RE: [Linux-ia64] Re: [PATCH] head.S fix for unusual load addrs

From: Luck, Tony <tony.luck_at_intel.com>
Date: 2003-05-16 02:33:12
> The patch looks mostly fine to me.  I'm not too fond of the __tpa()
> and __imva() macros, but that may be mainly a matter of preference.
> What I definitely don't like is that the casting seems rather confused
> and that the patch is adding __tpa() when we already have ia64_tpa().
> As an example of the casting issues: __imva() returns a long, but at
> times it's cast to "unsigned long" which doesn't make a lot of sense
> (for assignments, anyhow).  Moreover, we should stick to the Linux
> principle that kernel-space pointers have a type of "void *".

I can get rid of the __tpa() easily by using ia64_tpa() ... I'll
fix the type-casting fiasco while  making those changes.

Do you have some direction for the __imva() macro?  The acronym
stands for "Identity Mapped Virtual Address", and its purpose is
to provide the region 7 address for an object, so that the existing
code can continue to work without a whole lot of run-on changes.
If you just don't like the name, then it's easy to change to something
else.  If it's the typecast issue, then I can switch over to void *.
If it just makes your head hurt coping with schizophrenic dual
mapping that the kernel gets with this patch, then you'll just have
to take a couple of aspirin and look at the patch again in the
morning :-)

-Tony
 
Received on Thu May 15 12:12:21 2003

This archive was generated by hypermail 2.1.8 : 2005-08-02 09:20:14 EST