Yao Qi
2017-01-29 22:06:29 UTC
# Progress #
* TCWG-984, PR 20939, GDB aborts if there is an error in disassembly.
Done. All patches are committed to mainline. [3/10]
* Patch reviews, a lot, [5/10]
** Hold one of ILP32 BFD patch because it may affect aarch64 GDB,
Reviewed ILP32 GDB patch, and convince them not to add new target
descriptions for ILP32, to make our life easier in the future.
Otherwise, we need two copies of target description for each new
hardware feature, such as v8.3, SVE, etc.
** Try to hold one disassembler options patch, which affects ARM and
GDB. Gave some comments on the interface design, but looks I am
late.
** Finish the review on DW_OP_implicit_value. Met some grey area in
DWARF spec.
* Atomicity of concurrent modification on ptrace. [1/10]
It is unclear weather debugging can modify memory for one thread while
other threads are running. After the discussions with some kernel guy,
PTRACE_POKETEXT is atomic, but GDB use it in a non-atomic way when
writing 2-byte aligned 32-bit thumb-2 instruction. Suggested a
workaround upstreams.
* Misc, meeting [1/10]
# Plan #
* ILP32 patch review.
* Carefully read IBM's kernel debugging patches.
* Set up AArch64 bare debugging with GDB, OpenOCD, JTAG and
HiKey. Soldering is needed.
* TCWG-984, PR 20939, GDB aborts if there is an error in disassembly.
Done. All patches are committed to mainline. [3/10]
* Patch reviews, a lot, [5/10]
** Hold one of ILP32 BFD patch because it may affect aarch64 GDB,
Reviewed ILP32 GDB patch, and convince them not to add new target
descriptions for ILP32, to make our life easier in the future.
Otherwise, we need two copies of target description for each new
hardware feature, such as v8.3, SVE, etc.
** Try to hold one disassembler options patch, which affects ARM and
GDB. Gave some comments on the interface design, but looks I am
late.
** Finish the review on DW_OP_implicit_value. Met some grey area in
DWARF spec.
* Atomicity of concurrent modification on ptrace. [1/10]
It is unclear weather debugging can modify memory for one thread while
other threads are running. After the discussions with some kernel guy,
PTRACE_POKETEXT is atomic, but GDB use it in a non-atomic way when
writing 2-byte aligned 32-bit thumb-2 instruction. Suggested a
workaround upstreams.
* Misc, meeting [1/10]
# Plan #
* ILP32 patch review.
* Carefully read IBM's kernel debugging patches.
* Set up AArch64 bare debugging with GDB, OpenOCD, JTAG and
HiKey. Soldering is needed.
--
Yao Qi
Yao Qi