commit 0d1622d7f5 upstream.
The Intel Architecture Optimization Reference Manual states that a short
load that follows a long store to the same object will suffer a store
forwading penalty, particularly if the two accesses use different addresses.
Trivially, a long load that follows a short store will also suffer a penalty.
__downgrade_write() in rwsem incurs both penalties: the increment operation
will not be able to reuse a recently-loaded rwsem value, and its result will
not be reused by any recently-following rwsem operation.
A comment in the code states that this is because 64-bit immediates are
special and expensive; but while they are slightly special (only a single
instruction allows them), they aren't expensive: a test shows that two loops,
one loading a 32-bit immediate and one loading a 64-bit immediate, both take
1.5 cycles per iteration.
Fix this by changing __downgrade_write to use the same add instruction on
i386 and on x86_64, so that it uses the same operand size as all the other
rwsem functions.
Signed-off-by: Avi Kivity <avi@redhat.com>
LKML-Reference: <1266049992-17419-1-git-send-email-avi@redhat.com>
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
commit 1838ef1d78 upstream.
For x86-64, 32767 threads really is not enough. Change rwsem_count_t
to a signed long, so that it is 64 bits on x86-64.
This required the following changes to the assembly code:
a) %z0 doesn't work on all versions of gcc! At least gcc 4.4.2 as
shipped with Fedora 12 emits "ll" not "q" for 64 bits, even for
integer operands. Newer gccs apparently do this correctly, but
avoid this problem by using the _ASM_ macros instead of %z.
b) 64 bits immediates are only allowed in "movq $imm,%reg"
constructs... no others. Change some of the constraints to "e",
and fix the one case where we would have had to use an invalid
immediate -- in that case, we only care about the upper half
anyway, so just access the upper half.
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
LKML-Reference: <tip-bafaecd11df15ad5b1e598adc7736afcd38ee13d@git.kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
commit 5d0b7235d8 upstream.
The fast version of the rwsems (the code that uses xadd) has
traditionally only worked on x86-32, and as a result it mixes different
kinds of types wildly - they just all happen to be 32-bit. We have
"long", we have "__s32", and we have "int".
To make it work on x86-64, the types suddenly matter a lot more. It can
be either a 32-bit or 64-bit signed type, and both work (with the caveat
that a 32-bit counter will only have 15 bits of effective write
counters, so it's limited to 32767 users). But whatever type you
choose, it needs to be used consistently.
This makes a new 'rwsem_counter_t', that is a 32-bit signed type. For a
64-bit type, you'd need to also update the BIAS values.
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
LKML-Reference: <alpine.LFD.2.00.1001121755220.17145@localhost.localdomain>
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
commit 59c33fa779 upstream.
This makes gcc use the right register names and instruction operand sizes
automatically for the rwsem inline asm statements.
So instead of using "(%%eax)" to specify the memory address that is the
semaphore, we use "(%1)" or similar. And instead of forcing the operation
to always be 32-bit, we use "%z0", taking the size from the actual
semaphore data structure itself.
This doesn't actually matter on x86-32, but if we want to use the same
inline asm for x86-64, we'll need to have the compiler generate the proper
64-bit names for the registers (%rax instead of %eax), and if we want to
use a 64-bit counter too (in order to avoid the 15-bit limit on the
write counter that limits concurrent users to 32767 threads), we'll need
to be able to generate instructions with "q" accesses rather than "l".
Since this header currently isn't enabled on x86-64, none of that matters,
but we do want to use the xadd version of the semaphores rather than have
to take spinlocks to do a rwsem. The mm->mmap_sem can be heavily contended
when you have lots of threads all taking page faults, and the fallback
rwsem code that uses a spinlock performs abysmally badly in that case.
[ hpa: modified the patch to skip size suffixes entirely when they are
redundant due to register operands. ]
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
LKML-Reference: <alpine.LFD.2.00.1001121613560.17145@localhost.localdomain>
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
commit a7b480e7f3 upstream.
Add wbinvd_on_cpu and wbinvd_on_all_cpus stubs for executing wbinvd on a
particular CPU.
[ hpa: renamed lib/smp.c to lib/cache-smp.c ]
[ hpa: wbinvd_on_all_cpus() returns int, but wbinvd() returns
void. Thus, the former cannot be a macro for the latter,
replace with an inline function. ]
Signed-off-by: Borislav Petkov <borislav.petkov@amd.com>
LKML-Reference: <1264172467-25155-2-git-send-email-bp@amd64.org>
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
commit c697518a86 upstream
Add proper error and permission checking. This patch also change task
switching code to load segment selectors before segment descriptors, like
SDM requires, otherwise permission checking during segment descriptor
loading will be incorrect.
Signed-off-by: Gleb Natapov <gleb@redhat.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
Signed-off-by: Stefan Bader <stefan.bader@canonical.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
commit 1871c6020d upstream
Currently when x86 emulator needs to access memory, page walk is done with
broadest permission possible, so if emulated instruction was executed
by userspace process it can still access kernel memory. Fix that by
providing correct memory access to page walker during emulation.
Signed-off-by: Gleb Natapov <gleb@redhat.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
Signed-off-by: Stefan Bader <stefan.bader@canonical.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
commit a004475567 upstream
For some instructions CPU behaves differently for real-mode and
virtual 8086. Let emulator know which mode cpu is in, so it will
not poke into vcpu state directly.
Signed-off-by: Gleb Natapov <gleb@redhat.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
Signed-off-by: Stefan Bader <stefan.bader@canonical.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
commit 035a02c1e1 upstream.
Currently c1e_idle returns true for all CPUs greater than or equal to
family 0xf model 0x40. This covers too many CPUs.
Meanwhile a respective erratum for the underlying problem was filed
(#400). This patch adds the logic to check whether erratum #400
applies to a given CPU.
Especially for CPUs where SMI/HW triggered C1e is not supported,
c1e_idle() doesn't need to be used. We can check this by looking at
the respective OSVW bit for erratum #400.
Signed-off-by: Andreas Herrmann <andreas.herrmann3@amd.com>
LKML-Reference: <20100319110922.GA19614@alberich.amd.com>
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
commit ff30a0543e upstream.
Ever for 32-bit with sufficiently high NR_CPUS, and starting
with commit 789d03f584 also for
64-bit, the statically allocated early fixmap page tables were
not covering FIX_OHCI1394_BASE, leading to a boot time crash
when "ohci1394_dma=early" was used. Despite this entry not being
a permanently used one, it needs to be moved into the permanent
range since it has to be close to FIX_DBGP_BASE and
FIX_EARLYCON_MEM_BASE.
Reported-bisected-and-tested-by: Justin P. Mattock <justinmattock@gmail.com>
Fixes-bug: http://bugzilla.kernel.org/show_bug.cgi?id=14487
Signed-off-by: Jan Beulich <jbeulich@novell.com>
LKML-Reference: <4B9E15D30200007800034D23@vpn.id2.novell.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
commit 1431559200 upstream.
Distros generally (I looked at Debian, RHEL5 and SLES11) seem to
enable CONFIG_HIGHPTE for any x86 configuration which has highmem
enabled. This means that the overhead applies even to machines which
have a fairly modest amount of high memory and which therefore do not
really benefit from allocating PTEs in high memory but still pay the
price of the additional mapping operations.
Running kernbench on a 4G box I found that with CONFIG_HIGHPTE=y but
no actual highptes being allocated there was a reduction in system
time used from 59.737s to 55.9s.
With CONFIG_HIGHPTE=y and highmem PTEs being allocated:
Average Optimal load -j 4 Run (std deviation):
Elapsed Time 175.396 (0.238914)
User Time 515.983 (5.85019)
System Time 59.737 (1.26727)
Percent CPU 263.8 (71.6796)
Context Switches 39989.7 (4672.64)
Sleeps 42617.7 (246.307)
With CONFIG_HIGHPTE=y but with no highmem PTEs being allocated:
Average Optimal load -j 4 Run (std deviation):
Elapsed Time 174.278 (0.831968)
User Time 515.659 (6.07012)
System Time 55.9 (1.07799)
Percent CPU 263.8 (71.266)
Context Switches 39929.6 (4485.13)
Sleeps 42583.7 (373.039)
This patch allows the user to control the allocation of PTEs in
highmem from the command line ("userpte=nohigh") but retains the
status-quo as the default.
It is possible that some simple heuristic could be developed which
allows auto-tuning of this option however I don't have a sufficiently
large machine available to me to perform any particularly meaningful
experiments. We could probably handwave up an argument for a threshold
at 16G of total RAM.
Assuming 768M of lowmem we have 196608 potential lowmem PTE
pages. Each page can map 2M of RAM in a PAE-enabled configuration,
meaning a maximum of 384G of RAM could potentially be mapped using
lowmem PTEs.
Even allowing generous factor of 10 to account for other required
lowmem allocations, generous slop to account for page sharing (which
reduces the total amount of RAM mappable by a given number of PT
pages) and other innacuracies in the estimations it would seem that
even a 32G machine would not have a particularly pressing need for
highmem PTEs. I think 32G could be considered to be at the upper bound
of what might be sensible on a 32 bit machine (although I think in
practice 64G is still supported).
It's seems questionable if HIGHPTE is even a win for any amount of RAM
you would sensibly run a 32 bit kernel on rather than going 64 bit.
Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
LKML-Reference: <1266403090-20162-1-git-send-email-ian.campbell@citrix.com>
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
commit 18dce6ba5c upstream.
Thomas Renninger <trenn@suse.de> reported on IBM x3330
booting a latest kernel on this machine results in:
PCI: PCI BIOS revision 2.10 entry at 0xfd61c, last bus=1
PCI: Using configuration type 1 for base access bio: create slab <bio-0> at 0
ACPI: SCI (IRQ30) allocation failed
ACPI Exception: AE_NOT_ACQUIRED, Unable to install System Control Interrupt handler (20090903/evevent-161)
ACPI: Unable to start the ACPI Interpreter
Later all kind of devices fail...
and bisect it down to this commit:
commit b9c61b7007
x86/pci: update pirq_enable_irq() to setup io apic routing
it turns out we need to set irq routing for the sci on ioapic1 early.
-v2: make it work without sparseirq too.
-v3: fix checkpatch.pl warning, and cc to stable
Reported-by: Thomas Renninger <trenn@suse.de>
Bisected-by: Thomas Renninger <trenn@suse.de>
Tested-by: Thomas Renninger <trenn@suse.de>
Signed-off-by: Yinghai Lu <yinghai@kernel.org>
LKML-Reference: <1265793639-15071-2-git-send-email-yinghai@kernel.org>
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
commit f532509437 upstream
This patch moves the initialization of the iommu-api out of
the dma-ops initialization code. This ensures that the
iommu-api is initialized even with iommu=pt.
Signed-off-by: Joerg Roedel <joerg.roedel@amd.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
(cherry picked from afbcf7ab8d)
When we migrate a kvm guest that uses pvclock between two hosts, we may
suffer a large skew. This is because there can be significant differences
between the monotonic clock of the hosts involved. When a new host with
a much larger monotonic time starts running the guest, the view of time
will be significantly impacted.
Situation is much worse when we do the opposite, and migrate to a host with
a smaller monotonic clock.
This proposed ioctl will allow userspace to inform us what is the monotonic
clock value in the source host, so we can keep the time skew short, and
more importantly, never goes backwards. Userspace may also need to trigger
the current data, since from the first migration onwards, it won't be
reflected by a simple call to clock_gettime() anymore.
[marcelo: future-proof abi with a flags field]
[jan: fix KVM_GET_CLOCK by clearing flags field instead of checking it]
Signed-off-by: Glauber Costa <glommer@redhat.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
commit 05d43ed8a8 upstream.
Now that the previous commit made it possible to do the personality
setting at the point of no return, we do just that for ELF binaries.
And suddenly all the reasons for that insane TIF_ABI_PENDING bit go
away, and we can just make SET_PERSONALITY() just do the obvious thing
for a 32-bit compat process.
Everything becomes much more straightforward this way.
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
commit b160091802 upstream.
CONFIG_X86_CPU_DEBUG, which provides some parsed versions of the x86
CPU configuration via debugfs, has caused boot failures on real
hardware. The value of this feature has been marginal at best, as all
this information is already available to userspace via generic
interfaces.
Causes crashes that have not been fixed + minimal utility -> remove.
See the referenced LKML thread for more information.
Reported-by: Ozan Çağlayan <ozan@pardus.org.tr>
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
LKML-Reference: <alpine.LFD.2.00.1001221755320.13231@localhost.localdomain>
Cc: Jaswinder Singh Rajput <jaswinder@kernel.org>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
Cc: Rafael J. Wysocki <rjw@sisk.pl>
Cc: Yinghai Lu <yinghai@kernel.org>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
commit 6ede31e030 upstream.
Randy Dunlap reported the following build error:
"When CONFIG_SMP=n, CONFIG_X86_MSR=m:
ERROR: "msrs_free" [drivers/edac/amd64_edac_mod.ko] undefined!
ERROR: "msrs_alloc" [drivers/edac/amd64_edac_mod.ko] undefined!"
This is due to the fact that <arch/x86/lib/msr.c> is conditioned on
CONFIG_SMP and in the UP case we have only the stubs in the header.
Fork off SMP functionality into a new file (msr-smp.c) and build
msrs_{alloc,free} unconditionally.
Reported-by: Randy Dunlap <randy.dunlap@oracle.com>
Cc: H. Peter Anvin <hpa@zytor.com>
Signed-off-by: Borislav Petkov <petkovbb@gmail.com>
LKML-Reference: <20091216231625.GD27228@liondog.tnic>
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
commit 505422517d upstream.
The current rd/wrmsr_on_cpus helpers assume that the supplied
cpumasks are contiguous. However, there are machines out there
like some K8 multinode Opterons which have a non-contiguous core
enumeration on each node (e.g. cores 0,2 on node 0 instead of 0,1), see
http://www.gossamer-threads.com/lists/linux/kernel/1160268.
This patch fixes out-of-bounds writes (see URL above) by adding per-CPU
msr structs which are used on the respective cores.
Additionally, two helpers, msrs_{alloc,free}, are provided for use by
the callers of the MSR accessors.
Cc: H. Peter Anvin <hpa@zytor.com>
Cc: Mauro Carvalho Chehab <mchehab@redhat.com>
Cc: Aristeu Rozanski <aris@redhat.com>
Cc: Randy Dunlap <randy.dunlap@oracle.com>
Cc: Doug Thompson <dougthompson@xmission.com>
Signed-off-by: Borislav Petkov <borislav.petkov@amd.com>
LKML-Reference: <20091211171440.GD31998@aftab>
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
commit b8a4754147 upstream.
Since rdmsr_on_cpus and wrmsr_on_cpus are almost identical, unify them
into a common __rwmsr_on_cpus helper thus avoiding code duplication.
While at it, convert cpumask_t's to const struct cpumask *.
Signed-off-by: Borislav Petkov <borislav.petkov@amd.com>
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
commit 39d3077099 upstream.
The wrong address was being used to write the SCIR led regs on
remote hubs. Also, there was an inconsistency between how BIOS
and the kernel indexed these regs. Standardize on using the
lower 6 bits of the APIC ID as the index.
This patch fixes the problem of writing to an errant address to
a cpu # >= 64.
Signed-off-by: Mike Travis <travis@sgi.com>
Reviewed-by: Jack Steiner <steiner@sgi.com>
Cc: Robin Holt <holt@sgi.com>
Cc: Linus Torvalds <torvalds@linux-foundation.org>
LKML-Reference: <4B3922F9.3060905@sgi.com>
[ v2: fix a number of annoying checkpatch artifacts and whitespace noise ]
Signed-off-by: Ingo Molnar <mingo@elte.hu>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
commit 45a94d7cd4 upstream.
xsave_cntxt_init() does something like:
cpuid(0xd, ..); // find out what features FP/SSE/.. etc are supported
xsetbv(); // enable the features known to OS
cpuid(0xd, ..); // find out the size of the context for features enabled
Depending on what features get enabled in xsetbv(), value of the
cpuid.eax=0xd.ecx=0.ebx changes correspondingly (representing the
size of the context that is enabled).
As we don't have volatile keyword for native_cpuid(), gcc 4.1.2
optimizes away the second cpuid and the kernel continues to use
the cpuid information obtained before xsetbv(), ultimately leading to kernel
crash on processors supporting more state than the legacy FP/SSE.
Add "volatile" for native_cpuid().
Signed-off-by: Suresh Siddha <suresh.b.siddha@intel.com>
LKML-Reference: <1261009542.2745.55.camel@sbs-t61.sc.intel.com>
Signed-off-by: H. Peter Anvin <hpa@zytor.com>
Signed-off-by: Greg Kroah-Hartman <gregkh@suse.de>
