Commit graph

1386 commits

Author SHA1 Message Date
Rupert Shuttleworth
344b8da46a Mark sh_binary target as being bp2build_available.
Test: build/bazel/scripts/bp2build-sync.sh write.
Change-Id: I8e8c3c95aa2c7ff6d7db3d92291f224bdf7cf255
2021-02-09 11:53:36 +00:00
Ryan Prichard
ef1478765d Fix -Wl,--exclude-libs typo: x86->i686
These lines shouldn't actually matter, because the DSOs are using
version scripts to allow-list exported symbols.

Bug: none
Test: bionic unit tests
Change-Id: I39d3df8c4f8053624f862b3c6994e30c693e928c
2021-01-28 15:41:47 -08:00
Ryan Prichard
058eb8fa4e Ensure same order of global group members in all NS's
During "step 1" of find_libraries, the linker finds the transitive
closure of dependencies, in BFS order. As it finds each library, it
adds the library to its primary namespace (so that, if some other
library also depends on it, find_loaded_library_by_soname can find the
library in the process of being loaded).

LD_PRELOAD libraries are automatically marked DF_1_GLOBAL, and any
DF_1_GLOBAL library is added to every linker namespace. Previously,
this secondary namespace registration happened after step 1. The result
is that across different namespaces, the order of libraries could vary.
In general, a namespace's primary members will all appear before
secondary members. This is undesirable for libsigchain.so, which we
want to have appear before any other non-preloaded library.

Instead, when an soinfo is added to its primary namespace, immediately
add it to all the other namespaces, too. This ensures that the order of
soinfo objects is the same across namespaces.

Expand the dl.exec_with_ld_config_file_with_ld_preload and
dl.exec_with_ld_config_file tests to cover the new behavior. Mark
lib1.so DF_1_GLOBAL and use a "foo" symbol to mimic the behavior of a
signal API interposed by (e.g.) libsigchain.so and a ASAN preload.

Test: bionic unit tests
Bug: http://b/143219447
Change-Id: I9fd90f6f0d14caf1aca6d414b3e9aab77deca3ff
2021-01-22 03:27:43 -08:00
Ryan Prichard
fcd9c78534 Merge "Switch libc.so and linker to prebuilt LLVM libunwind" 2021-01-20 00:10:58 +00:00
Elliott Hughes
01be44d2f8 Inline call_array for clearer stack traces.
No-one seems to understand that a crash in a random .so from call_array()
in the linker isn't a linker bug. They _seem_ to understand (or at least
claim to) when we explain that this is just the linker calling their ELF
constructors --- despite the fact that the caller of call_array() is
call_constructors().

One experiment we can try though is to inline call_array() to elide that
frame from the crash dumps. I do also wonder whether renaming
call_constructors() to call_elf_constructors() would help/hinder/make no
difference. For now I'm leaning toward "hinder" because I suspect most
people don't understand "ELF constructor" and C++ folks at least will
probably be influenced in a not wholly incorrect direction when they
hear "constructor" (whereas "ELF constructor" might mislead them back in
the direction of "strange linker magic, not my fault" again)...

(The reformatting is clang-format's decision, not mine.)

Test: treehugger
Change-Id: I65ab95ceb2e988fd053c48c66f51afba17ccfa61
2021-01-19 09:47:50 -08:00
Ryan Prichard
cdf7175fa5 Switch libc.so and linker to prebuilt LLVM libunwind
For libc.so, use a special build of libunwind.a whose symbols aren't
hidden ("libunwind-exported"), because libc.so exports the _Unwind_*
APIs.

Bug: http://b/153025717
Test: bionic unit tests
Change-Id: I7435e076ec8cc4410e3e6086d3cf5d2c6011c80c
2021-01-14 17:35:05 -08:00
Ryan Prichard
bb1e37358f Delay setting linker soname until post-reloc and post-ctor
Setting the linker's soname ("ld-android.so") can allocate heap memory
now that the name uses an std::string, and it's probably a good idea to
defer doing this until after the linker has relocated itself (and after
it has called C++ constructors for global variables.)

Bug: none
Test: bionic unit tests
Test: verify that dlopen("ld-android.so", RTLD_NOLOAD) works
Change-Id: I6b9bd7552c3ae9b77e3ee9e2a98b069b8eef25ca
2021-01-13 17:48:05 -08:00
Elliott Hughes
f9dd1a760a Store soname as a std::string.
Once upon a time (and, indeed, to this very day if you're on LP32) the
soinfo struct used a fixed-length buffer for the soname. This caused
some issues, mainly with app developers who accidentally included a full
Windows "C:\My Computer\...\libfoo.so" style path. To avoid all this we
switched to just pointing into the ELF file itself, where the DT_SONAME
is already stored as a NUL-terminated string. And all was well for many
years.

Now though, we've seen a bunch of slow startup traces from dogfood where
`dlopen("libnativebridge.so")` in a cold start takes 125-200ms on a recent
device, despite no IO contention. Even though libnativebridge.so is only
20KiB.

Measurement showed that every library whose soname we check required
pulling in a whole page just for the (usually) very short string. Worse,
there's readahead. In one trace we saw 18 pages of libhwui.so pulled
in just for `"libhwui.so\0"`. In fact, there were 3306 pages (~13MiB)
added to the page cache during `dlopen("libnativebridge.so")`. 13MiB for
a 20KiB shared library!

This is the obvious change to use a std::string to copy the sonames
instead. This will dirty slightly more memory, but massively improve
locality.

Testing with the same pathological setup took `dlopen("libnativebridge.so")`
down from 192ms to 819us.

Bug: http://b/177102905
Test: tested with a pathologically modified kernel
Change-Id: I33837f4706adc25f93c6fa6013e8ba970911dfb9
2021-01-11 09:57:46 -08:00
Evgenii Stepanov
8564b8d9e6 Use ELF notes to set the desired memory tagging level.
Use a note in executables to specify
(none|sync|async) heap tagging level. To be extended with (heap x stack x
globals) in the future. A missing note disables all tagging.

Bug: b/135772972
Test: bionic-unit-tests (in a future change)

Change-Id: Iab145a922c7abe24cdce17323f9e0c1063cc1321
2021-01-06 16:08:18 -08:00
Yi Kong
6f6daaae9d linker: Enable PGO
linker-reloc-bench shows 1.2% improvement on flame.

                        PGO         No PGO
BM_linker_relocation/
real_time_mean          52215 us    52916 us

Test: linker-reloc-bench --benchmark_repetitions=25
Bug: 175205560
Change-Id: I6abbbd51912f86fcf8112c024d1cf2519056bea7
2020-12-09 19:03:53 +00:00
Elliott Hughes
658b938700 linker: add ld.config.txt parse time to systrace.
Bug: http://b/172640358
Test: treehugger
Change-Id: I9d6b7c1e200d2effb737508d3c282063edefa888
2020-11-06 09:28:16 -08:00
Yifan Hong
02bee7182e Merge "Make vendor_ramdisk_available." 2020-10-29 17:17:29 +00:00
Yifan Hong
b04490d7e5 Make vendor_ramdisk_available.
Test: pass
Bug: 156098440
Change-Id: I8fe2deb8fa3b24c227202be02d8af648629cffeb
2020-10-27 17:06:45 -07:00
Yi Kong
a7e363f873 No global ThinLTO for linker
Previous CL aosp/1432874 missed one project to exclude.

Test: make
Change-Id: Ib39e6a86c410e35141cba17948a0f30e197ea9a3
2020-10-01 04:10:01 +08:00
Yi Kong
15a05a77b4 No global ThinLTO for bionic
There are multiple build breakages with bionic when we enable ThinLTO
globally. Opt bionic out of ThinLTO for now.

#global-thinlto-opt-out

Test: TreeHugger
Bug: 169004486
Change-Id: I546a8074f9c3e0ddbd01d3b7cd730e215e3c0c49
2020-09-29 17:24:04 +00:00
Tamas Petz
8d55d1872a Enable BTI in bionic linker
This patch adds support to load BTI-enabled objects.

According to the ABI, BTI is recorded in the .note.gnu.property section.
The new parser evaluates the property section, if exists.
It searches for .note section with NT_GNU_PROPERTY_TYPE_0.
Once found it tries to find GNU_PROPERTY_AARCH64_FEATURE_1_AND.
The results are cached.

The main change in linker is when protection of loaded ranges gets
applied. When BTI is requested and the platform also supports it
the prot flags have to be amended with PROT_BTI for executable ranges.
Failing to add PROT_BTI flag would disable BTI protection.
Moreover, adding the new PROT flag for shared objects without BTI
compatibility would break applications.

Kernel does not add PROT_BTI to a loaded ELF which has interpreter.
Linker handles this case too.

Test: 1. Flame boots
      2. Tested on FVP with BTI enabled

Change-Id: Iafdf223b74c6e75d9f17ca90500e6fe42c4c1218
2020-09-23 17:53:28 -07:00
Jiyong Park
3b47d603fa Add arm64 source to linker_wrapper
As we enable arm64-based Bionic host target (linux_bionic_arm64),
linker_wrapper is added with the corresponding source.

Bug: 159685774
Test: HOST_CROSS_OS=linux_bionic HOST_CROSS_ARCH=arm64 m
Test: copy out/soong/host/linux_bionic_arm64/ to an ARM64 emulator
running Linux and execute the binaries

Change-Id: I4f367a349f7e0015318352cb7f2870fc856eab05
2020-09-22 11:28:22 +09:00
Vy Nguyen
d500751443 Add a thread-properties API
(Based on proposal at https://sourceware.org/glibc/wiki/ThreadPropertiesAPI)

This includes API to:
 - locate static and dynamic TLS
 - register thread-exit and  dynamic TLS creation/destruction callbacks

Change-Id: Icd9d29a5b2f47495395645e19d3b2c96826f19c8
2020-08-11 16:51:43 +00:00
Ryan Prichard
aff9a34bd8 Rename linker greylist to exempt-list
Update a comment in android-changes-for-ndk-developers.md about the
removed debug.ld.greylist_disabled system property.

Update language to comply with Android's inclusive language guidance

  #inclusivefixit

See https://source.android.com/setup/contribute/respectful-code for reference

Bug: http://b/162536543
Test: bionic-unit-tests
Change-Id: I760ee14bce14d9d799926c43d2c14fd8ffbc6968
2020-08-03 16:17:00 -07:00
Treehugger Robot
0f4bdb04fb Merge "linker: Cleanup for Android's inclusive language guidance" 2020-08-02 14:48:01 +00:00
Luke Huang
30f2f053f2 linker: Cleanup for Android's inclusive language guidance
1. Cleanup for #inclusivefixit. (whitelisted -> allowed_libs)
2. Support the old term for backwards compatibility. (Also update test.)
3. Fix the formatting errors found by clang-format.

See https://source.android.com/setup/contribute/respectful-code
for reference.

Bug: 161896447
Test: atest linker-unit-tests linker-benchmarks
Change-Id: I19dbed27a6d874ac0049cb7b67d2cb0f75369c1b
2020-07-31 11:37:28 +08:00
Ryan Prichard
238609328f Remove debug.ld.greylist_disabled property
This property provided a way to disable the greylist, for testing
whether an app targeting < 24 still works. Instead of turning off the
greylist, though, an app developer should simply target a newer API.

(If app developers really need this property for testing, they can
still use it on versions of Android between N and R, inclusive.)

Update language to comply with Android's inclusive language guidance

See https://source.android.com/setup/contribute/respectful-code for reference

#inclusivefixit

Bug: http://b/162536543
Test: bionic-unit-tests
Change-Id: Id1eb2807fbb7436dc9ed7fe47e15b7d165a26789
2020-07-30 19:29:17 -07:00
Elliott Hughes
68ae6ad12e Changes for #inclusivefixit.
Test: treehugger
Change-Id: I7ff0496c5c2792a41781e74634247f55b0548213
2020-07-21 16:34:58 -07:00
Evgenii Stepanov
e0848bbf89 Pagetable-friendly shared library address randomization.
Add inaccessible gaps between shared libraries to make it harder for the
attackers to defeat ASLR by random probing.

To avoid excessive page table bloat, only do this when a library is
about to cross a huge page boundary, effectively allowing several
smaller libraries to be lumped together.

Bug: 158113540
Test: look at /proc/$$/maps
Change-Id: I39c0100b81f72447e8b3c6faafa561111492bf8c
2020-07-16 13:57:20 -07:00
Evgenii Stepanov
b5f301eaaf Merge "Revert "Add randomized padding around shared library mappings."" 2020-07-06 22:18:45 +00:00
Evgenii Stepanov
474f2f5c8b Revert "Add randomized padding around shared library mappings."
This reverts commit a8cf3fef2a.

Reason for revert: memory regression due to the fragmentation of the page tables
Bug: 159810641
Bug: 158113540

Change-Id: I6212c623ff440c7f6889f0a1e82cf7a96200a411
2020-07-06 19:27:55 +00:00
Kiyoung Kim
46e731caf5 Disable warning message for missing linker config in some cases
There are some special cases - such as init process - when linker
configuration is not expected to exist. This change disables warning
message that generated linker configuration does not exist in those
cases.

Bug: 158800902
Test: Tested from cuttlefish that warning message is not generated from
init

Change-Id: Ie2fbb5210175cf1e6f2b7e638f57c3b74d395368
2020-06-30 09:21:07 +09:00
Evgenii Stepanov
a8cf3fef2a Add randomized padding around shared library mappings.
Improve ASLR by increasing the randomly sized gaps between shared
library mappings, and keep them mapped PROT_NONE.

Bug: 158113540
Test: look at /proc/$$/maps
Change-Id: Ie72c84047fb624fe2ac8b7744b2a2d0d255ea974
2020-06-17 17:28:28 -07:00
Evgenii Stepanov
071416b700 Use PROT_NONE on the unused parts of CFI shadow.
This replaces a single 2Gb readable memory region with a bunch of tiny
regions, and leaves the bulk of 2Gb mapped but unaccessible. This makes
it harder to defeat ASLR by probing for the CFI shadow region.

Sample CFI shadow mapping with this change:
7165151000-716541f000 ---p 00000000 00:00 0                              [anon:cfi shadow]
716541f000-7165420000 r--p 00000000 00:00 0                              [anon:cfi shadow]
7165420000-71654db000 ---p 00000000 00:00 0                              [anon:cfi shadow]
71654db000-71654dc000 r--p 00000000 00:00 0                              [anon:cfi shadow]
71654dc000-71654dd000 r--p 00000000 00:00 0                              [anon:cfi shadow]
71654dd000-71654f0000 ---p 00000000 00:00 0                              [anon:cfi shadow]
71654f0000-71654f1000 r--p 00000000 00:00 0                              [anon:cfi shadow]
71654f1000-71e5151000 ---p 00000000 00:00 0                              [anon:cfi shadow]

This change degrades CFI diagnostics for wild jumps and casts (i.e. when
the target of a CFI check is outside of any known library bounds). This
is acceptable, because CFI does not have much to tell about those cases
anyway. Such bugs will show up as SEGV_ACCERR crashes inside
__cfi_slowpath in libdl.so from now on.

Bug: 158113540
Test: bionic-unit-tests/cfi_test.*
Test: adb shell cat /proc/$PID/maps | grep cfi

Change-Id: I57cbd0d3f87eb1610ad99b48d98ffd497ba214b4
Merged-In: I57cbd0d3f87eb1610ad99b48d98ffd497ba214b4
2020-06-12 12:22:32 -07:00
Elliott Hughes
3205cddff0 linker: CHECK() or async_safe_fatal() rather than abort().
In particular, add the strerror() output if mprotect() fails.

Fix the CHECK macro so that you can make assertions involving operator%
without that being confused for a printf format specifier.

Bug: https://issuetracker.google.com/158645318
Test: treehugger
Change-Id: I6817f8ca5f094c52dc2c9067bfac90385a8743f5
2020-06-10 14:49:28 -07:00
Jooyung Han
61a9a4016c Make .whitelisted property "append"-able
.whitelisted is a list of sonames seperated by ":" just like
.shared_libs and .paths properties. It makes sense to make it appendable
using += as well ass .shared_libs and .paths.

Bug: n/a
Test: atest linker-unit-tests
Change-Id: If61cc553c8080e8a58de8a3a6051c1853f7bfe5f
2020-06-02 16:05:01 +09:00
Victor Khimenko
9b3e026c8f Refactor translateSystemPathToApexPath
This is more efficient and easier to read.

Test: bionic-unit-tests --gtest_filter=*icu*

Change-Id: Iddeed7cd4a1d48d8968f97951a9af004ccce52e8
2020-05-19 02:32:07 +02:00
Victor Khimenko
01790fdc69 Allow native_bridge linker to use different APEX from native one
Bug: http://b/156397945

Test: dlfcn.dlopen_system_libicuuc_android_api_level_28 with native_bridge

Change-Id: I3ecd929a0c31dda8cd424795f21f6ef472776557
2020-05-15 20:56:22 +00:00
Victor Khimenko
d15229df76 Copy translateSystemPathToApexPath to linker_translate_path.cpp
Bug: http://b/156397945

Test: m

Change-Id: I33f11b8a6a41475b6cdd65e1ff49b2084c33b197
2020-05-14 22:17:45 +02:00
vichang
10bad3e882 Merge "Move libicuuc.so into com.android.i18n module." 2020-05-11 11:37:06 +00:00
Victor Chang
f248d2de84 Move libicuuc.so into com.android.i18n module.
Change the location set in the linker

Bug: 130219528
Bug: 138994281
Test: atest CtsBionicTestCases
Test: atest CtsJniTestCases
Change-Id: I215a8e023ccc4d5ffdd7df884c809f8d12050c8f
2020-05-07 23:15:01 +00:00
Peter Collingbourne
144a6ab341 Merge "Introduce a new heap tagging level, M_HEAP_TAGGING_LEVEL_SYNC." 2020-05-07 17:51:51 +00:00
Peter Collingbourne
d3060019de Introduce a new heap tagging level, M_HEAP_TAGGING_LEVEL_SYNC.
The SYNC tagging level enables stack trace collection for allocations and
deallocations, which allows allocation and deallocation stack traces to
appear in tombstones when encountering a tag check fault in synchronous tag
checking mode.

Bug: 135772972
Change-Id: Ibda9f51b29d2c8e2c993fc74425dea7bfa23ab1e
2020-05-05 13:28:16 -07:00
Martin Stjernholm
82d84bcd7e Make Bionic and kernel system includes explicit in the blueprints.
Before this Soong added hardcoded system include paths to
bionic/libc/{include,kernel}, which won't work when Bionic libs are
packaged up as prebuilts in an SDK module snapshot.

Test: Build and boot
Test: Check in out/verbose.log.gz that a C file doesn't get any bionic
  include path for linux_glibc.
Bug: 153590472
Change-Id: I13c8eb3dd7150d6e0fee001b290b53fcebebcfea
2020-05-05 13:34:14 +01:00
Ryan Prichard
172611f5ce __loader_cfi_fail: acquire g_dl_mutex
CfiFail calls find_containing_library, which searches the linker's
internal soinfo list, which could be modified by another thread.

Bug: http://b/150372650
Test: bionic-unit-tests
Change-Id: I59024a0a47913caa75702f15ec058b0a360727b5
2020-04-20 18:14:51 -07:00
Ryan Prichard
c2a93792fc android_get_exported_namespace: acquire g_dl_mutex
android_get_exported_namespace searches g_exported_namespaces, which
isn't modified after process initialization, but it does the search
using a new std::string object, and the linker's malloc/free functions
aren't thread-safe. (They're protected by the same lock (g_dl_mutex) as
the rest of the linker's state.)

Bug: http://b/150372650
Test: bionic-unit-tests
Change-Id: Iafd12e5ab36ae61f0642aad59939f528d31bda16
2020-04-20 17:59:43 -07:00
Ryan Prichard
df3dc8039d Merge changes Ia359d9f2,Ibabbf559,I1b542e47,I4b6f48af,I82e442f0
* changes:
  Fix bootstrap linker library searching
  Remove dangling soinfo* from elf_readers_map_
  Fix DL_WARN_documented_change URL
  remove search_linked_namespaces param
  Refactor linker lookup code a bit
2020-03-31 21:02:18 +00:00
Peter Collingbourne
aed48830ac Merge "Switch to debugger_process_info in bionic." 2020-03-30 21:36:41 +00:00
Ryan Prichard
40494406b8 Fix bootstrap linker library searching
For the bootstrap linker, insert /system/${LIB}/bootstrap in front of
/system/${LIB} in any namespace search path.

Bug: http://b/152572170
Test: bionic unit tests
Change-Id: Ia359d9f2063f4b6fff3f79b51b500ba968a18247
2020-03-27 18:46:28 -07:00
Ryan Prichard
4fa6d9978a Remove dangling soinfo* from elf_readers_map_
If ElfReader::Read fails, then it is hazardous to leave the invalid
ElfReader in the soinfo*->ElfReader table, because a future soinfo
object could happen to have the same address, then reuse the invalid
ElfReader. I'm not sure whether this can break anything, because the
linker would call ElfReader::Read on the invalid object and overwrite
its previous value.

Test: bionic unit tests
Bug: none
Change-Id: Ibabbf559443441b9caeacc34ca165feaafe5e3a7
2020-03-27 18:46:28 -07:00
Ryan Prichard
146620b644 Fix DL_WARN_documented_change URL
The doc_link argument is really a URL fragment within the
android-changes-for-ndk-developers.md document, not a filename at the
root of the bionic repository.

Test: manual
Bug: none
Change-Id: I1b542e47aca132ce43ba1d50d83db1bf3c7b10c6
2020-03-27 18:46:28 -07:00
Ryan Prichard
323d7dfe90 remove search_linked_namespaces param
The search_linked_namespaces parameter to find_library_internal is
always true.

Bug: none
Test: bionic tests
Change-Id: I4b6f48afefca4f52b34ca2c9e0f4335fa895ff34
2020-03-27 18:46:28 -07:00
Ryan Prichard
aa2db969dc Refactor linker lookup code a bit
Add a few _Nonnull / _Nullable annotations. Clang may use them to issue
warnings but violating the annotation isn't undefined behavior.

Bug: none
Test: bionic tests
Change-Id: I82e442f06a2c59c69bc0008f0f686785695cdd02
2020-03-27 18:46:28 -07:00
Treehugger Robot
2a069b24df Merge "Revert "Statically link libdl.a to linker"" 2020-03-18 16:05:54 +00:00
Jooyung Han
274e21fdb6 Revert "Statically link libdl.a to linker"
This reverts commit 8dba7fefb0.

Reason for revert: This workaround is not necessary with ag/10700799

Merged-In: I2bf469bfe00f3a70e67085abfb3822db6908d522
Change-Id: If351aedfce23d60582f0c5ff965f94356d1900d9
Bug: 150860940
2020-03-18 08:45:25 +00:00