Make io_mapping_create_wc and io_mapping_free go through PAT to make sure
that there are no memory type aliases.
Signed-off-by: Venkatesh Pallipadi <venkatesh.pallipadi@intel.com>
Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com>
Cc: Dave Airlie <airlied@redhat.com>
Cc: Jesse Barnes <jbarnes@virtuousgeek.org>
Cc: Eric Anholt <eric@anholt.net>
Cc: Keith Packard <keithp@keithp.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
io_mapping_create_wc should take a resource_size_t parameter in place of
unsigned long. With unsigned long, there will be no way to map greater than 4GB
address in i386/32 bit.
On x86, greater than 4GB addresses cannot be mapped on i386 without PAE. Return
error for such a case.
Patch also adds a structure for io_mapping, that saves the base, size and
type on HAVE_ATOMIC_IOMAP archs, that can be used to verify the offset on
io_mapping_map calls.
Signed-off-by: Venkatesh Pallipadi <venkatesh.pallipadi@intel.com>
Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com>
Cc: Dave Airlie <airlied@redhat.com>
Cc: Jesse Barnes <jbarnes@virtuousgeek.org>
Cc: Eric Anholt <eric@anholt.net>
Cc: Keith Packard <keithp@keithp.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Add a function to check and keep identity maps in sync, when changing
any memory type. One of the follow on patches will also use this
routine.
Signed-off-by: Venkatesh Pallipadi <venkatesh.pallipadi@intel.com>
Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com>
Cc: Dave Airlie <airlied@redhat.com>
Cc: Jesse Barnes <jbarnes@virtuousgeek.org>
Cc: Eric Anholt <eric@anholt.net>
Cc: Keith Packard <keithp@keithp.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
- print test pattern instead of pattern number,
- show pattern as stored in memory,
- use proper priority flags,
- consistent use of u64 throughout the code
Signed-off-by: Andreas Herrmann <andreas.herrmann3@amd.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Impact: fix unexpected behaviour when pattern number is out of range
Current implementation provides 4 patterns for memtest. The code doesn't
check whether the memtest parameter value exceeds the maximum pattern number.
Instead the memtest code pretends to test with non-existing patterns, e.g.
when booting with memtest=10 I've observed the following
...
early_memtest: pattern num 10
0000001000 - 0000006000 pattern 0
...
0000001000 - 0000006000 pattern 1
...
0000001000 - 0000006000 pattern 2
...
0000001000 - 0000006000 pattern 3
...
0000001000 - 0000006000 pattern 4
...
0000001000 - 0000006000 pattern 5
...
0000001000 - 0000006000 pattern 6
...
0000001000 - 0000006000 pattern 7
...
0000001000 - 0000006000 pattern 8
...
0000001000 - 0000006000 pattern 9
...
But in fact Linux didn't test anything for patterns > 4 as the default
case in memtest() is to leave the function.
I suggest to use the memtest parameter as the number of tests to be
performed and to re-iterate over all existing patterns.
Signed-off-by: Andreas Herrmann <andreas.herrmann3@amd.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Impact: make more types of copies non-temporal
This change makes the following simple fix:
30d697f: x86: fix performance regression in write() syscall
A bit more sophisticated: we check the 'total' number of bytes
written to decide whether to copy in a cached or a non-temporal
way.
This will for example cause the tail (modulo 4096 bytes) chunk
of a large write() to be non-temporal too - not just the page-sized
chunks.
Cc: Salman Qazi <sqazi@google.com>
Cc: Nick Piggin <npiggin@suse.de>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Impact: cleanup, enable future change
Add a 'total bytes copied' parameter to __copy_from_user_*nocache(),
and update all the callsites.
The parameter is not used yet - architecture code can use it to
more intelligently decide whether the copy should be cached or
non-temporal.
Cc: Salman Qazi <sqazi@google.com>
Cc: Nick Piggin <npiggin@suse.de>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Impact: cleanup
Unused macro parameters cause spurious unused variable warnings.
Convert all cacheflush macros to inline functions to avoid the
warnings and achieve better type checking.
Signed-off-by: Tejun Heo <tj@kernel.org>
Recent changes in setup_percpu.c made a now meaningless DBG()
statement fail to compile and introduced a
comparison-of-different-types warning. Fix them.
Compile failure is reported by Ingo Molnar.
Signed-off-by: Tejun Heo <tj@kernel.org>
Reported-by: Ingo Molnar <mingo@elte.hu>
Impact: defconfig change
Enable MCE in the 64-bit defconfig.
Signed-off-by: Andi Kleen <ak@linux.intel.com>
Signed-off-by: H. Peter Anvin <hpa@linux.intel.com>
Impact: Fix marginal race condition
One the first CPU the machine checks are enabled early before
the local APIC is enabled. This could in theory lead
to some lost CMCI events very early during boot because
CMCIs cannot be delivered with disabled LAPIC.
The poller also doesn't recover from this because it doesn't
check CMCI banks.
Add an explicit CMCI banks check after the LAPIC is enabled.
This is only done for CPU #0, the other CPUs only initialize
machine checks after the LAPIC is on.
Signed-off-by: Andi Kleen <ak@linux.intel.com>
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Impact: Avoids confusing other OSes.
Disable the CMCI vector on reboot to avoid confusing other OS.
Signed-off-by: Andi Kleen <ak@linux.intel.com>
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Impact: Bug fix on UP
The MCE code is reinitialized from resume, so we can't use
__cpuinit/__cpuexit for most of the code. Remove those annotations
for anything downstream of mce_init().
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Impact: Major new feature
Intel CMCI (Corrected Machine Check Interrupt) is a new
feature on Nehalem CPUs. It allows the CPU to trigger
interrupts on corrected events, which allows faster
reaction to them instead of with the traditional
polling timer.
Also use CMCI to discover shared banks. Machine check banks
can be shared by CPU threads or even cores. Using the CMCI enable
bit it is possible to detect the fact that another CPU already
saw a specific bank. Use this to assign shared banks only
to one CPU to avoid reporting duplicated events.
On CPU hot unplug bank sharing is re discovered. This is done
using a thread that cycles through all the CPUs.
To avoid races between the poller and CMCI we only poll
for banks that are not CMCI capable and only check CMCI
owned banks on a interrupt.
The shared banks ownership information is currently only used for
CMCI interrupts, not polled banks.
The sharing discovery code follows the algorithm recommended in the
IA32 SDM Vol3a 14.5.2.1
The CMCI interrupt handler just calls the machine check poller to
pick up the machine check event that caused the interrupt.
I decided not to implement a separate threshold event like
the AMD version has, because the threshold is always one currently
and adding another event didn't seem to add any value.
Some code inspired by Yunhong Jiang's Xen implementation,
which was in term inspired by a earlier CMCI implementation
by me.
Signed-off-by: Andi Kleen <ak@linux.intel.com>
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Impact: New register definitions only
CMCI means support for raising an interrupt on a corrected machine
check event instead of having to poll for it. It's a new feature in
Intel Nehalem CPUs available on some machine check banks.
For details see the IA32 SDM Vol3a 14.5
Define the registers for it as a preparation for further patches.
Signed-off-by: Andi Kleen <ak@linux.intel.com>
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Define a per cpu bitmap that contains the banks polled by the machine
check poller. This is needed for the CMCI code in the next patches
to be able to disable polling on specific banks.
The bank by default contains all banks, so there is no behaviour
change. Only future code will remove some banks from the polling
set.
Signed-off-by: Andi Kleen <ak@linux.intel.com>
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Impact: behavior change, use common code
Use a standard leaky bucket ratelimit for the machine check
warning print interval instead of waiting every check_interval.
Also decrease the limit to twice per minute.
This interacts better with threshold interrupts because
they can happen more often than check_interval.
Signed-off-by: Andi Kleen <ak@linux.intel.com>
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Impact: minor bugfix
The threshold handler on AMD (and soon on Intel) could be theoretically
reentered by the hardware. This could lead to corrupted events
because the machine check poll code assumes it is not reentered.
Move the APIC ACK to the end of the interrupt handler to let
the hardware avoid that.
Signed-off-by: Andi Kleen <ak@linux.intel.com>
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Impact: cleanup; preparation for feature
The mce_amd_64 code has an own private MC threshold vector with an own
interrupt handler. Since Intel needs a similar handler
it makes sense to share the vector because both can not
be active at the same time.
I factored the common APIC handler code into a separate file which can
be used by both the Intel or AMD MC code.
This is needed for the next patch which adds an Intel specific
CMCI handler.
This patch should be a nop for AMD, it just moves some code
around.
Signed-off-by: Andi Kleen <ak@linux.intel.com>
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Impact: Cleanup (code movement)
Move MAX_NR_BANKS into mce.h because it's needed there
for followup patches.
Signed-off-by: Andi Kleen <ak@linux.intel.com>
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
The ones which go only into struct genapic are de-inlined
by compiler anyway, so remove the inline specifier from them.
Afterwards, remove summit_setup_portio_remap completely as it
is unused.
Remove inline also from summit_cpu_mask_to_apicid, since it's
not worth it (it is used in struct genapic too).
Signed-off-by: Jiri Slaby <jirislaby@gmail.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Use BAD_APICID instead of 0xFF constants in summit_cpu_mask_to_apicid.
Also remove bogus comments about what we actually return.
Signed-off-by: Jiri Slaby <jirislaby@gmail.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
This was changed to a physmap_t giving a clashing symbol redefinition,
but actually using a physmap_t consumes rather a lot of space on x86,
so stick with a private copy renamed with a voyager_ prefix and made
static. Nothing outside of the Voyager code uses it, anyway.
Signed-off-by: James Bottomley <James.Bottomley@HansenPartnership.com>
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
native_usergs_sysret64 is described as
extern void native_usergs_sysret64(void)
so lets add ENDPROC here
Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
Cc: heukelum@fastmail.fm
Cc: Jeremy Fitzhardinge <jeremy@xensource.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Impact: cleanup
NEXT_PAGE already has 'balign' so no
need to keep this redundant one.
Signed-off-by: Cyrill Gorcunov <gorcunov@openvz.org>
Cc: heukelum@fastmail.fm
Signed-off-by: Ingo Molnar <mingo@elte.hu>
While the introduction of __copy_from_user_nocache (see commit:
0812a579c92fefa57506821fa08e90f47cb6dbdd) may have been an improvement
for sufficiently large writes, there is evidence to show that it is
deterimental for small writes. Unixbench's fstime test gives the
following results for 256 byte writes with MAX_BLOCK of 2000:
2.6.29-rc6 ( 5 samples, each in KB/sec ):
283750, 295200, 294500, 293000, 293300
2.6.29-rc6 + this patch (5 samples, each in KB/sec):
313050, 3106750, 293350, 306300, 307900
2.6.18
395700, 342000, 399100, 366050, 359850
See w_test() in src/fstime.c in unixbench version 4.1.0. Basically, the above test
consists of counting how much we can write in this manner:
alarm(10);
while (!sigalarm) {
for (f_blocks = 0; f_blocks < 2000; ++f_blocks) {
write(f, buf, 256);
}
lseek(f, 0L, 0);
}
Note, there are other components to the write syscall regression
that are not addressed here.
Signed-off-by: Salman Qazi <sqazi@google.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Impact: add better first percpu allocation for NUMA
On NUMA, embedding allocator can't be used as different units can't be
made to fall in the correct NUMA nodes. To use large page mapping,
each unit needs to be remapped. However, percpu areas are usually
much smaller than large page size and unused space hurts a lot as the
number of cpus grow. This allocator remaps large pages for each chunk
but gives back unused part to the bootmem allocator making the large
pages mapped twice.
This adds slightly to the TLB pressure but is much better than using
4k mappings while still being NUMA-friendly.
Ingo suggested that this would be the correct approach for NUMA.
Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: Ingo Molnar <mingo@elte.hu>
Impact: add better first percpu allocation for !NUMA
On !NUMA, we can simply allocate contiguous memory and use it for the
first chunk without mapping it into vmalloc area. As the memory area
is covered by the large page physical memory mapping, it allows the
dynamic perpcu allocator to not add any TLB overhead for the static
percpu area and whatever falls into the first chunk and the
implementation is very simple too.
Signed-off-by: Tejun Heo <tj@kernel.org>
Impact: modularize percpu first chunk allocation
x86 is gonna have a few different strategies for the first chunk
allocation. Modularize it by separating out the current allocation
mechanism into pcpu_alloc_bootmem() and setup_pcpu_4k().
Signed-off-by: Tejun Heo <tj@kernel.org>
Impact: more latitude for first percpu chunk allocation
The first percpu chunk serves the kernel static percpu area and may or
may not contain extra room for further dynamic allocation.
Initialization of the first chunk needs to be done before normal
memory allocation service is up, so it has its own init path -
pcpu_setup_static().
It seems archs need more latitude while initializing the first chunk
for example to take advantage of large page mapping. This patch makes
the following changes to allow this.
* Define PERCPU_DYNAMIC_RESERVE to give arch hint about how much space
to reserve in the first chunk for further dynamic allocation.
* Rename pcpu_setup_static() to pcpu_setup_first_chunk().
* Make pcpu_setup_first_chunk() much more flexible by fetching page
pointer by callback and adding optional @unit_size, @free_size and
@base_addr arguments which allow archs to selectively part of chunk
initialization to their likings.
Signed-off-by: Tejun Heo <tj@kernel.org>
Impact: minor change to populate_extra_pte() and addition of pmd flavor
Update populate_extra_pte() to return pointer to the pte_t for the
specified address and add populate_extra_pmd() which only populates
till the pmd and returns pointer to the pmd entry for the address.
For 64bit, pud/pmd/pte fill functions are separated out from
set_pte_vaddr[_pud]() and used for set_pte_vaddr[_pud]() and
populate_extra_{pte|pmd}().
Signed-off-by: Tejun Heo <tj@kernel.org>
Impact: cleaner and consistent bootmem wrapping
By setting CONFIG_HAVE_ARCH_BOOTMEM_NODE, archs can define
arch-specific wrappers for bootmem allocation. However, this is done
a bit strangely in that only the high level convenience macros can be
changed while lower level, but still exported, interface functions
can't be wrapped. This not only is messy but also leads to strange
situation where alloc_bootmem() does what the arch wants it to do but
the equivalent __alloc_bootmem() call doesn't although they should be
able to be used interchangeably.
This patch updates bootmem such that archs can override / wrap the
backend function - alloc_bootmem_core() instead of the highlevel
interface functions to allow simpler and consistent wrapping. Also,
HAVE_ARCH_BOOTMEM_NODE is renamed to HAVE_ARCH_BOOTMEM.
Signed-off-by: Tejun Heo <tj@kernel.org>
Cc: Johannes Weiner <hannes@saeurebad.de>
