On Wed, Sep 24, 2025 at 12:28:22PM +0200, Martin Pieuchot wrote:
> Diff below is a small refactoring about UVM vnode descriptors.  These
> descriptors are used to keep track of mmap(2)ed vnodes.  They are
> currently created for each and every vnode.
> 
> This introduces a new function to do that and move the pool init in the
> right place.
> 
> ok?

On minor comment below.
 
> Index: kern/vfs_subr.c
> ===================================================================
> RCS file: /cvs/src/sys/kern/vfs_subr.c,v
> diff -u -p -r1.331 vfs_subr.c
> --- kern/vfs_subr.c	20 Sep 2025 13:53:36 -0000	1.331
> +++ kern/vfs_subr.c	24 Sep 2025 09:45:08 -0000
> @@ -67,7 +67,6 @@
>  
>  #include <netinet/in.h>
>  
> -#include <uvm/uvm_extern.h>
>  #include <uvm/uvm_vnode.h>
>  
>  #include "softraid.h"
> @@ -129,7 +128,6 @@ void printlockedvnodes(void);
>  #endif
>  
>  struct pool vnode_pool;
> -struct pool uvm_vnode_pool;
>  
>  static inline int rb_buf_compare(const struct buf *b1, const struct buf *b2);
>  RBT_GENERATE(buf_rb_bufs, buf, b_rbbufs, rb_buf_compare);
> @@ -154,8 +152,6 @@ vntblinit(void)
>  	maxvnodes = 2 * initialvnodes;
>  	pool_init(&vnode_pool, sizeof(struct vnode), 0, IPL_NONE,
>  	    PR_WAITOK, "vnodes", NULL);
> -	pool_init(&uvm_vnode_pool, sizeof(struct uvm_vnode), 0, IPL_NONE,
> -	    PR_WAITOK, "uvmvnodes", NULL);
>  	TAILQ_INIT(&vnode_hold_list);
>  	TAILQ_INIT(&vnode_free_list);
>  	TAILQ_INIT(&mountlist);
> @@ -424,9 +420,7 @@ getnewvnode(enum vtagtype tag, struct mo
>  	    ((TAILQ_FIRST(listhd = &vnode_hold_list) == NULL) || toggle))) {
>  		splx(s);
>  		vp = pool_get(&vnode_pool, PR_WAITOK | PR_ZERO);
> -		vp->v_uvm = pool_get(&uvm_vnode_pool, PR_WAITOK | PR_ZERO);
> -		vp->v_uvm->u_vnode = vp;
> -		uvm_obj_init(&vp->v_uvm->u_obj, &uvm_vnodeops, 0);
> +		uvm_vnp_obj_alloc(vp);
>  		RBT_INIT(buf_rb_bufs, &vp->v_bufs_tree);
>  		cache_tree_init(&vp->v_nc_tree);
>  		TAILQ_INIT(&vp->v_cache_dst);
> Index: uvm/uvm_vnode.c
> ===================================================================
> RCS file: /cvs/src/sys/uvm/uvm_vnode.c,v
> diff -u -p -r1.140 uvm_vnode.c
> --- uvm/uvm_vnode.c	15 Aug 2025 08:21:41 -0000	1.140
> +++ uvm/uvm_vnode.c	24 Sep 2025 09:44:36 -0000
> @@ -47,6 +47,7 @@
>  
>  #include <sys/param.h>
>  #include <sys/systm.h>
> +#include <sys/pool.h>
>  #include <sys/proc.h>
>  #include <sys/malloc.h>
>  #include <sys/vnode.h>
> @@ -67,7 +68,7 @@
>   * we keep a list of writeable active vnode-backed VM objects for sync op.
>   * we keep a simpleq of vnodes that are currently being sync'd.
>   */
> -
> +struct pool			uvm_vnode_pool;
>  LIST_HEAD(, uvm_vnode)		uvn_wlist;	/* [K] writeable uvns */
>  SIMPLEQ_HEAD(, uvm_vnode)	uvn_sync_q;	/* [S] sync'ing uvns */
>  struct rwlock uvn_sync_lock;			/* locks sync operation */
> @@ -113,7 +114,8 @@ const struct uvm_pagerops uvm_vnodeops =
>  void
>  uvn_init(void)
>  {
> -
> +	pool_init(&uvm_vnode_pool, sizeof(struct uvm_vnode), 0, IPL_NONE,
> +	    PR_WAITOK, "uvmvnodes", NULL);
>  	LIST_INIT(&uvn_wlist);
>  	/* note: uvn_sync_q init'd in uvm_vnp_sync() */
>  	rw_init_flags(&uvn_sync_lock, "uvnsync", RWL_IS_VNODE);
> @@ -1539,4 +1541,17 @@ uvm_vnp_sync(struct mount *mp)
>  	}
>  
>  	rw_exit_write(&uvn_sync_lock);
> +}
> +
> +void
> +uvm_vnp_obj_alloc(struct vnode *vp)
> +{
> +	struct uvm_vnode *uvn;
> +
> +	KASSERT(vp->v_uvm == NULL);
> +
> +	uvn = pool_get(&uvm_vnode_pool, PR_WAITOK | PR_ZERO);
> +	uvm_obj_init(&uvn->u_obj, &uvm_vnodeops, 0);
> +	uvn->u_vnode = vp;
> +	vp->v_uvm = uvn;
>  }
> Index: uvm/uvm_vnode.h
> ===================================================================
> RCS file: /cvs/src/sys/uvm/uvm_vnode.h,v
> diff -u -p -r1.21 uvm_vnode.h
> --- uvm/uvm_vnode.h	20 Oct 2022 13:31:52 -0000	1.21
> +++ uvm/uvm_vnode.h	24 Sep 2025 09:46:04 -0000
> @@ -94,4 +94,7 @@ struct uvm_vnode {
>   */
>  #define UVM_VNODE_BLOCKED (UVM_VNODE_ALOCK|UVM_VNODE_DYING|UVM_VNODE_RELKILL)
>  
> +void			uvm_vnp_obj_alloc(struct vnode *);
> +void			uvm_vnp_terminate(struct vnode *);

Should uvm_vnp_terminate() really be exposed here? It is already in
sys/vnode.h, or should uvm_vnp_obj_alloc() be added to sys/vnode.h or
should uvm_vnp_terminate() be removed from sys/vnode.h?

Apart from that OK claudio@
-- 
:wq Claudio