On Thu, Nov 19, 2015 at 02:35:59PM -0500, Dave Anderson wrote:
>
>
> ----- Original Message -----
> >
> > Below is a (hardly tested, pretty messy, incomplete - needs to address
> > other architectures verify_symbol() prototypes) patch that illustrates
> > this idea. I'm OK with special casing _kernel_flags_le, but generally
> > prefer to look for general solutions. Let me know if you like anything
> > in the patch below, if so, then I can clean it up and send it properly.
> > The patch takes care of the two cases pointed above, i.e. 'sym -l' and
> > 'rd -S'.
> >
> > Thanks,
> > drew
>
> I appreciate the effort, but I really don't want to make this into a
> project that seeps into the other architectures, or into the general symbol
> handling code. And although it probably wouldn't be an issue in this
> case, I don't like to change interfaces defined in defs.h because they
> can-be/are used by extension modules, and something like this could
> conceivably break one.
>
> All of the architectures have their own xxxx_verify_symbol() function
> to perform whatever machine-specific hacks that they require. And since
> it's a simple STREQ("_kernel_flags_le") check, please let's just keep it
> in-house so to speak.
>
> Thanks,
> Dave
It's certainly easier to do the STREQ("_kernel_flags_le") check, but
then 'sym some-interesting-absolute' won't work in crash. I suppose it's
pretty unlikely that that will ever matter except for this one case though,
so I'll send a v2 stashing _kernel_flags_le in a new member of
machdep->machspec later today or tomorrow.
Thanks,
drew
--
Crash-utility mailing list
Crash-utility@xxxxxxxxxx
https://www.redhat.com/mailman/listinfo/crash-utility
