On Mon, 2011-03-28 at 16:55 +0200, Karel KlÃÄ wrote: > Mark Wielaard <mjw@xxxxxxxxxx> writes: > > eu-readelf does a lot more than you seem to need (you want to at least > > use --numeric-addresses, so it doesn't try to map all addresses it > > prints out to the associated symbol names). > > > > Attached is a little C program that just gets the associated source > > files and nothing else. It might help go through the whole repository > > quicker. It simply goes through all Compile Units, prints the comp_dir > > and name, and then everything from the associated files table. > > thank you for the program, I'll use it to speed up the check. Please let me know if I can help in any way, or if there are other places in your script that rely on eu-readelf output parsing that could use some speed up. > I'm also going to verify how well .debug_info sections match the source > files included in debuginfo packages using `eu-readelf -winfo` data > (-wline checking would be too complicated for this script). I have > observed GDB displaying wrong source file lines when stepping through a > program several times, and I think such a check could discover some > issues. I am not exactly sure what you are thinking of here. Could you give an example of what you would be looking for and how that would be wrong? We are working on a dwarflint tool, which is somewhat lower level (deals only with the .debug_* sections in the ELF files directly, not with any higher level cross file/package things) that might be able to detect some things. It isn't currently finished, but we hope to polish it for f16. See https://fedorahosted.org/elfutils/wiki/DwarfLint Maybe it could be combined with your scans over all packages somehow to produce reports on "the health of debuginfo in Fedora"? Thanks, Mark -- devel mailing list devel@xxxxxxxxxxxxxxxxxxxxxxx https://admin.fedoraproject.org/mailman/listinfo/devel
