Hi,
I was trying to understand the issue, but there is one point I didn't get:
In the char/mem.c driver code, the VM_RESERVED flag is set in the vma structure. The swap_out_pmd() function uses the PageReserved() macro, to check if a page structure's
PG_reserved flag is set. Did I overlook something, or is there any relation between a vma structure's VM_RESERVED flag and the page structures's PG_reserved flag.
Also I noticed, that pte_remap_page_range() which is called during execution of remap_page_range() does explicitely check the PG_reserved flag of the page to be mapped (as
shown in the subsequent lines of code).
if ((!VALID_PAGE(page)) || PageReserved(page))
set_pte(pte, mk_pte_phys(phys_addr, prot));
So why does the memory mapping with the remap_page_range() call of the char/mem.c driver succeed although it doesn't set the reserved bit of the corresponding page
structure (but only of the corresponding vm area)?
thanks for your help
regards
Martin
Nat Ersoz wrote:
> Thanks for the response. I think I understand now.
>
> -----Original Message-----
> From: Momchil Velikov [mailto:velco@fadata.bg]
> Sent: Tuesday, January 29, 2002 12:19 AM
> To: Nat Ersoz
> Cc: kernelnewbies@nl.linux.org
> Subject: Re: mmap() and remap_page_range()
>
> >>>>> "Nat" == Nat Ersoz <nat.ersoz@myrio.com> writes:
> Nat> He references the char/mem.c device driver in the linux kernel,
> Nat> and using that code provides an example of phsyical memory
> Nat> mapping to a process virtual address. The code I'm interested in
> Nat> is attached to the bottom. Specifically, I'm confused as to the
> Nat> usage of the line of code:
>
> Nat> /* Don't try to swap out physical pages.. */
> vma-> vm_flags |= VM_RESERVED;
>
> Nat> I want to do this in my own code: map a physical device page of
> Nat> memory to user space, and be sure its not swapped out. What
> Nat> confuses me is that vma->vm_flags never seems to get used by any
> Nat> of char/mem.c or any or Rubini's discussion examples.
>
> Nat> Does this code actually do anything? Can someone explain how?
>
> It does, e.g. see swap_out_pmd below:
>
> /* mm->page_table_lock is held. mmap_sem is not held */
> static inline int swap_out_pmd(struct mm_struct * mm, struct vm_area_struct * vma, pmd_t *dir, unsigned long address, unsigned long end, int count, zone_t * classzone)
> {
> [...]
> do {
> if (pte_present(*pte)) {
> struct page *page = pte_page(*pte);
>
> if (VALID_PAGE(page) && !PageReserved(page)) {
> count -= try_to_swap_out(mm, vma, address, pte, page, classzone);
> [...]
> }
> }
> address += PAGE_SIZE;
> pte++;
> } while (address && (address < end));
> mm->swap_address = address;
> return count;
> }
> --
> Kernelnewbies: Help each other learn about the Linux kernel.
> Archive: http://mail.nl.linux.org/kernelnewbies/
> IRC Channel: irc.openprojects.net / #kernelnewbies
> Web Page: http://www.kernelnewbies.org/
>
> --
> Kernelnewbies: Help each other learn about the Linux kernel.
> Archive: http://mail.nl.linux.org/kernelnewbies/
> IRC Channel: irc.openprojects.net / #kernelnewbies
> Web Page: http://www.kernelnewbies.org/
--
Supercomputing System AG email: maletinsky@scs.ch
Martin Maletinsky phone: +41 (0)1 445 16 05
Technoparkstrasse 1 fax: +41 (0)1 445 16 10
CH-8005 Zurich
--
Kernelnewbies: Help each other learn about the Linux kernel.
Archive: http://mail.nl.linux.org/kernelnewbies/
IRC Channel: irc.openprojects.net / #kernelnewbies
Web Page: http://www.kernelnewbies.org/
