Re: [linuxsh-dev] Question on SH/kernel discrepancies

[NIIBE Y. <gn...@m1...>]

Jesper Skov wrote:
 > NIIBE> I believe it's needed for SH-4.  SH-4 has virtually indexed and
 > NIIBE> physically tagged cache system with write back.  For such
 > NIIBE> architecture the changes are required.
 > 
 > Such changes should be wrapped in #ifdef/#endif to make it clear why
 > it differs from the main sources. Otherwise it'll just be considered a
 > merge bug at some time and get ditched.

Let me explain clearly.  It is *NOT* SH-4 specific bug.  It is the bug
which is hit by SH-4, but the fix is *ALWAYS* needed, always.  I've
commit the bug report to Bugzilla at linux-mm.org today.

 > Same goes for the SolutionEngine hack.

For SolutionEngine hack, I agree.  While IDE-timing patch will be
introduced, we can clean it up.
--