commit_msg
stringlengths 1
24.2k
| commit_hash
stringlengths 2
84
⌀ | project
stringlengths 2
40
| source
stringclasses 4
values | labels
int64 0
1
| repo_url
stringlengths 26
70
⌀ | commit_url
stringlengths 74
118
⌀ | commit_date
stringlengths 25
25
⌀ |
|---|---|---|---|---|---|---|---|
Disallow dims input of 0 in tf.raw_ops.UnravelIndex
PiperOrigin-RevId: 384284198
Change-Id: Ia1804ef1aec57b4d857ea507e6891bcccde18e9b | a776040a5e7ebf76eeb7eb923bf1ae417dd4d233 | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/a776040a5e7ebf76eeb7eb923bf1ae417dd4d233 | 2021-07-12 12:00:39-07:00 |
Fix segmentation fault in shape inference logic.
When running shape functions, some functions (such as `MutableHashTableShape`)
produce extra output information in the form of a `ShapeAndType` struct. The
shapes embedded in this struct are owned by an inference context that is
cleaned up almost immediately; if the upstream code attempts to access this
shape information, it can trigger a segfault.
`ShapeRefiner` is mitigating this for normal output shapes by cloning them
(and thus putting the newly created shape under ownership of an inference
context that will not die), but we were not doing the same for shapes and
types. This commit fixes that by doing similar logic on output shapes and
types.
PiperOrigin-RevId: 384761124
Change-Id: I07c0c42d29dfbb55bfa13ec1f09ef825fb0a1a1d | ee119d4a498979525046fba1c3dd3f13a039fbb1 | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/ee119d4a498979525046fba1c3dd3f13a039fbb1 | 2021-07-14 12:50:47-07:00 |
Fix crash in MatrixSolve when inputs have different batch dimensions.
Before, the process would crash or certain elements would be silently ignored. Now an InvalidArgument is raised.
PiperOrigin-RevId: 384844020
Change-Id: Iba44417e383bdd0e1abc4012bfca83b2377dd335 | 579261dcd446385831fe4f7457d802a59685121d | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/579261dcd446385831fe4f7457d802a59685121d | 2021-07-14 20:50:03-07:00 |
Fix crash in softmax-xent when some input dimensions are 1.
Before, tf.nn.softmax_cross_entropy_with_logits would fail a CHECK if one input tensor had shape (1, 1) and the other did not.
In particular, the call to ToIndexArray<2> here https://github.com/tensorflow/tensorflow/blob/1f3da84a89702d3b4f234ee83762d738caffe098/tensorflow/core/kernels/xent_op.cc#L99 would fail, since the call assumed the array had two dimensions. If both dimensions were 1, BCast would merge the two dimensions into a single dimension. Passing fewer_dims_optimization=false stops this optimization
PiperOrigin-RevId: 384844496
Change-Id: Ifb02dc74964132c3ed3f3bc98b0858dbe4e258b7 | 4d74d8a00b07441cba090a02e0dd9ed385145bf4 | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/4d74d8a00b07441cba090a02e0dd9ed385145bf4 | 2021-07-14 21:04:00-07:00 |
Prevent division by 0 in `fully_connected.cc`
PiperOrigin-RevId: 385137282
Change-Id: If201e69b6e0048f0be001330b4b977e2b46db2cb | 718721986aa137691ee23f03638867151f74935f | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/718721986aa137691ee23f03638867151f74935f | 2021-07-16 06:57:28-07:00 |
Fix a null pointer exception in SVDF
This is due to not checking that `GetVariableInput` returns non-null tensor.
Also fix a potential null pointer exception in `GetVariableInput`.
PiperOrigin-RevId: 385160147
Change-Id: Iadf3f0705b036a9014d27caa5a8bbd91f4c4c401 | 5b048e87e4e55990dae6b547add4dae59f4e1c76 | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/5b048e87e4e55990dae6b547add4dae59f4e1c76 | 2021-07-16 09:20:22-07:00 |
Fix a null pointer exception caused by branching on uninitialized data.
This is due to not checking that the params for the quantization exists. If there is no quantization, we should not access the `.params` field.
PiperOrigin-RevId: 385168337
Change-Id: I28661e4f12ba1c92cfeae23d22a3fb2df2a2c6a4 | 4a91f2069f7145aab6ba2d8cfe41be8a110c18a5 | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/4a91f2069f7145aab6ba2d8cfe41be8a110c18a5 | 2021-07-16 10:03:12-07:00 |
Fix a null pointer exception caused by branching on uninitialized data.
This is due to not checking that the params for the quantization exists. If there is no quantization, we should not access the `.params` field.
PiperOrigin-RevId: 385173491
Change-Id: I8fc476c4b274fdb21ba741caa0fbc6d1b8840663 | 8933b8a21280696ab119b63263babdb54c298538 | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/8933b8a21280696ab119b63263babdb54c298538 | 2021-07-16 10:27:37-07:00 |
Prevent a division by 0 in average ops.
PiperOrigin-RevId: 385184660
Change-Id: I7affd4554f9b336fca29ac68f633232c094d0bd3 | dfa22b348b70bb89d6d6ec0ff53973bacb4f4695 | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/dfa22b348b70bb89d6d6ec0ff53973bacb4f4695 | 2021-07-16 11:15:34-07:00 |
Prevent a division by 0 in division ops.
PiperOrigin-RevId: 385223169
Change-Id: Ia4228960b5d2aa44480385f74bdd70d21a3613c3 | 1e206baedf8bef0334cca3eb92bab134ef525a28 | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/1e206baedf8bef0334cca3eb92bab134ef525a28 | 2021-07-16 14:28:13-07:00 |
Disallow negative ngram_widths values in tf.raw_ops.StringNGrams
PiperOrigin-RevId: 387148179
Change-Id: I641395a09a208be72ef9b3ceb128cf8a83a0775b | c283e542a3f422420cfdb332414543b62fc4e4a5 | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/c283e542a3f422420cfdb332414543b62fc4e4a5 | 2021-07-27 11:08:02-07:00 |
Disallow empty node_id_range in tf.raw_ops.BoostedTreesCalculateBestFeatureSplitV2 and tf.raw_ops.BoostedTreesCalculateBestGainsPerFeature
PiperOrigin-RevId: 387165936
Change-Id: I2f70341af96236b2776c2a592c917d549c1fc1e2 | 9c87c32c710d0b5b53dc6fd3bfde4046e1f7a5ad | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/9c87c32c710d0b5b53dc6fd3bfde4046e1f7a5ad | 2021-07-27 12:21:52-07:00 |
Ensure non-empty input_splits in tf.raw_ops.UnicodeEncode
PiperOrigin-RevId: 387170080
Change-Id: I3b489acc51c5cb4124c535b9df7cc6e62ef21766 | 2e0ee46f1a47675152d3d865797a18358881d7a6 | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/2e0ee46f1a47675152d3d865797a18358881d7a6 | 2021-07-27 12:36:17-07:00 |
In tf.raw_ops.BoostedTreesSparseCalculateBestFeatureSplit, limit stat_dim in stats_summary_indices to under stats_dims in stats_summary_shape
PiperOrigin-RevId: 387171191
Change-Id: I83ca8a75b22aa78c037e8b98779da6cced16bfaa | e84c975313e8e8e38bb2ea118196369c45c51378 | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/e84c975313e8e8e38bb2ea118196369c45c51378 | 2021-07-27 12:42:22-07:00 |
Prevent an OOB read in `expand_dims.cc`
The for loop that follows this check assumes that `axis` is between `0` and `input_dims.size`. If user supplied `axis` is negative, the if code before this check is supposed to bring it back to positive (similar to how in Python one can do `l[-3]` to mean `l[-3 + len(l)]`).
PiperOrigin-RevId: 387200206
Change-Id: I162f4feba12d547c3a4340833ae682016a2ebfab | d94ffe08a65400f898241c0374e9edc6fa8ed257 | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/d94ffe08a65400f898241c0374e9edc6fa8ed257 | 2021-07-27 14:52:03-07:00 |
Prevent nullptr dereference in MLIR TFLite dialect/optimizer.
PiperOrigin-RevId: 387220762
Change-Id: Id136ef04bb3d36123b4685d316ae81a9ec924d6b | d6b57f461b39fd1aa8c1b870f1b974aac3554955 | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/d6b57f461b39fd1aa8c1b870f1b974aac3554955 | 2021-07-27 16:32:11-07:00 |
Prevent division by 0 in LSH projection.
PiperOrigin-RevId: 387225857
Change-Id: Iaeb572a763618c64f503e0026f6dd9fd769bf50c | 0575b640091680cfb70f4dd93e70658de43b94f9 | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/0575b640091680cfb70f4dd93e70658de43b94f9 | 2021-07-27 16:50:05-07:00 |
Prevent heap OOB read in TFLite's `gather.cc`.
Passing negative indices is illegal but there was a missing check so that resulted in OOB accesses.
PiperOrigin-RevId: 387231300
Change-Id: I3111b54b2f232638d795be17efc46abe4ede6bf8 | eb921122119a6b6e470ee98b89e65d721663179d | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/eb921122119a6b6e470ee98b89e65d721663179d | 2021-07-27 17:20:51-07:00 |
Secure tf.raw_ops.QuantizeV2
Validate size and shape of min_range and max_range
Ensure axis is within input dims limits
PiperOrigin-RevId: 387232799
Change-Id: I36975281f7b5758e9e31a8dcc73fe610ef456318 | 6da6620efad397c85493b8f8667b821403516708 | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/6da6620efad397c85493b8f8667b821403516708 | 2021-07-27 17:27:17-07:00 |
Prevent dereferencing of null pointers in TFLite's `add.cc`.
PiperOrigin-RevId: 387244946
Change-Id: I56094233327fbd8439b92e1dbb1262176e00eeb9 | 15691e456c7dc9bd6be203b09765b063bf4a380c | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/15691e456c7dc9bd6be203b09765b063bf4a380c | 2021-07-27 18:47:05-07:00 |
Add remaining missing validation to `BoostedTreesCalculateBestFeatureSplit`
PiperOrigin-RevId: 387423006
Change-Id: I8eaf30efb223011519e60707bfa751b275d3a443 | 429f009d2b2c09028647dd4bb7b3f6f414bbaad7 | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/429f009d2b2c09028647dd4bb7b3f6f414bbaad7 | 2021-07-28 13:43:55-07:00 |
Ensure num_streams >= 0 in tf.raw_ops.BoostedTreesCreateQuantileStreamResource
PiperOrigin-RevId: 387452765
Change-Id: I9990c760e177fabca6a3b9b4612ceeaeeba51495 | 8a84f7a2b5a2b27ecf88d25bad9ac777cd2f7992 | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/8a84f7a2b5a2b27ecf88d25bad9ac777cd2f7992 | 2021-07-28 15:49:31-07:00 |
Fix bug that could cause map_fn to produce incorrect results (rather than an error)
when mapping over a ragged tensor with an inappropriate fn_output_signature. (Note: there are cases where the default value for fn_output_signature is not appropriate, so the user needs to explicitly specify the correct output signature.)
PiperOrigin-RevId: 387606546
Change-Id: Ib4ea27b9634e6ab413f211cfe809a69a90f0e2cd | 4e2565483d0ffcadc719bd44893fb7f609bb5f12 | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/4e2565483d0ffcadc719bd44893fb7f609bb5f12 | 2021-07-29 10:06:49-07:00 |
Ensure non-empty rt_nested_splits in tf.raw_ops.RaggedTensorToVariant
PiperOrigin-RevId: 387664237
Change-Id: Ia1700c34b5610873d63561abc86e23b46ead93b3 | be7a4de6adfbd303ce08be4332554dff70362612 | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/be7a4de6adfbd303ce08be4332554dff70362612 | 2021-07-29 14:09:24-07:00 |
Fix segfault/heap buffer overflow in `{Experimental,}DatasetToTFRecord` where dataset is numeric.
Code assumes only strings inputs and then interprets numbers as valid `tstring`s. Then, when trying to compute the CRC of the record this results in heap buffer overflow.
PiperOrigin-RevId: 387675909
Change-Id: I7396b9b8afc1ac744112af7c0b1cd7bb41e0f556 | e0b6e58c328059829c3eb968136f17aa72b6c876 | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/e0b6e58c328059829c3eb968136f17aa72b6c876 | 2021-07-29 15:08:55-07:00 |
Add more validation to `RequantizationRangePerChannel`.
PiperOrigin-RevId: 387693946
Change-Id: Ife8dcbdb021bec4787eef6a4361dd08f17c14bd6 | 9e62869465573cb2d9b5053f1fa02a81fce21d69 | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/9e62869465573cb2d9b5053f1fa02a81fce21d69 | 2021-07-29 16:35:05-07:00 |
Add missing validation to `RaggedTensorToSparse`.
There needs to be a check that the splits allow for valid ragged tensors.
PiperOrigin-RevId: 387712169
Change-Id: I2499175324b82b65d159a260c7f83b98ceb5cc7d | 1071f554dbd09f7e101324d366eec5f4fe5a3ece | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/1071f554dbd09f7e101324d366eec5f4fe5a3ece | 2021-07-29 18:27:51-07:00 |
Prevent division by 0 in common shape functions.
PiperOrigin-RevId: 387712197
Change-Id: Id25c7460e35b68aeeeac23b9a88e455b443ee149 | 8a793b5d7f59e37ac7f3cd0954a750a2fe76bad4 | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/8a793b5d7f59e37ac7f3cd0954a750a2fe76bad4 | 2021-07-29 18:33:51-07:00 |
Fix a shape inference issue leading to nullptr deref.
PiperOrigin-RevId: 387712259
Change-Id: I7e670772b259c068a501a187cd89f18773bb95a1 | da857cfa0fde8f79ad0afdbc94e88b5d4bbec764 | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/da857cfa0fde8f79ad0afdbc94e88b5d4bbec764 | 2021-07-29 18:39:14-07:00 |
Prevent a CHECK-fail due to empty tensor input in `map_stage_op.cc`
PiperOrigin-RevId: 387737906
Change-Id: Idc52df0c71c7ed6e2dd633b651a581932f277c8a | d7de67733925de196ec8863a33445b73f9562d1d | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/d7de67733925de196ec8863a33445b73f9562d1d | 2021-07-29 22:38:10-07:00 |
Add remaining validation to `sdca_internal.cc`
PiperOrigin-RevId: 387738010
Change-Id: I28eedcfd87a53aaf34deb075acea1f8c95470808 | a4e138660270e7599793fa438cd7b2fc2ce215a6 | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/a4e138660270e7599793fa438cd7b2fc2ce215a6 | 2021-07-29 22:49:47-07:00 |
Prevent nullptr deref in validation of indexes in map ops.
PiperOrigin-RevId: 387738023
Change-Id: I83d18d36a7b82ffd2a40b5124a4e5b4c72238f27 | 532f5c5a547126c634fefd43bbad1dc6417678ac | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/532f5c5a547126c634fefd43bbad1dc6417678ac | 2021-07-29 22:54:59-07:00 |
Prevent overflow due to integer conversion to unsigned.
PiperOrigin-RevId: 387738045
Change-Id: Id7e95bc07e02df1c66b72bd09f389608c87bdebe | b5cdbf12ffcaaffecf98f22a6be5a64bb96e4f58 | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/b5cdbf12ffcaaffecf98f22a6be5a64bb96e4f58 | 2021-07-29 23:04:06-07:00 |
Prevent CHECK-fail/heap OOB in UpperBound and LowerBound
PiperOrigin-RevId: 387738073
Change-Id: Iee74de95ddad18440d052a75a5a1cb67544f490a | 42459e4273c2e47a3232cc16c4f4fff3b3a35c38 | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/42459e4273c2e47a3232cc16c4f4fff3b3a35c38 | 2021-07-29 23:10:08-07:00 |
Reorganize and add more validation to MKL requantization
PiperOrigin-RevId: 387901341
Change-Id: I2515b9034c64e113db0bcec8337d30643ab0a0f1 | 203214568f5bc237603dbab6e1fd389f1572f5c9 | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/203214568f5bc237603dbab6e1fd389f1572f5c9 | 2021-07-30 16:11:14-07:00 |
Add one missing valdiation to `matrix_set_diag_op.cc`
PiperOrigin-RevId: 387923408
Change-Id: If6a97b9098c13879400f56c22f91555cdf0ce5d7 | ff8894044dfae5568ecbf2ed514c1a37dc394f1b | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/ff8894044dfae5568ecbf2ed514c1a37dc394f1b | 2021-07-30 19:06:58-07:00 |
Add missing validation to `matrix_diag_op.cc`
PiperOrigin-RevId: 387923533
Change-Id: Idfffeb328d5f9c6748d992d28a56d6e9e45103a0 | f2a673bd34f0d64b8e40a551ac78989d16daad09 | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/f2a673bd34f0d64b8e40a551ac78989d16daad09 | 2021-07-30 19:12:41-07:00 |
Ensure validation sticks in `save_restore_v2_ops.cc`
PiperOrigin-RevId: 387924206
Change-Id: I6156842eb3230076b5812c0815f3e66bd5241454 | 9728c60e136912a12d99ca56e106b7cce7af5986 | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/9728c60e136912a12d99ca56e106b7cce7af5986 | 2021-07-30 19:18:28-07:00 |
Prevent use after free.
A very old version of the code used `result` as a simple pointer to a resource. Two years later, the pointer got changed to a `unique_ptr` but author forgot to remove the call to `Unref`. Three years after that, we finally uncover the UAF.
PiperOrigin-RevId: 387924872
Change-Id: I70fb6f199164de49fac20c168132a07b84903f9b | 5ecec9c6fbdbc6be03295685190a45e7eee726ab | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/5ecec9c6fbdbc6be03295685190a45e7eee726ab | 2021-07-30 19:27:22-07:00 |
Add missing validation to `maxpooling_op.cc`
PiperOrigin-RevId: 387932441
Change-Id: I43a0b24e6a12cc965611144ba035accd384594b9 | 136b51f10903e044308cf77117c0ed9871350475 | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/136b51f10903e044308cf77117c0ed9871350475 | 2021-07-30 20:55:05-07:00 |
Prevent heap OOB in sparse reduction ops.
PiperOrigin-RevId: 387934524
Change-Id: I894aa30f1e454f09b471d565b4a325da49322c1a | 87158f43f05f2720a374f3e6d22a7aaa3a33f750 | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/87158f43f05f2720a374f3e6d22a7aaa3a33f750 | 2021-07-30 21:16:13-07:00 |
Prevent heap oob access in `resource_variable_ops.cc`
PiperOrigin-RevId: 387936433
Change-Id: I9e71ddaa8dbd51ec6afbf163a6b3b591f193b4f6 | bc9c546ce7015c57c2f15c168b3d9201de679a1d | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/bc9c546ce7015c57c2f15c168b3d9201de679a1d | 2021-07-30 21:50:06-07:00 |
Fix nullptr deref and heap OOB access in binary cwise ops.
PiperOrigin-RevId: 387936777
Change-Id: I608b8074cec36a982cca622b7144cb2c43e6e19f | 93f428fd1768df147171ed674fee1fc5ab8309ec | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/93f428fd1768df147171ed674fee1fc5ab8309ec | 2021-07-30 21:55:05-07:00 |
Prevent crash/heap OOB due to integer conversion to unsigned in NMS kernels
PiperOrigin-RevId: 387938262
Change-Id: Id361a715307e7179977cf5c64391c199a966f2ad | 3a7362750d5c372420aa8f0caf7bf5b5c3d0f52d | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/3a7362750d5c372420aa8f0caf7bf5b5c3d0f52d | 2021-07-30 22:11:06-07:00 |
Prevent division by 0 in `resource_variable_ops.cc`
PiperOrigin-RevId: 387939939
Change-Id: Ib04902d63756633999959a70613f2eaa30c2c151 | ac117ee8a8ea57b73d34665cdf00ef3303bc0b11 | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/ac117ee8a8ea57b73d34665cdf00ef3303bc0b11 | 2021-07-30 22:27:12-07:00 |
Validate dimensions of input tensor in `FractionalAvgPoolGrad`
PiperOrigin-RevId: 388286227
Change-Id: Ieb7566155e92acc8993a2212c76deacadc0edc8a | 0f931751fb20f565c4e94aa6df58d54a003cdb30 | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/0f931751fb20f565c4e94aa6df58d54a003cdb30 | 2021-08-02 13:09:46-07:00 |
Validate axis input in tf.raw_ops.QuantizeAndDequantizeV4Grad
PiperOrigin-RevId: 388291385
Change-Id: I3bab68dc61d935afa96c0da021a7b722c6dc8dc8 | 96f364a1ca3009f98980021c4b32be5fdcca33a1 | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/96f364a1ca3009f98980021c4b32be5fdcca33a1 | 2021-08-02 13:32:13-07:00 |
Fix heap OOB due to dimension mismatch in `ResourceScatterUpdate`
PiperOrigin-RevId: 388292801
Change-Id: Id9bd7244d98d41b1517d4771850b32782c0cc949 | 01cff3f986259d661103412a20745928c727326f | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/01cff3f986259d661103412a20745928c727326f | 2021-08-02 13:37:41-07:00 |
Don't do any work when reshaping 0 elements sparse tensor.
If reshaping to 0 elements tensor, check that input has no elements.
If reshaping no elements input, check that output has no elements.
PiperOrigin-RevId: 388296986
Change-Id: Iadc9fe7252e14313ca987e69bf0d7042fd10232a | 4923de56ec94fff7770df259ab7f2288a74feb41 | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/4923de56ec94fff7770df259ab7f2288a74feb41 | 2021-08-02 13:57:19-07:00 |
Fix FPE in inpace update ops.
PiperOrigin-RevId: 388303197
Change-Id: Ib48309b6213ffe53eba81004b00e889d653e4b83 | e86605c0a336c088b638da02135ea6f9f6753618 | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/e86605c0a336c088b638da02135ea6f9f6753618 | 2021-08-02 14:25:59-07:00 |
Fix NPE in restoring code.
PiperOrigin-RevId: 388303253
Change-Id: Ia8c68568cb854bca538909a182b31a618d68ce55 | 9e82dce6e6bd1f36a57e08fa85af213e2b2f2622 | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/9e82dce6e6bd1f36a57e08fa85af213e2b2f2622 | 2021-08-02 14:34:00-07:00 |
Ensure non-empty padding_value input to tf.raw_ops.MatrixDiagPartV2, if a padding_value is input
PiperOrigin-RevId: 388314614
Change-Id: If0b51ad58d5d8543a6be6ce8f42ae4755c80d55f | 482da92095c4d48f8784b1f00dda4f81c28d2988 | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/482da92095c4d48f8784b1f00dda4f81c28d2988 | 2021-08-02 15:21:23-07:00 |
Fix heap OOB in `tf.raw_ops.RaggedGather`
PiperOrigin-RevId: 388355464
Change-Id: If14d96231d1cd7aad7c4d1c22c1bab1576b75717 | a2b743f6017d7b97af1fe49087ae15f0ac634373 | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/a2b743f6017d7b97af1fe49087ae15f0ac634373 | 2021-08-02 19:10:10-07:00 |
Disallow division by zero FPE in `tf.raw_ops.ResourceScatterDiv`
Had to update a test that was broken.
PiperOrigin-RevId: 388516976
Change-Id: Ic358e6bf0559e011539974d453fc7aa18b427e9c | 4aacb30888638da75023e6601149415b39763d76 | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/4aacb30888638da75023e6601149415b39763d76 | 2021-08-03 12:33:45-07:00 |
Prevent nullptr deref in SparseTensorSliceDataset
The arguments must determine a valid sparse tensor. This means that when indices are empty then the values must be empty too (and the reverse).
Also added test, by modifying existing test with empty sparse tensor to now run with an invalid sparse tensor input.
PiperOrigin-RevId: 388562757
Change-Id: Id8b54cd7c2316025b4f9a77292c8fb5344d17609 | 02cc160e29d20631de3859c6653184e3f876b9d7 | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/02cc160e29d20631de3859c6653184e3f876b9d7 | 2021-08-03 15:57:40-07:00 |
Fix EinsumHelper::ParseEquation to avoid uninitialized accesses.
EinsumHelper::ParseEquation is supposed to return true or false in
input_has_ellipsis and output_has_ellipsis to indicate whether there is
ellipsis in the inputs and output. Previously, when there is no ellipsis in the
inputs or output, the routine doesn't assign false to the variables. This
change initializes the two variables with false to fix the problem.
PiperOrigin-RevId: 391772004
Change-Id: I17b6c88aadef4131470378e48cced054bf252e86 | f09caa532b6e1ac8d2aa61b7832c78c5b79300c6 | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/f09caa532b6e1ac8d2aa61b7832c78c5b79300c6 | 2021-08-19 09:10:18-07:00 |
Don't constant-fold DT_RESOURCE constants.
PiperOrigin-RevId: 391803952
Change-Id: I0ea3ec31d3e7dfda0f03b4027a237f08d00a3091 | 7731e8dfbe4a56773be5dc94d631611211156659 | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/7731e8dfbe4a56773be5dc94d631611211156659 | 2021-08-19 11:29:55-07:00 |
Prevent crashes when loading tensor slices with unsupported types.
Also fix the `Tensor(const TensorShape&)` constructor swapping the LOG(FATAL)
messages for the unset and unsupported types.
PiperOrigin-RevId: 392695027
Change-Id: I4beda7db950db951d273e3259a7c8534ece49354 | abcced051cb1bd8fb05046ac3b6023a7ebcc4578 | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/abcced051cb1bd8fb05046ac3b6023a7ebcc4578 | 2021-08-24 10:51:32-07:00 |
Use BuildTensorShapeBase when parsing unverified TensorShapes during checkpoint loading.
This avoids crashing when the TensorShape has negative dimensions.
PiperOrigin-RevId: 392769882
Change-Id: Id1f7ae7fcf8142193556af47abfda81b13d3cce4 | b619c6f865715ca3b15ef1842b5b95edbaa710ad | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/b619c6f865715ca3b15ef1842b5b95edbaa710ad | 2021-08-24 16:17:29-07:00 |
Add BuildTensorSlice for building from unvalidated TensorSliceProtos.
This avoids several sources of crashes and undefined behavior when loading
invalid checkpoints.
PiperOrigin-RevId: 392785704
Change-Id: Icd9713c768b882f3b58b427eddac376060696833 | e8dc63704c88007ee4713076605c90188d66f3d2 | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/e8dc63704c88007ee4713076605c90188d66f3d2 | 2021-08-24 17:45:52-07:00 |
Avoid buffer overflow when loading tensors with insufficient data from checkpoints.
`CopyDataFromTensorSliceToTensorSlice` does not (and cannot conveniently)
provide any bounds checking on its own, so the size is instead checked prior
to passing unvalidated data to that function.
PiperOrigin-RevId: 392971286
Change-Id: If2073b36d4d5eedd386329f56729395fd7effee1 | 368af875869a204b4ac552b9ddda59f6a46a56ec | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/368af875869a204b4ac552b9ddda59f6a46a56ec | 2021-08-25 13:49:17-07:00 |
PR #51732: Fix crash of tf.image.crop_and_resize when input is large number
Imported from GitHub PR https://github.com/tensorflow/tensorflow/pull/51732
This PR is part of the effort in #46890 where
tf.image.crop_and_resize will crash if shape consists of large number.
Signed-off-by: Yong Tang <[email protected]>
Copybara import of the project:
--
c8d87055a56d8740d27ad8bdc74a7459ede6900e by Yong Tang <[email protected]>:
Fix crash of tf.image.crop_and_resize when input is large number
This PR is part of the effort in 46890 where
tf.image.crop_and_resize will crash if shape consists of large number.
Signed-off-by: Yong Tang <[email protected]>
COPYBARA_INTEGRATE_REVIEW=https://github.com/tensorflow/tensorflow/pull/51732 from yongtang:46890-tf.image.crop_and_resize c8d87055a56d8740d27ad8bdc74a7459ede6900e
PiperOrigin-RevId: 394109830
Change-Id: If049dad0844df9353722029ee95bc76819eda1f4 | 7c1692bd417eb4f9b33ead749a41166d6080af85 | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/7c1692bd417eb4f9b33ead749a41166d6080af85 | 2021-08-31 16:29:59-07:00 |
Prevent unitialized variable use in grappler.
PiperOrigin-RevId: 399702928
Change-Id: Id7e75451fbff297692dfb687f60ea04b25c96b24 | 68867bf01239d9e1048f98cbad185bf4761bedd3 | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/68867bf01239d9e1048f98cbad185bf4761bedd3 | 2021-09-29 09:31:27-07:00 |
Add shape checks to FusedBatchNorm kernels.
PiperOrigin-RevId: 399755576
Change-Id: If8049fde109cc33badb5509d174b9b95aee1ea5e | aab9998916c2ffbd8f0592059fad352622f89cda | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/aab9998916c2ffbd8f0592059fad352622f89cda | 2021-09-29 13:17:15-07:00 |
Prevent out-of-bound accesses in SparseBincount.
PiperOrigin-RevId: 399918616
Change-Id: I11d154f4444d3fde1f09c5c40628b8671791a30d | f410212e373eb2aec4c9e60bf3702eba99a38aba | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/f410212e373eb2aec4c9e60bf3702eba99a38aba | 2021-09-30 06:43:41-07:00 |
Make SparseFillEmptyRows validate that the length of `values` must be equal to the number of index tuples.
PiperOrigin-RevId: 399969549
Change-Id: I3c2f2ca1c1d2cc88bb5951c6958b38c16e9436c8 | 67bfd9feeecfb3c61d80f0e46d89c170fbee682b | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/67bfd9feeecfb3c61d80f0e46d89c170fbee682b | 2021-09-30 10:49:53-07:00 |
Fix null pointer exception in shape inference function when tf.ragged.cross() is called with invalid inputs.
PiperOrigin-RevId: 400045848
Change-Id: Ia65501583b85cf1ec14a252d83fbdd716817a516 | fa6b7782fbb14aa08d767bc799c531f5e1fb3bb8 | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/fa6b7782fbb14aa08d767bc799c531f5e1fb3bb8 | 2021-09-30 16:10:47-07:00 |
Update TPU AllToAll op to avoid divide by 0.
PiperOrigin-RevId: 400259638
Change-Id: Ic4cfe4fe7159da38caed8044ee005f898e42cd86 | a8ad3e5e79c75f36edb81e0ba3f3c0c5442aeddc | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/a8ad3e5e79c75f36edb81e0ba3f3c0c5442aeddc | 2021-10-01 12:15:18-07:00 |
Prevent OOB access in QuantizeV2 shape inference
PiperOrigin-RevId: 400309614
Change-Id: I31412c71b05b4f21b677f7fa715a61499cbee39d | a0d64445116c43cf46a5666bd4eee28e7a82f244 | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/a0d64445116c43cf46a5666bd4eee28e7a82f244 | 2021-10-01 16:02:45-07:00 |
Fix access to undefined memory during shape inference of Cudnn*.
PiperOrigin-RevId: 400324259
Change-Id: Ie3b7859d19ae24ee9ac2adf413bdc1e851bbc604 | af5fcebb37c8b5d71c237f4e59c6477015c78ce6 | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/af5fcebb37c8b5d71c237f4e59c6477015c78ce6 | 2021-10-01 17:19:05-07:00 |
Fix macros for converting little endian to host for TF_TSRT_OFFSET GetSize
Make the macro that converts little endian data do nothing on little endian hosts,
and byte swap otherwise.
This only affects getting the size of TStrings of type "Offset".
Added a test for TStrings of type "Offset" that checks if type and size are consistent.
PiperOrigin-RevId: 400789721
Change-Id: I1398bffd842ab1631614b212b7c3a2af88d99538 | 3712a2d3455e6ccb924daa5724a3652a86f6b585 | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/3712a2d3455e6ccb924daa5724a3652a86f6b585 | 2021-10-04 12:54:06-07:00 |
Fix undefined behavior in CollectiveReduceV2 and others
We should not call done after it's moved.
PiperOrigin-RevId: 400838185
Change-Id: Ifc979740054b8f8c6f4d50acc89472fe60c4fdb1 | ca38dab9d3ee66c5de06f11af9a4b1200da5ef75 | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/ca38dab9d3ee66c5de06f11af9a4b1200da5ef75 | 2021-10-04 16:15:01-07:00 |
A negative size in one of the split sizes allowed the computed size of another
to exceed the total dimension, leading to a segfault and security vulnerability.
Adding a check for negative sizes prevents this.
PiperOrigin-RevId: 401035665
Change-Id: I79bbe329787dac82aa4bf60397a9129b716aedab | 25d622ffc432acc736b14ca3904177579e733cc6 | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/25d622ffc432acc736b14ca3904177579e733cc6 | 2021-10-05 11:41:35-07:00 |
Add error checking to ImmutableConst OP that strings are not yet supported.
PiperOrigin-RevId: 401065359
Change-Id: I9dd2bd2a2c36f22f4a05153daf6ebdc4613469d2 | 1cb6bb6c2a6019417c9adaf9e6843ba75ee2580b | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/1cb6bb6c2a6019417c9adaf9e6843ba75ee2580b | 2021-10-05 13:46:10-07:00 |
Validate that matrix dimension sizes in SparseMatMul are positive.
PiperOrigin-RevId: 401149683
Change-Id: Ib33eafc561a39c8741ece80b2edce6d4aae9a57d | e6cf28c72ba2eb949ca950d834dd6d66bb01cfae | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/e6cf28c72ba2eb949ca950d834dd6d66bb01cfae | 2021-10-05 22:02:57-07:00 |
Add shape checks to GPU TridiagonalMatMul.
When given invalid shapes, the GPU TridiagonalMatMul op could read invalid or uninitialized GPU memory.
PiperOrigin-RevId: 401775483
Change-Id: Ib5500aeb8225e50d4ce790b06d2c34751f544ad8 | 68422b215e618df5ad375bcdc6d2052e9fd3080a | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/68422b215e618df5ad375bcdc6d2052e9fd3080a | 2021-10-08 08:27:17-07:00 |
Roll forward https://github.com/tensorflow/tensorflow/commit/ab0ca4bbc66a476aea305f81c69e0201b5876d0a. The internal test that it broke has been fixed.
PiperOrigin-RevId: 401913101
Change-Id: I67f095899187e38101fbb10289c5e444b0a9e8c0 | da4aad5946be30e5f049920fa076e1f7ef021261 | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/da4aad5946be30e5f049920fa076e1f7ef021261 | 2021-10-08 20:30:17-07:00 |
Address QuantizeAndDequantizeV* heap oob. Added additional checks for the 'axis' attribute.
PiperOrigin-RevId: 402446942
Change-Id: Id2f6b82e4e740d0550329be02621c46466b5a5b9 | 7cf73a2274732c9d82af51c2bc2cf90d13cd7e6d | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/7cf73a2274732c9d82af51c2bc2cf90d13cd7e6d | 2021-10-11 19:07:32-07:00 |
Adding more validation checks to _ParallelConcatUpdate to avoid NPE.
PiperOrigin-RevId: 402569467
Change-Id: I2db122dab68be2a5e4e8dd3375f5a70c4d2307ec | f2c3931113eaafe9ef558faaddd48e00a6606235 | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/f2c3931113eaafe9ef558faaddd48e00a6606235 | 2021-10-12 08:36:20-07:00 |
Make Transpose's shape inference function validate that negative `perm` values are within the tensor's rank.
PiperOrigin-RevId: 403252853
Change-Id: Ia6b31b45b237312668bb31c2c3b3c7bbce2d2610 | c79ba87153ee343401dbe9d1954d7f79e521eb14 | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/c79ba87153ee343401dbe9d1954d7f79e521eb14 | 2021-10-14 19:42:55-07:00 |
Fix a NPE issue in invalid Exit op. Now it will report an error instead of crash.
PiperOrigin-RevId: 404089902
Change-Id: Ia6ec55445ea70ad045a4d339d354959ad0618f2a | 05cbebd3c6bb8f517a158b0155debb8df79017ff | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/05cbebd3c6bb8f517a158b0155debb8df79017ff | 2021-10-18 16:34:24-07:00 |
Fix segfault in pools on empty shapes when certain dimension were very large.
Pooling ops multiply certain components of the input shape, e.g. by multiplying input.shape[1] * input.shape[2] * input.shape[3]. This multiplication could overflow an int64 value if shape[0] was 0 but shape[1], shape[2], and shape[3] were very large, e.g. by passing an input with shape (0, 2**25, 2**25, 2**25).
PiperOrigin-RevId: 404644978
Change-Id: Ic79f89c970357ca2962b1f231449066db9403146 | 4dddb2fd0b01cdd196101afbba6518658a2c9e07 | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/4dddb2fd0b01cdd196101afbba6518658a2c9e07 | 2021-10-20 15:01:56-07:00 |
Fix segfault on OOM in Conv2D.
PiperOrigin-RevId: 404655317
Change-Id: I33588dbd3f5d0fef980e3c908bf5515a9ee09ce7 | e7f497570abb6b4ae5af4970620cd880e4c0c904 | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/e7f497570abb6b4ae5af4970620cd880e4c0c904 | 2021-10-20 15:48:54-07:00 |
Fixing security fixes in boosted trees ops
PiperOrigin-RevId: 405669548
Change-Id: Iae224d240d1779bcc02405c2fff99785644fbd0d | 5c8c9a8bfe750f9743d0c859bae112060b216f5c | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/5c8c9a8bfe750f9743d0c859bae112060b216f5c | 2021-10-26 10:03:28-07:00 |
Prevent `CHECK`-fail when decoding resource handles from proto
In certain scenarios, the proto might contain tensors that have too many elements (overflow). This is a `CHECK`-fail in general, but we should prevent this, given how many CVEs caused by that we have received this year (a large fraction of 200).
PiperOrigin-RevId: 408049766
Change-Id: I2ac20b247aa8ed9110846fbdb7a0a9401f2c168c | 14fea662350e7c26eb5fe1be2ac31704e5682ee6 | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/14fea662350e7c26eb5fe1be2ac31704e5682ee6 | 2021-11-06 10:27:15-07:00 |
Prevent a crash due to heap OOB write in grappler.
PiperOrigin-RevId: 408318417
Change-Id: If095feb8c001e3a8ac4a85b7387b81e8309df47d | 97282c6d0d34476b6ba033f961590b783fa184cd | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/97282c6d0d34476b6ba033f961590b783fa184cd | 2021-11-08 05:54:06-08:00 |
Validate `proto.dtype()` before calling `set_dtype()`.
This prevents a `DCHECK`-fail when the proto contains an invalid dtype for a tensor shape with 0 elements or for an incomplete tensor shape.
PiperOrigin-RevId: 408369083
Change-Id: Ia21a3e3d62a90d642a4561f08f3b543e5ad00c46 | 5b491cd5e41ad63735161cec9c2a568172c8b6a3 | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/5b491cd5e41ad63735161cec9c2a568172c8b6a3 | 2021-11-08 10:03:25-08:00 |
Remove a `DCHECK`-fail, log an error instead.
`DCHECK` in debug mode results in crashes. TensorFlow has had multiple vulnerabilities due to this.
Outside of debug mode, `DCHECK` is a no-op.
A better alternative is to report an error to the log buffer and continue. This should happen both in debug mode and in prod mode.
PiperOrigin-RevId: 408375925
Change-Id: Id5b3e19c73f3fbe0cc4bba26ca44ff9607bb6356 | c2b31ff2d3151acb230edc3f5b1832d2c713a9e0 | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/c2b31ff2d3151acb230edc3f5b1832d2c713a9e0 | 2021-11-08 10:21:15-08:00 |
Properly handle the case where `SpecializeType()` returns an error `Status`.
If the error case in `SpecializeType()` is reached, then we would get a crash when trying to access the value of an errorenous `StatusOr` object
PiperOrigin-RevId: 408380069
Change-Id: If3c3fc876dcf9384d5ec7a4985adc68c23ea7318 | cb164786dc891ea11d3a900e90367c339305dc7b | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/cb164786dc891ea11d3a900e90367c339305dc7b | 2021-11-08 10:35:56-08:00 |
Prevent null dereference read in `SpecializeType()`
For some adversarial protos, the attribute for a key might not exist.
PiperOrigin-RevId: 408382090
Change-Id: Ie7eabe532c9ff280fce5dce1f6cdb93c76c2e040 | 8a513cec4bec15961fbfdedcaa5376522980455c | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/8a513cec4bec15961fbfdedcaa5376522980455c | 2021-11-08 10:49:08-08:00 |
Fix heap OOB read/write due to incorrect indexing.
PiperOrigin-RevId: 408578046
Change-Id: Ifc9ffea49e5890f55fcb2c27568611052c3ddcfa | 0657c83d08845cc434175934c642299de2c0f042 | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/0657c83d08845cc434175934c642299de2c0f042 | 2021-11-09 04:49:27-08:00 |
Prevent integer overflow in `OpLevelCostEstimator::CalculateOutputSize`.
In order to not change the API, we return a negative value in case of overflow. A better fix is to change the API to return a status instead.
PiperOrigin-RevId: 408701427
Change-Id: Idf31e7f0bf18ca824d084fdd355e1f653f145c20 | b9bd6cfd1c50e6807846af9a86f9b83cafc9c8ae | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/b9bd6cfd1c50e6807846af9a86f9b83cafc9c8ae | 2021-11-09 14:11:04-08:00 |
Prevent integer overflow in `OpLevelCostEstimator::CalculateTensorSize`.
In order to not change the API, we return a negative value in case of overflow. A better fix is to change the API to return a status instead.
PiperOrigin-RevId: 408713061
Change-Id: I3771475b0c72a2844a3854086966562fd33f2da5 | fcd18ce3101f245b083b30655c27b239dc72221e | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/fcd18ce3101f245b083b30655c27b239dc72221e | 2021-11-09 15:00:15-08:00 |
Fix memory leak when a graph node is invalid.
If a graph node is invalid but a kernel is created then we set the kernel back to `nullptr` but we forget to delete it. Hence, we get a memory leak.
PiperOrigin-RevId: 408968108
Change-Id: I1d8a9d0d8988ed5e08be8b9f2004ce1b4cd11b7c | c79ccba517dbb1a0ccb9b01ee3bd2a63748b60dd | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/c79ccba517dbb1a0ccb9b01ee3bd2a63748b60dd | 2021-11-10 14:03:13-08:00 |
Fix abort caused by allocating a too large vector.
We need to make sure that the number of dimensions in a shape is within limits.
PiperOrigin-RevId: 408997911
Change-Id: If59e1c23f2ec9c2d4ff4d8632fd62b2a7773a4eb | 1361fb7e29449629e1df94d44e0427ebec8c83c7 | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/1361fb7e29449629e1df94d44e0427ebec8c83c7 | 2021-11-10 15:59:39-08:00 |
Make `IsSimplifiableReshape` return `Status` instead of `bool`.
This is to allow remove `CHECK`-fails in subsequent commits.
PiperOrigin-RevId: 409160987
Change-Id: I3f050218a3832271395c4372a0b8ea05f1c03d80 | ebc1a2ffe5a7573d905e99bd0ee3568ee07c12c1 | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/ebc1a2ffe5a7573d905e99bd0ee3568ee07c12c1 | 2021-11-11 09:04:14-08:00 |
Remove `CHECK`-fails from `IsSimplifiableReshape`
PiperOrigin-RevId: 409164987
Change-Id: I58c7dd459ff348c3dbae95e00c4c5e63b30a4e65 | 1fb27733f943295d874417630edd3b38b34ce082 | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/1fb27733f943295d874417630edd3b38b34ce082 | 2021-11-11 09:21:05-08:00 |
Eliminate `CHECK`-fails from `IsSimplifiableReshape` via `MakeShape(<invalid shape>)`
PiperOrigin-RevId: 409166738
Change-Id: I7f0a3590b8acae3f3e3e2fe636e1f5ef285693cf | 240655511cd3e701155f944a972db71b6c0b1bb6 | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/240655511cd3e701155f944a972db71b6c0b1bb6 | 2021-11-11 09:29:48-08:00 |
Prevent a null-pointer dereference / `CHECK`-fail in grappler.
PiperOrigin-RevId: 409187354
Change-Id: I369c249cca32e6c56ec193f0ebbf2f2768fc7d43 | 92dba16749fae36c246bec3f9ba474d9ddeb7662 | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/92dba16749fae36c246bec3f9ba474d9ddeb7662 | 2021-11-11 10:49:11-08:00 |
Use `PartialTensorShape` instead of `TensorShape`.
`TensorShape` constructor throws a CHECK-fail if shape is partial/overflows which the other doesn't. We are only determining the number of elements in the shape and partial shape should be used as it returns negative number when needed.
PiperOrigin-RevId: 409205384
Change-Id: Ia56542ff9ec758f2c9ffc7e4dcc9fa7eecd86e7b | c2426bba00a01de6913738df8fa78e0215fcce02 | tensorflow | cvefixes | 1 | https://github.com/tensorflow/tensorflow | https://github.com/tensorflow/tensorflow/commit/c2426bba00a01de6913738df8fa78e0215fcce02 | 2021-11-11 12:19:44-08:00 |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.