Ashish Joshi wrote: > James Carlson wrote: >> Gary Winiger writes: >> >>> Evolving to Uncommitted is requested in light of a future planned but, >>> not yet scheduled, project to integrate the device allocation >>> functionality >>> under Tamarack: Removable Media Enhancements in Solaris (PSARC/2005/399) >>> for all of the Solaris configurations. >> >> >> Does this mean that Sun Ray will need and get a contract? I don't see >> one in the case directory. >> > > > Because Sun Ray is using now Uncommitted interfaces? I guess. I'll > check with Gary when he's back (on Monday?). > > >>> + On a system configured with Trusted Extensions, the >>> + following additional exit values are returned: >>> + >>> + 3 Mounting of device failed. Caller shall not >>> + place device in error state. >>> + >>> + 4 Mounting of device succeeded. >> >> >> This seems a little over-complicated. I would suggest returning 0 for >> the mount-succeeded case and adding just: >> >> 3 Allocation succeeded, but mounting failed. Media may >> be in unformatted state. Caller should not assume >> device has failed. >> >> (In other words, I don't see a difference between codes 4 and 0.) >> > > Mounting/unmounting during allocation/deallocation (done through > device_clean) is done only in TX. > In TX, if allocate does not mount the device (for non-mountable > devices, and for mountable devices for which user opts to not mount > them), allocate needs to create device nodes in the non-global zone in > which the device is being allocated. > So, there needs to be a distinction between general success (0) and > mount success (4) because allocate needs to know that it has to create > device nodes in the non-global zone after general success from > device_clean, but not after mount success from device_clean.
Yes, we were running into the situation where the user plugs in a USB device that doesn't not have a mountable filesystem on it, and without the ability for allocate to be told the difference between "device error" and "device is OK, but just doesn't have anything that the system can mount", these devices were all winding up in the device error state which required admin intervention to clean up. mike
