Age | Commit message (Collapse) | Author |
|
qt-qpr1-dev-plus-aosp-without-vendor@5915889 into stage-aosp-master am: e98cd94d29
am: 405d67056b -s ours
am skip reason: subject contains skip directive
Change-Id: I3ef8628d2053cebccfb3de363c1ed59e0d8444e5
|
|
stage-aosp-master
am: e98cd94d29
Change-Id: I11069259a9ccc37fcc17413188966dcc9f3077fe
|
|
stage-aosp-master
Bug: 142003500
Change-Id: I7d391c8c55bc597b5d55cfae725ec5a941b32bd8
|
|
e67e7eb0a0 am: 1561b1d708
am: aa5ac19173
Change-Id: I0fa26c01134a5520ca431f55b296b79e63726e4b
|
|
e67e7eb0a0
am: 1561b1d708
Change-Id: Ib46f08ba242e42cd98013782910c2b0e015c1261
|
|
am: e67e7eb0a0
Change-Id: I83409f51cc0f0f106e3e70e58f845ba28b27f6d9
|
|
am: 4065871cda
Change-Id: I65ddfa0b1539d059c17cda8b15bef9c64231ad62
|
|
am: 3acae328c6
Change-Id: Ieb61184dc0e8af59a00af0910381ba9a6b11c51e
|
|
We could accidentally generate a NaN/infinity exponent for an overflow
result without noticing it. This meant that we occasionally produced
a NaN instead of infinity.
Fix bit-rot in README.txt testing instructions.
Test: Ran as part of art/test/1972-math-prec. Ran crcalc tests.
Change-Id: I3b218c22d8ec80557115a74528edfa8747ce6acc
|
|
am: 64a1c14118
Change-Id: Ib1a0ff27e2cbf61f9c1dc57a0224510287182253
|
|
am: 24c1121304
Change-Id: Ia8dd89967b8c1071281555af42e7f5961e86a9e9
|
|
am: 577c7451c3
Change-Id: I880b4aae0d1e63c8ef417b866a610efcc7cc2946
|
|
Change-Id: Iebff2debebae11673761f1f2a5bb88525e83f419
BUG: 129345239
|
|
Change-Id: Ib89ae777206affc0bf4744b11497c7d65e376664
BUG: 129345239
|
|
am: dfa22ffcbf
Change-Id: I622f1253b03f741dfc0ccfb57f67bcb18a3f7c08
|
|
am: b882fddd0b
Change-Id: Ice16e397ec1fb5b9f1447d5d437740704546f430
|
|
am: 3e8c12aba8
Change-Id: I92dd1613c1a4fba2691a20c0610bf657a123626a
|
|
See build/soong/README.md for more information.
Fixes: 122331934
Test: atest CRTests
Change-Id: I28874e7c5327914929cdd0fa292664dd27031307
|
|
am: 207e36f6b3
Change-Id: I35691d121a9c9a08f11b7bdc14f8049634f294ba
|
|
am: b1d4a0d26b
Change-Id: I9b5747e24bd3334df0e853f01a62703ccaa8bbe5
|
|
am: 3dfe34c7e0
Change-Id: Ic531a42df90f52b9bf20e0ae9f23b670f7323832
|
|
It used to be the case that we could be interrupted or run out of
memory after pushing nulls onto the two arrays capturing past
computations, but before changing the nulls to real values. This
left us in a corrupted state.
Bug: 117705512
Test: crcalc tests, Calculator tests
Change-Id: I35a5cf97967ede6fa298c178b5056461906bea1e
|
|
am: d38d20e635 -s ours
am skip reason: subject contains skip directive
Change-Id: I0ad82acf572df5ba94d7895e85688e867288d101
|
|
Bug: 120848293
Change-Id: Ib5b3ddabb65966690423ace9f29a82daac7891b0
|
|
am: a5e68d85b4
Change-Id: I844691b7b9985d85c2e53983dc4926791f0f0e2e
|
|
am: 8079f08863
Change-Id: I56cbb4762b1aa2df0d4a536ea442cc8cc99bb8dc
|
|
am: b80dbf3b74
Change-Id: Iaff517d873f5e946f367a2ba4edd823609007ef6
|
|
Bug: 33166666
Test: gerrit uploader
Change-Id: I944df6d9b26af998ad2dc688586868966f732f88
|
|
ours
am: f38945fef1 -s ours
Change-Id: I4def28d0b17583d36ebdae00e43bd6a166e0e1e0
|
|
am: 723478fca1 -s ours
Change-Id: Id06a04c25fbf31933d75c2d2f3761f54be0fe82e
|
|
am: 9e6e7043bf -s ours
Change-Id: Iacbb8549996d3135feffcd845d5c81bdaa435641
|
|
Merge pie-platform-release (PPRL.181205.001, history only) into master
Bug: 120502534
Change-Id: Ide8e527af362dda9cde7f0c24753f912c4b1050f
|
|
am: a68492948b
Change-Id: I60b69b89ed9d59016ecef00a0e702577b711cf7c
|
|
am: 2e0a80b8e8
Change-Id: I004e9d972d2a6048130f59ce62470c69c94e219f
|
|
am: 6bf2e1b8f2
Change-Id: I3a805067939fb91476c58541b7465e35705a9906
|
|
max_msd + r could overflow. Don't believe the original comment.
Also fixed a couple of unused variable warnings discovered in the
process.
Test: Ran provided tests on Android device, and ran various
calculator-related tests.
Change-Id: I7de61a894267a80cbf3b561616fd8504afc247df
|
|
Change-Id: Ifbe967b66de58bc4099139d65548e5af6d29e651
|
|
am: 4383b03a02
Change-Id: Id3ef8a677400eb05580cca238ea6eb1da10d78b5
|
|
am: 41a06cabfc
Change-Id: I365ff8da49e5543bc8f1f99b556d0e57217d3aa6
|
|
am: 5ee5a59192
Change-Id: I46a83d0fb874a51d29a123b82350b9de2e955406
|
|
Bug: 71686706
This one actually fixes the above issue in AOSP.
Without this, sqrt() tended to force evaluation of the argument
to twice the the requested precision, which clearly didn't scale.
For some reason, that didn't seem to be reproducible for all
calculator versions.
Test: Calculator tests + CR tests
Change-Id: I2cc0053f5e27d5d0364c707fe580aa246bbf99d0
|
|
Change-Id: Ieb4932390bd149aeb074063509768d942bdd1f4c
|
|
8d8c9f82a4
am: dd41fd08d7
Change-Id: Ie5fade5ea1555362d2db51eb281a89fdccd911af
|
|
am: 8d8c9f82a4
Change-Id: I8588cbb390482bbcd7065003ce54b7f41e28469c
|
|
am: 568c8e610b
Change-Id: I10293a6dd97fa86e7a7a7cabe3c763e821448eb3
|
|
Bug: 71686706
Always set appr_valid when we set the approximation. The sqrt_CR
constructor used by the pi computation failed to do that.
Have square root evaluate the argument to full precision before
the recursion. That way we evaluate the argument only once and compute
the lesser approximations via simple shifts.
Fix the comment summarizing the Newton iteration computation.
Rename max_prec_needed, since the name confused me while trying to
understand this code again.
Test: Calculator tests + CR tests
Change-Id: I6d704f24579bae5a5edb4c42487acf1be8fdcee6
|
|
Change-Id: I2e769546c2a2c0fc2d3b1bc5d97640e09ca3b079
|
|
am: 0c9acfc0d1
am: c0ef89f21b
Change-Id: I6943aaed41937d6226347d40e3379da1f1cc0125
|
|
am: 0c9acfc0d1
Change-Id: I0f2d9fe2bd643603704084a40b108b6764a84e3a
|
|
am: 56d4e91360
Change-Id: I6215c90f65b1fb8d21161c5be340b83013d90046
|