)]}'
{
  "log": [
    {
      "commit": "0c9e63fd38a2fb2181668a0cdd622a3c23cfd567",
      "tree": "8fdb91603347b1da2e83a095ebcaab44b2c3c237",
      "parents": [
        "d8733c2956968a01394a4d2a9e97a8b431a78776"
      ],
      "author": {
        "name": "Eric Dumazet",
        "email": "dada1@cosmosbay.com",
        "time": "Thu Mar 23 03:00:12 2006 -0800"
      },
      "committer": {
        "name": "Linus Torvalds",
        "email": "torvalds@g5.osdl.org",
        "time": "Thu Mar 23 07:38:09 2006 -0800"
      },
      "message": "[PATCH] Shrinks sizeof(files_struct) and better layout\n\n1) Reduce the size of (struct fdtable) to exactly 64 bytes on 32bits\n   platforms, lowering kmalloc() allocated space by 50%.\n\n2) Reduce the size of (files_struct), using a special 32 bits (or\n   64bits) embedded_fd_set, instead of a 1024 bits fd_set for the\n   close_on_exec_init and open_fds_init fields.  This save some ram (248\n   bytes per task) as most tasks dont open more than 32 files.  D-Cache\n   footprint for such tasks is also reduced to the minimum.\n\n3) Reduce size of allocated fdset.  Currently two full pages are\n   allocated, that is 32768 bits on x86 for example, and way too much.  The\n   minimum is now L1_CACHE_BYTES.\n\nUP and SMP should benefit from this patch, because most tasks will touch\nonly one cache line when open()/close() stdin/stdout/stderr (0/1/2),\n(next_fd, close_on_exec_init, open_fds_init, fd_array[0 ..  2] being in the\nsame cache line)\n\nSigned-off-by: Eric Dumazet \u003cdada1@cosmosbay.com\u003e\nSigned-off-by: Andrew Morton \u003cakpm@osdl.org\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@osdl.org\u003e\n"
    },
    {
      "commit": "529bf6be5c04f2e869d07bfdb122e9fd98ade714",
      "tree": "38514bb3941c4ac2a79266e4483663b79efa2f22",
      "parents": [
        "21a1ea9eb40411d4ee29448c53b9e4c0654d6ceb"
      ],
      "author": {
        "name": "Dipankar Sarma",
        "email": "dipankar@in.ibm.com",
        "time": "Tue Mar 07 21:55:35 2006 -0800"
      },
      "committer": {
        "name": "Linus Torvalds",
        "email": "torvalds@g5.osdl.org",
        "time": "Wed Mar 08 14:14:01 2006 -0800"
      },
      "message": "[PATCH] fix file counting\n\nI have benchmarked this on an x86_64 NUMA system and see no significant\nperformance difference on kernbench.  Tested on both x86_64 and powerpc.\n\nThe way we do file struct accounting is not very suitable for batched\nfreeing.  For scalability reasons, file accounting was\nconstructor/destructor based.  This meant that nr_files was decremented\nonly when the object was removed from the slab cache.  This is susceptible\nto slab fragmentation.  With RCU based file structure, consequent batched\nfreeing and a test program like Serge\u0027s, we just speed this up and end up\nwith a very fragmented slab -\n\nllm22:~ # cat /proc/sys/fs/file-nr\n587730  0       758844\n\nAt the same time, I see only a 2000+ objects in filp cache.  The following\npatch I fixes this problem.\n\nThis patch changes the file counting by removing the filp_count_lock.\nInstead we use a separate percpu counter, nr_files, for now and all\naccesses to it are through get_nr_files() api.  In the sysctl handler for\nnr_files, we populate files_stat.nr_files before returning to user.\n\nCounting files as an when they are created and destroyed (as opposed to\ninside slab) allows us to correctly count open files with RCU.\n\nSigned-off-by: Dipankar Sarma \u003cdipankar@in.ibm.com\u003e\nCc: \"Paul E. McKenney\" \u003cpaulmck@us.ibm.com\u003e\nCc: \"David S. Miller\" \u003cdavem@davemloft.net\u003e\nSigned-off-by: Andrew Morton \u003cakpm@osdl.org\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@osdl.org\u003e\n"
    },
    {
      "commit": "95e861db3eaba7bc99f8605db70103ec3d078203",
      "tree": "70f86b1acf1c8b1fb49f8bd11aaf5d81c39d0e97",
      "parents": [
        "d6c7ac081bf6cafcf780b919ee97978f1d01a0d7"
      ],
      "author": {
        "name": "Eric Dumazet",
        "email": "dada1@cosmosbay.com",
        "time": "Sun Nov 13 16:06:24 2005 -0800"
      },
      "committer": {
        "name": "Linus Torvalds",
        "email": "torvalds@g5.osdl.org",
        "time": "Sun Nov 13 18:14:10 2005 -0800"
      },
      "message": "[PATCH] reorder struct files_struct\n\nThe file_lock spinlock sits close to mostly read fields of \u0027struct\nfiles_struct\u0027\n\nIn SMP (and NUMA) environments, each time a thread wants to open or close\na file, it has to acquire the spinlock, thus invalidating the cache line\ncontaining this spinlock on other CPUS.  So other threads doing\nread()/write()/...  calls that use RCU to access the file table are going\nto ask further memory (possibly NUMA) transactions to read again this\nmemory line.\n\nMove the spinlock to another cache line, so that concurrent threads can\nshare the cache line containing \u0027count\u0027 and \u0027fdt\u0027 fields.\n\nIt\u0027s worth up to 9% on a microbenchmark using a 4-thread 2-package x86\nmachine.  See\nhttp://marc.theaimsgroup.com/?l\u003dlinux-kernel\u0026m\u003d112680448713342\u0026w\u003d2\n\nSigned-off-by: Eric Dumazet \u003cdada1@cosmosbay.com\u003e\nSigned-off-by: Andrew Morton \u003cakpm@osdl.org\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@osdl.org\u003e\n"
    },
    {
      "commit": "2109a2d1b175dfcffbfdac693bdbe4c4ab62f11f",
      "tree": "07b47953a74adaa7490ef24aab10c45a5a4148a6",
      "parents": [
        "4f12bb4f7715f418a9c80f89447948790f476958"
      ],
      "author": {
        "name": "Pekka J Enberg",
        "email": "penberg@cs.Helsinki.FI",
        "time": "Mon Nov 07 00:58:01 2005 -0800"
      },
      "committer": {
        "name": "Linus Torvalds",
        "email": "torvalds@g5.osdl.org",
        "time": "Mon Nov 07 07:53:24 2005 -0800"
      },
      "message": "[PATCH] mm: rename kmem_cache_s to kmem_cache\n\nThis patch renames struct kmem_cache_s to kmem_cache so we can start using\nit instead of kmem_cache_t typedef.\n\nSigned-off-by: Pekka Enberg \u003cpenberg@cs.helsinki.fi\u003e\nSigned-off-by: Andrew Morton \u003cakpm@osdl.org\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@osdl.org\u003e\n"
    },
    {
      "commit": "ab2af1f5005069321c5d130f09cce577b03f43ef",
      "tree": "73a70ba486f522cd9eeeef376ede2b5a1c1b473b",
      "parents": [
        "6e72ad2c581de121cc7e772469e2a8f6b1fd4379"
      ],
      "author": {
        "name": "Dipankar Sarma",
        "email": "dipankar@in.ibm.com",
        "time": "Fri Sep 09 13:04:13 2005 -0700"
      },
      "committer": {
        "name": "Linus Torvalds",
        "email": "torvalds@g5.osdl.org",
        "time": "Fri Sep 09 13:57:55 2005 -0700"
      },
      "message": "[PATCH] files: files struct with RCU\n\nPatch to eliminate struct files_struct.file_lock spinlock on the reader side\nand use rcu refcounting rcuref_xxx api for the f_count refcounter.  The\nupdates to the fdtable are done by allocating a new fdtable structure and\nsetting files-\u003efdt to point to the new structure.  The fdtable structure is\nprotected by RCU thereby allowing lock-free lookup.  For fd arrays/sets that\nare vmalloced, we use keventd to free them since RCU callbacks can\u0027t sleep.  A\nglobal list of fdtable to be freed is not scalable, so we use a per-cpu list.\nIf keventd is already handling the current cpu\u0027s work, we use a timer to defer\nqueueing of that work.\n\nSince the last publication, this patch has been re-written to avoid using\nexplicit memory barriers and use rcu_assign_pointer(), rcu_dereference()\npremitives instead.  This required that the fd information is kept in a\nseparate structure (fdtable) and updated atomically.\n\nSigned-off-by: Dipankar Sarma \u003cdipankar@in.ibm.com\u003e\nSigned-off-by: Andrew Morton \u003cakpm@osdl.org\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@osdl.org\u003e\n"
    },
    {
      "commit": "badf16621c1f9d1ac753be056fce11b43d6e0be5",
      "tree": "3fdf833fdf2e3d3a439090743539680449ec3428",
      "parents": [
        "c0dfb2905126e9e94edebbce8d3e05001301f52d"
      ],
      "author": {
        "name": "Dipankar Sarma",
        "email": "dipankar@in.ibm.com",
        "time": "Fri Sep 09 13:04:10 2005 -0700"
      },
      "committer": {
        "name": "Linus Torvalds",
        "email": "torvalds@g5.osdl.org",
        "time": "Fri Sep 09 13:57:55 2005 -0700"
      },
      "message": "[PATCH] files: break up files struct\n\nIn order for the RCU to work, the file table array, sets and their sizes must\nbe updated atomically.  Instead of ensuring this through too many memory\nbarriers, we put the arrays and their sizes in a separate structure.  This\npatch takes the first step of putting the file table elements in a separate\nstructure fdtable that is embedded withing files_struct.  It also changes all\nthe users to refer to the file table using files_fdtable() macro.  Subsequent\napplciation of RCU becomes easier after this.\n\nSigned-off-by: Dipankar Sarma \u003cdipankar@in.ibm.com\u003e\nSigned-Off-By: David Howells \u003cdhowells@redhat.com\u003e\nSigned-off-by: Andrew Morton \u003cakpm@osdl.org\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@osdl.org\u003e\n"
    },
    {
      "commit": "1da177e4c3f41524e886b7f1b8a0c1fc7321cac2",
      "tree": "0bba044c4ce775e45a88a51686b5d9f90697ea9d",
      "parents": [],
      "author": {
        "name": "Linus Torvalds",
        "email": "torvalds@ppc970.osdl.org",
        "time": "Sat Apr 16 15:20:36 2005 -0700"
      },
      "committer": {
        "name": "Linus Torvalds",
        "email": "torvalds@ppc970.osdl.org",
        "time": "Sat Apr 16 15:20:36 2005 -0700"
      },
      "message": "Linux-2.6.12-rc2\n\nInitial git repository build. I\u0027m not bothering with the full history,\neven though we have it. We can create a separate \"historical\" git\narchive of that later if we want to, and in the meantime it\u0027s about\n3.2GB when imported into git - space that would just make the early\ngit days unnecessarily complicated, when we don\u0027t have a lot of good\ninfrastructure for it.\n\nLet it rip!\n"
    }
  ]
}