get_pgd_slow() function

Borlizzi Giacomo giacomo.borlizzi at tei.ericsson.se
Fri Feb 9 03:08:07 EST 2001


I'm using an 860 processor running at 80MHz.
The problem that I have is that the shell don't start.

During the debugging phase made on the 2.2.13
the code executed the function exec_mmap(). Inside on it was
executed the macro SET_PAGE_DIR and after this execution the code
was crashed.
Investigating on that I discovered that code was crash immediately when
the
MTWB register was changed.
.....
That because MMU interrupt routine was accessing to some bad memory area.

Investigating on reasons for that I have made the Hypothesis that MMU
table was not well initialized, from this I made some changes on the
get_pgd_slow function.

Some day ago during the porting on linux-2.4.1 the same problem happens
so now it is time to resolve and understand why it happen

I like your hypothesis on cache, I have the cache configured with
COPYBACK
disabled so:

# CONFIG_8xx_COPYBACK is not set

and I'll try to change it....

But many doubt remaining....

Ciao,
/Giacomo

Dan Malek wrote:

> Borlizzi Giacomo wrote:
>
> > but I've modified with:
>
> What made you think you need to do this?
> Something isn't right.  The 'memcpy' you added just copies zero values
> from the init pgd into the new pgd, which the 'memset' already did
> for you.  Either the 'memset' you are using isn't correct, or the
> act of copying that much data flushed cache lines.  What kind of
> processor do you have?
>
>         -- Dan


** Sent via the linuxppc-embedded mail list. See http://lists.linuxppc.org/






More information about the Linuxppc-embedded mailing list