This series contain fixes for winedump for debug information: - it fixes a bunch of incorrect information dumping, - it fixes a case of crash for .DBG files, - it allows dumping very large PDB files (>4G)
Note: for the later item, current tools' read_file() doesn't work with such large files. Instead of fixing read_file(), I opted for letting a dumper for a given file format to only use a fd to the target file, instead of forcing all dumpers to work from a full in-memory image of the target file. This is useful for PDB files, as their underlying storage requires to reassemble non-contiguous blocks, so this saves memory (the in-memory image of the target file) and one copy of the whole file. This means that all functions to guess the file format now uses a fd instead of the full image.
-- v2: tools/winedump: Support dumping large PDB files (>4G). tools/winedump: Support more than 64K files in PDB DBI module source substream. tools/winedump: Don't miss hash entries in PDB files. tools/winedump: Don't miss PDB_SYMBOL_RANGE* in PDB files. tools/winedump: Use correct field when dumping CodeView symbols. tools/winedump: Use correct computation for first section out of a .DBG file.