d/freebsd-skq
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Remove unnecessary soft busy of the page before to do vn_rdwr() in

Browse Source 
kern_sendfile() which is unnecessary.
The page is already wired so it will not be subjected to pagefault.
The content cannot be effectively protected as it is full of races
already.
Multiple accesses to the same indexes are serialized through vn_rdwr().

Sponsored by:	EMC / Isilon storage division
Reviewed by:	alc, jeff
Tested by:	pho
This commit is contained in:
Attilio Rao 2013-08-04 15:56:19 +00:00
parent e44e975c1b
commit 878a788734
2 changed files with 5 additions and 14 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

8
sys/fs/tmpfs/tmpfs_vnops.c
View File

@ -448,10 +448,8 @@ tmpfs_nocacheread(vm_object_t tobj, vm_pindex_t idx,
VM_OBJECT_WLOCK(tobj);
/*
* The kern_sendfile() code calls vn_rdwr() with the page
* soft-busied. Ignore the soft-busy state here. Parallel
* reads of the page content from disk are prevented by
* VPO_BUSY.
* Parallel reads of the page content from disk are prevented
* by VPO_BUSY.
*
* Although the tmpfs vnode lock is held here, it is
* nonetheless safe to sleep waiting for a free page. The
@ -460,7 +458,7 @@ tmpfs_nocacheread(vm_object_t tobj, vm_pindex_t idx,
* type object.
*/
m = vm_page_grab(tobj, idx, VM_ALLOC_NORMAL | VM_ALLOC_RETRY |
VM_ALLOC_IGN_SBUSY | VM_ALLOC_NOBUSY);
VM_ALLOC_NOBUSY);
if (m->valid != VM_PAGE_BITS_ALL) {
vm_page_busy(m);
if (vm_pager_has_page(tobj, idx, NULL, NULL)) {

11
sys/kern/uipc_syscalls.c
View File

@ -2246,11 +2246,6 @@ retry_space:
ssize_t resid;
int readahead = sfreadahead * MAXBSIZE;
/*
* Ensure that our page is still around
* when the I/O completes.
*/
vm_page_io_start(pg);
VM_OBJECT_WUNLOCK(obj);
/*
@ -2264,11 +2259,9 @@ retry_space:
trunc_page(off), UIO_NOCOPY, IO_NODELOCKED |
IO_VMIO | ((readahead / bsize) << IO_SEQSHIFT),
td->td_ucred, NOCRED, &resid, td);
VM_OBJECT_WLOCK(obj);
vm_page_io_finish(pg);
if (!error)
VM_OBJECT_WUNLOCK(obj);
SFSTAT_INC(sf_iocnt);
if (error)
VM_OBJECT_WLOCK(obj);
}
if (error) {
vm_page_lock(pg);