)]}'
{
  "log": [
    {
      "commit": "33f0f88f1c51ae5c2d593d26960c760ea154c2e2",
      "tree": "f53a38cf49406863f079d74d0e8f91b276f7c1a9",
      "parents": [
        "6ed80991a2dce4afc113be35089c564d62fa1f11"
      ],
      "author": {
        "name": "Alan Cox",
        "email": "alan@lxorguk.ukuu.org.uk",
        "time": "Mon Jan 09 20:54:13 2006 -0800"
      },
      "committer": {
        "name": "Linus Torvalds",
        "email": "torvalds@g5.osdl.org",
        "time": "Tue Jan 10 08:01:59 2006 -0800"
      },
      "message": "[PATCH] TTY layer buffering revamp\n\nThe API and code have been through various bits of initial review by\nserial driver people but they definitely need to live somewhere for a\nwhile so the unconverted drivers can get knocked into shape, existing\ndrivers that have been updated can be better tuned and bugs whacked out.\n\nThis replaces the tty flip buffers with kmalloc objects in rings. In the\nnormal situation for an IRQ driven serial port at typical speeds the\nbehaviour is pretty much the same, two buffers end up allocated and the\nkernel cycles between them as before.\n\nWhen there are delays or at high speed we now behave far better as the\nbuffer pool can grow a bit rather than lose characters. This also means\nthat we can operate at higher speeds reliably.\n\nFor drivers that receive characters in blocks (DMA based, USB and\nespecially virtualisation) the layer allows a lot of driver specific\ncode that works around the tty layer with private secondary queues to be\nremoved. The IBM folks need this sort of layer, the smart serial port\npeople do, the virtualisers do (because a virtualised tty typically\noperates at infinite speed rather than emulating 9600 baud).\n\nFinally many drivers had invalid and unsafe attempts to avoid buffer\noverflows by directly invoking tty methods extracted out of the innards\nof work queue structs. These are no longer needed and all go away. That\nfixes various random hangs with serial ports on overflow.\n\nThe other change in here is to optimise the receive_room path that is\nused by some callers. It turns out that only one ldisc uses receive room\nexcept asa constant and it updates it far far less than the value is\nread. We thus make it a variable not a function call.\n\nI expect the code to contain bugs due to the size alone but I\u0027ll be\nwatching and squashing them and feeding out new patches as it goes.\n\nBecause the buffers now dynamically expand you should only run out of\nbuffering when the kernel runs out of memory for real.  That means a lot of\nthe horrible hacks high performance drivers used to do just aren\u0027t needed any\nmore.\n\nDescription:\n\ntty_insert_flip_char is an old API and continues to work as before, as does\ntty_flip_buffer_push() [this is why many drivers dont need modification].  It\ndoes now also return the number of chars inserted\n\nThere are also\n\ntty_buffer_request_room(tty, len)\n\nwhich asks for a buffer block of the length requested and returns the space\nfound.  This improves efficiency with hardware that knows how much to\ntransfer.\n\nand tty_insert_flip_string_flags(tty, str, flags, len)\n\nto insert a string of characters and flags\n\nFor a smart interface the usual code is\n\n    len \u003d tty_request_buffer_room(tty, amount_hardware_says);\n    tty_insert_flip_string(tty, buffer_from_card, len);\n\nMore description!\n\nAt the moment tty buffers are attached directly to the tty.  This is causing a\nlot of the problems related to tty layer locking, also problems at high speed\nand also with bursty data (such as occurs in virtualised environments)\n\nI\u0027m working on ripping out the flip buffers and replacing them with a pool of\ndynamically allocated buffers.  This allows both for old style \"byte I/O\"\ndevices and also helps virtualisation and smart devices where large blocks of\ndata suddenely materialise and need storing.\n\nSo far so good.  Lots of drivers reference tty-\u003eflip.*.  Several of them also\ncall directly and unsafely into function pointers it provides.  This will all\nbreak.  Most drivers can use tty_insert_flip_char which can be kept as an API\nbut others need more.\n\nAt the moment I\u0027ve added the following interfaces, if people think more will\nbe needed now is a good time to say\n\n int tty_buffer_request_room(tty, size)\n\nTry and ensure at least size bytes are available, returns actual room (may be\nzero).  At the moment it just uses the flipbuf space but that will change.\nRepeated calls without characters being added are not cumulative.  (ie if you\ncall it with 1, 1, 1, and then 4 you\u0027ll have four characters of space.  The\nother functions will also try and grow buffers in future but this will be a\nmore efficient way when you know block sizes.\n\n int tty_insert_flip_char(tty, ch, flag)\n\nAs before insert a character if there is room.  Now returns 1 for success, 0\nfor failure.\n\n int tty_insert_flip_string(tty, str, len)\n\nInsert a block of non error characters.  Returns the number inserted.\n\n int tty_prepare_flip_string(tty, strptr, len)\n\nAdjust the buffer to allow len characters to be added.  Returns a buffer\npointer in strptr and the length available.  This allows for hardware that\nneeds to use functions like insl or mencpy_fromio.\n\nSigned-off-by: Alan Cox \u003calan@redhat.com\u003e\nCc: Paul Fulghum \u003cpaulkf@microgate.com\u003e\nSigned-off-by: Hirokazu Takata \u003ctakata@linux-m32r.org\u003e\nSigned-off-by: Serge Hallyn \u003cserue@us.ibm.com\u003e\nSigned-off-by: Jeff Dike \u003cjdike@addtoit.com\u003e\nSigned-off-by: John Hawkes \u003chawkes@sgi.com\u003e\nSigned-off-by: Martin Schwidefsky \u003cschwidefsky@de.ibm.com\u003e\nSigned-off-by: Adrian Bunk \u003cbunk@stusta.de\u003e\nSigned-off-by: Andrew Morton \u003cakpm@osdl.org\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@osdl.org\u003e\n"
    },
    {
      "commit": "ff55fe2075e3901db4dbdc00e0b44a71bef97afd",
      "tree": "c6dfc8ba5d04fda4c5cb15cdebd5425ab674f8df",
      "parents": [
        "69ac59647e66c1b53fb98fe8b6d0f2099cffad60"
      ],
      "author": {
        "name": "Jason Baron",
        "email": "jbaron@redhat.com",
        "time": "Fri Sep 09 13:01:57 2005 -0700"
      },
      "committer": {
        "name": "Linus Torvalds",
        "email": "torvalds@g5.osdl.org",
        "time": "Fri Sep 09 13:57:31 2005 -0700"
      },
      "message": "[PATCH] pty_chars_in_buffer oops fix\n\nThe idea of this patch is to lock both sides of a ptmx/pty pair during line\ndiscipline changing.  This is needed to ensure that say a poll on one side of\nthe pty doesn\u0027t occur while the line discipline is actively being changed.\nThis resulted in an oops reported on lkml, see:\n\n\thttp://marc.theaimsgroup.com/?l\u003dlinux-kernel\u0026m\u003d111342171410005\u0026w\u003d2\n\nA \u0027hacky\u0027 approach was previously implmemented which served to eliminate the\npoll vs.  line discipline changing race.  However, this patch takes a more\ngeneral approach to the issue.  The patch only adds locking on a less often\nused path, the line-discipline changing path, as opposed to locking the\nptmx/pty pair on read/write/poll paths.\n\nThe patch below, takes both ldisc locks in either order b/c the locks are both\ntaken under the same spinlock().  I thought about locking the ptmx/pty\nseparately, such as master always first but that introduces a 3 way deadlock.\nFor example, process 1 does a blocking read on the slave side.  Then, process\n2 does an ldisc change on the slave side, which acquires the master ldisc lock\nbut not the slave\u0027s.  Finally, process 3 does a write which blocks on the\nprocess 2\u0027s ldisc reference.\n\nThis patch does introduce some changes in semantics.  For example, a line\ndiscipline change on side \u0027a\u0027 of a ptmx/pty pair, will now wait for a\nread/write to complete on the other side, or side \u0027b\u0027.  The current behavior\nis to simply wait for any read/writes on only side \u0027a\u0027, not both sides \u0027a\u0027 and\n\u0027b\u0027.  I think this behavior makes sense, but I wanted to point it out.\n\nI\u0027ve tested the patch with a bunch of read/write/poll while changing the line\ndiscipline out from underneath.\n\nThis patch obviates the need for the above \"hide the problem\" patch.\n\nSigned-off-by: Jason Baron \u003cjbaron@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"
    }
  ]
}