DragonFly BSD
DragonFly kernel List (threaded) for 2004-06
[Date Prev][Date Next]  [Thread Prev][Thread Next]  [Date Index][Thread Index]

Re: ATA interrupt code broken in new ISO?


To: Matthew Dillon <dillon@xxxxxxxxxxxxxxxxxxxx>
From: David Xu <davidxu@xxxxxxxxxxxxxx>
Date: Wed, 30 Jun 2004 09:14:58 +0800

Matthew Dillon wrote:
:I'm trying to find out if the current ISO for download has the broken ATA
:interrupt code or not.
:
:jm

I haven't generated another ISO yet. I'll do so tonight.

   I think the 'trick' for the boot code is going to be to relocate it to
   high memory right off the bat, and possibly also to get rid of the
   BTX junk.  BTX offers an ability to debug the boot code and the early
   kernel boot, but it doesn't play very nice with BIOSes.

   Unfortunately, relocating it is going to require a major rewrite. I
   actually was able to rewrite boot0 today, but it took all day, and its
   going to take even longer to rewrite boot1, boot2, and the loader.  There
   are about a dozen different hardwired addresses that need to be relocated
   and most of the existing code is not segment-friendly (which is why it took
   so long for me to rewrite boot0).

   Bad, on a sad note, there is just no way any of this stuff will make it
   to the first release, because we absolutely have to be released by
   Sunday.  Once USENIX is over I will take up the boot code again.  We are
   just going to have to release with the mess that we inherited from
   FreeBSD-5.

Is dfbsd kernel still loaded at physical 1M address ?
Intel document says first 4M phyiscal mem may have performance hit
with MTRR turned on, FreeBSD moved it to 4M if I remembered it
correctly.

-Matt
Matthew Dillon <dillon@xxxxxxxxxxxxx>

David Xu





[Date Prev][Date Next]  [Thread Prev][Thread Next]  [Date Index][Thread Index]