Awesome Privacy

Recent Commits

• maade93791 (09 Sept 24)

  android: use more basic CPU target for memtag This is required for hardened_malloc to work in microdroid on MTE-enabled devices (currently, 8th and 9th generation Pixels) since PVMFW only supports ARMv8 cores. https://android.googlesource.com/platform/packages/modules/Virtualization/+/refs/tags/android-15.0.0_r1/pvmfw/platform.dts#100

• Daniel Micay (05 Aug 24)

  update libdivide to 5.1

• Daniel Micay (15 Feb 24)

  update copyright notice

• Dmitry Muhomor (23 Jan 24)

  mte: use tag 0 for freed slots, stop reserving tag 15

• Dmitry Muhomor (23 Jan 24)

  amend memory tagging README section Memory tagging is enabled by default in bionic, but can be disabled at any point. Memory tagging can't be re-enabled after it's disabled.

• Dmitry Muhomor (23 Jan 24)

  memtag_test: add test for MADV_DONTNEED behavior

• Dmitry Muhomor (23 Jan 24)

  mte: remove util.h dependency from arm_mte.h It's needed for including arm_mte.h into memtag_test.cc

• Dmitry Muhomor (23 Jan 24)

  memtag_test: move SEGV code checks to device-side binary

• Daniel Micay (03 Jan 24)

  remove trailing whitespace

• Julien Voisin (03 Jan 24)

  Improve a bit the formulation of the MTE documentation

• Daniel Micay (03 Jan 24)

  update memory tagging documentation

• Dmitry Muhomor (31 Dec 23)

  android: restore the default SIGABRT handler in fatal_error() async_safe_fatal() calls abort() at the end, which can be intercepted by a custom SIGABRT handler. In particular, crashlytics installs such a handler and tries to fork() after catching SIGABRT. hardened_malloc uses pthread_atfork() to register fork handlers. These handlers try to lock internal hardened_malloc mutexes. If at least one of those mutexes is already locked, which is usually the case, thread that called fatai_error() gets deadlocked, while the other threads (if there are any) continue to run.

• Christian Göttsche (14 Dec 23)

  README: add note about AppArmor constraint on Debian

• jvoisin (16 Nov 23)

  Clarify a bit why a particular magic number was chosen

• Daniel Micay (14 Nov 23)

  Revert "use safe_flag for -fstack-clash-protection" This reverts commit 4171bd164e2ec4cf2546daa2b0f6f95af0d782df.

• Daniel Micay (08 Nov 23)

  use safe_flag for -fstack-clash-protection

• jvoisin (05 Nov 23)

  Run the testsuite on multiple compiler versions

• Dmitry Muhomor (01 Nov 23)

  memtag_test: fix sporadic failures of overflow/underflow tests

• Daniel Micay (30 Oct 23)

  update README now that MTE is implemented

• Dmitry Muhomor (30 Oct 23)

  mte: add scudo to CREDITS file

• Dmitry Muhomor (30 Oct 23)

  mte: add untag_pointer() variant for const pointers

• Dmitry Muhomor (30 Oct 23)

  mte: update comment about skipped tag array update in deallocate_small()

• Dmitry Muhomor (30 Oct 23)

  remove an always-true sizeof(u8) assert

• Dmitry Muhomor (30 Oct 23)

  mte: refactor tag_and_clear_slab_slot() Explicitly call is_memtag_enabled() before calling tag_and_clear_slab_slot() to make it clearer that memory is not zeroed when MTE is disabled.

• Dmitry Muhomor (30 Oct 23)

  mte: note why 0 tag is excluded

• Dmitry Muhomor (30 Oct 23)

  mte: note alignment requirements of arm_mte_tag_and_clear_mem()

• Dmitry Muhomor (30 Oct 23)

  mte: rename arm_mte_store_tags_and_clear() to arm_mte_tag_and_clear_mem()

• Dmitry Muhomor (29 Oct 23)

  mte: add comment about the reserved slab canary value

• Dmitry Muhomor (29 Oct 23)

  memtag_test: improve capturing of test results Using debuggerd + logcat parsing is unreliable and slow, print SEGV signal code to stderr instead.

• Dmitry Muhomor (29 Oct 23)

  memtag_test: improve tag_distinctness test - check that tag distinctess checks are actually reached (it was previously verified manually by looking at the now-removed printf output) - check that only non-reserved tags are used - check that all of non-reserved tags are used - print tag usage statistics at the end of run