Date: Sat, 15 Jun 2013 10:09:20 +0000 (UTC) From: Attilio Rao <attilio@FreeBSD.org> To: src-committers@freebsd.org, svn-src-user@freebsd.org Subject: svn commit: r251788 - user/attilio/vmobj-readlock/sys/kern Message-ID: <201306151009.r5FA9K2T022015@svn.freebsd.org>
next in thread | raw e-mail | index | archive | help
Author: attilio Date: Sat Jun 15 10:09:20 2013 New Revision: 251788 URL: http://svnweb.freebsd.org/changeset/base/251788 Log: Avoid to soft busy the page before to do vn_rdwr() in kern_sendfile(). That is unnecessary. The page is infact already wired and concurrent accesses to the specified page and off should be serialized already by vn_rdwr(). Sponsored by: EMC / Isilon storage division Discussed with: jeff, dillon Modified: user/attilio/vmobj-readlock/sys/kern/uipc_syscalls.c Modified: user/attilio/vmobj-readlock/sys/kern/uipc_syscalls.c ============================================================================== --- user/attilio/vmobj-readlock/sys/kern/uipc_syscalls.c Sat Jun 15 10:08:34 2013 (r251787) +++ user/attilio/vmobj-readlock/sys/kern/uipc_syscalls.c Sat Jun 15 10:09:20 2013 (r251788) @@ -2219,11 +2219,6 @@ retry_space: else { ssize_t resid; - /* - * Ensure that our page is still around - * when the I/O completes. - */ - vm_page_io_start(pg); VM_OBJECT_WUNLOCK(obj); /* @@ -2237,11 +2232,9 @@ retry_space: trunc_page(off), UIO_NOCOPY, IO_NODELOCKED | IO_VMIO | ((MAXBSIZE / bsize) << IO_SEQSHIFT), td->td_ucred, NOCRED, &resid, td); - VM_OBJECT_WLOCK(obj); - vm_page_io_finish(pg); - if (!error) - VM_OBJECT_WUNLOCK(obj); mbstat.sf_iocnt++; + if (error) + VM_OBJECT_WLOCK(obj); } if (error) { vm_page_lock(pg);
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?201306151009.r5FA9K2T022015>