On Tuesday, December 06, 2005 08:16:52 PM -0500 chas williams - CONTRACTOR <[EMAIL PROTECTED]> wrote:
i suspect the author of probe syscall will have a better idea about possible fixes.
That would be me. And indeed, the change in section is going to be a problem, since on x86 the scanner will expect the sys_call_table to be in the same section as init_mm, which will no longer be the case. What we need to do is find some exported symbol in the same section, preferably near the beginning.
Given a new symbol, there are a couple of approaches we can take to determine when to use it. One is to simply always use the new symbol on kernels 2.6.14 and newer. The other is to extend the control structure so we can give the probe code more than one section to look at.
it also looks like it might be time to start using the keyring. the ia64 does have a readonly sys_call_table that cannot be modified.
Well, does 2.6.14 have the exported symbols we're waiting for in the keyring interface?
-- Jeff _______________________________________________ OpenAFS-devel mailing list [email protected] https://lists.openafs.org/mailman/listinfo/openafs-devel
