pwndbg/pwndbg 2023.03.19 on GitHub

Here is the 2023.03.19 release. Thanks to everyone who contributed!

TL;DR what this release brings

New or edited commands:
- kbase, kchecksec, slab - kernel debugging only commands to get kernel base address, checksec for kernel and list kernel slabs
- valist - dumps arguments of a va_list structure on given address
- ai - ask AI about the current debugging session (requires openAI API key)
- cunwatch now operates on the index/number of expression instead of requiring the user to pass the whole expression to unwatch
- cyclic - improved UX by adding nicer info/error messages
- the tls command now leverages GDB's scheduler locking so now when it calls a function to obtain TLS address it won't allow any other target threads to run, which could previously cause issues on targets with multiple threads
We added command categories and refactored the display of pwndbg help :)
Added support for riscv:rv64 architecture
Lots of glibc heap commands fixes and improvements
- Better support for 32-bit and big-endian architectures
- Better heap heuristics
- largebins now display bin size ranges instead of indexes
- arena displays thread id
- vis_heap_chunks has now --all-chunks to display all chunks
- we renamed two flags in the vis_heap_chunks command (--naive -> --beyond-top and --display_all -> --no_truncate)
It is now possible to display instruction opcodes in the disasm context by setting set nearpc-num-opcode-bytes 9
Fixes for vmmap recursion errors
Fixes for Rust binaries debugging
Support for Rizin
Unicorn emulation is now disabled if mmap(1GB, RWX) fails due Unicorn Engine aborts on environments with low memory
Improved virtualenv detection and warnings
Fixed message when locales are wrong: in fact, Pwndbg requires LC_CTYPE=C.UTF-8 environment variable and not LC_ALL=en_US.UTF-8 PYTHONIOENCODING=UTF-8 as it previously suggested

What's Changed

Fail tests if an error occurs by @gsingh93 in #1445
Add ArchOps by @gsingh93 in #1444
Make it easier to remove cwatch expressions by @gsingh93 in #1421
Move some event hooks to hooks.py by @gsingh93 in #1431
Start a env_parallel session to avoid argument too long errors by @gsingh93 in #1446
Use RawTextHelpFormatter for more commands by @gsingh93 in #1448
Add kchecksec command by @gsingh93 in #1441
Add x86_64ArchOps by @gsingh93 in #1449
Add kbase command by @gsingh93 in #1450
Remove redundant struct.unpack formats by @gsingh93 in #1451
Fix gdb-pt-dump exception when attaching to x86 QEMU by @gsingh93 in #1452
Added command to list slabs by @gsingh93 in #1442
Add paging_enabled function and OnlyWhenPagingEnabled decorator by @gsingh93 in #1454
Cleanup command docstrings by @gsingh93 in #1461
Annotate some vmmap methods with @OnlyWhenRunning by @gsingh93 in #1463
Don't print aliases as commands in pwndbg output by @gsingh93 in #1460
Better error handling in bugreport command when no command history by @gsingh93 in #1464
Fix issue #1466 by @QueenEleanor in #1467
Add mypy configuration to pyproject.toml by @gsingh93 in #1468
Add safe type hints by @gsingh93 in #1469
Remove dependency to python-future (#1250) by @hamarituc in #1470
Automatically add types with pyannotate by @gsingh93 in #1471
More type hints by @gsingh93 in #1472
More type fixes by @gsingh93 in #1473
Fix type issue in ida.py by @gsingh93 in #1474
Upgrade dependencies by @gsingh93 in #1476
Add support for command categories by @gsingh93 in #1477
Add commands to categories by @gsingh93 in #1478
Add slab info command by @gsingh93 in #1447
Rename elfheader command to elfsections by @gsingh93 in #1481
Fix reload command by @disconnect3d in #1491
Fix #1457 - warn if monitor info mem is invalid by @disconnect3d in #1480
Add commands to Linux category by @gsingh93 in #1483
gdbinit.py: improve virtualenv detection and warnings by @disconnect3d in #1489
gdblib/symbol.py: remove symbol autofetch by @disconnect3d in #1494
Minor improvements (some of #1465) by @disconnect3d in #1493
Fix #855: Rust binaries debugging by @disconnect3d in #1495
gdb-tests/tests.sh: don't do make clean on each run by @disconnect3d in #1497
Remove commands/gdbinit.py by @gsingh93 in #1499
Add start command category by @gsingh93 in #1500
Add commands to memory category by @gsingh93 in #1492
Don't assign result of fetch_lazy by @gsingh93 in #1501
Fix some easy type errors by @hauntsaninja in #1502
Fix is_memory_mapped_file by @gsingh93 in #1503
Refactor nearpc by @gsingh93 in #1498
Implement contains for reg module by @gsingh93 in #1504
Run lint CI job and Ubuntu 22.04 and 18.04 by @gsingh93 in #1505
More type fixes by @gsingh93 in #1506
pwndbg.gdb.regs type fixes by @gsingh93 in #1508
Fix some misc type errors by @gsingh93 in #1509
Fix memoize debug path by @disconnect3d in #1510
Add gdbserver dependency to Ubuntu build by @CptGibbon in #1514
Fix optimize color memoization by @disconnect3d in #1511
Remove Arch.signed and Arch.unsigned by @gsingh93 in #1516
Fix some type and pylint errors by @gsingh93 in #1515
Type fixes by @gsingh93 in #1517
Fix memoize type issues and enable mypy in lint script by @gsingh93 in #1518
Update pylint config and fix lint issues by @gsingh93 in #1519
Fix more pylint warnings by @gsingh93 in #1520
Remove useless statement in procinfo.py by @gsingh93 in #1521
cyclic command: improve UX by @disconnect3d in #1522
Add valist command by @gsingh93 in #1524
Improve cwatch formatting by @gsingh93 in #1525
Improve hexdump collapse display by @disconnect3d in #1526
requirements.txt: bump pwntools to 4.0.9 by @disconnect3d in #1527
Update qemu test scripts and images by @gsingh93 in #1529
Don't fail CI if codecov upload fails by @gsingh93 in #1532
Finalize QEMU testing scripts by @gsingh93 in #1531
TLS related enhancements by @lebr0nli in #1536
Run QEMU tests on CI by @gsingh93 in #1533
Support showing opcodes in the disassembly by @lebr0nli in #1537
Bug fix for error handling of the heap commands by @lebr0nli in #1543
Fix procinfo command (#1544) on QEMU targets and with abstract UDS by @disconnect3d in #1545
lint.sh vermin: use -vvv --no-tips by @disconnect3d in #1552
Fix #1534: disable emulation if mmap(1G,RWX) fails by @disconnect3d in #1546
lint.sh: vermin -vvv instead of -vvvv by @disconnect3d in #1556
Fix configfile by @gsingh93 in #1569
Enhance the checks before accessing the memory for the heap heuristics by @lebr0nli in #1558
gdbinit.py: fix message when locales are wrong by @disconnect3d in #1573
Fix OnlyWithArch invocations (cymbol command) by @disconnect3d in #1574
Fix #1572: fsbase/gsbase commands on x86 32-bit archs by @disconnect3d in #1575
Add support for Rizin by @mariaWitch in #1571
Support using the rich module to print the full stacktrace by @lebr0nli in #1578
Remove /foo file by @disconnect3d in #1582
arena command: display thread id (fixes #1562) by @disconnect3d in #1583
Use pwnlib.util.misc.which instead of gdblib.which.which by @disconnect3d in #1584
Fix plt and gotplt commands by @disconnect3d in #1576
Fix auxv memory access error by @disconnect3d in #1586
Added support for riscv:rv64 by @zommiommy in #1588
Rewrite some unreliable methods for the heap heuristics by @lebr0nli in #1579
GPT-3 Command for Pwndbg by @oblivia-simplex in #1589
Make the output of the heap-related commands consistent by @lebr0nli in #1590
Avoid downloading files from remote server in some cases by @lebr0nli in #1593
Hopefully fix vmmap recursion issues by @disconnect3d in #1585
Add i686 heap alignment heuristic by @CptGibbon in #1595
A few last-minute fixes for the ai command by @oblivia-simplex in #1594
Hopefully fix the bug when using LD_PRELOAD to load libc by @lebr0nli in #1602
Fixes #1600, #752: invalid symbol.get results by @disconnect3d in #1605
tests.sh: check for parallel_env requirement (closes #1606) by @disconnect3d in #1607
symbol.py: remove _add_main_symbol_to_exe by @disconnect3d in #1609
Fix #1610: symbol.get with duplicated symbols by @disconnect3d in #1611
Add print all chunks in vis_heap_chunks by @neryaz in #1604
feature: Add headers to arenas command output by @CptGibbon in #1612
Display bin size range in largebins command output by @CptGibbon in #1613
ai tool now uses ChatGPT model by @oblivia-simplex in #1614
Add largebin_index_32_big macro by @CptGibbon in #1617
Fix #808: set search-memory-packet off only on GDB<=9 by @disconnect3d in #1619
Correct smallbins alignment on i386 by @CptGibbon in #1621
Correct largebin size lookups on i386 by @CptGibbon in #1623
Fix #1394: Improve installation process by @Cycatz in #1620
Bump Zig version by @CptGibbon in #1629
Fix the bug when using the heap heuristic on big-endian architectures by @lebr0nli in #1627
Fix the issues when finding tcache for some 32-bit architectures by @lebr0nli in #1625
fetch GDB version via regex to fit Fedora format by @disconnect3d in #1632

New Contributors

@QueenEleanor made their first contribution in #1467
@hamarituc made their first contribution in #1470
@hauntsaninja made their first contribution in #1502
@mariaWitch made their first contribution in #1571
@oblivia-simplex made their first contribution in #1589
@neryaz made their first contribution in #1604
@Cycatz made their first contribution in #1620

Full Changelog: 2022.12.19...2023.03.19