1. 11 May, 2009 1 commit
    • Yinghai Lu's avatar
      x86/acpi: remove irq-compression trick on 32-bit · b9e0353f
      Yinghai Lu authored
      
      
      We already have a per cpu vector on 32-bit via recent changes, and
      don't need this trick any more (which trick obfuscates the real GSI
      mappings and which only triggers on larger systems to begin with):
      
      On 3 ioapic system (24 per ioapic) before patch I got:
      
      ACPI: PCI Interrupt Link [ILSB] enabled at IRQ 71
      IOAPIC[2]: Set routing entry (10-23 -> 0xa9 -> IRQ 64 Mode:1 Active:1)
      pci 0000:80:01.1: PCI INT A -> Link[ILSB] -> GSI 71 (level, low) -> IRQ 64
      ACPI: PCI Interrupt Link [LE5B] enabled at IRQ 67
      IOAPIC[2]: Set routing entry (10-19 -> 0xb1 -> IRQ 65 Mode:1 Active:1)
      pci 0000:83:00.0: PCI INT B -> Link[LE5B] -> GSI 67 (level, low) -> IRQ 65
      ACPI: PCI Interrupt Link [LE5A] enabled at IRQ 66
      IOAPIC[2]: Set routing entry (10-18 -> 0xb9 -> IRQ 66 Mode:1 Active:1)
      pci 0000:83:00.1: PCI INT A -> Link[LE5A] -> GSI 66 (level, low) -> IRQ 66
      ACPI: PCI Interrupt Link [LE5D] enabled at IRQ 65
      IOAPIC[2]: Set routing entry (10-17 -> 0xc1 -> IRQ 67 Mode:1 Active:1)
      pci 0000:84:00.0: PCI INT B -> Link[LE5D] -> GSI 65 (level, low) -> IRQ 67
      ACPI: PCI Interrupt Link [LE5C] enabled at IRQ 64
      IOAPIC[2]: Set routing entry (10-16 -> 0xc9 -> IRQ 68 Mode:1 Active:1)
      pci 0000:84:00.1: PCI INT A -> Link[LE5C] -> GSI 64 (level, low) -> IRQ 68
      pci 0000:87:00.0: PCI INT B -> Link[LE5A] -> GSI 66 (level, low) -> IRQ 66
      pci 0000:87:00.1: PCI INT A -> Link[LE5D] -> GSI 65 (level, low) -> IRQ 67
      pci 0000:88:00.0: PCI INT B -> Link[LE5C] -> GSI 64 (level, low) -> IRQ 68
      pci 0000:88:00.1: PCI INT A -> Link[LE5B] -> GSI 67 (level, low) -> IRQ 65
      pci 0000:8b:00.0: PCI INT B -> Link[LE5A] -> GSI 66 (level, low) -> IRQ 66
      pci 0000:8b:00.1: PCI INT A -> Link[LE5D] -> GSI 65 (level, low) -> IRQ 67
      pci 0000:8c:00.0: PCI INT B -> Link[LE5C] -> GSI 64 (level, low) -> IRQ 68
      pci 0000:8c:00.1: PCI INT A -> Link[LE5B] -> GSI 67 (level, low) -> IRQ 65
      
      after the patch we get:
      
      ACPI: PCI Interrupt Link [ILSB] enabled at IRQ 71
      IOAPIC[2]: Set routing entry (10-23 -> 0xa9 -> IRQ 71 Mode:1 Active:1)
      pci 0000:80:01.1: PCI INT A -> Link[ILSB] -> GSI 71 (level, low) -> IRQ 71
      ACPI: PCI Interrupt Link [LE5B] enabled at IRQ 67
      IOAPIC[2]: Set routing entry (10-19 -> 0xb1 -> IRQ 67 Mode:1 Active:1)
      pci 0000:83:00.0: PCI INT B -> Link[LE5B] -> GSI 67 (level, low) -> IRQ 67
      ACPI: PCI Interrupt Link [LE5A] enabled at IRQ 66
      IOAPIC[2]: Set routing entry (10-18 -> 0xb9 -> IRQ 66 Mode:1 Active:1)
      pci 0000:83:00.1: PCI INT A -> Link[LE5A] -> GSI 66 (level, low) -> IRQ 66
      ACPI: PCI Interrupt Link [LE5D] enabled at IRQ 65
      IOAPIC[2]: Set routing entry (10-17 -> 0xc1 -> IRQ 65 Mode:1 Active:1)
      pci 0000:84:00.0: PCI INT B -> Link[LE5D] -> GSI 65 (level, low) -> IRQ 65
      ACPI: PCI Interrupt Link [LE5C] enabled at IRQ 64
      IOAPIC[2]: Set routing entry (10-16 -> 0xc9 -> IRQ 64 Mode:1 Active:1)
      pci 0000:84:00.1: PCI INT A -> Link[LE5C] -> GSI 64 (level, low) -> IRQ 64
      pci 0000:87:00.0: PCI INT B -> Link[LE5A] -> GSI 66 (level, low) -> IRQ 66
      pci 0000:87:00.1: PCI INT A -> Link[LE5D] -> GSI 65 (level, low) -> IRQ 65
      pci 0000:88:00.0: PCI INT B -> Link[LE5C] -> GSI 64 (level, low) -> IRQ 64
      pci 0000:88:00.1: PCI INT A -> Link[LE5B] -> GSI 67 (level, low) -> IRQ 67
      pci 0000:8b:00.0: PCI INT B -> Link[LE5A] -> GSI 66 (level, low) -> IRQ 66
      pci 0000:8b:00.1: PCI INT A -> Link[LE5D] -> GSI 65 (level, low) -> IRQ 65
      pci 0000:8c:00.0: PCI INT B -> Link[LE5C] -> GSI 64 (level, low) -> IRQ 64
      pci 0000:8c:00.1: PCI INT A -> Link[LE5B] -> GSI 67 (level, low) -> IRQ 67
      
      As it can be seen that GSIs now get mapped lineary.
      
      [ Impact: simplify irq number mapping on bigger 32-bit systems ]
      Signed-off-by: default avatarYinghai Lu <yinghai@kernel.org>
      Cc: Andrew Morton <akpm@linux-foundation.org>
      Cc: Jesse Barnes <jbarnes@virtuousgeek.org>
      Cc: Len Brown <lenb@kernel.org>
      LKML-Reference: <4A01C35C.7060207@kernel.org>
      Signed-off-by: default avatarIngo Molnar <mingo@elte.hu>
      b9e0353f
  2. 06 May, 2009 1 commit
    • Ingo Molnar's avatar
      irq: change ->set_affinity() to return status, fix · b2e5d858
      Ingo Molnar authored
      This build failure:
      
       arch/powerpc/sysdev/mpic.c:810: error: conflicting types for 'mpic_set_affinity'
       arch/powerpc/sysdev/mpic.h:39: error: previous declaration of 'mpic_set_affinity' was here
       make[2]: *** [arch/powerpc/sysdev/mpic.o] Error 1
       make[2]: *** Waiting for unfinished jobs....
      
      Triggers because the function prototype was not updated when the
      function call signature got changed by:
      
         d5dedd45
      
      : irq: change ->set_affinity() to return status
      
      [ Impact: build fix on powerpc ]
      
      Cc: Benjamin Herrenschmidt <benh@kernel.crashing.org>
      Cc: Yinghai Lu <yinghai@kernel.org>
      Cc: Andrew Morton <akpm@linux-foundation.org>
      Cc: Rusty Russell <rusty@rustcorp.com.au>
      Cc: linux-arch@vger.kernel.org
      LKML-Reference: <49F654E9.4070809@kernel.org>
      Signed-off-by: default avatarIngo Molnar <mingo@elte.hu>
      b2e5d858
  3. 01 May, 2009 2 commits
    • Ingo Molnar's avatar
      Merge branch 'x86/apic' into irq/numa · 4420471f
      Ingo Molnar authored
      
      
      Conflicts:
      	arch/x86/kernel/apic/io_apic.c
      
      Merge reason: non-trivial interaction between ongoing work in io_apic.c
                    and the NUMA migration feature in the irq tree.
      Signed-off-by: default avatarIngo Molnar <mingo@elte.hu>
      4420471f
    • Yinghai Lu's avatar
      x86/irq: use move_irq_desc() in create_irq_nr() · 15e957d0
      Yinghai Lu authored
      
      
      move_irq_desc() will try to move irq_desc to the home node if
      the allocated one is not correct, in create_irq_nr().
      
      ( This can happen on devices that are on different nodes that
        are using MSI, when drivers are loaded and unloaded randomly. )
      
      v2: fix non-smp build
      v3: add NUMA_IRQ_DESC to eliminate #ifdefs
      
      [ Impact: improve irq descriptor locality on NUMA systems ]
      Signed-off-by: default avatarYinghai Lu <yinghai@kernel.org>
      Cc: Andrew Morton <akpm@linux-foundation.org>
      Cc: Suresh Siddha <suresh.b.siddha@intel.com>
      Cc: "Eric W. Biederman" <ebiederm@xmission.com>
      Cc: Rusty Russell <rusty@rustcorp.com.au>
      LKML-Reference: <49F95EAE.2050903@kernel.org>
      Signed-off-by: default avatarIngo Molnar <mingo@elte.hu>
      15e957d0
  4. 28 Apr, 2009 9 commits
    • Yinghai Lu's avatar
      irq: make ht irq_desc more numa aware · 56b581ea
      Yinghai Lu authored
      
      
      Try to get irq_desc on the same node as create_irq_nr().
      
      [ Impact: optimization, make HT IRQs more NUMA-aware ]
      Signed-off-by: default avatarYinghai Lu <yinghai@kernel.org>
      Cc: Andrew Morton <akpm@linux-foundation.org>
      Cc: Suresh Siddha <suresh.b.siddha@intel.com>
      Cc: "Eric W. Biederman" <ebiederm@xmission.com>
      Cc: Rusty Russell <rusty@rustcorp.com.au>
      LKML-Reference: <49F655B6.8020109@kernel.org>
      Signed-off-by: default avatarIngo Molnar <mingo@elte.hu>
      56b581ea
    • Yinghai Lu's avatar
      x86/irq: change MSI irq_desc to be more numa aware · d047f53a
      Yinghai Lu authored
      
      
      Try to get irq_desc on the home node in create_irq_nr().
      
      v2: don't check if we can move it when sparse_irq is not used
      v3: use move_irq_des, if that node is not what we want
      
      [ Impact: optimization, make MSI IRQ descriptors more NUMA aware ]
      Signed-off-by: default avatarYinghai Lu <yinghai@kernel.org>
      Cc: Andrew Morton <akpm@linux-foundation.org>
      Cc: Suresh Siddha <suresh.b.siddha@intel.com>
      Cc: "Eric W. Biederman" <ebiederm@xmission.com>
      Cc: Rusty Russell <rusty@rustcorp.com.au>
      LKML-Reference: <49F6559F.7070005@kernel.org>
      Signed-off-by: default avatarIngo Molnar <mingo@elte.hu>
      d047f53a
    • Yinghai Lu's avatar
      irq: change io_apic_set_pci_routing() to use device parameter · 024154cf
      Yinghai Lu authored
      
      
      Make actual use of the device parameter passed down to
      io_apic_set_pci_routing() - to have the IRQ descriptor
      on the home node of the device.
      
      If no device has been passed down, we assume it's a platform
      device and use the boot node ID for the IRQ descriptor.
      
      [ Impact: optimization, make IO-APIC code more NUMA aware ]
      Signed-off-by: default avatarYinghai Lu <yinghai@kernel.org>
      Cc: Andrew Morton <akpm@linux-foundation.org>
      Cc: Suresh Siddha <suresh.b.siddha@intel.com>
      Cc: "Eric W. Biederman" <ebiederm@xmission.com>
      Cc: Rusty Russell <rusty@rustcorp.com.au>
      LKML-Reference: <49F6557E.3080101@kernel.org>
      Signed-off-by: default avatarIngo Molnar <mingo@elte.hu>
      024154cf
    • Yinghai Lu's avatar
      irq: change ACPI GSI APIs to also take a device argument · a2f809b0
      Yinghai Lu authored
      
      
      We want to use dev_to_node() later on, to be aware of the 'home node'
      of the GSI in question.
      
      [ Impact: cleanup, prepare the IRQ code to be more NUMA aware ]
      Signed-off-by: default avatarYinghai Lu <yinghai@kernel.org>
      Acked-by: default avatarLen Brown <lenb@kernel.org>
      Cc: Andrew Morton <akpm@linux-foundation.org>
      Cc: Suresh Siddha <suresh.b.siddha@intel.com>
      Cc: "Eric W. Biederman" <ebiederm@xmission.com>
      Cc: Rusty Russell <rusty@rustcorp.com.au>
      Cc: Len Brown <lenb@kernel.org>
      Cc: Bjorn Helgaas <bjorn.helgaas@hp.com>
      Cc: Tony Luck <tony.luck@intel.com>
      Cc: linux-acpi@vger.kernel.org
      Cc: linux-ia64@vger.kernel.org
      LKML-Reference: <49F65560.20904@kernel.org>
      Signed-off-by: default avatarIngo Molnar <mingo@elte.hu>
      a2f809b0
    • Yinghai Lu's avatar
      x86/irq: change irq_desc_alloc() to take node instead of cpu · 85ac16d0
      Yinghai Lu authored
      
      
      This simplifies the node awareness of the code. All our allocators
      only deal with a NUMA node ID locality not with CPU ids anyway - so
      there's no need to maintain (and transform) a CPU id all across the
      IRq layer.
      
      v2: keep move_irq_desc related
      
      [ Impact: cleanup, prepare IRQ code to be NUMA-aware ]
      Signed-off-by: default avatarYinghai Lu <yinghai@kernel.org>
      Cc: Andrew Morton <akpm@linux-foundation.org>
      Cc: Suresh Siddha <suresh.b.siddha@intel.com>
      Cc: "Eric W. Biederman" <ebiederm@xmission.com>
      Cc: Rusty Russell <rusty@rustcorp.com.au>
      Cc: Jeremy Fitzhardinge <jeremy@goop.org>
      LKML-Reference: <49F65536.2020300@kernel.org>
      Signed-off-by: default avatarIngo Molnar <mingo@elte.hu>
      85ac16d0
    • Yinghai Lu's avatar
      irq: only update affinity if ->set_affinity() is sucessfull · 57b150cc
      Yinghai Lu authored
      
      
      irq_set_affinity() and move_masked_irq() try to assign affinity
      before calling chip set_affinity(). Some archs are assigning it
      in ->set_affinity() again.
      
      We do something like:
      
       cpumask_cpy(desc->affinity, mask);
       desc->chip->set_affinity(mask);
      
      But in the failure path, affinity should not be touched - otherwise
      we'll end up with a different affinity mask despite the failure to
      migrate the IRQ.
      
      So try to update the afffinity only if set_affinity returns with 0.
      Also call irq_set_thread_affinity accordingly.
      
      v2: update after "irq, x86: Remove IRQ_DISABLED check in process context IRQ move"
      v3: according to Ingo, change set_affinity() in irq_chip should return int.
      v4: update comments by removing moving irq_desc code.
      
      [ Impact: fix /proc/irq/*/smp_affinity setting corner case bug ]
      Signed-off-by: default avatarYinghai Lu <yinghai@kernel.org>
      Cc: Andrew Morton <akpm@linux-foundation.org>
      Cc: Suresh Siddha <suresh.b.siddha@intel.com>
      Cc: "Eric W. Biederman" <ebiederm@xmission.com>
      Cc: Rusty Russell <rusty@rustcorp.com.au>
      LKML-Reference: <49F65509.60307@kernel.org>
      Signed-off-by: default avatarIngo Molnar <mingo@elte.hu>
      57b150cc
    • Yinghai Lu's avatar
      irq: change ->set_affinity() to return status · d5dedd45
      Yinghai Lu authored
      
      
      according to Ingo, change set_affinity() in irq_chip should return int,
      because that way we can handle failure cases in a much cleaner way, in
      the genirq layer.
      
      v2: fix two typos
      
      [ Impact: extend API ]
      Signed-off-by: default avatarYinghai Lu <yinghai@kernel.org>
      Cc: Andrew Morton <akpm@linux-foundation.org>
      Cc: Suresh Siddha <suresh.b.siddha@intel.com>
      Cc: "Eric W. Biederman" <ebiederm@xmission.com>
      Cc: Rusty Russell <rusty@rustcorp.com.au>
      Cc: linux-arch@vger.kernel.org
      LKML-Reference: <49F654E9.4070809@kernel.org>
      Signed-off-by: default avatarIngo Molnar <mingo@elte.hu>
      d5dedd45
    • Yinghai Lu's avatar
      x86/irq: remove leftover code from NUMA_MIGRATE_IRQ_DESC · fcef5911
      Yinghai Lu authored
      
      
      The original feature of migrating irq_desc dynamic was too fragile
      and was causing problems: it caused crashes on systems with lots of
      cards with MSI-X when user-space irq-balancer was enabled.
      
      We now have new patches that create irq_desc according to device
      numa node. This patch removes the leftover bits of the dynamic balancer.
      
      [ Impact: remove dead code ]
      Signed-off-by: default avatarYinghai Lu <yinghai@kernel.org>
      Cc: Andrew Morton <akpm@linux-foundation.org>
      Cc: Suresh Siddha <suresh.b.siddha@intel.com>
      Cc: "Eric W. Biederman" <ebiederm@xmission.com>
      Cc: Rusty Russell <rusty@rustcorp.com.au>
      LKML-Reference: <49F654AF.8000808@kernel.org>
      Signed-off-by: default avatarIngo Molnar <mingo@elte.hu>
      fcef5911
    • Yinghai Lu's avatar
      irq, cpumask: correct CPUMASKS_OFFSTACK typo and fix fallout · 9ec4fa27
      Yinghai Lu authored
      
      
      CPUMASKS_OFFSTACK is not defined anywhere (it is CPUMASK_OFFSTACK).
      It is a typo and init_allocate_desc_masks() is called before it set
      affinity to all cpus...
      
      Split init_alloc_desc_masks() into all_desc_masks() and init_desc_masks().
      
      Also use CPUMASK_OFFSTACK in alloc_desc_masks().
      
      [ Impact: fix smp_affinity copying/setup when moving irq_desc between CPUs ]
      Signed-off-by: default avatarYinghai Lu <yinghai@kernel.org>
      Acked-by: default avatarRusty Russell <rusty@rustcorp.com.au>
      Cc: Andrew Morton <akpm@linux-foundation.org>
      Cc: Suresh Siddha <suresh.b.siddha@intel.com>
      Cc: "Eric W. Biederman" <ebiederm@xmission.com>
      LKML-Reference: <49F6546E.3040406@kernel.org>
      Signed-off-by: default avatarIngo Molnar <mingo@elte.hu>
      9ec4fa27
  5. 27 Apr, 2009 11 commits
  6. 26 Apr, 2009 14 commits
  7. 24 Apr, 2009 2 commits
    • Theodore Ts'o's avatar
      ext4: Do not try to validate extents on special files · c4b5a614
      Theodore Ts'o authored
      
      
      The EXTENTS_FL flag should never be set on special files, but if it
      is, don't bother trying to validate that the extents tree is valid,
      since only files, directories, and non-fast symlinks will ever have an
      extent data structure.  We perhaps should flag the filesystem as being
      corrupted if we see a special file (named pipes, device nodes, Unix
      domain sockets, etc.) with the EXTENTS_FL flag, but e2fsck doesn't
      currently check this case, so we'll just ignore this for now, since
      it's harmless.
      
      Without this fix, a special device with the extents flag is flagged as
      an error by the kernel, so it is impossible to access or delete the
      inode, but e2fsck doesn't see it as a problem, leading to
      confused/frustrated users.
      Signed-off-by: default avatar"Theodore Ts'o" <tytso@mit.edu>
      c4b5a614
    • Rafael J. Wysocki's avatar
      PM/Hibernate: Fix waiting for image device to appear on resume · 0c8454f5
      Rafael J. Wysocki authored
      Commit c7510859
      
       ("PM/Hibernate: Wait for
      SCSI devices scan to complete during resume") added a call to
      scsi_complete_async_scans() to software_resume(), so that it waited for
      the SCSI scanning to complete, but the call was added at a wrong place.
      
      Namely, it should have been added after wait_for_device_probe(), which
      is called only if the image partition hasn't been specified yet.  Also,
      it's reasonable to check if the image partition is present and only wait
      for the device probing and SCSI scanning to complete if it is not the
      case.
      
      Additionally, since noresume is checked right at the beginning of
      software_resume() and the function returns immediately if it's set, it
      doesn't make sense to check it once again later.
      Signed-off-by: default avatarRafael J. Wysocki <rjw@sisk.pl>
      Signed-off-by: default avatarLinus Torvalds <torvalds@linux-foundation.org>
      0c8454f5