mlock() is the solution. I was actually trying to understand the mlock system call implementation to find out how the pages are locked in the RAM so that they are not pagged out. But I could not find any such thing in code walk through of sys_mlock(). Question: Page locking is handled in software or in MMU? I don't find any bit in the PTE (Page table entry) for page saying it to be locked. Same thing applies to pci_alloc_consistent(), and kernel text, data segment memory also. Can any one guide me to understand this? Regards, Parav Pandit --- Jiri Kosina <jikos@xxxxxxxx> wrote: > On Fri, 21 Jul 2006, Mulyadi Santosa wrote: > > > > Is it possible to do some kind of page locking > or pinning from > > > application, so that kernel doesn't swap out the > page as long as > > use mmap(MAP_NORESERVE) to make the user mode > pages unswappable... > > Wrong. MAP_NORESERVE tells kernel not to reserve > space in swap space for > this mapping (which is otherwise done, so that the > page could be safely > swapped out when needed). In case you have a > MAP_NORESERVE mapping and you > try to write into it when the system is low on > memory, you will get > segfault. > > mlock() makes user pages nonswappable. But it needs > root privileges to do > so. > > -- > JiKos. > > -- > Kernelnewbies: Help each other learn about the Linux > kernel. > Archive: > http://mail.nl.linux.org/kernelnewbies/ > FAQ: http://kernelnewbies.org/faq/ > > __________________________________________________ Do You Yahoo!? Tired of spam? Yahoo! Mail has the best spam protection around http://mail.yahoo.com -- Kernelnewbies: Help each other learn about the Linux kernel. Archive: http://mail.nl.linux.org/kernelnewbies/ FAQ: http://kernelnewbies.org/faq/
