linux-mips-fnet
[Top] [All Lists]

Re: TLB exception

To: Rabeeh Khoury <rabeeh@galileo.co.il>
Subject: Re: TLB exception
From: Ralf Baechle <ralf@uni-koblenz.de>
Date: Mon, 5 Jun 2000 21:14:11 +0200
Cc: "linux-mips@fnet.fr" <linux-mips@fnet.fr>
In-reply-to: <393C05C7.3E8EF447@galileo.co.il>; from rabeeh@galileo.co.il on Mon, Jun 05, 2000 at 03:55:51PM -0400
References: <393C05C7.3E8EF447@galileo.co.il>
On Mon, Jun 05, 2000 at 03:55:51PM -0400, Rabeeh Khoury wrote:

> I'v added a code to debug the TLB misses in my kernel (2.2)
> I'm running QED RM5271 in big mips mode ; I'v added the following code :
> 
>     8004de54:   3c010300        lui     $at,0x300
>     8004de58:   ac200000        sw      $zero,0($at)
> 
> which simply writes zero to virtual address 0x03000000, this code is
> invoked in kernel mode in kernel space and tries to write to user space.
> 
> after this I get a TLB exception with an error TLBL (TLB miss because of
> load or instruction fetch) !!!!!!!!!!!!!!!
> instead of TLBS.

Do you observe this tlbl exception code in or via the general exception
vector handler at 0x80000180?  If so you're taking a exception within
the other exception handler at 0x80000000.  There are special rules for
this nested exception that is the EPC and BadVAddr registers will keep
their old values.

> the CP0_STATUS = 00800008
> p.s. I have also strange values in the BADVADDR and CONEXT.

What is CONEXT?

> can any one speculate what is the problem ?
> do I need to do any thing special to access user space in kernel mode ?

There is nothing obviously missing.

  Ralf

<Prev in Thread] Current Thread [Next in Thread>
  • TLB exception, Rabeeh Khoury
    • Re: TLB exception, Ralf Baechle <=