Enable vm_object_mightbedirty() and vm_object_page_clean() for swap

objects backing tmpfs vnodes data.

The clean scan is limited to only remove write permissions from the
mapped pages of the objects.  This fixes the issue that tmpfs vnode
mtime is not updated from writes to the mmaped area after the initial
page-in.

Noted by:	mjg
Reviewed by:	markj
Discussed with:	jeff
Tested by:	pho
Sponsored by:	The FreeBSD Foundation
MFC after:	1 week
Differential revision:	https://reviews.freebsd.org/D23432
This commit is contained in:
Konstantin Belousov 2020-02-04 19:03:37 +00:00
parent a421e8786b
commit 8d34a3bf7d
Notes: svn2git 2020-12-20 02:59:44 +00:00
svn path=/head/; revision=357514
2 changed files with 48 additions and 29 deletions

View File

@ -1017,6 +1017,10 @@ vm_object_page_remove_write(vm_page_t p, int flags, boolean_t *allclean)
* write out pages with PGA_NOSYNC set (originally comes from MAP_NOSYNC),
* leaving the object dirty.
*
* For swap objects backing tmpfs regular files, do not flush anything,
* but remove write protection on the mapped pages to update mtime through
* mmaped writes.
*
* When stuffing pages asynchronously, allow clustering. XXX we need a
* synchronous clustering mode implementation.
*
@ -1038,8 +1042,7 @@ vm_object_page_clean(vm_object_t object, vm_ooffset_t start, vm_ooffset_t end,
VM_OBJECT_ASSERT_WLOCKED(object);
if (object->type != OBJT_VNODE || !vm_object_mightbedirty(object) ||
object->resident_page_count == 0)
if (!vm_object_mightbedirty(object) || object->resident_page_count == 0)
return (TRUE);
pagerflags = (flags & (OBJPC_SYNC | OBJPC_INVAL)) != 0 ?
@ -1072,32 +1075,36 @@ vm_object_page_clean(vm_object_t object, vm_ooffset_t start, vm_ooffset_t end,
vm_page_xunbusy(p);
continue;
}
if (object->type == OBJT_VNODE) {
n = vm_object_page_collect_flush(object, p, pagerflags,
flags, &allclean, &eio);
if (eio) {
res = FALSE;
allclean = FALSE;
}
if (object->generation != curgeneration &&
(flags & OBJPC_SYNC) != 0)
goto rescan;
n = vm_object_page_collect_flush(object, p, pagerflags,
flags, &allclean, &eio);
if (eio) {
res = FALSE;
allclean = FALSE;
}
if (object->generation != curgeneration &&
(flags & OBJPC_SYNC) != 0)
goto rescan;
/*
* If the VOP_PUTPAGES() did a truncated write, so
* that even the first page of the run is not fully
* written, vm_pageout_flush() returns 0 as the run
* length. Since the condition that caused truncated
* write may be permanent, e.g. exhausted free space,
* accepting n == 0 would cause an infinite loop.
*
* Forwarding the iterator leaves the unwritten page
* behind, but there is not much we can do there if
* filesystem refuses to write it.
*/
if (n == 0) {
/*
* If the VOP_PUTPAGES() did a truncated write, so
* that even the first page of the run is not fully
* written, vm_pageout_flush() returns 0 as the run
* length. Since the condition that caused truncated
* write may be permanent, e.g. exhausted free space,
* accepting n == 0 would cause an infinite loop.
*
* Forwarding the iterator leaves the unwritten page
* behind, but there is not much we can do there if
* filesystem refuses to write it.
*/
if (n == 0) {
n = 1;
allclean = FALSE;
}
} else {
n = 1;
allclean = FALSE;
vm_page_xunbusy(p);
}
np = vm_page_find_least(object, pi + n);
}
@ -1105,7 +1112,12 @@ vm_object_page_clean(vm_object_t object, vm_ooffset_t start, vm_ooffset_t end,
VOP_FSYNC(vp, (pagerflags & VM_PAGER_PUT_SYNC) ? MNT_WAIT : 0);
#endif
if (allclean)
/*
* Leave updating cleangeneration for tmpfs objects to tmpfs
* scan. It needs to update mtime, which happens for other
* filesystems during page writeouts.
*/
if (allclean && object->type == OBJT_VNODE)
object->cleangeneration = curgeneration;
return (res);
}

View File

@ -322,8 +322,15 @@ static __inline bool
vm_object_mightbedirty(vm_object_t object)
{
return (object->type == OBJT_VNODE &&
object->generation != object->cleangeneration);
if (object->type != OBJT_VNODE) {
if ((object->flags & OBJ_TMPFS_NODE) == 0)
return (false);
#ifdef KASSERT
KASSERT(object->type == OBJT_SWAP,
("TMPFS_NODE obj %p is not swap", object));
#endif
}
return (object->generation != object->cleangeneration);
}
void vm_object_clear_flag(vm_object_t object, u_short bits);