Patches to 0.5
I've uploaded a couple of small patches to the incoming directory at msede.
The file lvm-0.5aed.diff has a couple of cosmetic fixes plus a major fix to tools/lib/pv_move.c to change the LE sorting because it was pretty inefficient, so I benchmarked an improved bubble sort, insertion sort, and libc5 quicksort. The insertion sort won most of the time (using data from 6 AIX disks on a real system), but as expected qsort won on disks with many LVs. I looked at the glibc qsort code, and it should beat everything, but the insertion sort code is still there (#ifdefed out)... I also changed the way pe_src was allocated to avoid hundreds of small malloc() calls. In addition, some minor changes to the LVM_TO_* macros to reduce the number of multiplies done.
The file lvm.c.diff changes the internal lvm_strcmp() type functions to be lvm_strncmp(), to avoid the chance of buffer overflow when doing a copy. Not that I had any such problems, but with kernel code, it is best to be sure.
I also have some patches in the works to allow reading both v1 and v2 structs from disk, but until 0.6 was released there was not much need for this, but I will try to get them out.
Cheers, Andreas
Signature · 4 lines
Andreas Dilger University of Calgary \"If a man ate a pound of pasta and
Micronet Research Group \ a pound of antipasto, would they
Dept of Electrical & Computer Engineering \ cancel out, leaving him still
http://www-mddsp.enel.ucalgary.ca/People/adilger/ hungry?" -- Dogbert