Re: [Question] st_blocks of shm is number of pages, not 512B blocks

On Sat, Aug 15, 2026 at 10:59:24AM +0900, David Timber wrote:
> On 8/2/26 22:54, Konstantin Belousov wrote:
> > The page size as st_blocks multiplier is natural because shm is allocated
> > in pages.  I wanted to return the number of large pages for st_blocks for
> > the largepage shmfds as well, but somebody argued against it AFAIR.
> Perhaps shm can implement fstatfs() and return whatever largepage
> decided in f_bsize? Just a thought. Seems to me that the information is
> lost from userspace after creating the shm. I wonder if there's a
> standard interface to query it back after creating it. If not, f_bsize
> seems to be the candidate. Just a thought.
There is no filesystem to perform fstatfs() on.

BUt there is ioctl(shmfd, FIOGSHMLPGCNF, &struct shm_largepage_conf)
which returns psind, the index of the largepage size in the array
returned by getpagesizes().

> 
> Anyway, I wasn't really nitpicking why FreeBSD does things the way it
> does. I was just wondering if there's actual use case in userspace where
> returning st_blocks in system page size unit makes more sense. By that I
> mean, yes, the page size IS the allocation unit so it makes perfect
> sense. It saves division and multiplication to do the alignment dance.
> 
> Me thinks that not being aligned to the usual 512-byte unit with the
> rest of the implementation does more harm than good...

Ok, https://reviews.freebsd.org/D58942