TensorFlow is an end-to-end open source platform for machine learning. In affected versions the implementation for `tf.raw_ops.FractionalAvgPoolGrad` can be tricked into accessing data outside of bounds of heap allocated buffers. The [implementation](https://github.com/tensorflow/tensorflow/blob/f24faa153ad31a4b51578f8181d3aaab77a1ddeb/tensorflow/core/kernels/fractional_avg_pool_op.cc#L205) does not validate that the input tensor is non-empty. Thus, code constructs an empty `EigenDoubleMatrixMap` and then accesses this buffer with indices that are outside of the empty area. We have patched the issue in GitHub commit 0f931751fb20f565c4e94aa6df58d54a003cdb30. The fix will be included in TensorFlow 2.6.0. We will also cherrypick this commit on TensorFlow 2.5.1, TensorFlow 2.4.3, and TensorFlow 2.3.4, as these are also affected and still in supported range.

In getConfig of SoftVideoDecoderOMXComponent.cpp, there is a possible out of bounds write due to a missing validation check. This could lead to a local non-security issue with no additional execution privileges needed. User interaction is not needed for exploitation.

In pl330_dma_from_peri_start() of fp_spi_dma.c, there is a possible out of bounds write due to a missing bounds check. This could lead to local escalation of privilege with no additional execution privileges needed. User interaction is not needed for exploitation.

An issue was discovered on Samsung mobile devices with L(5.0/5.1) and M(6.0) (tablets) software. The lockscreen interface allows Add User actions, leading to an unintended ability to access user data in external storage. The Samsung ID is SVE-2016-7797 (March 2017).

An issue was discovered on LG mobile devices with Android OS 4.4 through 11 software. Attackers can leverage ISMS services to bypass access control on specific content providers. The LG ID is LVE-SMP-210003 (April 2021).

Improper access control of a component in CallBGProvider prior to SMR JUN-2021 Release 1 allows local attackers to access arbitrary files with an escalated privilege.

An improperly initialized 'migrationAuth' value in Google's go-tpm TPM1.2 library versions prior to 0.3.0 can lead an eavesdropping attacker to discover the auth value for a key created with CreateWrapKey. An attacker listening in on the channel can collect both 'encUsageAuth' and 'encMigrationAuth', and then can calculate 'usageAuth ^ encMigrationAuth' as the 'migrationAuth' can be guessed for all keys created with CreateWrapKey. TPM2.0 is not impacted by this. We recommend updating your library to 0.3.0 or later, or, if you cannot update, to call CreateWrapKey with a random 20-byte value for 'migrationAuth'.

In multiple files, there is a possible way to capture the device screen when disallowed by device policy due to a logic error in the code. This could lead to local escalation of privilege with no additional execution privileges needed. User interaction is not needed for exploitation.

An issue was discovered on Samsung mobile devices with O(8.0) and P(9.0) (Exynos8890 chipsets) software. A use-after-free occurs in the MALI GPU driver. The Samsung ID is SVE-2019-13921-1 (May 2019).

Use of Java's default temporary directory for file creation in `FileBackedOutputStream` in Google Guava versions 1.0 to 31.1 on Unix systems and Android Ice Cream Sandwich allows other users and apps on the machine with access to the default Java temporary directory to be able to access the files created by the class. Even though the security vulnerability is fixed in version 32.0.0, we recommend using version 32.0.1 as version 32.0.0 breaks some functionality under Windows.

TensorFlow is an end-to-end open source platform for machine learning. In affected versions it is possible to nest a `tf.map_fn` within another `tf.map_fn` call. However, if the input tensor is a `RaggedTensor` and there is no function signature provided, code assumes the output is a fully specified tensor and fills output buffer with uninitialized contents from the heap. The `t` and `z` outputs should be identical, however this is not the case. The last row of `t` contains data from the heap which can be used to leak other memory information. The bug lies in the conversion from a `Variant` tensor to a `RaggedTensor`. The [implementation](https://github.com/tensorflow/tensorflow/blob/460e000de3a83278fb00b61a16d161b1964f15f4/tensorflow/core/kernels/ragged_tensor_from_variant_op.cc#L177-L190) does not check that all inner shapes match and this results in the additional dimensions. The same implementation can result in data loss, if input tensor is tweaked. We have patched the issue in GitHub commit 4e2565483d0ffcadc719bd44893fb7f609bb5f12. The fix will be included in TensorFlow 2.6.0. We will also cherrypick this commit on TensorFlow 2.5.1, TensorFlow 2.4.3, and TensorFlow 2.3.4, as these are also affected and still in supported range.

Improper caller check vulnerability in Knox Core prior to SMR MAY-2021 Release 1 allows attackers to install arbitrary app.

Improper access control vulnerability in Samsung Dex for PC prior to SMR Aug-2022 Release 1 allows local attackers to scan and connect to PC by unprotected binder call.

A vulnerability using PendingIntent in Bixby Routines prior to version 3.1.21.8 in Android R(11.0) and 2.6.30.5 in Android Q(10.0) allows attackers to execute privileged action by hijacking and modifying the intent.

In query of DownloadManager.java, there is a possible read/write of arbitrary files due to a permissions bypass. This could lead to local information disclosure and file rewriting with no additional execution privileges needed. User interaction is not needed for exploitation.

PendingIntent hijacking vulnerability in KnoxPrivacyNoticeReceiver prior to SMR Feb-2022 Release 1 allows local attackers to access media files without permission via implicit Intent.

A vulnerability using PendingIntent in Reminder prior to version 12.2.05.0 in Android R(11.0) and 12.3.02.1000 in Android S(12.0) allows attackers to execute privileged action by hijacking and modifying the intent.

Tensorflow is an Open Source Machine Learning Framework. In multiple places, TensorFlow uses `tempfile.mktemp` to create temporary files. While this is acceptable in testing, in utilities and libraries it is dangerous as a different process can create the file between the check for the filename in `mktemp` and the actual creation of the file by a subsequent operation (a TOC/TOU type of weakness). In several instances, TensorFlow was supposed to actually create a temporary directory instead of a file. This logic bug is hidden away by the `mktemp` function usage. We have patched the issue in several commits, replacing `mktemp` with the safer `mkstemp`/`mkdtemp` functions, according to the usage pattern. Users are advised to upgrade as soon as possible.

Improper sanitization of incoming intent in Dressroom prior to SMR Jan-2022 Release 1 allows local attackers to read and write arbitrary files without permission.

Unprotected dynamic receiver in Telecom prior to SMR Feb-2022 Release 1 allows untrusted applications to launch arbitrary activity.

In set_secure_reg of sac_handler.c, there is a possible out of bounds read due to a missing bounds check. This could lead to local information disclosure of 4 bytes of stack memory with no additional execution privileges needed. User interaction is not needed for exploitation.

Heap buffer overflow in Tab Groups in Google Chrome prior to 126.0.6478.54 allowed a remote attacker who convinced a user to engage in specific UI gestures to potentially exploit heap corruption via a crafted HTML page. (Chromium security severity: High)

Type Confusion in V8 in Google Chrome prior to 126.0.6478.54 allowed a remote attacker to perform an out of bounds memory write via a crafted HTML page. (Chromium security severity: High)

Out of bounds access in SwiftShader in Google Chrome prior to 73.0.3683.75 allowed a remote attacker to potentially exploit heap corruption via a crafted HTML page.

Out of bounds write in Streams API in Google Chrome prior to 125.0.6422.141 allowed a remote attacker to execute arbitrary code inside a sandbox via a crafted HTML page. (Chromium security severity: High)

In a camera driver function in all Android releases from CAF using the Linux kernel, a bounds check is missing when writing into an array potentially leading to an out-of-bounds heap write.

In Exynos_parsing_user_data_registered_itu_t_t35 of VendorVideoAPI.cpp, there is a possible out of bounds write due to an incorrect bounds check. This could lead to local escalation of privilege with no additional execution privileges needed. User interaction is not needed for exploitation.

In prepare_response_locked of lwis_transaction.c, there is a possible out of bounds write due to improper input validation. This could lead to local escalation of privilege with no additional execution privileges needed. User interaction is not needed for exploitation.

Out of bounds access in SwiftShader in Google Chrome prior to 73.0.3683.75 allowed a remote attacker to potentially exploit heap corruption via a crafted HTML page.

Heap buffer overflow in ANGLE in Google Chrome on Windows prior to 74.0.3729.108 allowed a remote attacker to potentially exploit heap corruption via a crafted HTML page.

In all Qualcomm products with Android releases from CAF using the Linux kernel, in a driver function, a value from userspace is not properly validated potentially leading to an out of bounds heap write.

In all Qualcomm products with Android releases from CAF using the Linux kernel, due to a type downcast, a value may improperly pass validation and cause an out of bounds write later.

Integer overflow in PDFium in Google Chrome prior to 74.0.3729.108 allowed a remote attacker to potentially exploit heap corruption via a crafted PDF file.

Integer overflow in PDFium in Google Chrome prior to 76.0.3809.87 allowed a remote attacker to potentially exploit heap corruption via a crafted PDF file.

In Android before the 2018-06-05 security patch level, NVIDIA TLZ TrustZone contains a possible out of bounds write due to integer overflow which could lead to local escalation of privilege in the TrustZone with no additional execution privileges needed. User interaction is not needed for exploitation. This issue is rated as high. Version: N/A. Android: A-69480285. Reference: N-CVE-2017-6292.

Insufficient checks of pointer validity in WebRTC in Google Chrome prior to 72.0.3626.81 allowed a remote attacker to potentially exploit heap corruption via a crafted HTML page.

Integer overflow in ANGLE in Google Chrome on Windows prior to 74.0.3729.108 allowed a remote attacker to potentially exploit heap corruption via a crafted HTML page.

In Android before the 2018-06-05 security patch level, NVIDIA Tegra X1 TZ contains a possible out of bounds write due to missing bounds check which could lead to escalation of privilege from the kernel to the TZ. User interaction is not needed for exploitation. This issue is rated as high. Version: N/A. Android: A-69316825. Reference: N-CVE-2017-6294.

NVIDIA libnvmmlite_audio.so contains an elevation of privilege vulnerability when running in media server which may cause an out of bounds write and could lead to local code execution in a privileged process. This issue is rated as high. Product: Android. Version: N/A. Android: A-65023166. Reference: N-CVE-2017-6279.

Integer overflow in PDFium in Google Chrome prior to 76.0.3809.87 allowed a remote attacker to potentially exploit heap corruption via a crafted PDF file.

Out of bounds memory access in JavaScript in Google Chrome prior to 77.0.3865.75 allowed a remote attacker to potentially exploit heap corruption via a crafted HTML page.

NVIDIA libnvomx contains a possible out of bounds write due to a missing bounds check which could lead to local escalation of privilege. This issue is rated as high. Product: Android. Version: N/A. Android: A-64893247. Reference: N-CVE-2017-6286.

Heap buffer overflow in Dawn in Google Chrome prior to 125.0.6422.76 allowed a remote attacker to perform an out of bounds memory write via a crafted HTML page. (Chromium security severity: High)

Use after free in V8 in Google Chrome prior to 74.0.3729.108 allowed a remote attacker to potentially exploit heap corruption via a crafted HTML page.

In Android before the 2018-05-05 security patch level, NVIDIA Tegra X1 TZ contains a vulnerability in Widevine TA where the software writes data past the end, or before the beginning, of the intended buffer, which may lead to escalation of Privileges. This issue is rated as high. Android: A-69377364. Reference: N-CVE-2017-6293.

Heap buffer overflow in V8 Internationalization in Google Chrome prior to 101.0.4951.64 allowed a remote attacker to potentially exploit heap corruption via a crafted HTML page.

Off-by-one read/write on the heap in Blink in Google Chrome prior to 62.0.3202.62 allowed a remote attacker to corrupt memory and possibly leak information and potentially execute code via a crafted PDF file.

In writeInplace of Parcel.cpp, there is a possible out of bounds write. This could lead to local escalation of privilege with no additional execution privileges needed. User interaction is not needed for exploitation.

Stack overflow in PDFium in Google Chrome prior to 60.0.3112.78 for Linux, Windows, and Mac allowed a remote attacker to potentially exploit stack corruption via a crafted PDF file.

Heap buffer overflow in filter processing in Skia in Google Chrome prior to 57.0.2987.98 for Mac, Windows, and Linux and 57.0.2987.108 for Android allowed a remote attacker to perform an out of bounds memory read via a crafted HTML page.