* Use RBinSection, RDebugMap or RIOMap instead
* Remove omps command
* Fix segfault in ?E+
* Improve anal.eobjmp (bug spotted after almost killing iosections)
* bin/filter: correctly allocate strings when filtering them
Directly reallocate strings when filtering them, because it is very hard for
the caller to know before hand the correct size of the string otherwise. The
patch uses r_str_ APIs to makes this easy.
* r_bin_close should be r_bin_file_close
* moved bin/open.c into bin.c
* set BinFile id in r_bin_load_io
* use r_bin_open_io instead of r_bin_load_io
* also prefer RBinOptions instead of multiple arguments
* rename r_bin_load to r_bin_open and use RBinOptions
* remove R_API from functions not used outside of RBin module
* move not-exported functions to a separate private header
* move bin_private in i/private.h
* fix and use R_IPI
* fix mips_assemble
* More code cleanup and bring back bin.reload for the debugger
* Benchmark: swift: 23/9549 rcore: 17232/37147 (2-400x faster)
* For DVIA iOS app this is 1s less
* Demangle all the symbols, not just few
PR #11888
* bin: no need for 2 r_bin_load_io APIs
* bin: remove io-related unused functions
* r_bin.h: remove some old comments
* r_bin.h: group together related APIs
* extract print/set modes in r_types, to remove dups in r_bin and r_core
* clang-format-diff: fix ternary operator indent for pancake
* libr/bin: no need to allocate RBinOptions on the heap
* bin: start using r_return_* around, that's just the beginning
* bin: remove io_owned since it's not used anywhere
* io: make r_io_bind return nothing
* bin: remove unused functions and simplify r_bin_load_io
r_bin_load_io was calling r_bin_load_io2 with UT64_MAX as sz parameter,
but r_bin_load_io2 just returns false if (st64)sz is less than 0, so
that call is actually useless and can be removed.
* bin/bin: fix some preconditions
* bin/open: fix precondition to check for bin and filename too
* format/elf: extract function to avoid duplicated code
* format/elf: use ut64 to avoid errors in ELF32
Otherwise, while analyzing ELF32, -1 will be interpreted as the unsigned
value 0xffffffff, which is very different from -1 (0xffffffffffffffff).
* format/elf: fix symbols with shndx == SHT_NULL
Symbols with shndx == SHT_NULL should not be listed as symbols but they
should be available when other ELF structures (imports, relocs, etc.)
reference them.
This patch adds those symbols in the symbols_by_ord table but not to the
list of available symbols. Imports/relocs/etc. work by indexing symbols
by symbol index, so they are going to use the symbols_by_ord table and
make everything work.
Moreover, it refactors a bit the relocs function to avoid duplicated
code.
* bin: setimportd/setsymbold should clone the import/symbol
* bin/bin.c: free fields only when sym is not NULL
* format/elf: do not filter out symbols with size == 0 and consider sht_null symbols in shdr too.
* bin/bin.c: create R_STR_DUP macro and use it in _clone functions
* format/elf: when parsing symbols from phdr, stop at first UNK type/bind
* create some R_BIN defines to handle strings related to TYPE and BIND
in RBinSymbols
* format/elf: add a doc
