RTLD_NEXT/DEFAULT lookup should not skip handle lookup
in the case when the load group is RTLD_GLOBAL.
Note that there is a difference between load group is local group.
The local group includes externally referenced libraries where
the load group does not. The external reference in this context is a
DT_NEEDED library that belongs to a previously loaded group.
Bug: http://b/72237367
Test: bionic-unit-tests --gtest_filter=dl*
Change-Id: I8997cc961c13a5396f1756161798b45ed1cab16c
We've been using #pragma once for new internal files, but let's be more bold.
Bug: N/A
Test: builds
Change-Id: I7e2ee2730043bd884f9571cdbd8b524043030c07
Every other architecture already uses an assembly file here.
The previous code aligned ESP incorrectly, but it doesn't really matter
because everything is built with Clang's -mstackrealign, which realigns
ESP in every function prologue.
Bug: http://b/73140672#comment4
Test: lunch aosp_x86-eng; m; emulator; device boots
Test: manual
Change-Id: I921fd7848cdc611b4f8f13d1176d1983ffea952d
This change modifies the encoding used in SHT_RELR sections to a simpler
version that gives better results. This encoding was suggested by Andrew
Grieve and is described in this post on generic-abi@googlegroups.com:
https://groups.google.com/d/msg/generic-abi/bX460iggiKg/Pi9aSwwABgAJ
Bug: None
Test: Built image for marlin, flashed on device, ran arm and
aarch64 binaries containing '.relr.dyn' sections using
the new encoding.
Change-Id: I266affe0fbad91dc375995985a221cb02499447b
This commit allows users to create a link without soname filters between
two linker namespaces.
The motivation is to establish one-way shared library isolation. For
example, assume that there are two linker namespaces `default` and
`vndk`. We would like to limit the shared libraries that can be used by
the `default` namespace. In the meanwhile, we would like to allow the
`vndk` namespace to use shared libs from the `default` namespace if the
soname cannot be find in the search path or loaded sonames of the `vndk`
namespace.
shared_libs = %VNDK_CORE_LIBRARIES%
shared_libs += %VNDK_SAMEPROCESS_LIBRARIES%
vndk <-------------------------------------------- default
\_______________________________________________/^
allow_all_shared_libs = true
android_link_namespaces_all_libs() is added to libdl, but it is
versioned as LIBC_PRIVATE. android_link_namespaces_all_libs() is only
for unit tests.
Bug: 69824336
Test: adb shell /data/nativetest/linker-unit-tests/linker-unit-tests32
Test: adb shell /data/nativetest64/linker-unit-tests/linker-unit-tests64
Test: adb shell /data/nativetest/bionic-unit-tests/bionic-unit-tests
Test: adb shell /data/nativetest64/bionic-unit-tests/bionic-unit-tests
Test: Update /system/etc/ld.config*.txt and check whether the vndk
linker namespace of the vendor process can access the shared libs from
the default linker namespace.
Change-Id: I2879f0c5f5af60c7e56f8f743ebd2872e552286b
This change adds experimental support for SHT_RELR sections, proposed
here: https://groups.google.com/forum/#!topic/generic-abi/bX460iggiKg
Definitions for the new ELF section type and dynamic array tags, as well
as the encoding used in the new section are all under discussion and are
subject to change. Use with caution!
Bug: None
Test: Built image for marlin, flashed on device, ran arm and
aarch64 binaries containing '.relr.dyn' sections.
Change-Id: I2953ae932d3c42ae394e71f8fa058013758a1778
There were a bunch more unreasonable/incorrect ones, but these ones
seemed legit. Nothing very interesting, though.
Bug: N/A
Test: ran tests, benchmarks
Change-Id: If66971194d4a7b4bf6d0251bedb88e8cdc88a76f
Introduce new flag to mark soinfo as TLS_NODELETE when
there are thread_local dtors associated with dso_handle
belonging to it.
Test: bionic-unit-tests --gtest_filter=dl*
Test: bionic-unit-tests-glibc --gtest_filter=dl*
Bug: https://github.com/android-ndk/ndk/issues/360
Change-Id: I724ef89fc899788f95c47e6372c38b3313f18fed
ld-android.so exports exact same set of symbols
as linker. Since it is not supposed to be loaded
every symbol points to fail() method.
Test: make
Test: bionic-unit-tests --gtest_filter=dl*
Change-Id: I23bec365b302ce4ddf5f08832e665ae2b181cf8a
This will allow us to apply linker's version script to the "fake"
version of ld-android.so.
Test: make
Change-Id: I55645cc0fc90e6c65e2269ba2340a908aeed1eaa
C++17 removed a handful of things that are used in the libclang
headers, which are used by the versioner. Enable the flag that
re-exposes these until we have a C++17 compatible libclang.
Add a stub posix_memalign to the linker. libc++abi uses posix_memalign
when allocating exceptions, which the linker does not use.
Test: make checkbuild
Bug: None
Change-Id: I32f9d0591ef99a610f27efed90a5c9fd150f0d3e
This change addresses multiple problems introduced by
02586a2a34
1. In the case of unsuccessful dlopen the failure guard is triggered
for two namespaces which leads to double unload.
2. In the case where load_tasks includes libraries from 3 and more
namespaces it results in incorrect linking of libraries shared between
second and third/forth and so on namespaces.
The root cause of these problems was recursive call to find_libraries.
It does not do what it is expected to do. It does not form new load_tasks
list and immediately jumps to linking local_group. Not only this skips
reference counting it also will include unlinked but accessible library
from third (and fourth and fifth) namespaces in invalid local group. The
best case scenario here is that for 3 or more namesapces this will
fail to link. The worse case scenario it will link the library
incorrectly with will lead to very hard to catch bugs.
This change removes recursive call and replaces it with explicit list of
local_groups which should be linked. It also revisits the way we do
reference counting - with this change the reference counts are updated after
after libraries are successfully loaded.
Also update soinfo_free to abort in case when linker tries to free same
soinfo for the second time - this makes linker behavior less undefined.
Test: bionic-unit-tests
Bug: http://b/69787209
Change-Id: Iea25ced181a98c6503cce6e2b832c91d697342d5
When ro.vndk.version is set to a specific version, not "current",
use ld.config.$VER.txt as a linker namespace configuration file,
where $VER is the VNDK version set by ro.vndk.version.
Because ro.vndk.version is set by the vendor partition, the
configuration file will be automatically selected by the VNDK version
of vendor patition.
If ro.vndk.version is current or not set, ld.config.txt will be used
as before.
Bug: 69531793
Test: Build for a Pixel2 device.
In the out/target/product/<device> directory,
rename system/etc/ld.config.txt to system/etc/ld.config.27.1.0.txt
rename system/lib[64]/vndk to system/lib[64]/vndk-27.1.0
copy system/lib[64]/vndk-sp to system/lib[64]/vndk-sp-27.1.0
set ro.vndk.version to 27.1.0 in vendor/default.prop
Build system and vendor images with "make snod" and "make vnod".
Disble vbmeta using avbtool.
Flash a device and check boot.
Change-Id: Ic55bb0a741d434e5fa93e109be15df9d9de3f105
ld.config.txt currently does not support split line. As the file gets
larger, this limitation makes the file very unreadable. Now, long lines
can be avoided by breaking one line into multiple lines using +=
operator.
ex)
namespace.default.search.paths = /system/${LIB}
namespace.default.search.paths += /system/${LIB}
Delimitor (':' for *.paths and *.shared_libs, and ',' for *.links) is
automatically added.
Bug: 69888716
Test: linker-unit-tests passes
Change-Id: I4b94fd4e7f8a76d59db8d1096c86aa2118e46625
If libraries are on the odm partition they will fail to load
This patch adds the odm path as a search path for linker.
Test: libraries load from the odm partition
Change-Id: Ia7786e047cc565d74d25c025dacf9266b3763650
Signed-off-by: Alin Jerpelea <alin.jerpelea@son.com>
This reverts commit 58554ccb8a.
causes /vendor/bin/qseecomd to hit the new abort:
[ 8.983301] c5 603 DEBUG: Abort message: 'soinfo=0x7147894cd0 is not in soinfo_list (double unload?)'
Bug: http://b/69909887
Bug: http://b/69787209
Change-Id: Ied38f797e0a071a1acc5ed41adf1b45e855143c7
In the case of unsuccessful dlopen the failure guard is triggered
for two namespaces which leads to double unload.
Also update soinfo_free to abort in case when linker tries to free same
soinfo for the second time - this makes linker behavior less undefined.
Test: bionic-unit-tests
Bug: http://b/69787209
Change-Id: I886787ee021b050667f967bce7aa2708390886ea
Bug: http://b/29458203
Bug: http://b/68262627
Test: make && run bionic-unit-tests
Test: set debug.ld.all to dlopen and check the logs
Change-Id: I9a5495843a4145f267cc3a4714294d05b1e1fe90
Added a test to make sure linker produces correct error message
when user attempts to load a library in a linked namespace and fails.
Bug: http://b/67866190
Bug: http://b/64950640
Bug: http://b/64888291
Test: bionic-unit-test --gtest_filter=dlext*
Change-Id: I5b5c2070d1388eff123118350b2b5c8fc7571a29
vector::erase(iterator) erases the element that that iterator points
to, vector::erase(iterator a, iterator b) erases the range [a, b), with
a == b being a no-op.
Test: LD_PRELOAD=libc.so sh
Change-Id: I6a85c1cfaa8eb67756cb75d421f332d5c9a43a33
The permitted and search paths should be changed according to VNDK
version for VNDK snapshot. So, {VNDK_VER} value is added in ld.config.txt
file and this value is set through the linker.
If ro.vndk.version is not set, search paths are vndk and vndk-sp as
before.
Bug: 66074376
Test: build & run
Change-Id: I266b66fe1bc95d8925053ef497db11ac0a57c082
std::remove_if moves removed elements to the end, without actually
resizing the collection. To do so, you have to call erase on its
returned iterator.
Test: mma
Change-Id: Iae7f2f194166408f2b101d0c1cfc95202d8bbe63
I noticed that sometimes the old unwinder will add an extra PC 0 frame
after this change, but the new unwinder works in all cases. I'm not going
to fix the old unwinder since I plan to remove it very soon.
Bug: 67784501
Test: Forced a crash in the linker and verified that the unwind
Test: stops in __dl_start. Tested on arm/aarch64/x86/x86_64.
Change-Id: Id6585768023256be5c1d341df7b06b786a220b40
This patch resolves two issues:
(1) AOSP MIPS linker crash with SIGSEGV while relocating VDSO GOT
(2) Missing of MIPS_ABI_FP_ANY flag while ckecking & adjusting MIPS FP modes
(1): AOSP MIPS linker crash with SIGSEGV while relocating VDSO GOT
During bringup and testing of the kernel VDSO support on the emulator we
encountered an issue where all userspace processes were crashing due to
linker crash with SIGSEGV during VDSO GOT relocation.
The mentioned scenario will trigger SIGSEGV in the linker only if the kernel
code base contains VDSO implementation introduced with the following commits:
a7f4df4 MIPS: VDSO: Add implementations of gettimeofday() and clock_gettime()
c0a9f72 irqchip: irq-mips-gic: Provide function to map GIC user section
ebb5e78 MIPS: Initial implementation of a VDSO
Another prerequisite is that the linker contains the following commit from AOSP:
https://android-review.googlesource.com/#/c/264857/
The above commit introduces auxvec.h header containing definition of
AT_SYSINFO_EHDR in:
https://android.googlesource.com/platform/bionic/+/master/libc/kernel/uapi/asm-mips/asm/auxvec.h,
which in turn activates VDSO GOT relocations in mips_relocate_got():
https://android.googlesource.com/platform/bionic/+/master/linker/linker_mips.cpp#149
Since VDSO is mapped as a RO region, writing anything to its page will result
in SIGSEGV.
Removing this 0xdeadbeef cookies writes to the got[0]/got[1] solved SIGSEGV issue.
We also compared with the glibc linker code and we haven't seen anything similar
like writing some cookie values into GOT entries.
(2): Missing of MIPS_ABI_FP_ANY flag while checking & adjusting MIPS FP modes
This issue was found during testing of the patch:
https://android-review.googlesource.com/#/c/platform/bionic/+/494440/
This patch adds a bionic dlfcn.dlopen_vdso test which tries to open "linux-vdso.so.1"
(virtual ELF shared library) and expects to succeed.
dlopen fails in the mips linker part due to the following error:
"Unsupported MIPS32 FloatPt ABI 0 found in linux.vdso.so.1"
According to:
https://dmz-portal.imgtec.com/wiki/MIPS_O32_ABI_-_FR0_and_FR1_Interlinking#A.2._O32_FPXX_code_calling_FPXX,
ABI 0 is:
Val_GNU_MIPS_ABI_FP_ANY = 0 /* Not tagged or not using any ABIs affected by the differences. */
The issue appeared because MIPS_ABI_FP_ANY is not supposed to be checked as
a possible FP ABI-variant compatibility flag.
This patch adds logic to MIPS linker to handle this case appropriately,
after which dlfcn.dlopen_vdso test passes.
Test:
* bionic-unit-tests --gtest_filter=dlfcn.dlopen_vdso
* Boot android in emulator with kernel which supports VDSO optimizations
Change-Id: Icbcd9879beea1b38fbe8d97b3b205058eaa861f4
Signed-off-by: Goran Ferenc <goran.ferenc@imgtec.com>
