)]}' { "log": [ { "commit": "60b8267338aafde5315fc65ff385f3d4d75eccfe", "tree": "7e6e42c232a23b2e92ef2bedc9bd3a30b31ab92e", "parents": [ "930cc144a043ff95e56b6888fa51c618b33f89e7" ], "author": { "name": "David S. Miller", "email": "davem@davemloft.net", "time": "Wed Oct 22 22:09:59 2008 -0700" }, "committer": { "name": "David S. Miller", "email": "davem@davemloft.net", "time": "Wed Oct 22 22:09:59 2008 -0700" }, "message": "math-emu: Fix thinko in _FP_DIV\n\nIn commit 48d6c64311ddb6417b901639530ccbc47bdc7635 (\"math-emu: Add\nsupport for reporting exact invalid exception\") code was added to\nset the new FP_EX_INVALID_{IDI,ZDZ} exception flag bits.\n\nHowever there is a missing break statement for the\n_FP_CLS_COMBINE(FP_CLS_INF,FP_CLS_INF) switch case, the\ncode just falls into _FP_CLS_COMBINE(FP_CLS_ZERO,FP_CLS_ZERO)\nwhich then proceeds to overwrite all of the settings.\n\nFix by adding the missing break.\n\nSigned-off-by: David S. Miller \u003cdavem@davemloft.net\u003e\n" }, { "commit": "930cc144a043ff95e56b6888fa51c618b33f89e7", "tree": "95a446505d2e15ea2d1f4464c2a7a014402a4b95", "parents": [ "d41e2d7317cd55cc5135356a05c289537b0f6d70" ], "author": { "name": "Kumar Gala", "email": "galak@kernel.crashing.org", "time": "Tue Oct 21 22:19:00 2008 -0700" }, "committer": { "name": "David S. Miller", "email": "davem@davemloft.net", "time": "Wed Oct 22 21:53:53 2008 -0700" }, "message": "math-emu: Fix signalling of underflow and inexact while packing result.\n\nI\u0027m trying to move the powerpc math-emu code to use the include/math-emu bits.\n\nIn doing so I\u0027ve been using TestFloat to see how good or bad we are\ndoing. For the most part the current math-emu code that PPC uses has\na number of issues that the code in include/math-emu seems to solve\n(plus bugs we\u0027ve had for ever that no one every realized).\n\nAnyways, I\u0027ve come across a case that we are flagging underflow and\ninexact because we think we have a denormalized result from a double\nprecision divide:\n\n000.FFFFFFFFFFFFF / 3FE.FFFFFFFFFFFFE\n\tsoft: 001.0000000000000 ..... syst: 001.0000000000000 ...ux\n\nWhat it looks like is the results out of FP_DIV_D are:\n\nD:\nsign:\t 0\nmantissa: 01000000 00000000\nexp:\t -1023 (0)\n\nThe problem seems like we aren\u0027t normalizing the result and bumping the exp.\n\nNow that I\u0027m digging into this a bit I\u0027m thinking my issue has to do with\nthe fix DaveM put in place from back in Aug 2007 (commit\n405849610fd96b4f34cd1875c4c033228fea6c0f):\n\n[MATH-EMU]: Fix underflow exception reporting.\n\n 2) we ended up rounding back up to normal (this is the case where\n we set the exponent to 1 and set the fraction to zero), this\n should set inexact too\n...\n\n Another example, \"0x0.0000000000001p-1022 / 16.0\", should signal both\n inexact and underflow. The cpu implementations and ieee1754\n literature is very clear about this. This is case #2 above.\n\nHere is the distilled glibc test case from Jakub Jelinek which prompted that\ncommit:\n\n--------------------\n#include \u003cfloat.h\u003e\n#include \u003cfenv.h\u003e\n#include \u003cstdio.h\u003e\n\nvolatile double d \u003d DBL_MIN;\nvolatile double e \u003d 0x0.0000000000001p-1022;\nvolatile double f \u003d 16.0;\nint\nmain (void)\n{\n printf (\"%x\\n\", fetestexcept (FE_UNDERFLOW));\n d /\u003d f;\n printf (\"%x\\n\", fetestexcept (FE_UNDERFLOW));\n e /\u003d f;\n printf (\"%x\\n\", fetestexcept (FE_UNDERFLOW));\n return 0;\n}\n--------------------\n\nIt looks like the case I have we are exact before rounding, but think it\nlooks like the rounding case since it appears as if \"overflow is set\".\n\n000.FFFFFFFFFFFFF / 3FE.FFFFFFFFFFFFE \u003d 001.0000000000000\n\nI think the following adds the check for my case and still works for the\nissue your commit was trying to resolve.\n\nSigned-off-by: David S. Miller \u003cdavem@davemloft.net\u003e\n" }, { "commit": "48d6c64311ddb6417b901639530ccbc47bdc7635", "tree": "dfdb08412813343a6ee3bb8b57d34b99ec818a27", "parents": [ "40d3057ac036f2501c1930728a6179be4fca577b" ], "author": { "name": "Kumar Gala", "email": "galak@kernel.crashing.org", "time": "Fri Jun 27 09:39:00 2008 -0500" }, "committer": { "name": "Kumar Gala", "email": "galak@kernel.crashing.org", "time": "Tue Sep 16 10:01:37 2008 -0500" }, "message": "math-emu: Add support for reporting exact invalid exception\n\nSome architectures (like powerpc) provide status information on the exact\ntype of invalid exception. This is pretty straight forward as we already\nreport invalid exceptions via FP_SET_EXCEPTION.\n\nWe add new flags (FP_EX_INVALID_*) the architecture code can define if it\nwants the exact invalid exception reported.\n\nWe had to split out the INF/INF and 0/0 cases for divide to allow reporting\nthe two invalid forms properly.\n\nSigned-off-by: Kumar Gala \u003cgalak@kernel.crashing.org\u003e\nAcked-by: David S. Miller \u003cdavem@davemloft.net\u003e\n" }, { "commit": "40d3057ac036f2501c1930728a6179be4fca577b", "tree": "d6f53f7a35482e5ea83c0ada3e972664ceb3b66e", "parents": [ "82331ab15f14786f3d8e874efb76462685e3bfa0" ], "author": { "name": "Kumar Gala", "email": "galak@kernel.crashing.org", "time": "Fri Jun 27 09:33:59 2008 -0500" }, "committer": { "name": "Kumar Gala", "email": "galak@kernel.crashing.org", "time": "Tue Sep 16 10:01:37 2008 -0500" }, "message": "math-emu: Fix compiler warnings\n\nFix warnings of the form:\narch/powerpc/math-emu/fsubs.c:15: warning: \u0027R_f1\u0027 may be used uninitialized in this function\narch/powerpc/math-emu/fsubs.c:15: warning: \u0027R_f0\u0027 may be used uninitialized in this function\n\nSigned-off-by: Kumar Gala \u003cgalak@kernel.crashing.org\u003e\n" }, { "commit": "2a67789618abb74f0f97d4836a2b937bff2f1b2d", "tree": "bddf5671ce79a5e8cea24e18673ae26c9277293a", "parents": [ "d34fda4a84c18402640a1a2342d6e6d9829e6db7" ], "author": { "name": "Al Viro", "email": "viro@ftp.linux.org.uk", "time": "Sun Aug 19 01:03:07 2007 +0100" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@woody.linux-foundation.org", "time": "Sat Aug 18 17:15:17 2007 -0700" }, "message": "Fix \u003cmath-emu/soft-fp.h\u003e tpyo\n\nSigned-off-by: Al Viro \u003cviro@zeniv.linux.org.uk\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@linux-foundation.org\u003e\n" }, { "commit": "405849610fd96b4f34cd1875c4c033228fea6c0f", "tree": "a1438b7059f39f923d8b21337c0b242ef76f6059", "parents": [ "8b224b813aad0231af62dc75d056aae83c9d4d12" ], "author": { "name": "David S. Miller", "email": "davem@sunset.davemloft.net", "time": "Thu Aug 16 22:59:49 2007 -0700" }, "committer": { "name": "David S. Miller", "email": "davem@sunset.davemloft.net", "time": "Thu Aug 16 22:59:49 2007 -0700" }, "message": "[MATH-EMU]: Fix underflow exception reporting.\n\nThe underflow exception cases were wrong.\n\nThis is one weird area of ieee1754 handling in that the underflow\nbehavior changes based upon whether underflow is enabled in the trap\nenable mask of the FPU control register. As a specific case the Sparc\nV9 manual gives us the following description:\n\n--------------------\nIf UFM \u003d 0: Underflow occurs if a nonzero result is tiny and a\n loss of accuracy occurs. Tininess may be detected\n before or after rounding. Loss of accuracy may be\n either a denormalization loss or an inexact result.\n\nIf UFM \u003d 1: Underflow occurs if a nonzero result is tiny.\n Tininess may be detected before or after rounding.\n--------------------\n\nWhat this amounts to in the packing case is if we go subnormal,\nwe set underflow if any of the following are true:\n\n1) rounding sets inexact\n2) we ended up rounding back up to normal (this is the case where\n we set the exponent to 1 and set the fraction to zero), this\n should set inexact too\n3) underflow is set in FPU control register trap-enable mask\n\nThe initially discovered example was \"DBL_MIN / 16.0\" which\nincorrectly generated an underflow. It should not, unless underflow\nis set in the trap-enable mask of the FPU csr.\n\nAnother example, \"0x0.0000000000001p-1022 / 16.0\", should signal both\ninexact and underflow. The cpu implementations and ieee1754\nliterature is very clear about this. This is case #2 above.\n\nHowever, if underflow is set in the trap enable mask, only underflow\nshould be set and reported as a trap. That is handled properly by the\nprioritization logic in\n\narch/sparc{,64}/math-emu/math.c:record_exception().\n\nBased upon a report and test case from Jakub Jelinek.\n\nSigned-off-by: David S. Miller \u003cdavem@davemloft.net\u003e\n" }, { "commit": "757dea93e136b219af09d3cd56a81063fdbdef1a", "tree": "872f2db0b00716ed7a7e67cf0f0c0f83dbb689c4", "parents": [ "274ee1cd91800a7aa1ed34b7ab2db7c53f09c93a" ], "author": { "name": "Robert P. J. Day", "email": "rpjday@mindspring.com", "time": "Tue May 08 00:27:17 2007 -0700" }, "committer": { "name": "Linus Torvalds", "email": "torvalds@woody.linux-foundation.org", "time": "Tue May 08 11:15:05 2007 -0700" }, "message": "Delete unused header file math-emu/extended.h\n\nSigned-off-by: Robert P. J. Day \u003crpjday@mindspring.com\u003e\nSigned-off-by: Andrew Morton \u003cakpm@linux-foundation.org\u003e\nSigned-off-by: Linus Torvalds \u003ctorvalds@linux-foundation.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" } ] }