Branch: refs/heads/wuxx1279/amdgpu
Home: https://github.com/dyninst/dyninst
Commit: 5c01c39956034ce57f92d9327fac18b843d27a7a
https://github.com/dyninst/dyninst/commit/5c01c39956034ce57f92d9327fac18b843d27a7a
Author: Jonathon Anderson <17242663+blue42u@xxxxxxxxxxxxxxxxxxxxxxxx>
Date: 2022-02-04 (Fri, 04 Feb 2022)
Changed paths:
M parseAPI/h/CFG.h
M parseAPI/src/Block.C
M parseAPI/src/Parser.C
Log Message:
-----------
ParseAPI: Speed up the case where a function is called from many locations (#1190)
In some binaries a single function has many callers, so copying out the source edges for
the entry Block takes a while and causes significant contention on the Block's lock.
The case in Parser::finalize where this becomes a hotspot likely does not need to read
every edge, so this commit inlines those cases into methods on Block instead.
Commit: 3489263edbb3f8a4b10f35b8104cefbfab08a9f0
https://github.com/dyninst/dyninst/commit/3489263edbb3f8a4b10f35b8104cefbfab08a9f0
Author: Vanessasaurus <814322+vsoch@xxxxxxxxxxxxxxxxxxxxxxxx>
Date: 2022-02-04 (Fri, 04 Feb 2022)
Changed paths:
M .github/workflows/base-containers.yaml
M docker/Dockerfile
Log Message:
-----------
Dockerfile: use explicit packages for 'spack external find' (#1195)
currently gettext in ubuntu 20.04 is being added as an external dependency (just added about 24
hours ago) and it is breaking the build. This change will test removing from the OS (probably
not recommended but okay in the context of the Dyninst image)
Signed-off-by: vsoch <vsoch@xxxxxxxxxxxxxxxxxxxxxxxx>
Co-authored-by: vsoch <vsoch@xxxxxxxxxxxxxxxxxxxxxxxx>
Commit: 458d04f1130d7a678a80dec68bf6141f5b761d49
https://github.com/dyninst/dyninst/commit/458d04f1130d7a678a80dec68bf6141f5b761d49
Author: Tim Haines <thaines.astro@xxxxxxxxx>
Date: 2022-02-04 (Fri, 04 Feb 2022)
Changed paths:
M symtabAPI/doc/API/Symtab/Module.tex
Log Message:
-----------
Symtab::module documentation cleanup (#1189)
* Remove docs for hasLineInformation
This was removed by 0dee1c8 in 2016
* Remove getAllVariables
This was by b3ad134 in 2021. It was removed there because this function
was never implemented.
Co-authored-by: Tim Haines <thaines@xxxxxxxxxxx>
Commit: 2c94ce001356b55d6c83c2265b905de4f1e86a51
https://github.com/dyninst/dyninst/commit/2c94ce001356b55d6c83c2265b905de4f1e86a51
Author: Stan Cox <scox@xxxxxxxxxx>
Date: 2022-02-04 (Fri, 04 Feb 2022)
Changed paths:
M symtabAPI/src/emitElf.C
Log Message:
-----------
glibc r_debug extensions to support multiple namespaces (#1175)
* Remove interposed definition of _r_debug
Previously there was a dynamic relocation against _r_debug in the loader which picked up the interposed definition, but glibc now uses a direct internal hidden symbol reference and thus no longer updates the interposed object.
* Update the elf _DYNAMIC symbol
When the symbol table is created in emitElf<ElfTypes>::createSymbolTables, also update the elf _DYNAMIC symbol to reflect the new address of the .dynamic section.
Commit: 8afa925fd7b421ce6eb9ad3734d00c6907c8c610
https://github.com/dyninst/dyninst/commit/8afa925fd7b421ce6eb9ad3734d00c6907c8c610
Author: Xiaozhu Meng <mxz297@xxxxxxxxx>
Date: 2022-02-04 (Fri, 04 Feb 2022)
Changed paths:
M symtabAPI/src/Object-elf.C
M symtabAPI/src/Object-elf.h
Log Message:
-----------
Remove dead code Object::elf_vaddr_to_ptr (#1192)
Commit: fd8744ee667874c6780f4f3a4ce9297d0f8388bd
https://github.com/dyninst/dyninst/commit/fd8744ee667874c6780f4f3a4ce9297d0f8388bd
Author: Xiaozhu Meng <mxz297@xxxxxxxxx>
Date: 2022-02-04 (Fri, 04 Feb 2022)
Changed paths:
M symtabAPI/h/Symtab.h
M symtabAPI/src/Symtab.C
Log Message:
-----------
Remove dead code and variables related to Symtab::data_ptr_ and Symtab::code_ptr_ (#1192)
Commit: 645e38a33fa5a86153bd08d923355f432289f7d5
https://github.com/dyninst/dyninst/commit/645e38a33fa5a86153bd08d923355f432289f7d5
Author: Xiaozhu Meng <mxz297@xxxxxxxxx>
Date: 2022-02-04 (Fri, 04 Feb 2022)
Changed paths:
M symtabAPI/src/Object-elf.C
Log Message:
-----------
Allow zero-length ELF program headers (#1192)
CUDA-11.6 binaries do not have program headers, which causes ELF parsing to stop early and miss line info
Commit: 7c8d841b1321eefa8349423356b1f884cb3e60c1
https://github.com/dyninst/dyninst/commit/7c8d841b1321eefa8349423356b1f884cb3e60c1
Author: Tim Haines <thaines.astro@xxxxxxxxx>
Date: 2022-02-04 (Fri, 04 Feb 2022)
Changed paths:
M dyninstAPI/src/IAPI_to_AST.C
M dyninstAPI_RT/h/dyninstRTExport.h
Log Message:
-----------
Additional cleanup of memory emulation for hybrid analysis (#1172)
* Remove vestiges of runtime functions
* Remove translation for emulation shadow pages
Co-authored-by: Tim Haines <thaines@xxxxxxxxxxx>
Commit: b8fed71b204e98b016dc751f2273da5fa2d5967f
https://github.com/dyninst/dyninst/commit/b8fed71b204e98b016dc751f2273da5fa2d5967f
Author: kupsch <kupsch@xxxxxxxxxxx>
Date: 2022-02-09 (Wed, 09 Feb 2022)
Changed paths:
M CMakeLists.txt
M parseAPI/h/SymLiteCodeSource.h
M parseAPI/src/SymLiteCodeSource.C
M proccontrol/src/linux.C
Log Message:
-----------
fix building of symlite (#1197)
- add missing pure virtual functions isReadOnly
- fix isCode and isData methods to use program header permissions instead
of just returning true
- fix printf compiler warning
- disable parseThat if using symlite as it is not compatible
- remove #include <linux/elf.h> from proccontrol/src/linux.h on aarch64
if SYMLITE is configured as it is incompatible with <elf.h>
Commit: a85835269188c24a5c27a50bbd06751c1badc3d7
https://github.com/dyninst/dyninst/commit/a85835269188c24a5c27a50bbd06751c1badc3d7
Author: bbiiggppiigg <bbiiggppiigg@xxxxxxxxx>
Date: 2022-02-23 (Wed, 23 Feb 2022)
Changed paths:
A common/h/AMDGPU/cdna2/amdgpu_cdna2_sys_regs.h
A common/h/amdgpu_cdna2_op_table.h
A common/h/amdgpu_cdna2_sys_regs.h
A common/h/amdgpu_cdna_op_table.h
A common/h/amdgpu_cdna_sys_regs.h
M common/h/amdgpu_op_table.h
M common/h/dyn_regs.h
M common/h/entryIDs.h
M common/src/dyn_regs.C
M dwarf/src/dwarfHandle.C
M elf/src/Elf_X.C
M instructionAPI/CMakeLists.txt
M instructionAPI/h/Instruction.h
M instructionAPI/h/Operation_impl.h
M instructionAPI/h/Result.h
A instructionAPI/src/AMDGPU/cdna2/InstructionDecoder-amdgpu-cdna2.C
A instructionAPI/src/AMDGPU/cdna2/InstructionDecoder-amdgpu-cdna2.h
A instructionAPI/src/AMDGPU/cdna2/amdgpu_cdna2_decoder_impl.C
A instructionAPI/src/AMDGPU/cdna2/amdgpu_cdna2_decoder_impl.h
A instructionAPI/src/AMDGPU/cdna2/amdgpu_cdna2_insn_entry.h
A instructionAPI/src/AMDGPU/cdna2/amdgpu_cdna2_op_table.h
A instructionAPI/src/AMDGPU/cdna2/amdgpu_cdna2_opcode_tables.C
A instructionAPI/src/AMDGPU/cdna2/amdgpu_cdna2_sys_regs.h
A instructionAPI/src/AMDGPU/cdna2/decodeOperands.C
A instructionAPI/src/AMDGPU/cdna2/decodeOperands.h
A instructionAPI/src/AMDGPU/cdna2/finalizeOperands.C
A instructionAPI/src/AMDGPU/vega/InstructionDecoder-amdgpu-vega.C
A instructionAPI/src/AMDGPU/vega/InstructionDecoder-amdgpu-vega.h
A instructionAPI/src/AMDGPU/vega/amdgpu_vega_decoder_impl.C
A instructionAPI/src/AMDGPU/vega/amdgpu_vega_decoder_impl.h
A instructionAPI/src/AMDGPU/vega/amdgpu_vega_insn_entry.h
A instructionAPI/src/AMDGPU/vega/amdgpu_vega_opcode_tables.C
M instructionAPI/src/Instruction.C
M instructionAPI/src/InstructionCategories.C
R instructionAPI/src/InstructionDecoder-amdgpu-vega.C
R instructionAPI/src/InstructionDecoder-amdgpu-vega.h
M instructionAPI/src/InstructionDecoderImpl.C
M instructionAPI/src/Operand.C
M instructionAPI/src/Operation.C
M instructionAPI/src/Register.C
M instructionAPI/src/amdgpu_branchinsn_table.h
R instructionAPI/src/amdgpu_decoder_impl_vega.C
R instructionAPI/src/amdgpu_decoder_impl_vega.h
R instructionAPI/src/amdgpu_insn_entry.h
R instructionAPI/src/amdgpu_opcode_tables.C
M parseAPI/src/IA_IAPI.C
M parseAPI/src/IA_amdgpu.C
M parseAPI/src/SymbolicExpression.C
M proccontrol/src/process.C
Log Message:
-----------
Add Support for AMDGPU CDNA2 Architectures based on XML ISA spec (#1107)
Add Initial Support for cdna based on the lastest version of XML file ( 20210720 )
Added new register names based on the latest xml-isa drop
Decoder implementation updated based on the latest xml-isa drop
renumbered registers to make low registers have the same value as full register (or its super set)
add larger memoery types for representing consecutive registers used as a single operand
Modify the format function to output multiple registers as a single operand correctly, handling for certain registers are still being worked on
split register-vector into multiple registers
Commit: ee629d08b3a76d36d4fbded2395e7c8780cb65da
https://github.com/dyninst/dyninst/commit/ee629d08b3a76d36d4fbded2395e7c8780cb65da
Author: Tim Haines <thaines.astro@xxxxxxxxx>
Date: 2022-02-24 (Thu, 24 Feb 2022)
Changed paths:
M common/CMakeLists.txt
A common/h/AMDGPU/cdna2/amdgpu_cdna2_op_table.h
R common/h/amdgpu_cdna2_op_table.h
R common/h/amdgpu_cdna2_sys_regs.h
M common/h/entryIDs.h
Log Message:
-----------
Build fixes for amdgpu/cdna2 (#1203)
* Clean up duplicate cdna2 sys_reg headers (#1200)
* add install rule to install amdgpu headers under common/h (#1201)
Co-authored-by: bbiiggppiigg <bbiiggppiigg@xxxxxxxxx>
Commit: b5facb8f2a4f29d6fdedbcc11553c0844d33710c
https://github.com/dyninst/dyninst/commit/b5facb8f2a4f29d6fdedbcc11553c0844d33710c
Author: Tim Haines <thaines.astro@xxxxxxxxx>
Date: 2022-02-24 (Thu, 24 Feb 2022)
Changed paths:
M common/src/dyn_regs.C
M instructionAPI/src/AMDGPU/cdna2/InstructionDecoder-amdgpu-cdna2.C
M instructionAPI/src/AMDGPU/cdna2/decodeOperands.C
M instructionAPI/src/Register.C
Log Message:
-----------
Fix compiler warnings in amdgpu cdna2 code (#1198)
* Fix 'unused-variable' warnings from gcc
* Fix 'unused-parameter' warnings from gcc
Commit: 1c8a880e0b49ff4db1f44d3920903805fd77548a
https://github.com/dyninst/dyninst/commit/1c8a880e0b49ff4db1f44d3920903805fd77548a
Author: Tim Haines <thaines.astro@xxxxxxxxx>
Date: 2022-02-25 (Fri, 25 Feb 2022)
Changed paths:
M instructionAPI/src/AMDGPU/cdna2/InstructionDecoder-amdgpu-cdna2.C
M instructionAPI/src/AMDGPU/cdna2/InstructionDecoder-amdgpu-cdna2.h
M instructionAPI/src/AMDGPU/cdna2/amdgpu_cdna2_decoder_impl.h
Log Message:
-----------
Tidy up classes in AMDGPU (#1204)
* Remove dead code in makeAmdgpuRegID
* Fix possible no-return cdna2 functions
* Simplify code in bitfieldInsnAliasMap and condInsnAliasMap
* Correctly initialize all members of InstructionDecoder_amdgpu_cdna2
* Inline InstructionDecoder_amdgpu_cdna2 constructor
* Explicitly default ~InstructionDecoder_amdgpu_cdna2
* Fix possible non-return in decodeSGPRorM0
Commit: 4655b907f0dee6ad1062543ade0489e157ccdc58
https://github.com/dyninst/dyninst/commit/4655b907f0dee6ad1062543ade0489e157ccdc58
Author: bbiiggppiigg <bbiiggppiigg@xxxxxxxxx>
Date: 2022-02-25 (Fri, 25 Feb 2022)
Changed paths:
A common/h/AMDGPU/vega/amdgpu_vega_op_table.h
A common/h/AMDGPU/vega/amdgpu_vega_sys_regs.h
R common/h/amdgpu_cdna_op_table.h
R common/h/amdgpu_cdna_sys_regs.h
R common/h/amdgpu_op_table.h
R common/h/amdgpu_vega_sys_regs.h
M common/h/dyn_regs.h
M common/h/entryIDs.h
R instructionAPI/src/AMDGPU/cdna2/amdgpu_cdna2_op_table.h
R instructionAPI/src/AMDGPU/cdna2/amdgpu_cdna2_sys_regs.h
Log Message:
-----------
Code clean up for AMDGPU (#1205)
* Update the code layout for vega headers
* Remove unused cda headers
* remove duplicate cdna2 headers in instructionAPI
Compare: https://github.com/dyninst/dyninst/compare/f01e178a2c17...4655b907f0de
|