Re: Kernel BOOT (kernel 2.4.16)

Re: Kernel BOOT (kernel 2.4.16)

[LEROY Christophe]

Dan Malek <dan@embeddededge.com> wrote:
>LEROY Christophe wrote:
>
>
>> Can someone help me ?
>> io_block_mapping(IMAP_ADDR,...) fails because mem_pieces_find() can find

>> any piece.
>
>What code are you looking at?	The io_block_mapping is simply going to
>force a map_page() operation and update the PTE tables.  How do you see
>mem_pieces_find() involved in this path?
>
>Thanks.
>

map_page() calls pte_alloc()
pte_alloc() calls pte_alloc_one()
pte_alloc_one() calls early_get_page()
early_get_page() calls mem_pieces_find()
I'm sure that it come here because when I add a call to m8xx_restart() just
before the panic() call in mem_pieces_find(), the board restarts.

Thanks.

C.Leroy


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

Re: Kernel BOOT (kernel 2.4.16)

[Dan Malek]

LEROY Christophe wrote:


> I'm sure that it come here because when I add a call to m8xx_restart() just
> before the panic() call in mem_pieces_find(), the board restarts

Then something is wrong with the way your bootloader is telling the
kernel how much memory exists, although mapin_ram() should have
failed before you attempted any io mapping.


	-- Dan


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

Re: Kernel BOOT (kernel 2.4.16)

[LEROY Christophe]

Dan Malek <dan@embeddededge.com> wrote:
>LEROY Christophe wrote:
>
>
>> because it cannot find any piece for IMAP_ADDR,   as set_phys_avail()
>> only put map for RAM in phys_avail table.
>
>That's correct operation.  Only real memory is supposed to be in the
>phys_avail table.
>

Ok. So.

Can someone help me ?
io_block_mapping(IMAP_ADDR,...) fails because mem_pieces_find() can find
any piece.

How ever I cannot understand how it could find any free piece, because
mapin_ram has taken all free pieces i think.

I need some guidance here to understand what happens.

Thanks

C.Leroy


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

Re: Kernel BOOT (kernel 2.4.16)

[Dan Malek]

LEROY Christophe wrote:


> Can someone help me ?
> io_block_mapping(IMAP_ADDR,...) fails because mem_pieces_find() can find
> any piece.

What code are you looking at?  The io_block_mapping is simply going to
force a map_page() operation and update the PTE tables.  How do you see
mem_pieces_find() involved in this path?

Thanks.


	-- Dan


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

Kernel BOOT (kernel 2.4.16)

[LEROY Christophe]

Hello,

I'm trying to boot a 2.4.16 linux kernel on a MPC860 based board,
and it hangs up early in
MMU_init().....m8xx_map_io()....io_block_mapping(IMAP_ADDR,...).....mem_pieces_find()....

because it cannot find any piece for IMAP_ADDR,   as set_phys_avail()
only put map for RAM in phys_avail table.

Can someone help me ?

Thanks

C.Leroy

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