Jan, I've completed modifications to disk_parts.c so that the AI starting sector is not overwritten in GUI code.
I created a separate code path for AI and GUI in om_validate_and_resize_partitions(), since it was becoming overly complicated to have a single path. Ran spell(1) on some source files and corrected spelling. Added a cast when multiplying 32-bit values to compute a 64-bit value to prevent data loss. Tested AI: - tested new Solaris partition with best fit algorithm when best fit comes at start, middle, and end w.r.t. existing partitions - tested in absence of space requested in manifest Tested changes against GUI: - varied partition configuration before running GUI - tested deletion of DOS partitions in GUI Please review if you are still able. William jan damborsky wrote: > Hi William, > > I have just started to look at those code > changes. I would like to get back to you > tomorrow. > > Jan > > > William Schumann wrote: >> When a partition is created as specified in the AI engine manifest, >> the AI >> engine can determine the placement of the partition based on the unused >> regions in the partition map. The smallest free region accommodating >> the new >> partition will be automatically selected. Previously, the starting >> sector was required. >> >> http://cr.opensolaris.org/~wmsch/bug-5656/ >> http://defect.opensolaris.org/bz/show_bug.cgi?id=5656 >> >> The automatic placement can be indicated simply by not specifying a >> starting >> sector for the partition in the manifest: element partition_start_sector >> >> Changed static functions used by >> om_validate_and_resize_disk_partitions() that >> were assuming sorted partition tables. >> >> Added additional parameter to >> om_validate_and_resize_disk_partitions() to >> specify whether or not to remove gaps between partitions. Gaps will >> now be >> compressed for the GUI, but not for AI. >> >> Also, code for http://defect.opensolaris.org/bz/show_bug.cgi?id=5655 >> appears, but will not be invoked. This enhancement should be >> available soon. >> >> Tested by creating different combinations of FAT32 partitions with >> fdisk before >> using AI to create Solaris partition: >> - leaving free space of varying sizes between them >> - leaving free space at end, middle, beginning of disk >> - denying free space of requested size >> - exercised om_validate_and_resize_disk_partitions() functions >> - saved output of fdisk -W for testing identical conditions after >> code changes >> >> Made AI manifest element partition_number optional, since it is >> ignored when >> creating partitions. >> _______________________________________________ >> caiman-discuss mailing list >> caiman-discuss at opensolaris.org >> http://mail.opensolaris.org/mailman/listinfo/caiman-discuss >
