On Mon, Feb 24, 2020 at 03:56:58PM +1000, Allan McRae wrote:
On 20/2/20 11:51 am, Austin Lund wrote:
Currently only the file pointed to by the DW_AT_name is included as a source file in debug packages. This means many files that are useful for debugging are not included. For example, no header files are included but yet these may by referenced in the .debug_line section.
This sed script converts into shell variables the debug dump information from readelf about compilation units, directory tables and file tables. This can be used to get the full path of all the source files from within the package being compiled that are referenced in the debugging information. Also, placeholder symbols (e.g. <builtin>) and paths outside the current source (e.g. linked libraries) will be more consistently ignored from inclusion in the debug packages.
Signed-off-by: Austin Lund <austin.lund@gmail.com> ---
So... that sed script is horrendous! But let me see if I understand this correctly.
We currently only look at the .debug_info section, finding DW_AT_name/DW_AT_comp_dir pair to grab file names. That appears to get the main compilation units, but misses header files. It looks like your sed does something slightly different to get that info, although I can't tell if there is a functional difference.
Your patch additionally looks at the .debug_line section. This section has a table of directories that source files come from (which can be filtered to remove system directories), and a file name table with files from each directory. This does not include the files we currently grab.
But there must be something I am missing... For the example of "ls" I see "selinux.h" in from directory "./lib/selinux" in that .debug_line output. Your script does not include this file. Using the rpmtool "debugedit -l" does include that file in its file list. There are quite a few other examples.
Am I on the right track? Can you clarify what I am missing here?
It would seem I made a bad assmption about what was in and not in the current source tree. There will be references to libc files that appear in the output as a relative path. My thought was that if there wasn't a preceeding './' then it must refer to another directory. Anyway, yes. You are on the right track. My hope was that you can install the debug package and get all available source listings. Without these functions within the header files, that isn't possible. What's a better approach to parsing the readelf output? Awk?