pandora-kernel.git
13 years agoMerge branch 'kvm-updates/2.6.39' of git://git.kernel.org/pub/scm/virt/kvm/kvm
Linus Torvalds [Fri, 18 Mar 2011 01:40:35 +0000 (18:40 -0700)]
Merge branch 'kvm-updates/2.6.39' of git://git./virt/kvm/kvm

* 'kvm-updates/2.6.39' of git://git.kernel.org/pub/scm/virt/kvm/kvm: (55 commits)
  KVM: unbreak userspace that does not sets tss address
  KVM: MMU: cleanup pte write path
  KVM: MMU: introduce a common function to get no-dirty-logged slot
  KVM: fix rcu usage in init_rmode_* functions
  KVM: fix kvmclock regression due to missing clock update
  KVM: emulator: Fix permission checking in io permission bitmap
  KVM: emulator: Fix io permission checking for 64bit guest
  KVM: SVM: Load %gs earlier if CONFIG_X86_32_LAZY_GS=n
  KVM: x86: Remove useless regs_page pointer from kvm_lapic
  KVM: improve comment on rcu use in irqfd_deassign
  KVM: MMU: remove unused macros
  KVM: MMU: cleanup page alloc and free
  KVM: MMU: do not record gfn in kvm_mmu_pte_write
  KVM: MMU: move mmu pages calculated out of mmu lock
  KVM: MMU: set spte accessed bit properly
  KVM: MMU: fix kvm_mmu_slot_remove_write_access dropping intermediate W bits
  KVM: Start lock documentation
  KVM: better readability of efer_reserved_bits
  KVM: Clear async page fault hash after switching to real mode
  KVM: VMX: Initialize vm86 TSS only once.
  ...

13 years agoMerge branch 'stable/xen.pm.bug-fixes' of git://git.kernel.org/pub/scm/linux/kernel...
Linus Torvalds [Fri, 18 Mar 2011 01:37:42 +0000 (18:37 -0700)]
Merge branch 'stable/xen.pm.bug-fixes' of git://git./linux/kernel/git/konrad/xen

* 'stable/xen.pm.bug-fixes' of git://git.kernel.org/pub/scm/linux/kernel/git/konrad/xen:
  xen: use freeze/restore/thaw PM events for suspend/resume/chkpt
  xen: xenbus PM events support

13 years agoMerge branches 'stable/irq.fairness' and 'stable/irq.ween_of_nr_irqs' of git://git...
Linus Torvalds [Fri, 18 Mar 2011 01:27:49 +0000 (18:27 -0700)]
Merge branches 'stable/irq.fairness' and 'stable/irq.ween_of_nr_irqs' of git://git./linux/kernel/git/konrad/xen

* 'stable/irq.fairness' of git://git.kernel.org/pub/scm/linux/kernel/git/konrad/xen:
  xen: events: Remove redundant clear of l2i at end of round-robin loop
  xen: events: Make round-robin scan fairer by snapshotting each l2 word once only
  xen: events: Clean up round-robin evtchn scan.
  xen: events: Make last processed event channel a per-cpu variable.
  xen: events: Process event channels notifications in round-robin order.

* 'stable/irq.ween_of_nr_irqs' of git://git.kernel.org/pub/scm/linux/kernel/git/konrad/xen:
  xen: events: Fix compile error if CONFIG_SMP is not defined.
  xen: events: correct locking in xen_irq_from_pirq
  xen: events: propagate irq allocation failure instead of panicking
  xen: events: do not workaround too-small nr_irqs
  xen: events: remove use of nr_irqs as upper bound on number of pirqs
  xen: events: dynamically allocate irq info structures
  xen: events: maintain a list of Xen interrupts
  xen: events: push setup of irq<->{evtchn,ipi,virq,pirq} maps into irq_info init functions
  xen: events: turn irq_info constructors into initialiser functions
  xen: events: use per-cpu variable for cpu_evtchn_mask
  xen: events: refactor GSI pirq bindings functions
  xen: events: rename restore_cpu_pirqs -> restore_pirqs
  xen: events: remove unused public functions
  xen: events: fix xen_map_pirq_gsi error return
  xen: events: simplify comment
  xen: events: separate two unrelated halves of if condition

Fix up trivial conflicts in drivers/xen/events.c

13 years agoMerge branches 'stable/hvc-console', 'stable/gntalloc.v6' and 'stable/balloon' of...
Linus Torvalds [Fri, 18 Mar 2011 01:16:36 +0000 (18:16 -0700)]
Merge branches 'stable/hvc-console', 'stable/gntalloc.v6' and 'stable/balloon' of git://git./linux/kernel/git/konrad/xen

* 'stable/hvc-console' of git://git.kernel.org/pub/scm/linux/kernel/git/konrad/xen:
  xen/hvc: Disable probe_irq_on/off from poking the hvc-console IRQ line.

* 'stable/gntalloc.v6' of git://git.kernel.org/pub/scm/linux/kernel/git/konrad/xen:
  xen: gntdev: fix build warning
  xen/p2m/m2p/gnttab: do not add failed grant maps to m2p override
  xen-gntdev: Add cast to pointer
  xen-gntdev: Fix incorrect use of zero handle
  xen: change xen/[gntdev/gntalloc] to default m
  xen-gntdev: prevent using UNMAP_NOTIFY_CLEAR_BYTE on read-only mappings
  xen-gntdev: Avoid double-mapping memory
  xen-gntdev: Avoid unmapping ranges twice
  xen-gntdev: Use map->vma for checking map validity
  xen-gntdev: Fix unmap notify on PV domains
  xen-gntdev: Fix memory leak when mmap fails
  xen/gntalloc,gntdev: Add unmap notify ioctl
  xen-gntalloc: Userspace grant allocation driver
  xen-gntdev: Support mapping in HVM domains
  xen-gntdev: Add reference counting to maps
  xen-gntdev: Use find_vma rather than iterating our vma list manually
  xen-gntdev: Change page limit to be global instead of per-open

* 'stable/balloon' of git://git.kernel.org/pub/scm/linux/kernel/git/konrad/xen: (24 commits)
  xen-gntdev: Use ballooned pages for grant mappings
  xen-balloon: Add interface to retrieve ballooned pages
  xen-balloon: Move core balloon functionality out of module
  xen/balloon: Remove pr_info's and don't alter retry_count
  xen/balloon: Protect against CPU exhaust by event/x process
  xen/balloon: Migration from mod_timer() to schedule_delayed_work()
  xen/balloon: Removal of driver_pages

13 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi-misc-2.6
Linus Torvalds [Fri, 18 Mar 2011 00:54:40 +0000 (17:54 -0700)]
Merge git://git./linux/kernel/git/jejb/scsi-misc-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/jejb/scsi-misc-2.6: (170 commits)
  [SCSI] scsi_dh_rdac: Add MD36xxf into device list
  [SCSI] scsi_debug: add consecutive medium errors
  [SCSI] libsas: fix ata list corruption issue
  [SCSI] hpsa: export resettable host attribute
  [SCSI] hpsa: move device attributes to avoid forward declarations
  [SCSI] scsi_debug: Logical Block Provisioning (SBC3r26)
  [SCSI] sd: Logical Block Provisioning update
  [SCSI] Include protection operation in SCSI command trace
  [SCSI] hpsa: fix incorrect PCI IDs and add two new ones (2nd try)
  [SCSI] target: Fix volume size misreporting for volumes > 2TB
  [SCSI] bnx2fc: Broadcom FCoE offload driver
  [SCSI] fcoe: fix broken fcoe interface reset
  [SCSI] fcoe: precedence bug in fcoe_filter_frames()
  [SCSI] libfcoe: Remove stale fcoe-netdev entries
  [SCSI] libfcoe: Move FCOE_MTU definition from fcoe.h to libfcoe.h
  [SCSI] libfc: introduce __fc_fill_fc_hdr that accepts fc_hdr as an argument
  [SCSI] fcoe, libfc: initialize EM anchors list and then update npiv EMs
  [SCSI] Revert "[SCSI] libfc: fix exchange being deleted when the abort itself is timed out"
  [SCSI] libfc: Fixing a memory leak when destroying an interface
  [SCSI] megaraid_sas: Version and Changelog update
  ...

Fix up trivial conflicts due to whitespace differences in
drivers/scsi/libsas/{sas_ata.c,sas_scsi_host.c}

13 years agoMerge branch 'next' of git://git.kernel.org/pub/scm/linux/kernel/git/davej/cpufreq
Linus Torvalds [Fri, 18 Mar 2011 00:42:14 +0000 (17:42 -0700)]
Merge branch 'next' of git://git./linux/kernel/git/davej/cpufreq

* 'next' of git://git.kernel.org/pub/scm/linux/kernel/git/davej/cpufreq:
  [CPUFREQ] pcc-cpufreq: remove duplicate statements
  [CPUFREQ] Remove the pm_message_t argument from driver suspend
  [CPUFREQ] Remove unneeded locks
  [CPUFREQ] Remove old, deprecated per cpu ondemand/conservative sysfs files
  [CPUFREQ] Remove deprecated sysfs file sampling_rate_max
  [CPUFREQ] powernow-k8: The table index is not worth displaying
  [CPUFREQ] calculate delay after dbs_check_cpu
  [CPUFREQ] Add documentation for sampling_down_factor
  [CPUFREQ] drivers/cpufreq: Remove unnecessary semicolons

13 years agoMerge branch 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jack/linux...
Linus Torvalds [Fri, 18 Mar 2011 00:41:19 +0000 (17:41 -0700)]
Merge branch 'for_linus' of git://git./linux/kernel/git/jack/linux-fs-2.6

* 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jack/linux-fs-2.6:
  ext3: Always set dx_node's fake_dirent explicitly.
  ext3: Fix an overflow in ext3_trim_fs.
  jbd: Remove one to many n's in a word.
  ext3: skip orphan cleanup on rocompat fs
  ext2: Fix link count corruption under heavy link+rename load
  ext3: speed up group trim with the right free block count.
  ext3: Adjust trim start with first_data_block.
  quota: return -ENOMEM when memory allocation fails

13 years agoMerge branch 'nfs-for-2.6.39' of git://git.linux-nfs.org/projects/trondmy/nfs-2.6
Linus Torvalds [Fri, 18 Mar 2011 00:40:00 +0000 (17:40 -0700)]
Merge branch 'nfs-for-2.6.39' of git://git.linux-nfs.org/projects/trondmy/nfs-2.6

* 'nfs-for-2.6.39' of git://git.linux-nfs.org/projects/trondmy/nfs-2.6: (54 commits)
  RPC: killing RPC tasks races fixed
  xprt: remove redundant check
  SUNRPC: Convert struct rpc_xprt to use atomic_t counters
  SUNRPC: Ensure we always run the tk_callback before tk_action
  sunrpc: fix printk format warning
  xprt: remove redundant null check
  nfs: BKL is no longer needed, so remove the include
  NFS: Fix a warning in fs/nfs/idmap.c
  Cleanup: Factor out some cut-and-paste code.
  cleanup: save 60 lines/100 bytes by combining two mostly duplicate functions.
  NFS: account direct-io into task io accounting
  gss:krb5 only include enctype numbers in gm_upcall_enctypes
  RPCRDMA: Fix FRMR registration/invalidate handling.
  RPCRDMA: Fix to XDR page base interpretation in marshalling logic.
  NFSv4: Send unmapped uid/gids to the server when using auth_sys
  NFSv4: Propagate the error NFS4ERR_BADOWNER to nfs4_do_setattr
  NFSv4: cleanup idmapper functions to take an nfs_server argument
  NFSv4: Send unmapped uid/gids to the server if the idmapper fails
  NFSv4: If the server sends us a numeric uid/gid then accept it
  NFSv4.1: reject zero layout with zeroed stripe unit
  ...

13 years agoMerge branch 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jack/linux...
Linus Torvalds [Fri, 18 Mar 2011 00:29:38 +0000 (17:29 -0700)]
Merge branch 'for_linus' of git://git./linux/kernel/git/jack/linux-udf-2.6

* 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/jack/linux-udf-2.6:
  UDF: Fix compiler warning
  udf: Convert UDF to new truncate calling sequence

13 years agoMerge branch 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/bp/bp
Linus Torvalds [Fri, 18 Mar 2011 00:21:32 +0000 (17:21 -0700)]
Merge branch 'for-linus' of git://git./linux/kernel/git/bp/bp

* 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/bp/bp: (38 commits)
  amd64_edac: Fix decode_syndrome types
  amd64_edac: Fix DCT argument type
  amd64_edac: Fix ranges signedness
  amd64_edac: Drop local variable
  amd64_edac: Fix PCI config addressing types
  amd64_edac: Fix DRAM base macros
  amd64_edac: Fix node id signedness
  amd64_edac: Drop redundant declarations
  amd64_edac: Enable driver on F15h
  amd64_edac: Adjust ECC symbol size to F15h
  amd64_edac: Simplify scrubrate setting
  PCI: Rename CPU PCI id define
  amd64_edac: Improve DRAM address mapping
  amd64_edac: Sanitize ->read_dram_ctl_register
  amd64_edac: Adjust sys_addr to chip select conversion routine to F15h
  amd64_edac: Beef up early exit reporting
  amd64_edac: Revamp online spare handling
  amd64_edac: Fix channel interleave removal
  amd64_edac: Correct node interleaving removal
  amd64_edac: Add support for interleaved region swapping
  ...

Fix up trivial conflict in include/linux/pci_ids.h due to
AMD_15H_NB_MISC being renamed as AMD_15H_NB_F3 next to the new
AMD_15H_NB_LINK entry.

13 years agoMerge branch 'for-linus/2639/i2c-1' of git://git.fluff.org/bjdooks/linux
Linus Torvalds [Fri, 18 Mar 2011 00:10:19 +0000 (17:10 -0700)]
Merge branch 'for-linus/2639/i2c-1' of git://git.fluff.org/bjdooks/linux

* 'for-linus/2639/i2c-1' of git://git.fluff.org/bjdooks/linux:
  i2c-mpc: Add support for 64bit system
  i2c: add driver for Freescale i.MX28
  i2c: tegra: Add i2c support

13 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/wim/linux-2.6-watchdog
Linus Torvalds [Fri, 18 Mar 2011 00:09:29 +0000 (17:09 -0700)]
Merge git://git./linux/kernel/git/wim/linux-2.6-watchdog

* git://git.kernel.org/pub/scm/linux/kernel/git/wim/linux-2.6-watchdog:
  watchdog: booke_wdt: clean up status messages
  watchdog: cleanup spaces before tabs
  watchdog: convert to DEFINE_PCI_DEVICE_TABLE
  watchdog: Xen watchdog driver
  watchdog: Intel SCU Watchdog Timer Driver for Moorestown and Medfield platforms.
  watchdog: jz4740_wdt - fix magic character checking
  watchdog: add JZ4740 watchdog driver
  watchdog: it87_wdt: Add support for IT8721F watchdog
  watchdog: hpwdt: build hpwdt as module by default with NMI_DECODING enabled
  watchdog: hpwdt: Fix a couple of typos

13 years agoMerge branch 'hwmon-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/groec...
Linus Torvalds [Thu, 17 Mar 2011 23:59:38 +0000 (16:59 -0700)]
Merge branch 'hwmon-for-linus' of git://git./linux/kernel/git/groeck/staging

* 'hwmon-for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/groeck/staging: (44 commits)
  hwmon: (lineage-pem): Fix in1 voltage alarm sysfs attributes
  hwmon/f71882fg: Add support for f71808e
  hwmon/f71882fg: Add support for f71869f and f71869e
  hwmon/f71882fg: Add support for f71889ed
  hwmon/f71882fg: Break out test for auto pwm's controlled by digital readings
  hwmon/f71882fg: Separate temp beep sysfs attr from the other temp sysfs attr
  hwmon/f71882fg: Remove bogus temp2_type for certain models
  hwmon/f71882fg: Make number of temps configurable
  hwmon/f71882fg: Make creation of in sysfs attributes more generic
  hwmon/f71882fg: Only allow negative auto point temps if fan_neg_temp is enabled
  hwmon/f71882fg: Fix temp1 sensor type reporting
  hwmon: (w83627ehf) Display correct temperature sensor labels for systems with NCT6775F
  hwmon: (w83627ehf) Add fan debounce support for NCT6775F and NCT6776F
  hwmon: (w83627ehf) Update Kconfig for W83677HG-B, NCT6775F and NCT6776F
  hwmon: (w83627ehf) Store rpm instead of raw fan speed data
  hwmon: (w83627ehf) Use 16 bit fan count registers if supported
  hwmon: (w83627ehf) Add support for Nuvoton NCT6775F and NCT6776F
  hwmon: (w83627ehf) Permit enabling SmartFan IV mode if configured at startup
  hwmon: (w83627ehf) Convert register arrays to 16 bit, and convert access to pointers
  hwmon: (w83627ehf) Remove references to datasheets which no longer exist
  ...

13 years agosmp_call_function_interrupt: use typedef and %pf
Milton Miller [Tue, 15 Mar 2011 19:27:17 +0000 (13:27 -0600)]
smp_call_function_interrupt: use typedef and %pf

Use the newly added smp_call_func_t in smp_call_function_interrupt for
the func variable, and make the comment above the WARN more assertive
and explicit.  Also, func is a function pointer and does not need an
offset, so use %pf not %pS.

Signed-off-by: Milton Miller <miltonm@bga.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agosmp_call_function_many: handle concurrent clearing of mask
Milton Miller [Tue, 15 Mar 2011 19:27:17 +0000 (13:27 -0600)]
smp_call_function_many: handle concurrent clearing of mask

Mike Galbraith reported finding a lockup ("perma-spin bug") where the
cpumask passed to smp_call_function_many was cleared by other cpu(s)
while a cpu was preparing its call_data block, resulting in no cpu to
clear the last ref and unlock the block.

Having cpus clear their bit asynchronously could be useful on a mask of
cpus that might have a translation context, or cpus that need a push to
complete an rcu window.

Instead of adding a BUG_ON and requiring yet another cpumask copy, just
detect the race and handle it.

Note: arch_send_call_function_ipi_mask must still handle an empty
cpumask because the data block is globally visible before the that arch
callback is made.  And (obviously) there are no guarantees to which cpus
are notified if the mask is changed during the call; only cpus that were
online and had their mask bit set during the whole call are guaranteed
to be called.

Reported-by: Mike Galbraith <efault@gmx.de>
Reported-by: Jan Beulich <JBeulich@novell.com>
Acked-by: Jan Beulich <jbeulich@novell.com>
Cc: stable@kernel.org
Signed-off-by: Milton Miller <miltonm@bga.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agocall_function_many: add missing ordering
Milton Miller [Tue, 15 Mar 2011 19:27:16 +0000 (13:27 -0600)]
call_function_many: add missing ordering

Paul McKenney's review pointed out two problems with the barriers in the
2.6.38 update to the smp call function many code.

First, a barrier that would force the func and info members of data to
be visible before their consumption in the interrupt handler was
missing.  This can be solved by adding a smp_wmb between setting the
func and info members and setting setting the cpumask; this will pair
with the existing and required smp_rmb ordering the cpumask read before
the read of refs.  This placement avoids the need a second smp_rmb in
the interrupt handler which would be executed on each of the N cpus
executing the call request.  (I was thinking this barrier was present
but was not).

Second, the previous write to refs (establishing the zero that we the
interrupt handler was testing from all cpus) was performed by a third
party cpu.  This would invoke transitivity which, as a recient or
concurrent addition to memory-barriers.txt now explicitly states, would
require a full smp_mb().

However, we know the cpumask will only be set by one cpu (the data
owner) and any preivous iteration of the mask would have cleared by the
reading cpu.  By redundantly writing refs to 0 on the owning cpu before
the smp_wmb, the write to refs will follow the same path as the writes
that set the cpumask, which in turn allows us to keep the barrier in the
interrupt handler a smp_rmb instead of promoting it to a smp_mb (which
will be be executed by N cpus for each of the possible M elements on the
list).

I moved and expanded the comment about our (ab)use of the rcu list
primitives for the concurrent walk earlier into this function.  I
considered moving the first two paragraphs to the queue list head and
lock, but felt it would have been too disconected from the code.

Cc: Paul McKinney <paulmck@linux.vnet.ibm.com>
Cc: stable@kernel.org (2.6.32 and later)
Signed-off-by: Milton Miller <miltonm@bga.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agocall_function_many: fix list delete vs add race
Milton Miller [Tue, 15 Mar 2011 19:27:16 +0000 (13:27 -0600)]
call_function_many: fix list delete vs add race

Peter pointed out there was nothing preventing the list_del_rcu in
smp_call_function_interrupt from running before the list_add_rcu in
smp_call_function_many.

Fix this by not setting refs until we have gotten the lock for the list.
Take advantage of the wmb in list_add_rcu to save an explicit additional
one.

I tried to force this race with a udelay before the lock & list_add and
by mixing all 64 online cpus with just 3 random cpus in the mask, but
was unsuccessful.  Still, inspection shows a valid race, and the fix is
a extension of the existing protection window in the current code.

Cc: stable@kernel.org (v2.6.32 and later)
Reported-by: Peter Zijlstra <peterz@infradead.org>
Signed-off-by: Milton Miller <miltonm@bga.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agomm: PageBuddy and mapcount robustness
Andrea Arcangeli [Thu, 17 Mar 2011 23:16:35 +0000 (00:16 +0100)]
mm: PageBuddy and mapcount robustness

Change the _mapcount value indicating PageBuddy from -2 to -128 for
more robusteness against page_mapcount() undeflows.

Use reset_page_mapcount instead of __ClearPageBuddy in bad_page to
ignore the previous retval of PageBuddy().

Signed-off-by: Andrea Arcangeli <aarcange@redhat.com>
Reported-by: Hugh Dickins <hughd@google.com>
Signed-off-by: Linus Torvalds <torvalds@linux-foundation.org>
13 years agoMerge branch 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/epip/linux...
Linus Torvalds [Thu, 17 Mar 2011 17:11:25 +0000 (10:11 -0700)]
Merge branch 'for_linus' of git://git./linux/kernel/git/epip/linux-2.6-unicore32

* 'for_linus' of git://git.kernel.org/pub/scm/linux/kernel/git/epip/linux-2.6-unicore32: (40 commits)
  unicore32: rewrite arch-specific tlb.h to use asm-generic version
  unicore32: modify io_p2v and io_v2p macros, and adjust PKUNITY_mmio_BASEs
  unicore32: replace unicore32-specific iomap functions with generic lib implementation
  unicore32 machine related: add frame buffer driver for pkunity-v3 soc
  unicore32 machine related files: add i2c bus drivers for pkunity-v3 soc
  unicore32 io: redefine __REG(x) and re-use readl/writel funcs
  unicore32 i8042 upgrade and bugfix: adjust resource request region type
  unicore32 upgrade to v2.6.38-rc5: add one more paramter for pte_alloc_map call
  unicore32 i8042: adjust io funcs of i8042-unicore32io.h
  unicore32: rename PKUNITY_IOSPACE_BASE to PKUNITY_MMIO_BASE
  unicore32: modify function names and parameters for irq_chips
  unicore32: remove unused lines in arch/unicore32/include/asm/irq.h
  unicore32 time.c: change calculate method for clock_event_device
  unicore32: ADD MAINTAINER for unicore32 architecture
  unicore32 machine related files: ps2 driver
  unicore32 machine related files: pci bus handling
  unicore32 machine related files: hardware registers
  unicore32 machine related files: core files
  unicore32 additional architecture files: boot process
  unicore32 additional architecture files: low-level lib: misc
  ...

Acked-by: Arnd Bergmann <arnd@arndb.de>
13 years agoMerge branch 'for-linus' of git://git390.marist.edu/pub/scm/linux-2.6
Linus Torvalds [Thu, 17 Mar 2011 17:10:49 +0000 (10:10 -0700)]
Merge branch 'for-linus' of git://git390.marist.edu/linux-2.6

* 'for-linus' of git://git390.marist.edu/pub/scm/linux-2.6:
  [S390] kexec: Disable ftrace during kexec
  [S390] support XZ compressed kernel
  [S390] css_bus_type: make it static
  [S390] css_driver: remove duplicate members
  [S390] css: remove subchannel private
  [S390] css: move chsc_private to drv_data
  [S390] css: move io_private to drv_data
  [S390] cio: move cdev pointer to io_subchannel_private
  [S390] cio: move options to io_sch_private
  [S390] cio: move asms to generic header
  [S390] cio: move orb definitions to separate header
  [S390] Write protect module text and RO data
  [S390] dasd: get rid of compile warning
  [S390] remove superfluous check from do_IRQ
  [S390] remove redundant stack check option

13 years agoMerge branch 'sh-latest' of git://git.kernel.org/pub/scm/linux/kernel/git/lethal...
Linus Torvalds [Thu, 17 Mar 2011 16:57:10 +0000 (09:57 -0700)]
Merge branch 'sh-latest' of git://git./linux/kernel/git/lethal/sh-2.6

* 'sh-latest' of git://git.kernel.org/pub/scm/linux/kernel/git/lethal/sh-2.6: (34 commits)
  sh: Convert to generic show_interrupts.
  sh: Wire up new fhandle and clock_adjtime syscalls.
  sh: modify platform_device for sh_eth driver
  sh: add GETHER's platform_device in board-sh7757lcr
  sh: update sh7757lcr_defconfig
  sh: add platform_device of tmio_mmc and sh_mmcif to sh7757lcr
  sh: dmaengine support for SH7757
  sh: add mmc clock in clock-sh7757
  sh: add spi_board_info in sh7757lcr
  sh: add platform_device for SPI
  sh: add USB_ARCH_HAS_EHCI and OHCI for SH7757
  sh: Rename cpuidle states to fit general conventions
  serial: sh-sci: fix deadlock when resuming from S3 sleep
  sh: Enable CONFIG_GCOV_PROFILE_ALL for sh
  sh: Fix up async PCIe probing on SMP.
  serial: sh-sci: Kill off the special earlyprintk device.
  serial: sh-sci: Use dev_name() for region reservations.
  serial: sh-sci: Fix up earlyprintk port mapping.
  serial: sh-sci: Limit early console to one device.
  serial: sh-sci: Fix up break timer scheduling race.
  ...

13 years agoMerge git://git.kernel.org/pub/scm/linux/kernel/git/lethal/fbdev-2.6
Linus Torvalds [Thu, 17 Mar 2011 16:56:43 +0000 (09:56 -0700)]
Merge git://git./linux/kernel/git/lethal/fbdev-2.6

* git://git.kernel.org/pub/scm/linux/kernel/git/lethal/fbdev-2.6:
  fbdev: sh_mobile_lcdc: Add YUV framebuffer support
  viafb: split pll configs up
  viafb: remove duplicated clock storage
  viafb: always return the best possible clock
  viafb: remove duplicated clock information
  fbdev: sh_mobile_lcdcfb: add backlight support
  viafb: factor lcd scaling parameters out
  viafb: strip some structures
  viafb: remove unused data_mode and device_type
  viafb: kill lcd_panel_id
  video via: make local variables static
  video via: fix iomem access
  video/via: drop deprecated (and unused) i2c_adapter.id

13 years agoRPC: killing RPC tasks races fixed
Stanislav Kinsbursky [Thu, 17 Mar 2011 15:54:23 +0000 (18:54 +0300)]
RPC: killing RPC tasks races fixed

RPC task RPC_TASK_QUEUED bit is set must be checked before trying to wake up
task rpc_killall_tasks() because task->tk_waitqueue can not be set (equal to
NULL).
Also, as Trond Myklebust mentioned, such approach (instead of checking
tk_waitqueue to NULL) allows us to "optimise away the call to
rpc_wake_up_queued_task() altogether for those
tasks that aren't queued".

Here is an example of dereferencing of tk_waitqueue equal to NULL:

CPU 0                CPU 1 CPU 2
-------------------- --------------------- --------------------------
nfs4_run_open_task
rpc_run_task
rpc_execute
rpc_set_active
rpc_make_runnable
(waiting)
rpc_async_schedule
nfs4_open_prepare
nfs_wait_on_sequence
nfs_umount_begin
rpc_killall_tasks
rpc_wake_up_task
rpc_wake_up_queued_task
spin_lock(tk_waitqueue == NULL)
BUG()
rpc_sleep_on
spin_lock(&q->lock)
__rpc_sleep_on
task->tk_waitqueue = q

Signed-off-by: Stanislav Kinsbursky <skinsbursky@openvz.org>
Cc: stable@kernel.org
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agoxprt: remove redundant check
j223yang@asset.uwaterloo.ca [Wed, 16 Mar 2011 15:16:22 +0000 (11:16 -0400)]
xprt: remove redundant check

remove redundant check.

Signed-off-by: Jinqiu Yang <crindy646@gmail.com>
Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agoSUNRPC: Convert struct rpc_xprt to use atomic_t counters
Trond Myklebust [Tue, 15 Mar 2011 23:56:30 +0000 (19:56 -0400)]
SUNRPC: Convert struct rpc_xprt to use atomic_t counters

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
13 years agoSUNRPC: Ensure we always run the tk_callback before tk_action
Trond Myklebust [Tue, 15 Mar 2011 23:56:30 +0000 (19:56 -0400)]
SUNRPC: Ensure we always run the tk_callback before tk_action

This fixes a race in which the task->tk_callback() puts the rpc_task
to sleep, setting a new callback. Under certain circumstances, the current
code may end up executing the task->tk_action before it gets round to the
callback.

Signed-off-by: Trond Myklebust <Trond.Myklebust@netapp.com>
Cc: stable@kernel.org
13 years agoMerge branch 'drm-core-next' of git://git.kernel.org/pub/scm/linux/kernel/git/airlied...
Linus Torvalds [Thu, 17 Mar 2011 16:11:39 +0000 (09:11 -0700)]
Merge branch 'drm-core-next' of git://git./linux/kernel/git/airlied/drm-2.6

* 'drm-core-next' of git://git.kernel.org/pub/scm/linux/kernel/git/airlied/drm-2.6: (177 commits)
  drm/radeon: fixup refcounts in radeon dumb create ioctl.
  drm: radeon: *_cs_packet_parse_vline() cleanup
  radeon: merge list_del()/list_add_tail() to list_move_tail()
  drm: Retry i2c transfer of EDID block after failure
  drm/radeon/kms: fix typo in atom overscan setup
  drm: Hold the mode mutex whilst probing for sysfs status
  drm/nouveau: fix __nouveau_fence_wait performance
  drm/nv40: attempt to reserve just enough vram for all 32 channels
  drm/nv50: check for vm traps on every gr irq
  drm/nv50: decode vm faults some more
  drm/nouveau: add nouveau_enum_find() util function
  drm/nouveau: properly handle pushbuffer check failures
  drm/nvc0: remove vm hack forcing large/small pages to not share a PDE
  drm/i915: disable opregion lid detection for now.
  drm/i915: Only wait on a pending flip if we intend to write to the buffer
  drm/i915/dp: Sanity check eDP existence
  drm: add cap bit to denote if dumb ioctl is available or not.
  drm/core: add ioctl to query device/driver capabilities
  drm/radeon/kms: allow max clock of 340 Mhz on hdmi 1.3+
  drm/radeon/kms: add cayman pci ids
  ...

13 years agoKVM: unbreak userspace that does not sets tss address
Gleb Natapov [Sun, 13 Mar 2011 10:34:27 +0000 (12:34 +0200)]
KVM: unbreak userspace that does not sets tss address

Commit 6440e5967bc broke old userspaces that do not set tss address
before entering vcpu. Unbreak it by setting tss address to a safe
value on the first vcpu entry. New userspaces should set tss address,
so print warning in case it doesn't.

Signed-off-by: Gleb Natapov <gleb@redhat.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
13 years agoKVM: MMU: cleanup pte write path
Xiao Guangrong [Wed, 9 Mar 2011 07:43:51 +0000 (15:43 +0800)]
KVM: MMU: cleanup pte write path

This patch does:
- call vcpu->arch.mmu.update_pte directly
- use gfn_to_pfn_atomic in update_pte path

The suggestion is from Avi.

Signed-off-by: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
13 years agoKVM: MMU: introduce a common function to get no-dirty-logged slot
Xiao Guangrong [Wed, 9 Mar 2011 07:43:00 +0000 (15:43 +0800)]
KVM: MMU: introduce a common function to get no-dirty-logged slot

Cleanup the code of pte_prefetch_gfn_to_memslot and mapping_level_dirty_bitmap

Signed-off-by: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
13 years agoKVM: fix rcu usage in init_rmode_* functions
Xiao Guangrong [Wed, 9 Mar 2011 07:41:04 +0000 (15:41 +0800)]
KVM: fix rcu usage in init_rmode_* functions

fix:
[ 3494.671786] stack backtrace:
[ 3494.671789] Pid: 10527, comm: qemu-system-x86 Not tainted 2.6.38-rc6+ #23
[ 3494.671790] Call Trace:
[ 3494.671796]  [] ? lockdep_rcu_dereference+0x9d/0xa5
[ 3494.671826]  [] ? kvm_memslots+0x6b/0x73 [kvm]
[ 3494.671834]  [] ? gfn_to_memslot+0x16/0x4f [kvm]
[ 3494.671843]  [] ? gfn_to_hva+0x16/0x27 [kvm]
[ 3494.671851]  [] ? kvm_write_guest_page+0x31/0x83 [kvm]
[ 3494.671861]  [] ? kvm_clear_guest_page+0x1a/0x1c [kvm]
[ 3494.671867]  [] ? vmx_set_tss_addr+0x83/0x122 [kvm_intel]

and:
[ 8328.789599] stack backtrace:
[ 8328.789601] Pid: 18736, comm: qemu-system-x86 Not tainted 2.6.38-rc6+ #23
[ 8328.789603] Call Trace:
[ 8328.789609]  [] ? lockdep_rcu_dereference+0x9d/0xa5
[ 8328.789621]  [] ? kvm_memslots+0x6b/0x73 [kvm]
[ 8328.789628]  [] ? gfn_to_memslot+0x16/0x4f [kvm]
[ 8328.789635]  [] ? gfn_to_hva+0x16/0x27 [kvm]
[ 8328.789643]  [] ? kvm_write_guest_page+0x31/0x83 [kvm]
[ 8328.789699]  [] ? kvm_clear_guest_page+0x1a/0x1c [kvm]
[ 8328.789713]  [] ? vmx_create_vcpu+0x316/0x3c8 [kvm_intel]

Signed-off-by: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
13 years agoKVM: fix kvmclock regression due to missing clock update
Nikola Ciprich [Wed, 9 Mar 2011 22:36:51 +0000 (23:36 +0100)]
KVM: fix kvmclock regression due to missing clock update

commit 387b9f97750444728962b236987fbe8ee8cc4f8c moved kvm_request_guest_time_update(vcpu),
breaking 32bit SMP guests using kvm-clock. Fix this by moving (new) clock update function
to proper place.

Signed-off-by: Nikola Ciprich <nikola.ciprich@linuxbox.cz>
Acked-by: Zachary Amsden <zamsden@redhat.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
13 years agoKVM: emulator: Fix permission checking in io permission bitmap
Gleb Natapov [Mon, 7 Mar 2011 12:55:07 +0000 (14:55 +0200)]
KVM: emulator: Fix permission checking in io permission bitmap

Currently if io port + len crosses 8bit boundary in io permission bitmap the
check may allow IO that otherwise should not be allowed. The patch fixes that.

Signed-off-by: Gleb Natapov <gleb@redhat.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
13 years agoKVM: emulator: Fix io permission checking for 64bit guest
Gleb Natapov [Mon, 7 Mar 2011 12:55:06 +0000 (14:55 +0200)]
KVM: emulator: Fix io permission checking for 64bit guest

Current implementation truncates upper 32bit of TR base address during IO
permission bitmap check. The patch fixes this.

Reported-and-tested-by: Francis Moreau <francis.moro@gmail.com>
Signed-off-by: Gleb Natapov <gleb@redhat.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
13 years agoKVM: SVM: Load %gs earlier if CONFIG_X86_32_LAZY_GS=n
Avi Kivity [Tue, 8 Mar 2011 14:09:51 +0000 (16:09 +0200)]
KVM: SVM: Load %gs earlier if CONFIG_X86_32_LAZY_GS=n

With CONFIG_CC_STACKPROTECTOR, we need a valid %gs at all times, so disable
lazy reload and do an eager reload immediately after the vmexit.

Reported-by: IVAN ANGELOV <ivangotoy@gmail.com>
Acked-By: Joerg Roedel <joerg.roedel@amd.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
13 years agoKVM: x86: Remove useless regs_page pointer from kvm_lapic
Takuya Yoshikawa [Sat, 5 Mar 2011 03:40:20 +0000 (12:40 +0900)]
KVM: x86: Remove useless regs_page pointer from kvm_lapic

Access to this page is mostly done through the regs member which holds
the address to this page.  The exceptions are in vmx_vcpu_reset() and
kvm_free_lapic() and these both can easily be converted to using regs.

Signed-off-by: Takuya Yoshikawa <yoshikawa.takuya@oss.ntt.co.jp>
Signed-off-by: Avi Kivity <avi@redhat.com>
13 years agoKVM: improve comment on rcu use in irqfd_deassign
Michael S. Tsirkin [Sun, 6 Mar 2011 11:03:26 +0000 (13:03 +0200)]
KVM: improve comment on rcu use in irqfd_deassign

The RCU use in kvm_irqfd_deassign is tricky: we have rcu_assign_pointer
but no synchronize_rcu: synchronize_rcu is done by kvm_irq_routing_update
which we share a spinlock with.

Fix up a comment in an attempt to make this clearer.

Signed-off-by: Michael S. Tsirkin <mst@redhat.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
13 years agoKVM: MMU: remove unused macros
Xiao Guangrong [Fri, 4 Mar 2011 11:01:39 +0000 (19:01 +0800)]
KVM: MMU: remove unused macros

These macros are not used, so removed

Signed-off-by: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
13 years agoKVM: MMU: cleanup page alloc and free
Xiao Guangrong [Fri, 4 Mar 2011 11:01:10 +0000 (19:01 +0800)]
KVM: MMU: cleanup page alloc and free

Using __get_free_page instead of alloc_page and page_address,
using free_page instead of __free_page and virt_to_page

Signed-off-by: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
13 years agoKVM: MMU: do not record gfn in kvm_mmu_pte_write
Xiao Guangrong [Fri, 4 Mar 2011 11:00:00 +0000 (19:00 +0800)]
KVM: MMU: do not record gfn in kvm_mmu_pte_write

No need to record the gfn to verifier the pte has the same mode as
current vcpu, it's because we only speculatively update the pte only
if the pte and vcpu have the same mode

Signed-off-by: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
13 years agoKVM: MMU: move mmu pages calculated out of mmu lock
Xiao Guangrong [Fri, 4 Mar 2011 10:59:21 +0000 (18:59 +0800)]
KVM: MMU: move mmu pages calculated out of mmu lock

kvm_mmu_calculate_mmu_pages need to walk all memslots and it's protected by
kvm->slots_lock, so move it out of mmu spinlock

Signed-off-by: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
13 years agoKVM: MMU: set spte accessed bit properly
Xiao Guangrong [Fri, 4 Mar 2011 10:58:02 +0000 (18:58 +0800)]
KVM: MMU: set spte accessed bit properly

Set spte accessed bit only if guest_initiated == 1 that means the really
accessed

Signed-off-by: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
13 years agoKVM: MMU: fix kvm_mmu_slot_remove_write_access dropping intermediate W bits
Xiao Guangrong [Fri, 4 Mar 2011 10:56:41 +0000 (18:56 +0800)]
KVM: MMU: fix kvm_mmu_slot_remove_write_access dropping intermediate W bits

Only remove write access in the last sptes.

Signed-off-by: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
13 years agoKVM: Start lock documentation
Jan Kiszka [Wed, 9 Feb 2011 14:11:28 +0000 (15:11 +0100)]
KVM: Start lock documentation

The goal of this document shall be
- overview of all locks used in KVM core
- provide details on the scope of each lock
- explain the lock type, specifically of a raw spin locks
- provide a lock ordering guide

Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
13 years agoKVM: better readability of efer_reserved_bits
Lai Jiangshan [Mon, 21 Feb 2011 03:51:35 +0000 (11:51 +0800)]
KVM: better readability of efer_reserved_bits

use EFER_SCE, EFER_LME and EFER_LMA instead of magic numbers.

Signed-off-by: Lai Jiangshan <laijs@cn.fujitsu.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
13 years agoKVM: Clear async page fault hash after switching to real mode
Lai Jiangshan [Mon, 21 Feb 2011 03:21:30 +0000 (11:21 +0800)]
KVM: Clear async page fault hash after switching to real mode

The hash array of async gfns may still contain some left gfns after
kvm_clear_async_pf_completion_queue() called, need to clear them.

Signed-off-by: Lai Jiangshan <laijs@cn.fujitsu.com>
Acked-by: Gleb Natapov <gleb@redhat.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
13 years agoKVM: VMX: Initialize vm86 TSS only once.
Gleb Natapov [Mon, 21 Feb 2011 10:07:59 +0000 (12:07 +0200)]
KVM: VMX: Initialize vm86 TSS only once.

Currently vm86 task is initialized on each real mode entry and vcpu
reset. Initialization is done by zeroing TSS and updating relevant
fields. But since all vcpus are using the same TSS there is a race where
one vcpu may use TSS while other vcpu is initializing it, so the vcpu
that uses TSS will see wrong TSS content and will behave incorrectly.
Fix that by initializing TSS only once.

Signed-off-by: Gleb Natapov <gleb@redhat.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
13 years agoKVM: VMX: update live TR selector if it changes in real mode
Gleb Natapov [Mon, 21 Feb 2011 10:07:58 +0000 (12:07 +0200)]
KVM: VMX: update live TR selector if it changes in real mode

When rmode.vm86 is active TR descriptor is updated with vm86 task values,
but selector is left intact. vmx_set_segment() makes sure that if TR
register is written into while vm86 is active the new values are saved
for use after vm86 is deactivated, but since selector is not updated on
vm86 activation/deactivation new value is lost. Fix this by writing new
selector into vmcs immediately.

Signed-off-by: Gleb Natapov <gleb@redhat.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
13 years agoKVM: VMX: add the __noclone attribute to vmx_vcpu_run
Lai Jiangshan [Fri, 11 Feb 2011 06:29:40 +0000 (14:29 +0800)]
KVM: VMX: add the __noclone attribute to vmx_vcpu_run

The changelog of 104f226 said "adds the __noclone attribute",
but it was missing in its patch. I think it is still needed.

Signed-off-by: Lai Jiangshan <laijs@cn.fujitsu.com>
Acked-by: Andi Kleen <ak@linux.intel.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
13 years agoKVM: x86: Convert tsc_write_lock to raw_spinlock
Jan Kiszka [Fri, 4 Feb 2011 09:49:11 +0000 (10:49 +0100)]
KVM: x86: Convert tsc_write_lock to raw_spinlock

Code under this lock requires non-preemptibility. Ensure this also over
-rt by converting it to raw spinlock.

Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
13 years agoKVM: remove isr_ack logic from PIC
Gleb Natapov [Wed, 9 Feb 2011 10:09:46 +0000 (12:09 +0200)]
KVM: remove isr_ack logic from PIC

isr_ack logic was added by e48258009d to avoid unnecessary IPIs. Back
then it made sense, but now the code checks that vcpu is ready to accept
interrupt before sending IPI, so this logic is no longer needed. The
patch removes it.

Fixes a regression with Debian/Hurd.

Signed-off-by: Gleb Natapov <gleb@redhat.com>
Reported-and-tested-by: Jonathan Nieder <jrnieder@gmail.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
13 years agoKVM: VMX: fix detection of BIOS disabling VMX
Joseph Cihula [Tue, 8 Feb 2011 19:45:56 +0000 (11:45 -0800)]
KVM: VMX: fix detection of BIOS disabling VMX

This patch fixes the logic used to detect whether BIOS has disabled VMX, for
the case where VMX is enabled only under SMX, but tboot is not active.

Signed-off-by: Joseph Cihula <joseph.cihula@intel.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
13 years agoKVM: Convert kvm_lock to raw_spinlock
Jan Kiszka [Tue, 8 Feb 2011 11:55:33 +0000 (12:55 +0100)]
KVM: Convert kvm_lock to raw_spinlock

Code under this lock requires non-preemptibility. Ensure this also over
-rt by converting it to raw spinlock.

Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
13 years agoKVM: SVM: check for progress after IRET interception
Avi Kivity [Thu, 3 Feb 2011 13:29:52 +0000 (15:29 +0200)]
KVM: SVM: check for progress after IRET interception

When we enable an NMI window, we ask for an IRET intercept, since
the IRET re-enables NMIs.  However, the IRET intercept happens before
the instruction executes, while the NMI window architecturally opens
afterwards.

To compensate for this mismatch, we only open the NMI window in the
following exit, assuming that the IRET has by then executed; however,
this assumption is not always correct; we may exit due to a host interrupt
or page fault, without having executed the instruction.

Fix by checking for forward progress by recording and comparing the IRET's
rip.  This is somewhat of a hack, since an unchaging rip does not mean that
no forward progress has been made, but is the simplest fix for now.

Signed-off-by: Avi Kivity <avi@redhat.com>
13 years agoKVM: Fix race between nmi injection and enabling nmi window
Avi Kivity [Thu, 3 Feb 2011 13:07:07 +0000 (15:07 +0200)]
KVM: Fix race between nmi injection and enabling nmi window

The interrupt injection logic looks something like

  if an nmi is pending, and nmi injection allowed
    inject nmi
  if an nmi is pending
    request exit on nmi window

the problem is that "nmi is pending" can be set asynchronously by
the PIT; if it happens to fire between the two if statements, we
will request an nmi window even though nmi injection is allowed.  On
SVM, this has disasterous results, since it causes eflags.TF to be
set in random guest code.

The fix is simple; make nmi_pending synchronous using the standard
vcpu->requests mechanism; this ensures the code above is completely
synchronous wrt nmi_pending.

Signed-off-by: Avi Kivity <avi@redhat.com>
13 years agoKVM: use yield_to instead of sleep in kvm_vcpu_on_spin
Rik van Riel [Tue, 1 Feb 2011 14:53:28 +0000 (09:53 -0500)]
KVM: use yield_to instead of sleep in kvm_vcpu_on_spin

Instead of sleeping in kvm_vcpu_on_spin, which can cause gigantic
slowdowns of certain workloads, we instead use yield_to to get
another VCPU in the same KVM guest to run sooner.

This seems to give a 10-15% speedup in certain workloads.

Signed-off-by: Rik van Riel <riel@redhat.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
13 years agoKVM: keep track of which task is running a KVM vcpu
Rik van Riel [Tue, 1 Feb 2011 14:52:41 +0000 (09:52 -0500)]
KVM: keep track of which task is running a KVM vcpu

Keep track of which task is running a KVM vcpu.  This helps us
figure out later what task to wake up if we want to boost a
vcpu that got preempted.

Unfortunately there are no guarantees that the same task
always keeps the same vcpu, so we can only track the task
across a single "run" of the vcpu.

Signed-off-by: Rik van Riel <riel@redhat.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
13 years agoexport pid symbols needed for kvm_vcpu_on_spin
Rik van Riel [Tue, 1 Feb 2011 14:51:46 +0000 (09:51 -0500)]
export pid symbols needed for kvm_vcpu_on_spin

Export the symbols required for a race-free kvm_vcpu_on_spin.

Signed-off-by: Rik van Riel <riel@redhat.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
13 years agoKVM: Drop ad-hoc vendor specific instruction restriction
Avi Kivity [Tue, 1 Feb 2011 14:32:04 +0000 (16:32 +0200)]
KVM: Drop ad-hoc vendor specific instruction restriction

Use the new support in the emulator, and drop the ad-hoc code in x86.c.

Signed-off-by: Avi Kivity <avi@redhat.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
13 years agoKVM: x86 emulator: vendor specific instructions
Avi Kivity [Tue, 1 Feb 2011 14:32:03 +0000 (16:32 +0200)]
KVM: x86 emulator: vendor specific instructions

Mark some instructions as vendor specific, and allow the caller to request
emulation only of vendor specific instructions.  This is useful in some
circumstances (responding to a #UD fault).

Signed-off-by: Avi Kivity <avi@redhat.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
13 years agoKVM: Drop bogus x86_decode_insn() error check
Avi Kivity [Tue, 1 Feb 2011 14:32:02 +0000 (16:32 +0200)]
KVM: Drop bogus x86_decode_insn() error check

x86_decode_insn() doesn't return X86EMUL_* values, so the check
for X86EMUL_PROPOGATE_FAULT will always fail.  There is a proper
check later on, so there is no need for a replacement for this
code.

Signed-off-by: Avi Kivity <avi@redhat.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
13 years agoKVM: x86: Drop obsolete warning about INIT on runnable VCPU
Jan Kiszka [Tue, 1 Feb 2011 12:27:15 +0000 (13:27 +0100)]
KVM: x86: Drop obsolete warning about INIT on runnable VCPU

This warning was once used for debugging QEMU user space. Though
uncommon, it is actually possible to send an INIT request to a running
VCPU. So better drop this warning before someone misuses it to flood
kernel logs this way.

Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
13 years agoKVM: x86: release kvmclock page on reset
Glauber Costa [Tue, 1 Feb 2011 19:16:40 +0000 (14:16 -0500)]
KVM: x86: release kvmclock page on reset

When a vcpu is reset, kvmclock page keeps being written to this days.
This is wrong and inconsistent: a cpu reset should take it to its
initial state.

Signed-off-by: Glauber Costa <glommer@redhat.com>
CC: Jan Kiszka <jan.kiszka@siemens.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
13 years agomm: remove is_hwpoison_address
Huang Ying [Sun, 30 Jan 2011 03:15:49 +0000 (11:15 +0800)]
mm: remove is_hwpoison_address

Unused.

Signed-off-by: Huang Ying <ying.huang@intel.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
13 years agoKVM: Replace is_hwpoison_address with __get_user_pages
Huang Ying [Sun, 30 Jan 2011 03:15:49 +0000 (11:15 +0800)]
KVM: Replace is_hwpoison_address with __get_user_pages

is_hwpoison_address only checks whether the page table entry is
hwpoisoned, regardless the memory page mapped.  While __get_user_pages
will check both.

QEMU will clear the poisoned page table entry (via unmap/map) to make
it possible to allocate a new memory page for the virtual address
across guest rebooting.  But it is also possible that the underlying
memory page is kept poisoned even after the corresponding page table
entry is cleared, that is, a new memory page can not be allocated.
__get_user_pages can catch these situations.

Signed-off-by: Huang Ying <ying.huang@intel.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
13 years agomm: make __get_user_pages return -EHWPOISON for HWPOISON page optionally
Huang Ying [Sun, 30 Jan 2011 03:15:48 +0000 (11:15 +0800)]
mm: make __get_user_pages return -EHWPOISON for HWPOISON page optionally

Make __get_user_pages return -EHWPOISON for HWPOISON page only if
FOLL_HWPOISON is specified.  With this patch, the interested callers
can distinguish HWPOISON pages from general FAULT pages, while other
callers will still get -EFAULT for all these pages, so the user space
interface need not to be changed.

This feature is needed by KVM, where UCR MCE should be relayed to
guest for HWPOISON page, while instruction emulation and MMIO will be
tried for general FAULT page.

The idea comes from Andrew Morton.

Signed-off-by: Huang Ying <ying.huang@intel.com>
Cc: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
13 years agomm: export __get_user_pages
Huang Ying [Sun, 30 Jan 2011 03:15:47 +0000 (11:15 +0800)]
mm: export __get_user_pages

In most cases, get_user_pages and get_user_pages_fast should be used
to pin user pages in memory.  But sometimes, some special flags except
FOLL_GET, FOLL_WRITE and FOLL_FORCE are needed, for example in
following patch, KVM needs FOLL_HWPOISON.  To support these users,
__get_user_pages is exported directly.

There are some symbol name conflicts in infiniband driver, fixed them too.

Signed-off-by: Huang Ying <ying.huang@intel.com>
CC: Andrew Morton <akpm@linux-foundation.org>
CC: Michel Lespinasse <walken@google.com>
CC: Roland Dreier <roland@kernel.org>
CC: Ralph Campbell <infinipath@qlogic.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
13 years agoKVM: x86: handle guest access to BBL_CR_CTL3 MSR
john cooper [Fri, 21 Jan 2011 05:21:00 +0000 (00:21 -0500)]
KVM: x86: handle guest access to BBL_CR_CTL3 MSR

A correction to Intel cpu model CPUID data (patch queued)
caused winxp to BSOD when booted with a Penryn model.
This was traced to the CPUID "model" field correction from
6 -> 23 (as is proper for a Penryn class of cpu).  Only in
this case does the problem surface.

The cause for this failure is winxp accessing the BBL_CR_CTL3
MSR which is unsupported by current kvm, appears to be a
legacy MSR not fully characterized yet existing in current
silicon, and is apparently carried forward in MSR space to
accommodate vintage code as here.  It is not yet conclusive
whether this MSR implements any of its legacy functionality
or is just an ornamental dud for compatibility.  While I
found no silicon version specific documentation link to
this MSR, a general description exists in Intel's developer's
reference which agrees with the functional behavior of
other bootloader/kernel code I've examined accessing
BBL_CR_CTL3.  Regrettably winxp appears to be setting bit #19
called out as "reserved" in the above document.

So to minimally accommodate this MSR, kvm msr get will provide
the equivalent mock data and kvm msr write will simply toss the
guest passed data without interpretation.  While this treatment
of BBL_CR_CTL3 addresses the immediate problem, the approach may
be modified pending clarification from Intel.

Signed-off-by: john cooper <john.cooper@redhat.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
13 years agoKVM: make make_all_cpus_request() lockless
Xiao Guangrong [Wed, 12 Jan 2011 07:41:22 +0000 (15:41 +0800)]
KVM: make make_all_cpus_request() lockless

Now, we have 'vcpu->mode' to judge whether need to send ipi to other
cpus, this way is very exact, so checking request bit is needless,
then we can drop the spinlock let it's collateral

Signed-off-by: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
13 years agoKVM: Add "exiting guest mode" state
Xiao Guangrong [Wed, 12 Jan 2011 07:40:31 +0000 (15:40 +0800)]
KVM: Add "exiting guest mode" state

Currently we keep track of only two states: guest mode and host
mode.  This patch adds an "exiting guest mode" state that tells
us that an IPI will happen soon, so unless we need to wait for the
IPI, we can avoid it completely.

Also
1: No need atomically to read/write ->mode in vcpu's thread

2: reorganize struct kvm_vcpu to make ->mode and ->requests
   in the same cache line explicitly

Signed-off-by: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
13 years agoKVM: fix build warning within __kvm_set_memory_region() on s390
Heiko Carstens [Mon, 17 Jan 2011 20:21:08 +0000 (21:21 +0100)]
KVM: fix build warning within __kvm_set_memory_region() on s390

Get rid of this warning:

  CC      arch/s390/kvm/../../../virt/kvm/kvm_main.o
arch/s390/kvm/../../../virt/kvm/kvm_main.c:596:12: warning: 'kvm_create_dirty_bitmap' defined but not used

The only caller of the function is within a !CONFIG_S390 section, so add the
same ifdef around kvm_create_dirty_bitmap() as well.

Signed-off-by: Heiko Carstens <heiko.carstens@de.ibm.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
13 years agoKVM: x86: Remove user space triggerable MCE error message
Jan Kiszka [Sat, 15 Jan 2011 09:00:53 +0000 (10:00 +0100)]
KVM: x86: Remove user space triggerable MCE error message

This case is a pure user space error we do not need to record. Moreover,
it can be misused to flood the kernel log. Remove it.

Signed-off-by: Jan Kiszka <jan.kiszka@siemens.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
13 years agoKVM: fix rcu usage warning in kvm_arch_vcpu_ioctl_set_sregs()
Xiao Guangrong [Wed, 12 Jan 2011 07:39:18 +0000 (15:39 +0800)]
KVM: fix rcu usage warning in kvm_arch_vcpu_ioctl_set_sregs()

Fix:

[ 1001.499596] ===================================================
[ 1001.499599] [ INFO: suspicious rcu_dereference_check() usage. ]
[ 1001.499601] ---------------------------------------------------
[ 1001.499604] include/linux/kvm_host.h:301 invoked rcu_dereference_check() without protection!
......
[ 1001.499636] Pid: 6035, comm: qemu-system-x86 Not tainted 2.6.37-rc6+ #62
[ 1001.499638] Call Trace:
[ 1001.499644]  [] lockdep_rcu_dereference+0x9d/0xa5
[ 1001.499653]  [] gfn_to_memslot+0x8d/0xc8 [kvm]
[ 1001.499661]  [] gfn_to_hva+0x16/0x3f [kvm]
[ 1001.499669]  [] kvm_read_guest_page+0x1e/0x5e [kvm]
[ 1001.499681]  [] kvm_read_guest_page_mmu+0x53/0x5e [kvm]
[ 1001.499699]  [] load_pdptrs+0x3f/0x9c [kvm]
[ 1001.499705]  [] ? vmx_set_cr0+0x507/0x517 [kvm_intel]
[ 1001.499717]  [] kvm_arch_vcpu_ioctl_set_sregs+0x1f3/0x3c0 [kvm]
[ 1001.499727]  [] kvm_vcpu_ioctl+0x6a5/0xbc5 [kvm]

Signed-off-by: Xiao Guangrong <xiaoguangrong@cn.fujitsu.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
13 years agoKVM: VMX: Avoid atomic operation in vmx_vcpu_run
Avi Kivity [Thu, 6 Jan 2011 16:09:12 +0000 (18:09 +0200)]
KVM: VMX: Avoid atomic operation in vmx_vcpu_run

Instead of exchanging the guest and host rcx, have separate storage
for each.  This allows us to avoid using the xchg instruction, which
is is a little slower than normal operations.

Signed-off-by: Avi Kivity <avi@redhat.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
13 years agoKVM: VMX: Simplify saving guest rcx in vmx_vcpu_run
Avi Kivity [Thu, 6 Jan 2011 16:09:11 +0000 (18:09 +0200)]
KVM: VMX: Simplify saving guest rcx in vmx_vcpu_run

Change

  push top-of-stack
  pop guest-rcx
  pop dummy

to

  pop guest-rcx

which is the same thing, only simpler.

Signed-off-by: Avi Kivity <avi@redhat.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
13 years agoKVM: VMX: increase ple_gap default to 128
Rik van Riel [Tue, 4 Jan 2011 14:51:33 +0000 (09:51 -0500)]
KVM: VMX: increase ple_gap default to 128

On some CPUs, a ple_gap of 41 is simply insufficient to ever trigger
PLE exits, even with the minimalistic PLE test from kvm-unit-tests.

http://git.kernel.org/?p=virt/kvm/kvm-unit-tests.git;a=commitdiff;h=eda71b28fa122203e316483b35f37aaacd42f545

For example, the Xeon X5670 CPU needs a ple_gap of at least 48 in
order to get pause loop exits:

# modprobe kvm_intel ple_gap=47
# taskset 1 /usr/local/bin/qemu-system-x86_64 \
  -device testdev,chardev=log -chardev stdio,id=log \
  -kernel x86/vmexit.flat -append ple-round-robin -smp 2
VNC server running on `::1:5900'
enabling apic
enabling apic
ple-round-robin 58298446
# rmmod kvm_intel
# modprobe kvm_intel ple_gap=48
# taskset 1 /usr/local/bin/qemu-system-x86_64 \
   -device testdev,chardev=log -chardev stdio,id=log \
   -kernel x86/vmexit.flat -append ple-round-robin -smp 2
VNC server running on `::1:5900'
enabling apic
enabling apic
ple-round-robin 36616

Increase the ple_gap to 128 to be on the safe side.

Signed-off-by: Rik van Riel <riel@redhat.com>
Acked-by: Zhai, Edwin <edwin.zhai@intel.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
13 years agoKVM: SVM: Add support for perf-kvm
Joerg Roedel [Fri, 14 Jan 2011 15:45:02 +0000 (16:45 +0100)]
KVM: SVM: Add support for perf-kvm

This patch adds the necessary code to run perf-kvm on AMD
machines.

Signed-off-by: Joerg Roedel <joerg.roedel@amd.com>
Signed-off-by: Avi Kivity <avi@redhat.com>
13 years agoKVM: VMX: Avoid leaking fake realmode state to userspace
Avi Kivity [Mon, 3 Jan 2011 12:28:52 +0000 (14:28 +0200)]
KVM: VMX: Avoid leaking fake realmode state to userspace

When emulating real mode, we fake some state:

 - tr.base points to a fake vm86 tss
 - segment registers are made to conform to vm86 restrictions

change vmx_get_segment() not to expose this fake state to userspace;
instead, return the original state.

Signed-off-by: Avi Kivity <avi@redhat.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
13 years agoKVM: VMX: Save and restore tr selector across mode switches
Avi Kivity [Mon, 3 Jan 2011 12:28:51 +0000 (14:28 +0200)]
KVM: VMX: Save and restore tr selector across mode switches

When emulating real mode we play with tr hidden state, but leave
tr.selector alone.  That works well, except for save/restore, since
loading TR writes it to the hidden state in vmx->rmode.

Fix by also saving and restoring the tr selector; this makes things
more consistent and allows migration to work during the early
boot stages of Windows XP.

Signed-off-by: Avi Kivity <avi@redhat.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
13 years agoKVM: PPC: Fix SPRG get/set for Book3S and BookE
Peter Tyser [Wed, 29 Dec 2010 19:51:25 +0000 (13:51 -0600)]
KVM: PPC: Fix SPRG get/set for Book3S and BookE

Previously SPRGs 4-7 were improperly read and written in
kvm_arch_vcpu_ioctl_get_regs() and kvm_arch_vcpu_ioctl_set_regs();

Signed-off-by: Alexander Graf <agraf@suse.de>
Signed-off-by: Peter Tyser <ptyser@xes-inc.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
13 years agoKVM guest: Fix section mismatch derived from kvm_guest_cpu_online()
Sedat Dilek [Sun, 2 Jan 2011 23:01:29 +0000 (00:01 +0100)]
KVM guest: Fix section mismatch derived from kvm_guest_cpu_online()

WARNING: arch/x86/built-in.o(.text+0x1bb74): Section mismatch in reference from the function kvm_guest_cpu_online() to the function .cpuinit.text:kvm_guest_cpu_init()
The function kvm_guest_cpu_online() references
the function __cpuinit kvm_guest_cpu_init().
This is often because kvm_guest_cpu_online lacks a __cpuinit
annotation or the annotation of kvm_guest_cpu_init is wrong.

This patch fixes the warning.

Tested with linux-next (next-20101231)

Signed-off-by: Sedat Dilek <sedat.dilek@gmail.com>
Acked-by: Rik van Riel <riel@redhat.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
13 years agoKVM: MMU: Don't flush shadow when enabling dirty tracking
Avi Kivity [Mon, 27 Dec 2010 10:08:45 +0000 (12:08 +0200)]
KVM: MMU: Don't flush shadow when enabling dirty tracking

Instead, drop large mappings, which were the reason we dropped shadow.

Signed-off-by: Avi Kivity <avi@redhat.com>
Signed-off-by: Marcelo Tosatti <mtosatti@redhat.com>
13 years agoMerge branch 'for_next' into for_linus
Jan Kara [Thu, 17 Mar 2011 15:44:22 +0000 (16:44 +0100)]
Merge branch 'for_next' into for_linus

13 years agoamd64_edac: Fix decode_syndrome types
Borislav Petkov [Wed, 23 Feb 2011 16:41:50 +0000 (17:41 +0100)]
amd64_edac: Fix decode_syndrome types

Those should all be unsigned.

Signed-off-by: Borislav Petkov <borislav.petkov@amd.com>
13 years agoamd64_edac: Fix DCT argument type
Borislav Petkov [Wed, 23 Feb 2011 16:25:12 +0000 (17:25 +0100)]
amd64_edac: Fix DCT argument type

Fix amd64_debug_display_dimm_sizes() arguments order per convention (pvt
is always first). Also, the now second arg denotes the DCT so adjust its
type.

Signed-off-by: Borislav Petkov <borislav.petkov@amd.com>
13 years agoamd64_edac: Fix ranges signedness
Borislav Petkov [Mon, 21 Feb 2011 18:49:01 +0000 (19:49 +0100)]
amd64_edac: Fix ranges signedness

The dram ranges make sense only as an unsigned type.

Signed-off-by: Borislav Petkov <borislav.petkov@amd.com>
13 years agoamd64_edac: Drop local variable
Borislav Petkov [Mon, 21 Feb 2011 18:43:02 +0000 (19:43 +0100)]
amd64_edac: Drop local variable

Use the macro directly instead

Signed-off-by: Borislav Petkov <borislav.petkov@amd.com>
13 years agoamd64_edac: Fix PCI config addressing types
Borislav Petkov [Mon, 21 Feb 2011 18:37:24 +0000 (19:37 +0100)]
amd64_edac: Fix PCI config addressing types

Adjust argument types to the PCI config API's types.

Signed-off-by: Borislav Petkov <borislav.petkov@amd.com>
13 years agoamd64_edac: Fix DRAM base macros
Borislav Petkov [Mon, 21 Feb 2011 18:33:10 +0000 (19:33 +0100)]
amd64_edac: Fix DRAM base macros

Return unsigned u8 values only.

Signed-off-by: Borislav Petkov <borislav.petkov@amd.com>
13 years agoamd64_edac: Fix node id signedness
Borislav Petkov [Mon, 21 Feb 2011 17:55:00 +0000 (18:55 +0100)]
amd64_edac: Fix node id signedness

A node id can never be negative since we use it as an index into
the DRAM ranges array. This also makes one of the BUG_ON conditions
redundant.

Signed-off-by: Borislav Petkov <borislav.petkov@amd.com>
13 years agoamd64_edac: Drop redundant declarations
Borislav Petkov [Mon, 21 Feb 2011 17:05:57 +0000 (18:05 +0100)]
amd64_edac: Drop redundant declarations

Those were moved to the mce_amd.h header.

Signed-off-by: Borislav Petkov <borislav.petkov@amd.com>
13 years agoamd64_edac: Enable driver on F15h
Borislav Petkov [Wed, 19 Jan 2011 17:15:10 +0000 (18:15 +0100)]
amd64_edac: Enable driver on F15h

Add the PCI device ids required for driver registration. Remove
pvt->ctl_name and use the family descriptor directly, instead. Then,
bump driver version and fixup its format. Finally, enable DRAM ECC
decoding on F15h.

Signed-off-by: Borislav Petkov <borislav.petkov@amd.com>
13 years agoamd64_edac: Adjust ECC symbol size to F15h
Borislav Petkov [Wed, 19 Jan 2011 19:35:12 +0000 (20:35 +0100)]
amd64_edac: Adjust ECC symbol size to F15h

F15h has the same ECC symbol size options as F10h revD and later so
adjust checks to that. Simplify code a bit.

Signed-off-by: Borislav Petkov <borislav.petkov@amd.com>
13 years agoamd64_edac: Simplify scrubrate setting
Borislav Petkov [Wed, 19 Jan 2011 19:02:38 +0000 (20:02 +0100)]
amd64_edac: Simplify scrubrate setting

Drop per-instance variable and compute min scrubrate dynamically.

Signed-off-by: Borislav Petkov <borislav.petkov@amd.com>
13 years agoPCI: Rename CPU PCI id define
Borislav Petkov [Wed, 19 Jan 2011 17:22:11 +0000 (18:22 +0100)]
PCI: Rename CPU PCI id define

With increasing number of PCI function ids, add the PCI function id
in the define name instead of its symbolic name in the BKDG for more
clarity.

Acked-by: Ingo Molnar <mingo@elte.hu>
Acked-by: Jesse Barnes <jbarnes@virtuousgeek.org>
Signed-off-by: Borislav Petkov <borislav.petkov@amd.com>
13 years agoamd64_edac: Improve DRAM address mapping
Borislav Petkov [Tue, 18 Jan 2011 18:16:08 +0000 (19:16 +0100)]
amd64_edac: Improve DRAM address mapping

Drop static tables which map the bits in F2x80 to a chip select size in
favor of functions doing the mapping with some bit fiddling. Also, add
F15 support.

Signed-off-by: Borislav Petkov <borislav.petkov@amd.com>
13 years agoamd64_edac: Sanitize ->read_dram_ctl_register
Borislav Petkov [Mon, 17 Jan 2011 16:52:57 +0000 (17:52 +0100)]
amd64_edac: Sanitize ->read_dram_ctl_register

This function is relevant for F10h and higher, and it has only one
callsite so drop its function pointer from the low_ops struct.

Signed-off-by: Borislav Petkov <borislav.petkov@amd.com>
13 years agoamd64_edac: Adjust sys_addr to chip select conversion routine to F15h
Borislav Petkov [Mon, 17 Jan 2011 14:59:58 +0000 (15:59 +0100)]
amd64_edac: Adjust sys_addr to chip select conversion routine to F15h

F15h sys_addr to chip select mapping is almost identical to F10h's so
reuse that. Rename functions on that path accordingly.

Signed-off-by: Borislav Petkov <borislav.petkov@amd.com>
13 years agoamd64_edac: Beef up early exit reporting
Borislav Petkov [Mon, 17 Jan 2011 12:03:26 +0000 (13:03 +0100)]
amd64_edac: Beef up early exit reporting

Add paranoid checks for the sys address before going off and decoding
it.

Signed-off-by: Borislav Petkov <borislav.petkov@amd.com>
13 years agoamd64_edac: Revamp online spare handling
Borislav Petkov [Thu, 13 Jan 2011 17:02:22 +0000 (18:02 +0100)]
amd64_edac: Revamp online spare handling

Replace per-DCT macros with smarter ones, drop hack and look for the
spare rank on all chip selects on a channel.

Signed-off-by: Borislav Petkov <borislav.petkov@amd.com>