Your logsegment is sitting on the same device as the default/system segments (this is *normal* for the master database).
When you have log and data mixed on a device (see the 'mixed log and data' clause in the sp_helpdb output), then you'll get what appears to be some strange/funky values in the output of sp_helpsegment. The primary issue is that some of the pages will be in use by other segments, so those pages will be missing from the sp_helpsegment output.
In your case:
- total size = 26MB = this is the total size of the all the disk fragments that your logsegment is sitting on (which happens to be the entire 'master' device fragment in use by the master database)
- total pages = 6656 = 26MB / 4K (page size), ie, just another way of saying 26MB
- free pages = 128 = amount of space left on the devices that the logsegment resides on; in this case sp_helpdb is showing that you only have 512KB of space left on the master device (512K / 4K page size = 128 pages)
- used pages = 3315 = amount of pages actually in use by the logsegment
Soooo, for this particular case the numbers look ok.
As for the 3315 pages in use by the logsegment ... I'd suggest you 'dump transaction master with truncate_only' and then consider taking a database dump of the master database.
Going forward you'll want a regularly scheduled job that periodically truncates the log (to keep your master database from filling up). My preference would be a threshold on the master/logsegment segment that automatically truncs the log when the threshold is crossed.