| commit | b84ae4398f90e0bda5b61cdf08cdb67ea840abe0 | |
| author | Tim Abbott <tabbott@mit.edu> | |
| Mon, 15 Dec 2008 18:20:16 +0000 (15 13:20 -0500) | ||
| committer | Anders Kaseorg <andersk@mit.edu> | |
| Fri, 19 Dec 2008 19:13:43 +0000 (19 14:13 -0500) | ||
| tree | a3a71149ecba3681df197524a1112a60f499b569 | treesnapshot (tar.gz zip) |
| parent | 36919423b1af8cde6900bc1a8a1b521b21ab921f | commitdiff |
Make sure that one of the other_sect offsets is 0.
This is a hack to work around the fact that our .fixup splitting
algorithm is not correct when there are direct relocations (i.e. not
from __ex_table) into .fixup. In recent amd64 kernels, there are
direct relocations to the bad_from_user symbol at the start of the
.fixup section in arch/x86/lib/copy_user_64.S, and no indirect
relocations to that same place.
Signed-off-by: Tim Abbott <tabbott@mit.edu>
This is a hack to work around the fact that our .fixup splitting
algorithm is not correct when there are direct relocations (i.e. not
from __ex_table) into .fixup. In recent amd64 kernels, there are
direct relocations to the bad_from_user symbol at the start of the
.fixup section in arch/x86/lib/copy_user_64.S, and no indirect
relocations to that same place.
Signed-off-by: Tim Abbott <tabbott@mit.edu>
| objmanip.c | diffblobblamehistory |