Reason for TEA machine check exception

キャンセル
次の結果を表示 
表示  限定  | 次の代わりに検索 
もしかして: 

Reason for TEA machine check exception

ソリューションへジャンプ
1,712件の閲覧回数
mditto
Contributor I
I am using a custom MPC8248 board running Linux 2.4.  I occasionally get a
machine check exception when trying to read from an address that should be
mapped to PCI space.  The reason is supposedly a transfer error acknowledge
(SRR1=41030).  As far as I can tell from the 8272 reference maual, this means
that the !TEA signal was asserted during the bus cycle, but on this board that
pin is pulled up to VCC.  Is there some other condition that causes a TEA
machine check?

Thanks in advance.

Mike

タグ(1)
0 件の賞賛
返信
1 解決策
1,006件の閲覧回数
mditto
Contributor I
(Answering my own message for anyone interested.)

It seems that TEA is internally asserted by the PCI bridge when it raises a
machine check exception, or at least it fills in SRR1 as if there was a TEA.
The PCI registers, such as PCI_ECCR, can be consulted to get the
details about the PCI bus error in this case.

元の投稿で解決策を見る

0 件の賞賛
返信
1 返信
1,007件の閲覧回数
mditto
Contributor I
(Answering my own message for anyone interested.)

It seems that TEA is internally asserted by the PCI bridge when it raises a
machine check exception, or at least it fills in SRR1 as if there was a TEA.
The PCI registers, such as PCI_ECCR, can be consulted to get the
details about the PCI bus error in this case.

0 件の賞賛
返信