Recently, the maps for an elf in memory might show up looking like:
f0000-f1000 0 r-- /system/lib/libc.so
f1000-f2000 0 ---
f2000-f3000 1000 r-x /system/lib/libc.so
f3000-f4000 2000 rw- /system/lib/libc.so
The problem is that there is logic in the code that assumed that the
map before the execute map must be the read-only map. In the case
above, this is not true. Add a new prev_real_map that will point
to the previous map that is not one of these empty maps.
This will fix the backtraces that look like this:
#00 pc 0000000000050d58 /apex/com.android.runtime/lib64/bionic/libc.so!libc.so (offset 0x50000) (syscall+24) (BuildId: 5252408bf30e395d49ee270b54c77ca4)
To get rid of the !libc.so and the offset value, which is not correct.
Added new unit tests to verify this.
Added new offline test which an empty map between read-only and execute
map. Before this change, the backtraces had lines like
libc.so!libc.so (offset XXX) would be present.
Bug: 148075852
Test: Ran unit tests.
Change-Id: Ie04bfc96b8f91ed885cb1e655cf1e346efe48a45
The function StepIfSignalHandler assumed that the rel_pc passed
to it was actually an elf offset. A new version of clang created a libc.so
that has a load bias, so tests unwinding through a signal handler
would fail on arm. On other ABIs, there is unwind information that could
be used instead, so the unwind still worked.
The fix is to subtract the load bias from the rel_pc to get an elf
offset to pass to the Register StepIfSignalHandler functions. Change all
of the Register funtions to make it clear what the first parameter means.
Add a unit test for this new code. Also, add an offline test for
this case.
Bug: 145683525
Test: Ran unit tests using the new clang and the old clang.
Change-Id: I3e249653b79bcad6d3a56411a7911fde4888e9d6
The code was not properly getting the variable addresses and using
the offset and address fields of the .data section.
Fix all of that, and update the tests.
Bug: 145162678
Test: Unit tests pass.
Test: ./art/test/run-test --dex2oat-jobs 4 --host --prebuild --compact-dex-level fast --jit --no-relocate --runtime-option -Xcheck:jni 137-cfi
Test: ./art/test/testrunner/testrunner.py -t 137 --host
Change-Id: Ic61c4487334fd2273cda9c56eb1a3b525a03edb7
The bias for the PT_GNU_EH_FRAME was using the paddr instead of vaddr.
This doesn't match the way the load bias is calculated, which always
use vaddr - offset, so change to use vaddr.
Found on an old x86 device that has a vdso that sets vaddr differently
from paddr.
Add a new offline test to catch this case and update the elf interface
unit tests.
Also, fix a small bug in the unwind_for_offline tool.
Bug: 142365899
Test: Unit tests pass.
Change-Id: I5f0bf062dd8ee45aa8553189ba493ec962e0b059
The original code assumed that the load bias in the program headers
would be exactly the same as in eh_frame/eh_frame_hdr/debug_frame.
This isn't guaranteed, so add a section bias for use when creating
a DwarfSection. In addtion, make the load bias and section bias
a signed value. There is no reason that this value needs to be positive,
so don't force it to be.
Add a new offline test that has a different load bias in eh_frame than
in the executable load.
Add additional unit tests to verify the load bias values are set properly.
Clean up the tests in ElfInterfaceTest, making all tests names follow the
same convention.
Bug: 141888859
Bug: 142094469
Test: New units and old unit tests pass on host and taimen.
Change-Id: Ib878123ab5545f0f315c749cfe0d27b012d873ee
Changes:
- Change GetSoname to always returns a std::string.
- Added new unit tests for the soname printing.
- Modify the GetElf() function to save the same elf when we see rosegment
linkers that split the read-only and read-write across a map. This
avoids creating multiple elf objects for each map.
- Fixed a few offline unwind tests.
Bug: 29218999
Test: Unit tests pass.
Change-Id: Iad7c38b5c2957a8c5fd4ba94ebec335bafcad57d
The elf_start_offset field in a map_info was not set in one path.
Added new offline test for this, and update a few unit tests.
Bug: 128442792
Test: New unit tests pass.
Change-Id: I2c6ac1b5271a99aa0e8c4b6342a5970199185112
The linker was modified so that a shared library has a read-only
map, then a read-execute map to represent the whole shared library.
When backtraces are created, then the offsets are set incorrectly
for backtraces. For example, all backtraces wind up with an offset now,
and a shared library loaded out of an apk shows the wrong offset.
Changes:
- Fix the FormatFrame function which was putting the offset before the
map name.
- Refactor the Maps and MapInfo classes to keep track of the previous map
instead of all maps. This was the only map that was ever needed.
- Modify the unwind_for_offline tool to capture both the read-only and
read-execute map when using the read-only segment option.
Bug: 120981155
Test: Updated unit tests pass.
Test: Unwinds on device don't show the offsets everywhere any more.
Change-Id: I75b3727221be9c20316dfdcd7a1c6d242d7ce948
When a shared library is loaded directly from an apk, the new way the
linker splits a shared library into a read-only and execute segment
broke unwinding. Modify the code to handle this case.
Other changes:
- Modify the algorithm for finding read-only map entries. Before, the code
would search the entire map for the closest offset. Now it simply looks
at the previous map. I did this because the old code was too lenient and
might still work even if the linker changes. I want this to break if the
linker behavior changes so that I can analyze the change.
- Update the tools to use PTRACE_SEIZE instead of PTRACE_ATTACH since
PTRACE_ATTACH doesn't work in all cases.
- Small refactor of the GetFileMemory function.
- Add new unit test cases and new offline unwind test cases.
Bug: 120618231
Test: Ran new unit tests, ran original failing test.
Change-Id: I4bade55cf33220d52f1d5e9b0cbbbcc8419669d4
If multiple threads are unwinding at the same time, new maps that contain
the global variables for dex files and jit information are created. This
leads to threads creating more new maps that then get searched, then
more maps, then more searching until virtual address space exhaustion.
Fix this so that we only search maps that have a corresponding rw map that
could contain the global memory.
Small refactor to combine the code to search for global variables into
one class that both classes inherit from.
Modify unit tests for the new pattern checking.
Bug: 117761427
Test: Ran unit tests for libunwindstack/libbacktrace/simpleperf.
Test: Ran art 004-ThreadStress that used to fail.
Change-Id: I837ca6b9d0383100079de090bc7d019598e0cdfe
It turns out that for the dwarf information, if a FDE indicates it's pc
relative, then pc has to be incremented by the load bias. If not, then
it should not be incremented.
Previously, the code always subtracted load bias values from pcs, and assumed
that all fdes were incremented by load bias values. The new code actually
reads the fdes and adjusted the pcs in the fde and in the eh frame hdr so
that load bias values are already handled properly.
In addition, add dumping of arm exidx values in unwind_reg_info. This allowed
verifying that the debug frame in those elf files was being handled properly.
Added a new unit test that only has a debug frame that has a non-zero load
bias and has fde entries that do not have pc relative encoding.
Fix a couple of other small bugs.
Bug: 109824792
Test: All libbacktrace/libunwindstack unit tests pass.
Test: Ran ART 137-cfi test and 004-ThreadStress.
Test: Verify that displaying the fde start and end pc actually match the
Test: real data for fde that have pc relative set, and that don't.
Test: Verified that the unwind information for arm exidx matches the
Test: debug frame data.
Change-Id: I707555286b5cb05df9f25489e8c5ede753cfe0fb
The new lld linker uses all non-zero offset executable maps. There was
a bug when trying to find if the stack is in a signal handler that caused
the code to read the wrong place in the elf. Fixed by not adding the elf
offset to the relative pc.
Also fixed the unwind_for_offline tool to dump multiple stacks if necessary.
Added new offline unit test that would have failed with the old code.
Bug: 79936827
Test: Ran unit tests and libbacktrace unit tests.
Test: Dumped backtraces of system pids.
Test: Ran 137-cfi art test.
Change-Id: Iaca8c18c2a314902e64c3f72831234297e8dcb1b
Changes:
- Add a new flag to the libbacktrace and libunwindstack map data.
- Modify the unwinder to handle this map to use the raw pc when stepping.
- Add new unit tests for this case.
Bug: http://b/73127105
Test: Run simpleperf to unwind through jit symfiles.
Test: Run new unit tests.
Test: Run 137-cfi test on host.
Change-Id: I10bc0410680accc6d35fe51e9f1098911f667e01
There are a few places where it is assumed that this register is
set to the cfa value when interpreting DWARF information.
Add a testcase for unwinding art_quick_osr_stub on ARM.
Bug: 73954823
Test: Ran libunwindstack/libbacktrace unit tests.
Test: Random debuggerd -b of process on a hikey.
Test: Ran the 137 art test on host.
Change-Id: Ida6ccdc38c3cfeea6b57fe861a0cc127b150b790
- Add a new tool that can dump the register information given a pc.
- Add a new offline unwinder test that fails without this change.
- Update unit tests.
- Refactor offline unwind tests to make it easier to add tests.
Bug: 73048324
Test: Passes new unit tests.
Test: Passes libbacktrace unit tests.
Test: Ran debuggerd -b on random processes.
Change-Id: If6c70a044299505c2274ae6888b35bf9ac34b74b
Shrink a few files that are huge by removing all of the .debug_XXX
sections except .debug_frame since they aren't used.
Rename all of the arm32 to arm and x86_32 to x86.
Test: All unit tests pass.
Change-Id: Ia0f0baadf2a7fbc42a544aff2f14d5ed5f9287b7
- Be a little more lenient when reading the cies/fdes. If next entry data
winds up incorrect, don't fail, simply stop processing the entries. This
only applies when reading all of the cies/fdes at once.
- Fail to init an eh_frame with no entries and fallback to assuming the
eh_frame has no header instead.
- Change the step to always try debug_frame first which has the most
accurate information.
- Add small unit tests and a couple of offline unit tests to verify
this behavior.
These changes are needed to support offline unwinding since it depends
on this new behavior.
Bug: 65682279
Test: Ran new unit tests.
Change-Id: I3529f1b0c8e14cd7409494e5de2f3c9e78d0855e
Sometimes a process will have multiple shared libraries loaded
that have defined __jit_debug_descriptor. Specifically, art testing
will load libart.so and libartd.so, which would have broken unwinding
through jit'd code for art testing if libart.so winds up being found first.
In order to avoid duplicating the code for the linker to figure out
which one is live, change the algorithm to find the first non-zero
first_entry_ set in __jit_debug_descriptor.
Bug: 68396769
Test: Passes unit tests.
Test: Verified this can unwind 137-cfi on arm/arm64.
Change-Id: Ic4d403065d2c6f22476ef0171e7add17cd1464cd
Changes:
- New JitDebug class to handle all of the jit gdb interface.
- Add unit tests for all, along with new offline test using debug data.
- Add new Memory type called MemoryOfflineParts that has multiple
MemoryOffline objects to support the offline test.
- Update the tools to use the JitDebug object.
- Modify libbacktrace to use the JitDebug, but only looking in libart.so
and libartd.so.
- Change the Format32Bits to Is32Bit since it's more accurate and I use
it in a different context where original name didn't make sense.
- Add a new function to find global variables in an elf file
(GetGlobalVariable).
- Add a new function to determine if a pc is valid for this elf (IsValidPc).
Bug: 68396769
Test: Ran new unit tests. Added new offline test that uses jit debug data.
Test: Ran art test that generates jit data and verified a crash unwinds
Test: through the jit data.
Change-Id: I6e7ee2f5bab2242028a06feece156dff21c0a974
- Add a load_bias field in MapInfo so that it can be loaded offline,
and also so it can be cached.
- Add an Add function to the Maps class so that it's possible to manually
create a map.
- Remove the OfflineMaps class since I haven't found a reason for this to
exist.
- Add a pointer to the gnu debugdata compressed section in the interface
itself and modify the step path to try eh_frame, then debug_frame, then
gnu_debugdata. This way arm can add exidx as the last step behind
gnu_debugdata. Add an offline test to verify the order of unwind.
- Fix x86_64_ucontext_t since it was a different size on 32 bit and 64 bit
systems.
Test: Pass new unit tests.
Change-Id: I978b70d6c244bd307c62a29886d24c1a8cb2af23
When stepping, it's necessary to use both the unaltered relative pc
and the adjusted relative pc. If the adjusted pc is not used, the
wrong unwind information can be used.
Added new offline unit tests that take real data and verifies that it
unwinds properly.
Fix a bug in the map code that would not properly parse map data for
a 64 bit process when done in a 32 bit process.
Fix bug in eh_frame processing that didn't adjust the pc correctly.
Fix unit tests related to the pc adjustment.
Bug: 69475565
Test: Passes libbacktrace/libunwindstack unit tests.
Test: Run debuggerd -b on processes on a hikey.
Change-Id: Ic501a1c4549c5f61d2742a7105c42a960f2c892b
- Modify the tests to share some code for the generation of the elf data.
- Move the gnu_debugdata files into a separate directory.
- Add tests for GetElf().
- Move the CreateMemory() tests and the GetElf() tests into separate files.
Bug: 23762183
Test: New unit tests pass.
Change-Id: Ie3380296bb49753c2ac8801cfa11f93d6ff7121d
Christopher Ferris