)]}'
{
  "log": [
    {
      "commit": "6c9ae009b298753a3baf71298d676a68b5a10c8f",
      "tree": "95a7e4ca54fe8e96c95cedd4e8d6450659015212",
      "parents": [
        "558bbb2fc75fd9676e07e347c021865e326c56db"
      ],
      "author": {
        "name": "Eric Dumazet",
        "email": "eric.dumazet@gmail.com",
        "time": "Thu Jan 13 15:45:38 2011 -0800"
      },
      "committer": {
        "name": "Linus Torvalds",
        "email": "torvalds@linux-foundation.org",
        "time": "Thu Jan 13 17:32:31 2011 -0800"
      },
      "message": "irq: use per_cpu kstat_irqs\n\nUse modern per_cpu API to increment {soft|hard}irq counters, and use\nper_cpu allocation for (struct irq_desc)-\u003ekstats_irq instead of an array.\n\nThis gives better SMP/NUMA locality and saves few instructions per irq.\n\nWith small nr_cpuids values (8 for example), kstats_irq was a small array\n(less than L1_CACHE_BYTES), potentially source of false sharing.\n\nIn the !CONFIG_SPARSE_IRQ case, remove the huge, NUMA/cache unfriendly\nkstat_irqs_all[NR_IRQS][NR_CPUS] array.\n\nNote: we still populate kstats_irq for all possible irqs in\nearly_irq_init().  We probably could use on-demand allocations.  (Code\nincluded in alloc_descs()).  Problem is not all IRQS are used with a prior\nalloc_descs() call.\n\nkstat_irqs_this_cpu() is not used anymore, remove it.\n\nSigned-off-by: Eric Dumazet \u003ceric.dumazet@gmail.com\u003e\nReviewed-by: Christoph Lameter \u003ccl@linux.com\u003e\nCc: Ingo Molnar \u003cmingo@elte.hu\u003e\nCc: Andi Kleen \u003candi@firstfloor.org\u003e\nCc: Tejun Heo \u003ctj@kernel.org\u003e\nCc: Thomas Gleixner \u003ctglx@linutronix.de\u003e\nSigned-off-by: Andrew Morton \u003cakpm@linux-foundation.org\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@linux-foundation.org\u003e\n"
    },
    {
      "commit": "478735e38887077ac77a9756121b6ce0cb956e2f",
      "tree": "7b42110f71b1711f56cbbc0a01f00894b221337d",
      "parents": [
        "f2c66cd8eeddedb440f33bc0f5cec1ed7ae376cb"
      ],
      "author": {
        "name": "KAMEZAWA Hiroyuki",
        "email": "kamezawa.hiroyu@jp.fujitsu.com",
        "time": "Wed Oct 27 15:34:15 2010 -0700"
      },
      "committer": {
        "name": "Linus Torvalds",
        "email": "torvalds@linux-foundation.org",
        "time": "Wed Oct 27 18:03:13 2010 -0700"
      },
      "message": "/proc/stat: fix scalability of irq sum of all cpu\n\nIn /proc/stat, the number of per-IRQ event is shown by making a sum each\nirq\u0027s events on all cpus.  But we can make use of kstat_irqs().\n\nkstat_irqs() do the same calculation, If !CONFIG_GENERIC_HARDIRQ,\nit\u0027s not a big cost. (Both of the number of cpus and irqs are small.)\n\nIf a system is very big and CONFIG_GENERIC_HARDIRQ, it does\n\n\tfor_each_irq()\n\t\tfor_each_cpu()\n\t\t\t- look up a radix tree\n\t\t\t- read desc-\u003eirq_stat[cpu]\nThis seems not efficient. This patch adds kstat_irqs() for\nCONFIG_GENRIC_HARDIRQ and change the calculation as\n\n\tfor_each_irq()\n\t\tlook up radix tree\n\t\tfor_each_cpu()\n\t\t\t- read desc-\u003eirq_stat[cpu]\n\nThis reduces cost.\n\nA test on (4096cpusp, 256 nodes, 4592 irqs) host (by Jack Steiner)\n\n%time cat /proc/stat \u003e /dev/null\n\nBefore Patch:\t 2.459 sec\nAfter Patch :\t  .561 sec\n\n[akpm@linux-foundation.org: unexport kstat_irqs, coding-style tweaks]\n[akpm@linux-foundation.org: fix unused variable \u0027per_irq_sum\u0027]\nSigned-off-by: KAMEZAWA Hiroyuki \u003ckamezawa.hiroyu@jp.fujitsu.com\u003e\nTested-by: Jack Steiner \u003csteiner@sgi.com\u003e\nAcked-by: Jack Steiner \u003csteiner@sgi.com\u003e\nCc: Yinghai Lu \u003cyinghai@kernel.org\u003e\nCc: Ingo Molnar \u003cmingo@elte.hu\u003e\nCc: Thomas Gleixner \u003ctglx@linutronix.de\u003e\nSigned-off-by: Andrew Morton \u003cakpm@linux-foundation.org\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@linux-foundation.org\u003e\n"
    },
    {
      "commit": "c0a19ebc018222ffd1dd93af5b53d9efd779c19b",
      "tree": "f648fb93efb064b22000aca24e48dfda2cbfab0a",
      "parents": [
        "2ee39065988d632b403f8470942b0b5edee3632b"
      ],
      "author": {
        "name": "Thomas Gleixner",
        "email": "tglx@linutronix.de",
        "time": "Tue Oct 12 21:58:27 2010 +0200"
      },
      "committer": {
        "name": "Thomas Gleixner",
        "email": "tglx@linutronix.de",
        "time": "Tue Oct 12 21:59:55 2010 +0200"
      },
      "message": "genirq: Fix CONFIG_GENIRQ_NO_DEPRECATED\u003dy build\n\nThis option can be set to verify the full conversion to the new chip\nfunctions. Fix the fallout of the patch rework, so the core code\ncompiles and works with it.\n\nSigned-off-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\n"
    },
    {
      "commit": "baa0d233afe765daa6dc01ff233aea8c5944f534",
      "tree": "23b0793b6014b637680cddef4ccc4a2860cf6436",
      "parents": [
        "a05a900a51c7622ebd8ddb41f14f8bf9db599d8d"
      ],
      "author": {
        "name": "Thomas Gleixner",
        "email": "tglx@linutronix.de",
        "time": "Tue Oct 05 15:14:35 2010 +0200"
      },
      "committer": {
        "name": "Thomas Gleixner",
        "email": "tglx@linutronix.de",
        "time": "Tue Oct 12 16:53:45 2010 +0200"
      },
      "message": "genirq: Switch sparse_irq allocator to GFP_KERNEL\n\nThe allocator functions are now called outside of preempt disabled\nregions. Switch to GFP_KERNEL.\n\nSigned-off-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\nReviewed-by: Ingo Molnar \u003cmingo@elte.hu\u003e\n"
    },
    {
      "commit": "a05a900a51c7622ebd8ddb41f14f8bf9db599d8d",
      "tree": "3c0f7d5c6aff2af86c350c76cc44631e620d9333",
      "parents": [
        "c2f31c37b7303150ffcce53f6c6ed4ac62952b34"
      ],
      "author": {
        "name": "Thomas Gleixner",
        "email": "tglx@linutronix.de",
        "time": "Fri Oct 08 12:47:53 2010 +0200"
      },
      "committer": {
        "name": "Thomas Gleixner",
        "email": "tglx@linutronix.de",
        "time": "Tue Oct 12 16:53:45 2010 +0200"
      },
      "message": "genirq: Make sparse_lock a mutex\n\nNo callers from atomic regions. \n\nSigned-off-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\nReviewed-by: Ingo Molnar \u003cmingo@elte.hu\u003e\n"
    },
    {
      "commit": "78f90d91f395cd0dc1ef3f21e0c5cd6fd50d202c",
      "tree": "e9a6b0a5d8a0b99cd23dbc1c1f376a7e78a93428",
      "parents": [
        "b7b29338dc7111ed8bd4d6555d84afae13ebe752"
      ],
      "author": {
        "name": "Thomas Gleixner",
        "email": "tglx@linutronix.de",
        "time": "Wed Sep 29 17:18:47 2010 +0200"
      },
      "committer": {
        "name": "Thomas Gleixner",
        "email": "tglx@linutronix.de",
        "time": "Tue Oct 12 16:53:44 2010 +0200"
      },
      "message": "genirq: Remove the now unused sparse irq leftovers\n\nThe move_irq_desc() function was only used due to the problem that the\nallocator did not free the old descriptors. So the descriptors had to\nbe moved in create_irq_nr(). That\u0027s history.\n\nThe code would have never been able to move active interrupt\ndescriptors on affinity settings. That can be done in a completely\ndifferent way w/o all this horror.\n\nRemove all of it.\n\nSigned-off-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\nReviewed-by: Ingo Molnar \u003cmingo@elte.hu\u003e\n"
    },
    {
      "commit": "b7b29338dc7111ed8bd4d6555d84afae13ebe752",
      "tree": "4c3159ea8bb0489ba463a061d3e6446dbfb45af2",
      "parents": [
        "b7d0d8258a9f71949b810e0f82a3d75088f4d364"
      ],
      "author": {
        "name": "Thomas Gleixner",
        "email": "tglx@linutronix.de",
        "time": "Wed Sep 29 18:46:55 2010 +0200"
      },
      "committer": {
        "name": "Thomas Gleixner",
        "email": "tglx@linutronix.de",
        "time": "Tue Oct 12 16:53:44 2010 +0200"
      },
      "message": "genirq: Sanitize dynamic irq handling\n\nUse the cleanup functions of the dynamic allocator. No need to have\nseparate implementations.\n\nSigned-off-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\nReviewed-by: Ingo Molnar \u003cmingo@elte.hu\u003e\n"
    },
    {
      "commit": "b7d0d8258a9f71949b810e0f82a3d75088f4d364",
      "tree": "1e37a6d3e89816abab898e3646a4e23eeb457021",
      "parents": [
        "77dff1c755c3218691e95e7e38ee14323b35dbdb"
      ],
      "author": {
        "name": "Thomas Gleixner",
        "email": "tglx@linutronix.de",
        "time": "Wed Sep 29 18:44:23 2010 +0200"
      },
      "committer": {
        "name": "Thomas Gleixner",
        "email": "tglx@linutronix.de",
        "time": "Tue Oct 12 16:53:44 2010 +0200"
      },
      "message": "genirq: Remove arch_init_chip_data()\n\nThis function should have not been there in the first place.\n\nSigned-off-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\nReviewed-by: Ingo Molnar \u003cmingo@elte.hu\u003e\n"
    },
    {
      "commit": "b683de2b3cb17bb10fa6fd4af614dc75b5749fe0",
      "tree": "e1a799293b218f5c13d4903c57fab144b2f433b3",
      "parents": [
        "aa99ec0f3f26bf2bcd0fa5176de93598427f1e5e"
      ],
      "author": {
        "name": "Thomas Gleixner",
        "email": "tglx@linutronix.de",
        "time": "Mon Sep 27 20:55:03 2010 +0200"
      },
      "committer": {
        "name": "Thomas Gleixner",
        "email": "tglx@linutronix.de",
        "time": "Tue Oct 12 16:39:08 2010 +0200"
      },
      "message": "genirq: Query arch for number of early descriptors\n\nsparse irq sets up NR_IRQS_LEGACY irq descriptors and archs then go\nahead and allocate more.\n\nUse the unused return value of arch_probe_nr_irqs() to let the\narchitecture return the number of early allocations. Fix up all users.\n\nSigned-off-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\nReviewed-by: Ingo Molnar \u003cmingo@elte.hu\u003e\n"
    },
    {
      "commit": "aa99ec0f3f26bf2bcd0fa5176de93598427f1e5e",
      "tree": "bbba6c0fc1293f0bc854644aa177ea67306f8814",
      "parents": [
        "25ade601a0f97453c6f511ebfae9339e06a28d75"
      ],
      "author": {
        "name": "Thomas Gleixner",
        "email": "tglx@linutronix.de",
        "time": "Mon Sep 27 20:02:56 2010 +0200"
      },
      "committer": {
        "name": "Thomas Gleixner",
        "email": "tglx@linutronix.de",
        "time": "Tue Oct 12 16:39:07 2010 +0200"
      },
      "message": "genirq: Use sane sparse allocator\n\nMake irq_to_desc_alloc_node() a wrapper around the new allocator.\n\nSigned-off-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\nReviewed-by: Ingo Molnar \u003cmingo@elte.hu\u003e\n"
    },
    {
      "commit": "06f6c3399e9f9ff6eafc200e80f9226c3cee0eaf",
      "tree": "f4d8fc67194b1a50bfe501634088b3776ca6bbd4",
      "parents": [
        "a98d24b71b6e229965f18dc00d28dc71cb8fe324"
      ],
      "author": {
        "name": "Thomas Gleixner",
        "email": "tglx@linutronix.de",
        "time": "Tue Oct 12 12:31:46 2010 +0200"
      },
      "committer": {
        "name": "Thomas Gleixner",
        "email": "tglx@linutronix.de",
        "time": "Tue Oct 12 16:39:07 2010 +0200"
      },
      "message": "genirq: Implement irq reservation\n\nMark a range of interrupts as allocated. In the SPARSE_IRQ\u003dn case we\nneed this to update the bitmap for the legacy irqs so the enumerator\nvia irq_get_next_irq() works.\n\nSigned-off-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\n"
    },
    {
      "commit": "a98d24b71b6e229965f18dc00d28dc71cb8fe324",
      "tree": "213a6daf67e459b2078e630d5a8163c3b8fd3a9c",
      "parents": [
        "13bfe99e09123ef5edb6acb81ba337d2db600b53"
      ],
      "author": {
        "name": "Thomas Gleixner",
        "email": "tglx@linutronix.de",
        "time": "Thu Sep 30 10:45:07 2010 +0200"
      },
      "committer": {
        "name": "Thomas Gleixner",
        "email": "tglx@linutronix.de",
        "time": "Tue Oct 12 16:39:07 2010 +0200"
      },
      "message": "genirq: Implement sane enumeration\n\nUse the allocator bitmap to lookup active interrupts.\n\nSigned-off-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\nReviewed-by: Ingo Molnar \u003cmingo@elte.hu\u003e\n"
    },
    {
      "commit": "13bfe99e09123ef5edb6acb81ba337d2db600b53",
      "tree": "f96f5be41c90e1ab437bb40a02a8f137deb9e3ab",
      "parents": [
        "1f5a5b87f78fade3ae48dfd55e8765d1d622ea4e"
      ],
      "author": {
        "name": "Thomas Gleixner",
        "email": "tglx@linutronix.de",
        "time": "Thu Sep 30 02:46:07 2010 +0200"
      },
      "committer": {
        "name": "Thomas Gleixner",
        "email": "tglx@linutronix.de",
        "time": "Tue Oct 12 16:39:07 2010 +0200"
      },
      "message": "genirq: Prepare proc for real sparse irq support\n\n/proc/irq never removes any entries, but when irq descriptors can be\nfreed for real this is necessary. Otherwise we\u0027d reference a freed\ndescriptor in /proc/irq/N\n\nSigned-off-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\nReviewed-by: Ingo Molnar \u003cmingo@elte.hu\u003e\n"
    },
    {
      "commit": "1f5a5b87f78fade3ae48dfd55e8765d1d622ea4e",
      "tree": "762a5dbf40129ffd9667a170b2503a77c95320f7",
      "parents": [
        "1318a481fc37c503a901b96ae06b692ca2b21af5"
      ],
      "author": {
        "name": "Thomas Gleixner",
        "email": "tglx@linutronix.de",
        "time": "Mon Sep 27 17:48:26 2010 +0200"
      },
      "committer": {
        "name": "Thomas Gleixner",
        "email": "tglx@linutronix.de",
        "time": "Tue Oct 12 16:39:07 2010 +0200"
      },
      "message": "genirq: Implement a sane sparse_irq allocator\n\nThe current sparse_irq allocator has several short comings due to\nfailures in the design or the lack of it:\n\n - Requires iteration over the number of active irqs to find a free slot\n   (Some architectures have grown their own workarounds for this)\n - Removal of entries is not possible\n - Racy between create_irq_nr and destroy_irq (plugged by horrible\n   callbacks)\n - Migration of active irq descriptors is not possible\n - No bulk allocation of irq ranges\n - Sprinkeled irq_desc references all over the place outside of kernel/irq/\n   (The previous chip functions series is addressing this issue)\n\nImplement a sane allocator which fixes the above short comings (though\nmigration of active descriptors needs a full tree wide cleanup of the\ndirect and mostly unlocked access to irq_desc).\n\nThe new allocator still uses a radix_tree, but uses a bitmap for\nkeeping track of allocated irq numbers. That allows:\n\n - Fast lookup of a free slot\n - Allows the removal of descriptors\n - Prevents the create/destroy race\n - Bulk allocation of consecutive irq ranges\n - Basic design is ready for migration of life descriptors after\n   further cleanups\n\nThe bitmap is also used in the SPARSE_IRQ\u003dn case for lookup and\nraceless (de)allocation of irq numbers. So it removes the requirement\nfor looping through the descriptor array to find slots.\n\nRight now it uses sparse_irq_lock to protect the bitmap and the radix\ntree, but after cleaning up all users we should be able convert that\nto a mutex and to switch the radix_tree and decriptor allocations to\nGFP_KERNEL.\n\n[ Folded in a bugfix from Yinghai Lu ]\n\nSigned-off-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\nReviewed-by: Ingo Molnar \u003cmingo@elte.hu\u003e\n"
    },
    {
      "commit": "1318a481fc37c503a901b96ae06b692ca2b21af5",
      "tree": "2b1bb8f52af5938759af38efdfa8a868b02d5804",
      "parents": [
        "d895f51ebb54cefe367bda135fcf2cd734d51d03"
      ],
      "author": {
        "name": "Thomas Gleixner",
        "email": "tglx@linutronix.de",
        "time": "Mon Sep 27 21:01:37 2010 +0200"
      },
      "committer": {
        "name": "Thomas Gleixner",
        "email": "tglx@linutronix.de",
        "time": "Tue Oct 12 16:39:06 2010 +0200"
      },
      "message": "genirq: Provide default irq init flags\n\nArch code sets it\u0027s own irq_desc.status flags right after boot and for\ndynamically allocated interrupts. That might involve iterating over a\nhuge array.\n\nAllow ARCH_IRQ_INIT_FLAGS to set separate flags aside of IRQ_DISABLED\nwhich is the default.\n\nSigned-off-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\nReviewed-by: Ingo Molnar \u003cmingo@elte.hu\u003e\n"
    },
    {
      "commit": "d895f51ebb54cefe367bda135fcf2cd734d51d03",
      "tree": "bb70160ad291a7280ff8e8a2aba07d407f7f07dc",
      "parents": [
        "40f08a724fcc21285cf3a75aec957aef908605c6"
      ],
      "author": {
        "name": "Thomas Gleixner",
        "email": "tglx@linutronix.de",
        "time": "Mon Sep 27 17:45:49 2010 +0200"
      },
      "committer": {
        "name": "Thomas Gleixner",
        "email": "tglx@linutronix.de",
        "time": "Tue Oct 12 16:39:06 2010 +0200"
      },
      "message": "genirq: Remove export of kstat_irqs_cpu\n\nThe statistics accessor is only used by proc/stats and\nshow_interrupts(). Both are compiled in.\n\nSigned-off-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\nReviewed-by: Ingo Molnar \u003cmingo@elte.hu\u003e\n"
    },
    {
      "commit": "154cd387cdf0e5566ce523cbddf92dd2a062dfd6",
      "tree": "8ab329f26992c570b65fe62843275eb71c8bf505",
      "parents": [
        "3795de236d67a05994a1a12759db9d4dd9ffc42c"
      ],
      "author": {
        "name": "Thomas Gleixner",
        "email": "tglx@linutronix.de",
        "time": "Wed Sep 22 15:58:45 2010 +0200"
      },
      "committer": {
        "name": "Thomas Gleixner",
        "email": "tglx@linutronix.de",
        "time": "Tue Oct 12 16:39:06 2010 +0200"
      },
      "message": "genirq: Remove early_init_irq_lock_class()\n\nearly_init_irq_lock_class() is called way before anything touches the\nirq descriptors. In case of SPARSE_IRQ\u003dy this is a NOP operation\nbecause the radix tree is empty at this point. For the SPARSE_IRQ\u003dn\ncase it\u0027s sufficient to set the lock class in early_init_irq(). \n\nSigned-off-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\nReviewed-by: Ingo Molnar \u003cmingo@elte.hu\u003e\n"
    },
    {
      "commit": "3795de236d67a05994a1a12759db9d4dd9ffc42c",
      "tree": "75e5a2a8922e114de60f468494c879ba4f65ebee",
      "parents": [
        "f303a6dd127b5ec6de90d1cd79ed19820c7e9658"
      ],
      "author": {
        "name": "Thomas Gleixner",
        "email": "tglx@linutronix.de",
        "time": "Wed Sep 22 17:09:43 2010 +0200"
      },
      "committer": {
        "name": "Thomas Gleixner",
        "email": "tglx@linutronix.de",
        "time": "Tue Oct 12 16:39:05 2010 +0200"
      },
      "message": "genirq: Distangle kernel/irq/handle.c\n\nkernel/irq/handle.c has become a dumpground for random code in random\norder. Split out the irq descriptor management and the dummy irq_chip\nimplementation into separate files. Cleanup the include maze while at\nit.\n\nNo code change.\n\nSigned-off-by: Thomas Gleixner \u003ctglx@linutronix.de\u003e\nReviewed-by: Ingo Molnar \u003cmingo@elte.hu\u003e\n"
    }
  ]
}