public inbox for linux-kernel@vger.kernel.org
 help / color / mirror / Atom feed
* Too big EBDA issue
@ 2002-01-25 16:58 ASA
  2002-01-25 17:38 ` H. Peter Anvin
  0 siblings, 1 reply; 2+ messages in thread
From: ASA @ 2002-01-25 16:58 UTC (permalink / raw)
  To: linux-kernel

Hello all,

My system is embedded PC/104 and has C&T 65545 videochip and DiskOnChip
flash device. I'm developing a special linux-based application.

Today I had to upgrade DiskOnChip BIOS extender and after that I could not
boot linux anymore. After digging hard in problem I found that EBDA was
enlarged to 33KB so remaining conventional memory was reduced to 607KB but
normal booting proccess bzImage loading requires at least 608 KB. After
checking on other systems with DiskOnChip I found their EBDA have sizes
typically of 29-31 KB.

Yeah, it is very large EBDA (normal PC's I checked just have only 1 KB
EBDA). It seems DickOnChip BIOS requires much space on irder to store own
temporary data to implement their TrueFFS.

But I guess that there will be some other BIOS extensions that will require
another EBDA space. As far as bzImage loading model requires space of 32 K
between 576K (0x90000) and 608K (0x98000) but almost no other place I think
there is necessity to extend boot protocol in order to relocate 16-bit mode
loader closer to the lowest memory bound, not to the upper one.

I also reported that issue to DiskOnChip developers (www.m-sys.com) but
there is a possibility that other hardware developers can extend EBDA also.

So extending boot protocol in order to move far away of cancer of growing
EBDA would be worthly of note. As a new feature for 2.5/2.6 kernels by
example.

-- 
Best regards,
 ASA                          mailto:llb@udm.net.ru


^ permalink raw reply	[flat|nested] 2+ messages in thread

end of thread, other threads:[~2002-01-25 17:39 UTC | newest]

Thread overview: 2+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2002-01-25 16:58 Too big EBDA issue ASA
2002-01-25 17:38 ` H. Peter Anvin

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox