On September 15, 2001 10:52 pm, Ed Tomlinson wrote:
> On September 15, 2001 10:07 pm, dwchin@umich.edu wrote:
> > Oh, a couple of details: I'm running RedHat 7.1 (except for the kernel
> > update as mentioned in previous message). And gdb says that lvcreate is
> > getting a segmentation fault at lv_create_remove.c:42
> >
> > 42 inline int lv_create ( vg_t *vg, lv_t *lv, char *lv_name) {
>
> I suspect this _might_ have something to do with the mix/max changes in
> 2.4.10pre9. In 2.4.10pre Linus changed min/max to use three args - he had
> good reasons to do this. It created lots of flack. After days (weeks?) of
> debate a two arg version he accepted was found. Meanwhile the rest of the
> kernel (and patches like LVM) had adpated to the three arg mix/max.
>
> The jist of the above it to look at the LVM patch generated for 2.4.10pre9
> and change and lines begining with "+" that contain "min(" or "max(" to
> "min_t(" or "max_t(" respectivily.
>
> Suspect then it will work - I am testing this here now.
And it works fine... Here are the fragraments changed.
+ if (old_lv->lv_stripes < 2) { /* linear logical volume */
+ end = min_t(uint, old_lv->lv_current_le, new_lv->lv_current_le);
+ for (l = 0; l < end; l++) {
+ new_stripe_size = new_lv->lv_allocated_le / new_lv->lv_stripes;
+ end = min_t(uint, old_stripe_size, new_stripe_size);
+
+ for (i = source = dest = 0; i < new_lv->lv_stripes; i++) {
- min_blksize = min(blksize_org, blksize_snap);
+ max_blksize = max_t(int, blksize_org, blksize_snap);
+ min_blksize = min_t(int, blksize_org, blksize_snap);
max_sectors = KIO_MAX_SECTORS * (min_blksize>>9);
- nr_sectors = min(chunk_size, max_sectors);
+ nr_sectors = min_t(int, chunk_size, max_sectors);
chunk_size -= nr_sectors;
- buckets = min(buckets, max_buckets);
+ buckets = min_t(unsigned long, buckets, max_buckets);
while (buckets & (buckets-1))
Luck.
Ed Tomlinson
