Re: kern/60456: FFS_NO_SNAPSHOT option is not entirely effective

The following reply was made to PR kern/60456; it has been noted by GNATS.

From: Taylor R Campbell <[email protected]>
To: Jason Thorpe <[email protected]>
Cc: "J. Hannken-Illjes" <[email protected]>,
	[email protected], [email protected]
Subject: Re: kern/60456: FFS_NO_SNAPSHOT option is not entirely effective
Date: Sat, 18 Jul 2026 21:58:28 +0000

 > Date: Sat, 18 Jul 2026 14:53:08 -0700
 > From: Jason Thorpe <[email protected]>
 >=20
 > > On Jul 18, 2026, at 12:51=E2=80=AFPM, Taylor R Campbell <riastradh@NetB=
 SD.org> wrote:
 > >=20
 > > Perhaps we should have an option that elides all of the file system
 > > snapshot code, but adds a small check so that if you try to mount a
 > > file system with any snapshots, it refuses to mount read/write -- only
 > > allows mounting file systems with snapshots read-only, so that you
 > > won't inadvertently corrupt the snapshots.
 >=20
 > Indeed, vfs_trans.c also contributes a non-trivial bit of .text on
 > smol machines.
 
 vfs_trans.c is for a lot more than just snapshots -- e.g., it is
 needed for unmounting or changing mount options, or for revoking
 vnodes either with revoke(2) (closing ttys) or detaching devices.
 Perhaps you can #ifdef out the cow pastures of vfs_trans.c but not the
 other parts.