On Tue, Jun 16, 2009 at 4:19 AM, Scott Rohling<[email protected]> wrote: > Ditto.. especially coming from the wonders of z/VM link modes (e.g. MR > seems like what we want here) as we VMers have..
With all respect, the VM link mode has nothing to do with this. That's merely part of why you ended having the disk RW or RO. > On Mon, Jun 15, 2009 at 6:58 PM, Richard Troth <[email protected]> wrote: > >> I have never understood why the driver does not automatically set the >> R/O flag when a write operation is rejected. That would be late. At that point the page cache holds a page with (different) contents that is presented to processes as the contents of that block on disk, which it maybe is not. Given the lazy write, people would already have been using the wrong data for 30 seconds before you get notified. There's nothing in the S/390 I/O architecture that supports the concept of R/O disks (unlike tapes). It would be possible to create such an interface and define a bit in the device info diagnose that Linux is using already (to parse the output of Q V is not really nice). Clearly this would only work for Linux on z/VM, but the issue does not really apply to LPAR anyway. But maybe Carsten still knows why we did not think that was an option. Rob ---------------------------------------------------------------------- For LINUX-390 subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO LINUX-390 or visit http://www.marist.edu/htbin/wlvindex?LINUX-390
