linux-mm.kvack.org archive mirror
 help / color / mirror / Atom feed
* Why the kernel needs `split_mem_range` to split the physical address range?
@ 2018-04-01  7:01 Hao Lee
  2018-04-01 13:06 ` Matthew Wilcox
  0 siblings, 1 reply; 3+ messages in thread
From: Hao Lee @ 2018-04-01  7:01 UTC (permalink / raw)
  To: linux-mm

Hi,

I'm currently studying the memory management subsystem. When I read
the code of x86_64 memory mapping initialization, I encounter a
problem and can't find answers on Google.

I wonder why the kernel needs `split_mem_range()`[0] to split physical
address range. To make this question clear, I find an example from
dmesg. The arguments of `split_mem_range` are start=0x00100000,
end=0x80000000. The splitting result is:

[mem 0x00100000-0x001FFFFF] page 4k
[mem 0x00200000-0x7FFFFFFF] page 2M

I don't know why the first 1MiB range is separated out to use 4k page
frame. I think these two ranges can be merged and let the range
[0x00100000-0x7FFFFFFF] use 2M page frame completely. I can't
understand the purpose of this range splitting. Could someone please
explain this function to me? Many Thanks!

[0] https://github.com/torvalds/linux/blob/10b84daddbec72c6b440216a69de9a9605127f7a/arch/x86/mm/init.c#L325

Regards,
Hao Lee

^ permalink raw reply	[flat|nested] 3+ messages in thread

end of thread, other threads:[~2018-04-02 11:25 UTC | newest]

Thread overview: 3+ messages (download: mbox.gz / follow: Atom feed)
-- links below jump to the message on this page --
2018-04-01  7:01 Why the kernel needs `split_mem_range` to split the physical address range? Hao Lee
2018-04-01 13:06 ` Matthew Wilcox
2018-04-02 11:25   ` Hao Lee

This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox