On Mon, Sep 09, 2002 at 02:04:28PM -0600, Joseph A. Knapka wrote: > Haijin Yan wrote: > > > > Hi, > > when I was reading linux MM, I noticed the concept of page cache, buffer > > cache, and swap cache. I am confused about the followings: > > > > page->mapping points to the inode's address space(i_mapping), or NULL if > > it is anonymous page. so is this the concept of page cache? No global page > > cache but each inode has a page cache? > > > > what is the meaning of page->buffers? > > IIRC, it points to the list of buffer heads describing the disk > location of the data resident in the page. Buffer heads are the > structs used to schedule traditional (read()/write()) disk > I/O, whereas page->mapping is used for mmap() I/O. So a > "buffer" is just a logical block of data within a page > that correponds to a physical disk block somewhere; and > the buffer head is a description of that relationship. In the > older 2.4 kernels, they were complimentary: when a mmap()ed page > had to be written out, the VM created buffers describing the page > and then handed those buffers off to the disk I/O subsystem. I'm > not sure if things work the same now. (Probably so.) Each page in page cache belongs to an inode or is anonymous. But page_launder() writes the pages out independently of what they belong to. So it one cache. Mappings are there so page_launder knows where to write the page (it calls down to writepage method in mapping). With buffers, IIRC it's still the same way. When a page has to be written to disk, it's turned into buffers and these are queued for disk write. > > In page_launder() it was checked to > > under different context for different purpose, can someone kindly points > > how many cases are there and what they stand for? > > I don't know which version of the code you're looking at, but > basically what's going on is that if the page has buffers and > needs to be written out, page_launder() can hand it to disk I/O > immediately; if no buffers, it must create them before the > page can be written out. If the page is clean and has buffers, > we need to destroy them before reclaiming the page (which is > done either in page_launder or in reclaim_page in 2.4.<=5). > > > In linux 2.4, document said buffer cache and page cache are unified, > > where can I find evident in src code? > > I think that's what I was referring to above - all I/O is > handled by the disk buffer mechanism; there's no special-case > code for doing I/O on pagecache pages. Each mapping that can ever have a dirty page has a writepage method. This method is called from page_launder and it must either create and schedule buffers for the page, or clean it up by other means (network filesystems send a network request and no buffers are ever created). ------------------------------------------------------------------------------- Jan 'Bulb' Hudec <bulb@ucw.cz> -- Kernelnewbies: Help each other learn about the Linux kernel. Archive: http://mail.nl.linux.org/kernelnewbies/ FAQ: http://kernelnewbies.org/faq/
