13 листопада 2009 о 21:00 +0100 Kurt Roeckx написав(-ла): > On Fri, Nov 13, 2009 at 08:33:59PM +0100, Eugeniy Meshcheryakov wrote: > > 13 ????????? 2009 ? 19:03 +0100 Kurt Roeckx ???????(-??): > > > So I'm guessing that it opens the file in /lib/modules/ > > > calls the validate() function which says it was succesful, > > > while the file actually doesn't have any debug info? > > Also it is strange that validate() is successful because as I understand > > .gnu_debuglink aslo contains checksum, and it should be different that > > one of the original .ko. > > As I understand it that's in the .note.gnu.build-id, and > they should be identical. It's how you can be sure they > are about the same binary. > > In the log files you show, it's also trying > /usr/lib/debug/.build-id/8d/19f404d7db2242e69c87419b8fd32f67fca584.debug > > That "19f404d7db2242e69c87419b8fd32f67fca584" is the build-id > stored in the .note.gnu.build-id. Yes, but checking also crc for modules found using gnu_debuglink (unlike those that found under .build_id) should fix the bug... And break not stripped modules :/
signature.asc
Description: Digital signature
