX Deal with VMI. What's their calling convention?
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Impact: Optimization
Each asm paravirt-ops call says what registers are available for
clobbering. This patch makes use of this to selectively save/restore
registers around each pvops call. In many cases this significantly
shrinks code size.
Signed-off-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com>
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Impact: Optimization
Several paravirt ops implementations simply return their arguments,
the most obvious being the make_pte/pte_val class of operations on
native.
On 32-bit, the identity function is literally a no-op, as the calling
convention uses the same registers for the first argument and return.
On 64-bit, it can be implemented with a single "mov".
This patch adds special identity functions for 32 and 64 bit argument,
and machinery to recognize them and replace them with either nops or a
mov as appropriate.
At the moment, the only users for the identity functions are the
pagetable entry conversion functions.
The result is a measureable improvement on pagetable-heavy benchmarks
(2-3%, reducing the pvops overhead from 5 to 2%).
Signed-off-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com>
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
pte_flags() was introduced as a new pvop in order to extract just the
flags portion of a pte, which is a potentially cheaper operation than
extracting the page number as well. It turns out this operation is
not needed, because simply using a mask to extract the flags from a
pte is sufficient for all current users.
Signed-off-by: Jeremy Fitzhardinge <jeremy.fitzhardinge@citrix.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Impact: cleanup
Remove byte locks implementation, which was introduced by Jeremy in
8efcbab6 ("paravirt: introduce a "lock-byte" spinlock implementation"),
but turned out to be dead code that is not used by any in-kernel
virtualization guest (Xen uses its own variant of spinlocks implementation
and KVM is not planning to move to byte locks).
Signed-off-by: Jiri Kosina <jkosina@suse.cz>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Impact: reduce stack usage, use new cpumask API.
This is made a little more tricky by uv_flush_tlb_others which
actually alters its argument, for an IPI to be sent to the remaining
cpus in the mask.
I solve this by allocating a cpumask_var_t for this case and falling back
to IPI should this fail.
To eliminate temporaries in the caller, all flush_tlb_others implementations
now do the this-cpu-elimination step themselves.
Note also the curious "cpus_or(f->flush_cpumask, cpumask, f->flush_cpumask)"
which has been there since pre-git and yet f->flush_cpumask is always zero
at this point.
Signed-off-by: Rusty Russell <rusty@rustcorp.com.au>
Signed-off-by: Mike Travis <travis@sgi.com>
Change header guards named "ASM_X86__*" to "_ASM_X86_*" since:
a. the double underscore is ugly and pointless.
b. no leading underscore violates namespace constraints.
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
