)]}'
{
  "log": [
    {
      "commit": "6df10a82f8de89c66eb91c371d62d76e87b2cbba",
      "tree": "e8615f4c00a44f4e4a609d59c51e958f0300526b",
      "parents": [
        "a33a7d7309d79656bc19a0e96fc4547a1633283e"
      ],
      "author": {
        "name": "Mark Nutter",
        "email": "mnutter@us.ibm.com",
        "time": "Thu Mar 23 00:00:12 2006 +0100"
      },
      "committer": {
        "name": "Paul Mackerras",
        "email": "paulus@samba.org",
        "time": "Mon Mar 27 14:48:28 2006 +1100"
      },
      "message": "[PATCH] spufs: enable SPE problem state MMIO access.\n\nThis patch is layered on top of CONFIG_SPARSEMEM\nand is patterned after direct mapping of LS.\n\nThis patch allows mmap() of the following regions:\n\"mfc\", which represents the area from [0x3000 - 0x3fff];\n\"cntl\", which represents the area from [0x4000 - 0x4fff];\n\"signal1\" which begins at offset 0x14000; \"signal2\" which\nbegins at offset 0x1c000.\n\nThe signal1 \u0026 signal2 files may be mmap()\u0027d by regular user\nprocesses.  The cntl and mfc file, on the other hand, may\nonly be accessed if the owning process has CAP_SYS_RAWIO,\nbecause they have the potential to confuse the kernel\nwith regard to parallel access to the same files with\nregular file operations: the kernel always holds a spinlock\nwhen accessing registers in these areas to serialize them,\nwhich can not be guaranteed with user mmaps,\n\nSigned-off-by: Arnd Bergmann \u003carnd.bergmann@de.ibm.com\u003e\nSigned-off-by: Paul Mackerras \u003cpaulus@samba.org\u003e\n"
    },
    {
      "commit": "a33a7d7309d79656bc19a0e96fc4547a1633283e",
      "tree": "f3de6d139af1f1cbdf6d37800c9e13c07e9bc7f6",
      "parents": [
        "2dd14934c9138c562d93c501e88c6d6f061eb8ba"
      ],
      "author": {
        "name": "Arnd Bergmann",
        "email": "abergman@de.ibm.com",
        "time": "Thu Mar 23 00:00:11 2006 +0100"
      },
      "committer": {
        "name": "Paul Mackerras",
        "email": "paulus@samba.org",
        "time": "Mon Mar 27 14:48:26 2006 +1100"
      },
      "message": "[PATCH] spufs: implement mfc access for PPE-side DMA\n\nThis patch adds a new file called \u0027mfc\u0027 to each spufs directory.\nThe file accepts DMA commands that are a subset of what would\nbe legal DMA commands for problem state register access. Upon\nreading the file, a bitmask is returned with the completed\ntag groups set.\n\nThe file is meant to be used from an abstraction in libspe\nthat is added by a different patch.\n\nFrom the kernel perspective, this means a process can now\noffload a memory copy from or into an SPE local store\nwithout having to run code on the SPE itself.\n\nThe transfer will only be performed while the SPE is owned\nby one thread that is waiting in the spu_run system call\nand the data will be transferred into that thread\u0027s\naddress space, independent of which thread started the\ntransfer.\n\nSigned-off-by: Arnd Bergmann \u003carnd.bergmann@de.ibm.com\u003e\nSigned-off-by: Paul Mackerras \u003cpaulus@samba.org\u003e\n"
    },
    {
      "commit": "2dd14934c9138c562d93c501e88c6d6f061eb8ba",
      "tree": "2eda4f517dd319989908a80bfbe06ce629932f77",
      "parents": [
        "a7f31841a40776605c834053ad1eb82d539bd79f"
      ],
      "author": {
        "name": "Arnd Bergmann",
        "email": "abergman@de.ibm.com",
        "time": "Thu Mar 23 00:00:09 2006 +0100"
      },
      "committer": {
        "name": "Paul Mackerras",
        "email": "paulus@samba.org",
        "time": "Mon Mar 27 14:48:24 2006 +1100"
      },
      "message": "[PATCH] spufs: allow SPU code to do syscalls\n\nAn SPU does not have a way to implement system calls\nitself, but it can create intercepts to the kernel.\n\nThis patch uses the method defined by the JSRE interface\nfor C99 host library calls from an SPU to implement\nLinux system calls. It uses the reserved SPU stop code\n0x2104 for this, using the structure layout and syscall\nnumbers for ppc64-linux.\n\nI\u0027m still undecided wether it is better to have a list\nof allowed syscalls or a list of forbidden syscalls,\nsince we can\u0027t allow an SPU to call all syscalls that\nare defined for ppc64-linux.\n\nThis patch implements the easier choice of them, with a\nblacklist that only prevents an SPU from calling anything\nthat interacts with its own execution, e.g fork, execve,\nclone, vfork, exit, spu_run and spu_create and everything\nthat deals with signals.\n\nSigned-off-by: Arnd Bergmann \u003carnd.bergmann@de.ibm.com\u003e\nSigned-off-by: Paul Mackerras \u003cpaulus@samba.org\u003e\n"
    },
    {
      "commit": "2fb9d2063626374dd8a2514b3a730facac8235d8",
      "tree": "b410dcdbc5aee656c37951be36951130450549e7",
      "parents": [
        "aeb013772a2cc85a8d0baffd64977d2888bc781d"
      ],
      "author": {
        "name": "Arnd Bergmann",
        "email": "arnd@arndb.de",
        "time": "Thu Jan 05 14:05:29 2006 +0000"
      },
      "committer": {
        "name": "Paul Mackerras",
        "email": "paulus@samba.org",
        "time": "Mon Jan 09 15:44:57 2006 +1100"
      },
      "message": "[PATCH] spufs: set irq affinity for running threads\n\nFor far, all SPU triggered interrupts always end up on\nthe first SMT thread, which is a bad solution.\n\nThis patch implements setting the affinity to the\nCPU that was running last when entering execution on\nan SPU. This should result in a significant reduction\nin IPI calls and better cache locality for SPE thread\nspecific data.\n\nSigned-off-by: Arnd Bergmann \u003carndb@de.ibm.com\u003e\nSigned-off-by: Paul Mackerras \u003cpaulus@samba.org\u003e\n"
    },
    {
      "commit": "aeb013772a2cc85a8d0baffd64977d2888bc781d",
      "tree": "8ebf25d8ea0d2e0bb384f1399c1e124bd1c12044",
      "parents": [
        "6ff730c33b42a6c68217fc6660728676aa8eeb9c"
      ],
      "author": {
        "name": "Arnd Bergmann",
        "email": "arnd@arndb.de",
        "time": "Wed Jan 04 20:31:32 2006 +0100"
      },
      "committer": {
        "name": "Paul Mackerras",
        "email": "paulus@samba.org",
        "time": "Mon Jan 09 15:44:55 2006 +1100"
      },
      "message": "[PATCH] spufs: fix allocation on 64k pages\n\nThe size of the local store is architecture defined\nand independent from the page size, so it should\nnot be defined in terms of pages in the first place.\n\nThis mistake broke a few places when building for\n64kb pages.\n\nSigned-off-by: Arnd Bergmann \u003carndb@de.ibm.com\u003e\nSigned-off-by: Paul Mackerras \u003cpaulus@samba.org\u003e\n"
    },
    {
      "commit": "f0831acc4b78e2d9737e8ed91b8b7505b21ddb83",
      "tree": "0c901e45cdc932776d3953cfcdf66015d6853bec",
      "parents": [
        "ce8ab8541203f6c7be5b2eeaa97f14f1d8d44e4f"
      ],
      "author": {
        "name": "Arnd Bergmann",
        "email": "arnd@arndb.de",
        "time": "Wed Jan 04 20:31:30 2006 +0100"
      },
      "committer": {
        "name": "Paul Mackerras",
        "email": "paulus@samba.org",
        "time": "Mon Jan 09 15:44:49 2006 +1100"
      },
      "message": "[PATCH] spufs: abstract priv1 register access.\n\nIn a hypervisor based setup, direct access to the first\npriviledged register space can typically not be allowed\nto the kernel and has to be implemented through hypervisor\ncalls.\n\nAs suggested by Masato Noguchi, let\u0027s abstract the register\naccess trough a number of function calls. Since there is\ncurrently no public specification of actual hypervisor\ncalls to implement this, I only provide a place that\nmakes it easier to hook into.\n\nCc: Masato Noguchi \u003cMasato.Noguchi@jp.sony.com\u003e\nCc: Geoff Levand \u003cgeoff.levand@am.sony.com\u003e\nSigned-off-by: Arnd Bergmann \u003carndb@de.ibm.com\u003e\nSigned-off-by: Paul Mackerras \u003cpaulus@samba.org\u003e\n"
    },
    {
      "commit": "8837d9216f99048636fbb2c11347358e99e06181",
      "tree": "26e5cab20e59879f5e3271ea73c0320ae31da140",
      "parents": [
        "3f51dd91c80746a5cf76f8c4a77bfc88aa82bb9e"
      ],
      "author": {
        "name": "Arnd Bergmann",
        "email": "arnd@arndb.de",
        "time": "Wed Jan 04 20:31:28 2006 +0100"
      },
      "committer": {
        "name": "Paul Mackerras",
        "email": "paulus@samba.org",
        "time": "Mon Jan 09 15:44:43 2006 +1100"
      },
      "message": "[PATCH] spufs: clean up use of bitops\n\nchecking bits manually might not be synchonized with\nthe use of set_bit/clear_bit. Make sure we always use\nthe correct bitops by removing the unnecessary\nidentifiers.\n\nSigned-off-by: Arnd Bergmann \u003carndb@de.ibm.com\u003e\nSigned-off-by: Paul Mackerras \u003cpaulus@samba.org\u003e\n"
    },
    {
      "commit": "88ced0314938814e1772b4d0d7ab20c52e4472b6",
      "tree": "3e06de882c0bf5706ee7a8994e06eb8c9ed3feca",
      "parents": [
        "e1333803c3a8fb167ba67ffc5540dbb53fa7deb3"
      ],
      "author": {
        "name": "Arnd Bergmann",
        "email": "arnd@arndb.de",
        "time": "Fri Dec 16 22:43:46 2005 +0100"
      },
      "committer": {
        "name": "Paul Mackerras",
        "email": "paulus@samba.org",
        "time": "Mon Jan 09 15:13:08 2006 +1100"
      },
      "message": "[PATCH] powerpc: sanitize header files for user space includes\n\ninclude/asm-ppc/ had #ifdef __KERNEL__ in all header files that\nare not meant for use by user space, include/asm-powerpc does\nnot have this yet.\n\nThis patch gets us a lot closer there. There are a few cases\nwhere I was not sure, so I left them out. I have verified\nthat no CONFIG_* symbols are used outside of __KERNEL__\nany more and that there are no obvious compile errors when\nincluding any of the headers in user space libraries.\n\nSigned-off-by: Arnd Bergmann \u003carnd@arndb.de\u003e\nSigned-off-by: Paul Mackerras \u003cpaulus@samba.org\u003e\n"
    },
    {
      "commit": "2a911f0bb73e67826062b7d073dd7367ca449724",
      "tree": "c1d8d4f340cf7571722b2d9019f158acd345cff4",
      "parents": [
        "5110459f181ef1f11200bb3dec61953f08cc49e7"
      ],
      "author": {
        "name": "Arnd Bergmann",
        "email": "arnd@arndb.de",
        "time": "Mon Dec 05 22:52:26 2005 -0500"
      },
      "committer": {
        "name": "Paul Mackerras",
        "email": "paulus@samba.org",
        "time": "Mon Jan 09 14:52:58 2006 +1100"
      },
      "message": "[PATCH] spufs: Improved SPU preemptability [part 2].\n\nThis patch reduces lock complexity of SPU scheduler, particularly\nfor involuntary preemptive switches.  As a result the new code\ndoes a better job of mapping the highest priority tasks to SPUs.\n\nLock complexity is reduced by using the system default workqueue\nto perform involuntary saves.  In this way we avoid nasty lock\nordering problems that the previous code had.  A \"minimum timeslice\"\nfor SPU contexts is also introduced.  The intent here is to avoid\nthrashing.\n\nWhile the new scheduler does a better job at prioritization it\nstill does nothing for fairness.\n\nFrom: Mark Nutter \u003cmnutter@us.ibm.com\u003e\nSigned-off-by: Arnd Bergmann \u003carndb@de.ibm.com\u003e\nSigned-off-by: Paul Mackerras \u003cpaulus@samba.org\u003e\n"
    },
    {
      "commit": "5110459f181ef1f11200bb3dec61953f08cc49e7",
      "tree": "73356ce50b3fb5055b4a6f39f237f046615f797d",
      "parents": [
        "3b3d22cb84a0bb12f6bbb2b1158972894bec3f21"
      ],
      "author": {
        "name": "Arnd Bergmann",
        "email": "arnd@arndb.de",
        "time": "Mon Dec 05 22:52:25 2005 -0500"
      },
      "committer": {
        "name": "Paul Mackerras",
        "email": "paulus@samba.org",
        "time": "Mon Jan 09 14:52:55 2006 +1100"
      },
      "message": "[PATCH] spufs: Improved SPU preemptability.\n\nThis patch makes it easier to preempt an SPU context by\nhaving the scheduler hold ctx-\u003estate_sema for much shorter\nperiods of time.\n\nAs part of this restructuring, the control logic for the \"run\"\noperation is moved from arch/ppc64/kernel/spu_base.c to\nfs/spufs/file.c.  Of course the base retains \"bottom half\"\nhandlers for class{0,1} irqs.  The new run loop will re-acquire\nan SPU if preempted.\n\nFrom: Mark Nutter \u003cmnutter@us.ibm.com\u003e\nSigned-off-by: Arnd Bergmann \u003carndb@de.ibm.com\u003e\nSigned-off-by: Paul Mackerras \u003cpaulus@samba.org\u003e\n"
    },
    {
      "commit": "8b3d6663c6217e4f50cc3720935a96da9b984117",
      "tree": "5295c29787ac66c26ddf715868fda7fcd3ad5f97",
      "parents": [
        "05b841174c289ca62a6b42d883b8791d9ac3a4bd"
      ],
      "author": {
        "name": "Arnd Bergmann",
        "email": "arnd@arndb.de",
        "time": "Tue Nov 15 15:53:52 2005 -0500"
      },
      "committer": {
        "name": "Paul Mackerras",
        "email": "paulus@samba.org",
        "time": "Mon Jan 09 14:49:30 2006 +1100"
      },
      "message": "[PATCH] spufs: cooperative scheduler support\n\nThis adds a scheduler for SPUs to make it possible to use\nmore logical SPUs than physical ones are present in the\nsystem.\n\nCurrently, there is no support for preempting a running\nSPU thread, they have to leave the SPU by either triggering\nan event on the SPU that causes it to return to the\nowning thread or by sending a signal to it.\n\nThis patch also adds operations that enable accessing an SPU\nin either runnable or saved state. We use an RW semaphore\nto protect the state of the SPU from changing underneath\nus, while we are holding it readable. In order to change\nthe state, it is acquired writeable and a context save\nor restore is executed before downgrading the semaphore\nto read-only.\n\nFrom: Mark Nutter \u003cmnutter@us.ibm.com\u003e,\n      Uli Weigand \u003cUlrich.Weigand@de.ibm.com\u003e\nSigned-off-by: Arnd Bergmann \u003carndb@de.ibm.com\u003e\nSigned-off-by: Paul Mackerras \u003cpaulus@samba.org\u003e\n"
    },
    {
      "commit": "5473af049d8b3556874174e61ce1986c9b5e8fa6",
      "tree": "53da74c13eb9125b85e85f9fc44981d3d1b41b49",
      "parents": [
        "67207b9664a8d603138ef1556141e6d0a102bea7"
      ],
      "author": {
        "name": "Mark Nutter",
        "email": "mnutter@us.ibm.com",
        "time": "Tue Nov 15 15:53:49 2005 -0500"
      },
      "committer": {
        "name": "Paul Mackerras",
        "email": "paulus@samba.org",
        "time": "Mon Jan 09 14:49:16 2006 +1100"
      },
      "message": "[PATCH] spufs: switchable spu contexts\n\nAdd some infrastructure for saving and restoring the context of an\nSPE. This patch creates a new structure that can hold the whole\nstate of a physical SPE in memory. It also contains code that\navoids races during the context switch and the binary code that\nis loaded to the SPU in order to access its registers.\n\nThe actual PPE- and SPE-side context switch code are two separate\npatches.\n\nSigned-off-by: Arnd Bergmann \u003carndb@de.ibm.com\u003e\nSigned-off-by: Paul Mackerras \u003cpaulus@samba.org\u003e\n"
    },
    {
      "commit": "67207b9664a8d603138ef1556141e6d0a102bea7",
      "tree": "e98886778be65aeb6625a5f516873bbc5beeb978",
      "parents": [
        "d7a301033f1990188f65abf4fe8e5b90ef0e3888"
      ],
      "author": {
        "name": "Arnd Bergmann",
        "email": "arnd@arndb.de",
        "time": "Tue Nov 15 15:53:48 2005 -0500"
      },
      "committer": {
        "name": "Paul Mackerras",
        "email": "paulus@samba.org",
        "time": "Mon Jan 09 14:49:12 2006 +1100"
      },
      "message": "[PATCH] spufs: The SPU file system, base\n\nThis is the current version of the spu file system, used\nfor driving SPEs on the Cell Broadband Engine.\n\nThis release is almost identical to the version for the\n2.6.14 kernel posted earlier, which is available as part\nof the Cell BE Linux distribution from\nhttp://www.bsc.es/projects/deepcomputing/linuxoncell/.\n\nThe first patch provides all the interfaces for running\nspu application, but does not have any support for\ndebugging SPU tasks or for scheduling. Both these\nfunctionalities are added in the subsequent patches.\n\nSee Documentation/filesystems/spufs.txt on how to use\nspufs.\n\nSigned-off-by: Arnd Bergmann \u003carndb@de.ibm.com\u003e\nSigned-off-by: Paul Mackerras \u003cpaulus@samba.org\u003e\n"
    }
  ]
}