Skip to contentSkip to conversation

Page table mapped and identity segments!

DISCUSSION Ameya Mitragotri 3 participants 3 messages Nov 4, 2003

lore
Share & tools
lore
#0 lore raw

Hello,
I am refering to David's and Stephanes's book on the IA64 linux kernel. From my understanding,

1) I can see that the PAGE_OFFSET which designates the start of the identity mapped segment starts at 0xe000000000000000 (region 7).

2) This region has the linux kernel text/data/stack segments.

Now after compiling the linux-2.6.0-test5 on an itanium-2 machine, the System.map generated contains addresses only starting with 0xa0000..which means all these lie in the page table mapped segments (region 5).

Why is this happening?

Apologies if this is too trivial to be asked.

Thanks
Ameya

#1 lore raw
Quoted 1 line · Ameya Mitragotri

I am refering to David's and Stephanes's book on the IA64 linux kernel.

From my understanding,

Quoted 10 lines

1) I can see that the PAGE_OFFSET which designates the start of the identity mapped segment starts at 0xe000000000000000 (region 7).

2) This region has the linux kernel text/data/stack segments.

Now after compiling the linux-2.6.0-test5 on an itanium-2 machine, the System.map generated contains addresses only starting with 0xa0000..which means all these lie in the page table mapped segments (region 5).

The book refers to the 2.4 and earlier kernels. In 2.6 kernels the kernel code and data lives in region 5, not 7.

#2 lore raw
In reply to Ameya Mitragotri
Quoted 2 lines · Ameya Mitragotri

Hello,
I am refering to David's and Stephanes's book on the IA64 linux kernel.

From my understanding,

Quoted 4 lines

1) I can see that the PAGE_OFFSET which designates the start of the identity mapped segment starts at 0xe000000000000000 (region 7).

Quoted 9 lines

2) This region has the linux kernel text/data/stack segments.

Now after compiling the linux-2.6.0-test5 on an itanium-2 machine, the System.map generated contains addresses only starting with 0xa0000..which means all these lie in the page table mapped segments (region 5).

Why is this happening?

Kernel mapping has changed in the 2.6 kernel. The kernel is now relocatable (needed for new hardware). Because of that it was decided to move the kernel outside the identity mapped region. But the virtual to physical mappings are nevertheless done with the old mechanism of pinned large 64M pages for text and data. See head.S where these mappings are done.

Christian

Quoted 14 lines

Apologies if this is too trivial to be asked.

Thanks
Ameya

-
To unsubscribe from this list: send the line "unsubscribe linux-ia64" in the body of a message to [email protected]
More majordomo info at http://vger.kernel.org/majordomo-info.html