Skip site navigation (1)Skip section navigation (2)
Date:      Wed, 16 Nov 2016 18:53:43 +0200
From:      Konstantin Belousov <kostikbel@gmail.com>
To:        Ruslan Bukin <ruslan.bukin@cl.cam.ac.uk>
Cc:        Alan Cox <alc@FreeBSD.org>, src-committers@freebsd.org, svn-src-all@freebsd.org, svn-src-head@freebsd.org
Subject:   Re: svn commit: r308691 - in head/sys: cddl/compat/opensolaris/sys cddl/contrib/opensolaris/uts/common/fs/zfs fs/tmpfs kern vm
Message-ID:  <20161116165343.GX54029@kib.kiev.ua>
In-Reply-To: <20161116133718.GA10251@bsdpad.com>
References:  <201611151822.uAFIMoj2092581@repo.freebsd.org> <20161116133718.GA10251@bsdpad.com>

next in thread | previous in thread | raw e-mail | index | archive | help
On Wed, Nov 16, 2016 at 01:37:18PM +0000, Ruslan Bukin wrote:
> I have a panic with this on RISC-V. Any ideas ?
How did you checked that the revision you replied to, makes the problem ?
Note that the backtrace below is not reasonable.

> 
>               vvvvvvvvvvvvvvvvvvvvvvvvvvvvvvvv
>                   vvvvvvvvvvvvvvvvvvvvvvvvvvvv
> rrrrrrrrrrrrr       vvvvvvvvvvvvvvvvvvvvvvvvvv
> rrrrrrrrrrrrrrrr      vvvvvvvvvvvvvvvvvvvvvvvv
> rrrrrrrrrrrrrrrrrr    vvvvvvvvvvvvvvvvvvvvvvvv
> rrrrrrrrrrrrrrrrrr    vvvvvvvvvvvvvvvvvvvvvvvv
> rrrrrrrrrrrrrrrrrr    vvvvvvvvvvvvvvvvvvvvvvvv
> rrrrrrrrrrrrrrrr      vvvvvvvvvvvvvvvvvvvvvv  
> rrrrrrrrrrrrr       vvvvvvvvvvvvvvvvvvvvvv    
> rr                vvvvvvvvvvvvvvvvvvvvvv      
> rr            vvvvvvvvvvvvvvvvvvvvvvvv      rr
> rrrr      vvvvvvvvvvvvvvvvvvvvvvvvvv      rrrr
> rrrrrr      vvvvvvvvvvvvvvvvvvvvvv      rrrrrr
> rrrrrrrr      vvvvvvvvvvvvvvvvvv      rrrrrrrr
> rrrrrrrrrr      vvvvvvvvvvvvvv      rrrrrrrrrr
> rrrrrrrrrrrr      vvvvvvvvvv      rrrrrrrrrrrr
> rrrrrrrrrrrrrr      vvvvvv      rrrrrrrrrrrrrr
> rrrrrrrrrrrrrrrr      vv      rrrrrrrrrrrrrrrr
> rrrrrrrrrrrrrrrrrr          rrrrrrrrrrrrrrrrrr
> rrrrrrrrrrrrrrrrrrrr      rrrrrrrrrrrrrrrrrrrr
> rrrrrrrrrrrrrrrrrrrrrr  rrrrrrrrrrrrrrrrrrrrrr
> 
>        INSTRUCTION SETS WANT TO BE FREE
> KDB: debugger backends: ddb
> KDB: current backend: ddb
> Found 2 CPUs in the device tree
> Copyright (c) 1992-2016 The FreeBSD Project.
> Copyright (c) 1979, 1980, 1983, 1986, 1988, 1989, 1991, 1992, 1993, 1994
> 	The Regents of the University of California. All rights reserved.
> FreeBSD is a registered trademark of The FreeBSD Foundation.
> FreeBSD 12.0-CURRENT #4 0a3288b(br-riscv-isa-update)-dirty: Wed Nov 16 13:28:11 UTC 2016
>     rb743@vica.cl.cam.ac.uk:/home/rb743/obj/riscv.riscv64/home/rb743/dev/freebsd-riscv/sys/SPIKE riscv
> gcc version 6.1.0 (GCC) 
> Preloaded elf64 kernel "kernel" at 0xffffffc0026be360.
> CPU(0): Unknown Implementer Unknown Processor
> Starting CPU 1 (0)
> FreeBSD/SMP: Multiprocessor System Detected: 2 CPUs
> ULE: setup cpu 0
> ULE: setup cpu 1
> random: entropy device external interface
> crypto: <crypto core>
> mem: <memory>
> openfirm: <Open Firmware control device>
> null: <full device, null device, zero device>
> nfslock: pseudo-device
> random: harvesting attach, 8 bytes (4 bits) from nexus0
> ofwbus0: <Open Firmware Device Tree>
> simplebus0: <Flattened device tree simple bus> on ofwbus0
> random: harvesting attach, 8 bytes (4 bits) from simplebus0
> random: harvesting attach, 8 bytes (4 bits) from ofwbus0
> timer0: <RISC-V Timer> mem 0x40000000-0x40000007,0x40000008-0x40001007 irq 5 on simplebus0
> Timecounter "RISC-V Timecounter" frequency 1000000 Hz quality 1000
> Event timer "RISC-V Eventtimer" frequency 1000000 Hz quality 1000
> random: harvesting attach, 8 bytes (4 bits) from timer0
> cpulist0: <Open Firmware CPU Group> on ofwbus0
> cpu0: <Open Firmware CPU> on cpulist0
> cpu0: missing 'clock-frequency' property
> riscv64_cpu0: register <0>
> random: harvesting attach, 8 bytes (4 bits) from riscv64_cpu0
> random: harvesting attach, 8 bytes (4 bits) from cpu0
> cpu1: <Open Firmware CPU> on cpulist0
> cpu1: missing 'clock-frequency' property
> riscv64_cpu1: register <0>
> random: harvesting attach, 8 bytes (4 bits) from riscv64_cpu1
> random: harvesting attach, 8 bytes (4 bits) from cpu1
> random: harvesting attach, 8 bytes (4 bits) from cpulist0
> simplebus0: <pic@0> compat riscv,pic (no driver attached)
> rcons0: <RISC-V console> irq 1 on simplebus0
> random: harvesting attach, 8 bytes (4 bits) from rcons0
> cryptosoft0: <software crypto>
> crypto: assign cryptosoft0 driver id 0, flags 100663296
> crypto: cryptosoft0 registers alg 1 flags 0 maxoplen 0
> crypto: cryptosoft0 registers alg 2 flags 0 maxoplen 0
> crypto: cryptosoft0 registers alg 3 flags 0 maxoplen 0
> crypto: cryptosoft0 registers alg 4 flags 0 maxoplen 0
> crypto: cryptosoft0 registers alg 5 flags 0 maxoplen 0
> crypto: cryptosoft0 registers alg 16 flags 0 maxoplen 0
> crypto: cryptosoft0 registers alg 6 flags 0 maxoplen 0
> crypto: cryptosoft0 registers alg 7 flags 0 maxoplen 0
> crypto: cryptosoft0 registers alg 18 flags 0 maxoplen 0
> crypto: cryptosoft0 registers alg 19 flags 0 maxoplen 0
> crypto: cryptosoft0 registers alg 20 flags 0 maxoplen 0
> crypto: cryptosoft0 registers alg 8 flags 0 maxoplen 0
> crypto: cryptosoft0 registers alg 15 flags 0 maxoplen 0
> crypto: cryptosoft0 registers alg 9 flags 0 maxoplen 0
> crypto: cryptosoft0 registers alg 10 flags 0 maxoplen 0
> crypto: cryptosoft0 registers alg 13 flags 0 maxoplen 0
> crypto: cryptosoft0 registers alg 14 flags 0 maxoplen 0
> crypto: cryptosoft0 registers alg 11 flags 0 maxoplen 0
> crypto: cryptosoft0 registers alg 22 flags 0 maxoplen 0
> crypto: cryptosoft0 registers alg 23 flags 0 maxoplen 0
> crypto: cryptosoft0 registers alg 25 flags 0 maxoplen 0
> crypto: cryptosoft0 registers alg 24 flags 0 maxoplen 0
> crypto: cryptosoft0 registers alg 26 flags 0 maxoplen 0
> crypto: cryptosoft0 registers alg 27 flags 0 maxoplen 0
> crypto: cryptosoft0 registers alg 28 flags 0 maxoplen 0
> crypto: cryptosoft0 registers alg 21 flags 0 maxoplen 0
> crypto: cryptosoft0 registers alg 17 flags 0 maxoplen 0
> random: harvesting attach, 8 bytes (4 bits) from cryptosoft0
> Device configuration finished.
> procfs registered
> Timecounters tick every 1.000 msec
> lo0: bpf attached
> vlan: initialized, using hash tables with chaining
> tcp_init: net.inet.tcp.tcbhashsize auto tuned to 8192
> IPsec: Initialized Security Association Processing.
> t[0] == 0xffffffc00265bf50
> t[1] == 0xffffffc00016494c
> t[2] == 0x0000000000000050
> t[3] == 0x0000000000000000
> t[4] == 0x0000000000000000
> t[5] == 0x0000000000000000
> t[6] == 0x0000000000000000
> s[0] == 0xffffffc000003db0
> s[1] == 0x0000000000000000
> s[2] == 0xffffffc002c4c510
> s[3] == 0xffffffc002cea9c0
> s[4] == 0x000000000098967f
> s[5] == 0x00000000039386ff
> s[6] == 0xffffffc000574778
> s[7] == 0x0000000000000000
> s[8] == 0xffffffc00267c580
> s[9] == 0xffffffc000531218
> s[10] == 0x0000000000000412
> s[11] == 0x0000000000000000
> a[0] == 0x0000000000000000
> a[1] == 0x0000000000000000
> a[2] == 0xffffffc000531218
> a[3] == 0x0000000000000412
> a[4] == 0xffffffc00267c580
> a[5] == 0x0000000000000002
> a[6] == 0x0000000000000000
> a[7] == 0x0000000000000003
> sepc == 0xffffffc00013deac
> sstatus == 0x8000000000006100
> panic: vm_fault failed: ffffffc00013deac, va 0x0000000000000018
> cpuid = 0
> KDB: stack backtrace:
> db_trace_self() at db_read_token+0x704
> 	 pc = 0xffffffc000455b44 ra = 0xffffffc0000244f8
> 	 sp = 0xffffffc000003858 fp = 0xffffffc000003a78
> 
> db_read_token() at kdb_backtrace+0x3c
> 	 pc = 0xffffffc0000244f8 ra = 0xffffffc0001a5588
> 	 sp = 0xffffffc000003a78 fp = 0xffffffc000003a88
> 
> kdb_backtrace() at vpanic+0x158
> 	 pc = 0xffffffc0001a5588 ra = 0xffffffc00015bd74
> 	 sp = 0xffffffc000003a88 fp = 0xffffffc000003ac8
> 
> vpanic() at panic+0x34
> 	 pc = 0xffffffc00015bd74 ra = 0xffffffc00015c58c
> 	 sp = 0xffffffc000003ac8 fp = 0xffffffc000003ae8
> 
> panic() at sysarch+0x36c
> 	 pc = 0xffffffc00015c58c ra = 0xffffffc0004622b0
> 	 sp = 0xffffffc000003ae8 fp = 0xffffffc000003bf8
> 
> sysarch() at do_trap_supervisor+0xa0
> 	 pc = 0xffffffc0004622b0 ra = 0xffffffc000462498
> 	 sp = 0xffffffc000003bf8 fp = 0xffffffc000003c18
> 
> do_trap_supervisor() at cpu_exception_handler_supervisor+0xb0
> 	 pc = 0xffffffc000462498 ra = 0xffffffc000456440
> 	 sp = 0xffffffc000003c18 fp = 0xffffffc000003db0
> 
> cpu_exception_handler_supervisor() at ruxagg+0x34
> 	 pc = 0xffffffc000456440 ra = 0xffffffc000154458
> 	 sp = 0xffffffc000003db0 fp = 0xffffffc000003dd0
> 
> ruxagg() at rufetch+0x9c
> 	 pc = 0xffffffc000154458 ra = 0xffffffc000154784
> 	 sp = 0xffffffc000003dd0 fp = 0xffffffc000003e00
> 
> rufetch() at exec_shell_imgact+0x1204
> 	 pc = 0xffffffc000154784 ra = 0xffffffc0000f9d4c
> 	 sp = 0xffffffc000003e00 fp = 0xffffffc000003ee0
> 
> exec_shell_imgact() at mi_startup+0x190
> 	 pc = 0xffffffc0000f9d4c ra = 0xffffffc0000fa768
> 	 sp = 0xffffffc000003ee0 fp = 0xffffffc000003f20
> 
> mi_startup() at kernbase+0x248
> 	 pc = 0xffffffc0000fa768 ra = 0xffffffc000000248
> 	 sp = 0xffffffc000003f20 fp = 0x0000000085002ff8
> 
> KDB: enter: panic
> [ thread pid 0 tid 100000 ]
> Stopped at      kdb_enter+0x4c:
> db> 
> 
> Ruslan
> 
> On Tue, Nov 15, 2016 at 06:22:50PM +0000, Alan Cox wrote:
> > Author: alc
> > Date: Tue Nov 15 18:22:50 2016
> > New Revision: 308691
> > URL: https://svnweb.freebsd.org/changeset/base/308691
> > 
> > Log:
> >   Remove most of the code for implementing PG_CACHED pages.  (This change does
> >   not remove user-space visible fields from vm_cnt or all of the references to
> >   cached pages from comments.  Those changes will come later.)
> >   
> >   Reviewed by:	kib, markj
> >   Tested by:	pho
> >   Sponsored by:	Dell EMC Isilon
> >   Differential Revision:	https://reviews.freebsd.org/D8497
> > 
> > Modified:
> >   head/sys/cddl/compat/opensolaris/sys/vnode.h
> >   head/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/zfs_vnops.c
> >   head/sys/fs/tmpfs/tmpfs_subr.c
> >   head/sys/kern/kern_exec.c
> >   head/sys/kern/uipc_shm.c
> >   head/sys/vm/swap_pager.c
> >   head/sys/vm/vm_fault.c
> >   head/sys/vm/vm_mmap.c
> >   head/sys/vm/vm_object.c
> >   head/sys/vm/vm_object.h
> >   head/sys/vm/vm_page.c
> >   head/sys/vm/vm_page.h
> >   head/sys/vm/vm_reserv.c
> >   head/sys/vm/vm_reserv.h
> >   head/sys/vm/vnode_pager.c
> > 
> > Modified: head/sys/cddl/compat/opensolaris/sys/vnode.h
> > ==============================================================================
> > --- head/sys/cddl/compat/opensolaris/sys/vnode.h	Tue Nov 15 17:01:48 2016	(r308690)
> > +++ head/sys/cddl/compat/opensolaris/sys/vnode.h	Tue Nov 15 18:22:50 2016	(r308691)
> > @@ -75,8 +75,7 @@ vn_is_readonly(vnode_t *vp)
> >  #define	vn_mountedvfs(vp)	((vp)->v_mountedhere)
> >  #define	vn_has_cached_data(vp)	\
> >  	((vp)->v_object != NULL && \
> > -	 ((vp)->v_object->resident_page_count > 0 || \
> > -	  !vm_object_cache_is_empty((vp)->v_object)))
> > +	 (vp)->v_object->resident_page_count > 0)
> >  #define	vn_exists(vp)		do { } while (0)
> >  #define	vn_invalid(vp)		do { } while (0)
> >  #define	vn_renamepath(tdvp, svp, tnm, lentnm)	do { } while (0)
> > 
> > Modified: head/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/zfs_vnops.c
> > ==============================================================================
> > --- head/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/zfs_vnops.c	Tue Nov 15 17:01:48 2016	(r308690)
> > +++ head/sys/cddl/contrib/opensolaris/uts/common/fs/zfs/zfs_vnops.c	Tue Nov 15 18:22:50 2016	(r308691)
> > @@ -426,10 +426,6 @@ page_busy(vnode_t *vp, int64_t start, in
> >  				continue;
> >  			}
> >  			vm_page_sbusy(pp);
> > -		} else if (pp == NULL) {
> > -			pp = vm_page_alloc(obj, OFF_TO_IDX(start),
> > -			    VM_ALLOC_SYSTEM | VM_ALLOC_IFCACHED |
> > -			    VM_ALLOC_SBUSY);
> >  		} else {
> >  			ASSERT(pp != NULL && !pp->valid);
> >  			pp = NULL;
> > 
> > Modified: head/sys/fs/tmpfs/tmpfs_subr.c
> > ==============================================================================
> > --- head/sys/fs/tmpfs/tmpfs_subr.c	Tue Nov 15 17:01:48 2016	(r308690)
> > +++ head/sys/fs/tmpfs/tmpfs_subr.c	Tue Nov 15 18:22:50 2016	(r308691)
> > @@ -1372,12 +1372,9 @@ retry:
> >  					VM_WAIT;
> >  					VM_OBJECT_WLOCK(uobj);
> >  					goto retry;
> > -				} else if (m->valid != VM_PAGE_BITS_ALL)
> > -					rv = vm_pager_get_pages(uobj, &m, 1,
> > -					    NULL, NULL);
> > -				else
> > -					/* A cached page was reactivated. */
> > -					rv = VM_PAGER_OK;
> > +				}
> > +				rv = vm_pager_get_pages(uobj, &m, 1, NULL,
> > +				    NULL);
> >  				vm_page_lock(m);
> >  				if (rv == VM_PAGER_OK) {
> >  					vm_page_deactivate(m);
> > 
> > Modified: head/sys/kern/kern_exec.c
> > ==============================================================================
> > --- head/sys/kern/kern_exec.c	Tue Nov 15 17:01:48 2016	(r308690)
> > +++ head/sys/kern/kern_exec.c	Tue Nov 15 18:22:50 2016	(r308691)
> > @@ -1006,7 +1006,7 @@ exec_map_first_page(imgp)
> >  					break;
> >  			} else {
> >  				ma[i] = vm_page_alloc(object, i,
> > -				    VM_ALLOC_NORMAL | VM_ALLOC_IFNOTCACHED);
> > +				    VM_ALLOC_NORMAL);
> >  				if (ma[i] == NULL)
> >  					break;
> >  			}
> > 
> > Modified: head/sys/kern/uipc_shm.c
> > ==============================================================================
> > --- head/sys/kern/uipc_shm.c	Tue Nov 15 17:01:48 2016	(r308690)
> > +++ head/sys/kern/uipc_shm.c	Tue Nov 15 18:22:50 2016	(r308691)
> > @@ -455,12 +455,9 @@ retry:
> >  					VM_WAIT;
> >  					VM_OBJECT_WLOCK(object);
> >  					goto retry;
> > -				} else if (m->valid != VM_PAGE_BITS_ALL)
> > -					rv = vm_pager_get_pages(object, &m, 1,
> > -					    NULL, NULL);
> > -				else
> > -					/* A cached page was reactivated. */
> > -					rv = VM_PAGER_OK;
> > +				}
> > +				rv = vm_pager_get_pages(object, &m, 1, NULL,
> > +				    NULL);
> >  				vm_page_lock(m);
> >  				if (rv == VM_PAGER_OK) {
> >  					vm_page_deactivate(m);
> > 
> > Modified: head/sys/vm/swap_pager.c
> > ==============================================================================
> > --- head/sys/vm/swap_pager.c	Tue Nov 15 17:01:48 2016	(r308690)
> > +++ head/sys/vm/swap_pager.c	Tue Nov 15 18:22:50 2016	(r308691)
> > @@ -1126,7 +1126,7 @@ swap_pager_getpages(vm_object_t object, 
> >  	if (shift != 0) {
> >  		for (i = 1; i <= shift; i++) {
> >  			p = vm_page_alloc(object, m[0]->pindex - i,
> > -			    VM_ALLOC_NORMAL | VM_ALLOC_IFNOTCACHED);
> > +			    VM_ALLOC_NORMAL);
> >  			if (p == NULL) {
> >  				/* Shift allocated pages to the left. */
> >  				for (j = 0; j < i - 1; j++)
> > @@ -1144,8 +1144,7 @@ swap_pager_getpages(vm_object_t object, 
> >  	if (rahead != NULL) {
> >  		for (i = 0; i < *rahead; i++) {
> >  			p = vm_page_alloc(object,
> > -			    m[reqcount - 1]->pindex + i + 1,
> > -			    VM_ALLOC_NORMAL | VM_ALLOC_IFNOTCACHED);
> > +			    m[reqcount - 1]->pindex + i + 1, VM_ALLOC_NORMAL);
> >  			if (p == NULL)
> >  				break;
> >  			bp->b_pages[shift + reqcount + i] = p;
> > 
> > Modified: head/sys/vm/vm_fault.c
> > ==============================================================================
> > --- head/sys/vm/vm_fault.c	Tue Nov 15 17:01:48 2016	(r308690)
> > +++ head/sys/vm/vm_fault.c	Tue Nov 15 18:22:50 2016	(r308691)
> > @@ -559,8 +559,7 @@ fast_failed:
> >  				unlock_and_deallocate(&fs);
> >  				VM_WAITPFAULT;
> >  				goto RetryFault;
> > -			} else if (fs.m->valid == VM_PAGE_BITS_ALL)
> > -				break;
> > +			}
> >  		}
> >  
> >  readrest:
> > 
> > Modified: head/sys/vm/vm_mmap.c
> > ==============================================================================
> > --- head/sys/vm/vm_mmap.c	Tue Nov 15 17:01:48 2016	(r308690)
> > +++ head/sys/vm/vm_mmap.c	Tue Nov 15 18:22:50 2016	(r308691)
> > @@ -877,9 +877,6 @@ RestartScan:
> >  					pindex = OFF_TO_IDX(current->offset +
> >  					    (addr - current->start));
> >  					m = vm_page_lookup(object, pindex);
> > -					if (m == NULL &&
> > -					    vm_page_is_cached(object, pindex))
> > -						mincoreinfo = MINCORE_INCORE;
> >  					if (m != NULL && m->valid == 0)
> >  						m = NULL;
> >  					if (m != NULL)
> > 
> > Modified: head/sys/vm/vm_object.c
> > ==============================================================================
> > --- head/sys/vm/vm_object.c	Tue Nov 15 17:01:48 2016	(r308690)
> > +++ head/sys/vm/vm_object.c	Tue Nov 15 18:22:50 2016	(r308691)
> > @@ -178,9 +178,6 @@ vm_object_zdtor(void *mem, int size, voi
> >  	    ("object %p has reservations",
> >  	    object));
> >  #endif
> > -	KASSERT(vm_object_cache_is_empty(object),
> > -	    ("object %p has cached pages",
> > -	    object));
> >  	KASSERT(object->paging_in_progress == 0,
> >  	    ("object %p paging_in_progress = %d",
> >  	    object, object->paging_in_progress));
> > @@ -212,8 +209,6 @@ vm_object_zinit(void *mem, int size, int
> >  	object->paging_in_progress = 0;
> >  	object->resident_page_count = 0;
> >  	object->shadow_count = 0;
> > -	object->cache.rt_root = 0;
> > -	object->cache.rt_flags = 0;
> >  
> >  	mtx_lock(&vm_object_list_mtx);
> >  	TAILQ_INSERT_TAIL(&vm_object_list, object, object_list);
> > @@ -792,8 +787,6 @@ vm_object_terminate(vm_object_t object)
> >  	if (__predict_false(!LIST_EMPTY(&object->rvq)))
> >  		vm_reserv_break_all(object);
> >  #endif
> > -	if (__predict_false(!vm_object_cache_is_empty(object)))
> > -		vm_page_cache_free(object, 0, 0);
> >  
> >  	KASSERT(object->cred == NULL || object->type == OBJT_DEFAULT ||
> >  	    object->type == OBJT_SWAP,
> > @@ -1135,13 +1128,6 @@ shadowlookup:
> >  		} else if ((tobject->flags & OBJ_UNMANAGED) != 0)
> >  			goto unlock_tobject;
> >  		m = vm_page_lookup(tobject, tpindex);
> > -		if (m == NULL && advise == MADV_WILLNEED) {
> > -			/*
> > -			 * If the page is cached, reactivate it.
> > -			 */
> > -			m = vm_page_alloc(tobject, tpindex, VM_ALLOC_IFCACHED |
> > -			    VM_ALLOC_NOBUSY);
> > -		}
> >  		if (m == NULL) {
> >  			/*
> >  			 * There may be swap even if there is no backing page
> > @@ -1406,19 +1392,6 @@ retry:
> >  		swap_pager_copy(orig_object, new_object, offidxstart, 0);
> >  		TAILQ_FOREACH(m, &new_object->memq, listq)
> >  			vm_page_xunbusy(m);
> > -
> > -		/*
> > -		 * Transfer any cached pages from orig_object to new_object.
> > -		 * If swap_pager_copy() found swapped out pages within the
> > -		 * specified range of orig_object, then it changed
> > -		 * new_object's type to OBJT_SWAP when it transferred those
> > -		 * pages to new_object.  Otherwise, new_object's type
> > -		 * should still be OBJT_DEFAULT and orig_object should not
> > -		 * contain any cached pages within the specified range.
> > -		 */
> > -		if (__predict_false(!vm_object_cache_is_empty(orig_object)))
> > -			vm_page_cache_transfer(orig_object, offidxstart,
> > -			    new_object);
> >  	}
> >  	VM_OBJECT_WUNLOCK(orig_object);
> >  	VM_OBJECT_WUNLOCK(new_object);
> > @@ -1754,13 +1727,6 @@ vm_object_collapse(vm_object_t object)
> >  				    backing_object,
> >  				    object,
> >  				    OFF_TO_IDX(object->backing_object_offset), TRUE);
> > -
> > -				/*
> > -				 * Free any cached pages from backing_object.
> > -				 */
> > -				if (__predict_false(
> > -				    !vm_object_cache_is_empty(backing_object)))
> > -					vm_page_cache_free(backing_object, 0, 0);
> >  			}
> >  			/*
> >  			 * Object now shadows whatever backing_object did.
> > @@ -1889,7 +1855,7 @@ vm_object_page_remove(vm_object_t object
> >  	    (options & (OBJPR_CLEANONLY | OBJPR_NOTMAPPED)) == OBJPR_NOTMAPPED,
> >  	    ("vm_object_page_remove: illegal options for object %p", object));
> >  	if (object->resident_page_count == 0)
> > -		goto skipmemq;
> > +		return;
> >  	vm_object_pip_add(object, 1);
> >  again:
> >  	p = vm_page_find_least(object, start);
> > @@ -1946,9 +1912,6 @@ next:
> >  		vm_page_unlock(p);
> >  	}
> >  	vm_object_pip_wakeup(object);
> > -skipmemq:
> > -	if (__predict_false(!vm_object_cache_is_empty(object)))
> > -		vm_page_cache_free(object, start, end);
> >  }
> >  
> >  /*
> > 
> > Modified: head/sys/vm/vm_object.h
> > ==============================================================================
> > --- head/sys/vm/vm_object.h	Tue Nov 15 17:01:48 2016	(r308690)
> > +++ head/sys/vm/vm_object.h	Tue Nov 15 18:22:50 2016	(r308691)
> > @@ -118,7 +118,6 @@ struct vm_object {
> >  	vm_ooffset_t backing_object_offset;/* Offset in backing object */
> >  	TAILQ_ENTRY(vm_object) pager_object_list; /* list of all objects of this pager type */
> >  	LIST_HEAD(, vm_reserv) rvq;	/* list of reservations */
> > -	struct vm_radix cache;		/* (o + f) root of the cache page radix trie */
> >  	void *handle;
> >  	union {
> >  		/*
> > @@ -290,13 +289,6 @@ void vm_object_pip_wakeup(vm_object_t ob
> >  void vm_object_pip_wakeupn(vm_object_t object, short i);
> >  void vm_object_pip_wait(vm_object_t object, char *waitid);
> >  
> > -static __inline boolean_t
> > -vm_object_cache_is_empty(vm_object_t object)
> > -{
> > -
> > -	return (vm_radix_is_empty(&object->cache));
> > -}
> > -
> >  void umtx_shm_object_init(vm_object_t object);
> >  void umtx_shm_object_terminated(vm_object_t object);
> >  extern int umtx_shm_vnobj_persistent;
> > 
> > Modified: head/sys/vm/vm_page.c
> > ==============================================================================
> > --- head/sys/vm/vm_page.c	Tue Nov 15 17:01:48 2016	(r308690)
> > +++ head/sys/vm/vm_page.c	Tue Nov 15 18:22:50 2016	(r308691)
> > @@ -154,8 +154,7 @@ static int vm_pageout_pages_needed;
> >  
> >  static uma_zone_t fakepg_zone;
> >  
> > -static struct vnode *vm_page_alloc_init(vm_page_t m);
> > -static void vm_page_cache_turn_free(vm_page_t m);
> > +static void vm_page_alloc_check(vm_page_t m);
> >  static void vm_page_clear_dirty_mask(vm_page_t m, vm_page_bits_t pagebits);
> >  static void vm_page_enqueue(uint8_t queue, vm_page_t m);
> >  static void vm_page_free_wakeup(void);
> > @@ -1118,9 +1117,7 @@ void
> >  vm_page_dirty_KBI(vm_page_t m)
> >  {
> >  
> > -	/* These assertions refer to this operation by its public name. */
> > -	KASSERT((m->flags & PG_CACHED) == 0,
> > -	    ("vm_page_dirty: page in cache!"));
> > +	/* Refer to this operation by its public name. */
> >  	KASSERT(m->valid == VM_PAGE_BITS_ALL,
> >  	    ("vm_page_dirty: page is invalid!"));
> >  	m->dirty = VM_PAGE_BITS_ALL;
> > @@ -1459,142 +1456,6 @@ vm_page_rename(vm_page_t m, vm_object_t 
> >  }
> >  
> >  /*
> > - *	Convert all of the given object's cached pages that have a
> > - *	pindex within the given range into free pages.  If the value
> > - *	zero is given for "end", then the range's upper bound is
> > - *	infinity.  If the given object is backed by a vnode and it
> > - *	transitions from having one or more cached pages to none, the
> > - *	vnode's hold count is reduced.
> > - */
> > -void
> > -vm_page_cache_free(vm_object_t object, vm_pindex_t start, vm_pindex_t end)
> > -{
> > -	vm_page_t m;
> > -	boolean_t empty;
> > -
> > -	mtx_lock(&vm_page_queue_free_mtx);
> > -	if (__predict_false(vm_radix_is_empty(&object->cache))) {
> > -		mtx_unlock(&vm_page_queue_free_mtx);
> > -		return;
> > -	}
> > -	while ((m = vm_radix_lookup_ge(&object->cache, start)) != NULL) {
> > -		if (end != 0 && m->pindex >= end)
> > -			break;
> > -		vm_radix_remove(&object->cache, m->pindex);
> > -		vm_page_cache_turn_free(m);
> > -	}
> > -	empty = vm_radix_is_empty(&object->cache);
> > -	mtx_unlock(&vm_page_queue_free_mtx);
> > -	if (object->type == OBJT_VNODE && empty)
> > -		vdrop(object->handle);
> > -}
> > -
> > -/*
> > - *	Returns the cached page that is associated with the given
> > - *	object and offset.  If, however, none exists, returns NULL.
> > - *
> > - *	The free page queue must be locked.
> > - */
> > -static inline vm_page_t
> > -vm_page_cache_lookup(vm_object_t object, vm_pindex_t pindex)
> > -{
> > -
> > -	mtx_assert(&vm_page_queue_free_mtx, MA_OWNED);
> > -	return (vm_radix_lookup(&object->cache, pindex));
> > -}
> > -
> > -/*
> > - *	Remove the given cached page from its containing object's
> > - *	collection of cached pages.
> > - *
> > - *	The free page queue must be locked.
> > - */
> > -static void
> > -vm_page_cache_remove(vm_page_t m)
> > -{
> > -
> > -	mtx_assert(&vm_page_queue_free_mtx, MA_OWNED);
> > -	KASSERT((m->flags & PG_CACHED) != 0,
> > -	    ("vm_page_cache_remove: page %p is not cached", m));
> > -	vm_radix_remove(&m->object->cache, m->pindex);
> > -	m->object = NULL;
> > -	vm_cnt.v_cache_count--;
> > -}
> > -
> > -/*
> > - *	Transfer all of the cached pages with offset greater than or
> > - *	equal to 'offidxstart' from the original object's cache to the
> > - *	new object's cache.  However, any cached pages with offset
> > - *	greater than or equal to the new object's size are kept in the
> > - *	original object.  Initially, the new object's cache must be
> > - *	empty.  Offset 'offidxstart' in the original object must
> > - *	correspond to offset zero in the new object.
> > - *
> > - *	The new object must be locked.
> > - */
> > -void
> > -vm_page_cache_transfer(vm_object_t orig_object, vm_pindex_t offidxstart,
> > -    vm_object_t new_object)
> > -{
> > -	vm_page_t m;
> > -
> > -	/*
> > -	 * Insertion into an object's collection of cached pages
> > -	 * requires the object to be locked.  In contrast, removal does
> > -	 * not.
> > -	 */
> > -	VM_OBJECT_ASSERT_WLOCKED(new_object);
> > -	KASSERT(vm_radix_is_empty(&new_object->cache),
> > -	    ("vm_page_cache_transfer: object %p has cached pages",
> > -	    new_object));
> > -	mtx_lock(&vm_page_queue_free_mtx);
> > -	while ((m = vm_radix_lookup_ge(&orig_object->cache,
> > -	    offidxstart)) != NULL) {
> > -		/*
> > -		 * Transfer all of the pages with offset greater than or
> > -		 * equal to 'offidxstart' from the original object's
> > -		 * cache to the new object's cache.
> > -		 */
> > -		if ((m->pindex - offidxstart) >= new_object->size)
> > -			break;
> > -		vm_radix_remove(&orig_object->cache, m->pindex);
> > -		/* Update the page's object and offset. */
> > -		m->object = new_object;
> > -		m->pindex -= offidxstart;
> > -		if (vm_radix_insert(&new_object->cache, m))
> > -			vm_page_cache_turn_free(m);
> > -	}
> > -	mtx_unlock(&vm_page_queue_free_mtx);
> > -}
> > -
> > -/*
> > - *	Returns TRUE if a cached page is associated with the given object and
> > - *	offset, and FALSE otherwise.
> > - *
> > - *	The object must be locked.
> > - */
> > -boolean_t
> > -vm_page_is_cached(vm_object_t object, vm_pindex_t pindex)
> > -{
> > -	vm_page_t m;
> > -
> > -	/*
> > -	 * Insertion into an object's collection of cached pages requires the
> > -	 * object to be locked.  Therefore, if the object is locked and the
> > -	 * object's collection is empty, there is no need to acquire the free
> > -	 * page queues lock in order to prove that the specified page doesn't
> > -	 * exist.
> > -	 */
> > -	VM_OBJECT_ASSERT_WLOCKED(object);
> > -	if (__predict_true(vm_object_cache_is_empty(object)))
> > -		return (FALSE);
> > -	mtx_lock(&vm_page_queue_free_mtx);
> > -	m = vm_page_cache_lookup(object, pindex);
> > -	mtx_unlock(&vm_page_queue_free_mtx);
> > -	return (m != NULL);
> > -}
> > -
> > -/*
> >   *	vm_page_alloc:
> >   *
> >   *	Allocate and return a page that is associated with the specified
> > @@ -1610,9 +1471,6 @@ vm_page_is_cached(vm_object_t object, vm
> >   *	optional allocation flags:
> >   *	VM_ALLOC_COUNT(number)	the number of additional pages that the caller
> >   *				intends to allocate
> > - *	VM_ALLOC_IFCACHED	return page only if it is cached
> > - *	VM_ALLOC_IFNOTCACHED	return NULL, do not reactivate if the page
> > - *				is cached
> >   *	VM_ALLOC_NOBUSY		do not exclusive busy the page
> >   *	VM_ALLOC_NODUMP		do not include the page in a kernel core dump
> >   *	VM_ALLOC_NOOBJ		page is not associated with an object and
> > @@ -1626,8 +1484,6 @@ vm_page_is_cached(vm_object_t object, vm
> >  vm_page_t
> >  vm_page_alloc(vm_object_t object, vm_pindex_t pindex, int req)
> >  {
> > -	struct vnode *vp = NULL;
> > -	vm_object_t m_object;
> >  	vm_page_t m, mpred;
> >  	int flags, req_class;
> >  
> > @@ -1670,31 +1526,12 @@ vm_page_alloc(vm_object_t object, vm_pin
> >  		 * Allocate from the free queue if the number of free pages
> >  		 * exceeds the minimum for the request class.
> >  		 */
> > -		if (object != NULL &&
> > -		    (m = vm_page_cache_lookup(object, pindex)) != NULL) {
> > -			if ((req & VM_ALLOC_IFNOTCACHED) != 0) {
> > -				mtx_unlock(&vm_page_queue_free_mtx);
> > -				return (NULL);
> > -			}
> > -			if (vm_phys_unfree_page(m))
> > -				vm_phys_set_pool(VM_FREEPOOL_DEFAULT, m, 0);
> > -#if VM_NRESERVLEVEL > 0
> > -			else if (!vm_reserv_reactivate_page(m))
> > -#else
> > -			else
> > -#endif
> > -				panic("vm_page_alloc: cache page %p is missing"
> > -				    " from the free queue", m);
> > -		} else if ((req & VM_ALLOC_IFCACHED) != 0) {
> > -			mtx_unlock(&vm_page_queue_free_mtx);
> > -			return (NULL);
> >  #if VM_NRESERVLEVEL > 0
> > -		} else if (object == NULL || (object->flags & (OBJ_COLORED |
> > +		if (object == NULL || (object->flags & (OBJ_COLORED |
> >  		    OBJ_FICTITIOUS)) != OBJ_COLORED || (m =
> > -		    vm_reserv_alloc_page(object, pindex, mpred)) == NULL) {
> > -#else
> > -		} else {
> > +		    vm_reserv_alloc_page(object, pindex, mpred)) == NULL)
> >  #endif
> > +		{
> >  			m = vm_phys_alloc_pages(object != NULL ?
> >  			    VM_FREEPOOL_DEFAULT : VM_FREEPOOL_DIRECT, 0);
> >  #if VM_NRESERVLEVEL > 0
> > @@ -1720,33 +1557,9 @@ vm_page_alloc(vm_object_t object, vm_pin
> >  	 *  At this point we had better have found a good page.
> >  	 */
> >  	KASSERT(m != NULL, ("vm_page_alloc: missing page"));
> > -	KASSERT(m->queue == PQ_NONE,
> > -	    ("vm_page_alloc: page %p has unexpected queue %d", m, m->queue));
> > -	KASSERT(m->wire_count == 0, ("vm_page_alloc: page %p is wired", m));
> > -	KASSERT(m->hold_count == 0, ("vm_page_alloc: page %p is held", m));
> > -	KASSERT(!vm_page_busied(m), ("vm_page_alloc: page %p is busy", m));
> > -	KASSERT(m->dirty == 0, ("vm_page_alloc: page %p is dirty", m));
> > -	KASSERT(pmap_page_get_memattr(m) == VM_MEMATTR_DEFAULT,
> > -	    ("vm_page_alloc: page %p has unexpected memattr %d", m,
> > -	    pmap_page_get_memattr(m)));
> > -	if ((m->flags & PG_CACHED) != 0) {
> > -		KASSERT((m->flags & PG_ZERO) == 0,
> > -		    ("vm_page_alloc: cached page %p is PG_ZERO", m));
> > -		KASSERT(m->valid != 0,
> > -		    ("vm_page_alloc: cached page %p is invalid", m));
> > -		if (m->object != object || m->pindex != pindex)
> > -			m->valid = 0;
> > -		m_object = m->object;
> > -		vm_page_cache_remove(m);
> > -		if (m_object->type == OBJT_VNODE &&
> > -		    vm_object_cache_is_empty(m_object))
> > -			vp = m_object->handle;
> > -	} else {
> > -		KASSERT(m->valid == 0,
> > -		    ("vm_page_alloc: free page %p is valid", m));
> > -		vm_phys_freecnt_adj(m, -1);
> > -	}
> > +	vm_phys_freecnt_adj(m, -1);
> >  	mtx_unlock(&vm_page_queue_free_mtx);
> > +	vm_page_alloc_check(m);
> >  
> >  	/*
> >  	 * Initialize the page.  Only the PG_ZERO flag is inherited.
> > @@ -1778,9 +1591,6 @@ vm_page_alloc(vm_object_t object, vm_pin
> >  
> >  	if (object != NULL) {
> >  		if (vm_page_insert_after(m, object, pindex, mpred)) {
> > -			/* See the comment below about hold count. */
> > -			if (vp != NULL)
> > -				vdrop(vp);
> >  			pagedaemon_wakeup();
> >  			if (req & VM_ALLOC_WIRED) {
> >  				atomic_subtract_int(&vm_cnt.v_wire_count, 1);
> > @@ -1801,15 +1611,6 @@ vm_page_alloc(vm_object_t object, vm_pin
> >  		m->pindex = pindex;
> >  
> >  	/*
> > -	 * The following call to vdrop() must come after the above call
> > -	 * to vm_page_insert() in case both affect the same object and
> > -	 * vnode.  Otherwise, the affected vnode's hold count could
> > -	 * temporarily become zero.
> > -	 */
> > -	if (vp != NULL)
> > -		vdrop(vp);
> > -
> > -	/*
> >  	 * Don't wakeup too often - wakeup the pageout daemon when
> >  	 * we would be nearly out of memory.
> >  	 */
> > @@ -1819,16 +1620,6 @@ vm_page_alloc(vm_object_t object, vm_pin
> >  	return (m);
> >  }
> >  
> > -static void
> > -vm_page_alloc_contig_vdrop(struct spglist *lst)
> > -{
> > -
> > -	while (!SLIST_EMPTY(lst)) {
> > -		vdrop((struct vnode *)SLIST_FIRST(lst)-> plinks.s.pv);
> > -		SLIST_REMOVE_HEAD(lst, plinks.s.ss);
> > -	}
> > -}
> > -
> >  /*
> >   *	vm_page_alloc_contig:
> >   *
> > @@ -1873,8 +1664,6 @@ vm_page_alloc_contig(vm_object_t object,
> >      u_long npages, vm_paddr_t low, vm_paddr_t high, u_long alignment,
> >      vm_paddr_t boundary, vm_memattr_t memattr)
> >  {
> > -	struct vnode *drop;
> > -	struct spglist deferred_vdrop_list;
> >  	vm_page_t m, m_tmp, m_ret;
> >  	u_int flags;
> >  	int req_class;
> > @@ -1900,7 +1689,6 @@ vm_page_alloc_contig(vm_object_t object,
> >  	if (curproc == pageproc && req_class != VM_ALLOC_INTERRUPT)
> >  		req_class = VM_ALLOC_SYSTEM;
> >  
> > -	SLIST_INIT(&deferred_vdrop_list);
> >  	mtx_lock(&vm_page_queue_free_mtx);
> >  	if (vm_cnt.v_free_count + vm_cnt.v_cache_count >= npages +
> >  	    vm_cnt.v_free_reserved || (req_class == VM_ALLOC_SYSTEM &&
> > @@ -1922,17 +1710,7 @@ retry:
> >  		return (NULL);
> >  	}
> >  	if (m_ret != NULL)
> > -		for (m = m_ret; m < &m_ret[npages]; m++) {
> > -			drop = vm_page_alloc_init(m);
> > -			if (drop != NULL) {
> > -				/*
> > -				 * Enqueue the vnode for deferred vdrop().
> > -				 */
> > -				m->plinks.s.pv = drop;
> > -				SLIST_INSERT_HEAD(&deferred_vdrop_list, m,
> > -				    plinks.s.ss);
> > -			}
> > -		}
> > +		vm_phys_freecnt_adj(m_ret, -npages);
> >  	else {
> >  #if VM_NRESERVLEVEL > 0
> >  		if (vm_reserv_reclaim_contig(npages, low, high, alignment,
> > @@ -1943,6 +1721,8 @@ retry:
> >  	mtx_unlock(&vm_page_queue_free_mtx);
> >  	if (m_ret == NULL)
> >  		return (NULL);
> > +	for (m = m_ret; m < &m_ret[npages]; m++)
> > +		vm_page_alloc_check(m);
> >  
> >  	/*
> >  	 * Initialize the pages.  Only the PG_ZERO flag is inherited.
> > @@ -1975,8 +1755,6 @@ retry:
> >  		m->oflags = VPO_UNMANAGED;
> >  		if (object != NULL) {
> >  			if (vm_page_insert(m, object, pindex)) {
> > -				vm_page_alloc_contig_vdrop(
> > -				    &deferred_vdrop_list);
> >  				if (vm_paging_needed())
> >  					pagedaemon_wakeup();
> >  				if ((req & VM_ALLOC_WIRED) != 0)
> > @@ -2001,57 +1779,28 @@ retry:
> >  			pmap_page_set_memattr(m, memattr);
> >  		pindex++;
> >  	}
> > -	vm_page_alloc_contig_vdrop(&deferred_vdrop_list);
> >  	if (vm_paging_needed())
> >  		pagedaemon_wakeup();
> >  	return (m_ret);
> >  }
> >  
> >  /*
> > - * Initialize a page that has been freshly dequeued from a freelist.
> > - * The caller has to drop the vnode returned, if it is not NULL.
> > - *
> > - * This function may only be used to initialize unmanaged pages.
> > - *
> > - * To be called with vm_page_queue_free_mtx held.
> > + * Check a page that has been freshly dequeued from a freelist.
> >   */
> > -static struct vnode *
> > -vm_page_alloc_init(vm_page_t m)
> > +static void
> > +vm_page_alloc_check(vm_page_t m)
> >  {
> > -	struct vnode *drop;
> > -	vm_object_t m_object;
> >  
> >  	KASSERT(m->queue == PQ_NONE,
> > -	    ("vm_page_alloc_init: page %p has unexpected queue %d",
> > -	    m, m->queue));
> > -	KASSERT(m->wire_count == 0,
> > -	    ("vm_page_alloc_init: page %p is wired", m));
> > -	KASSERT(m->hold_count == 0,
> > -	    ("vm_page_alloc_init: page %p is held", m));
> > -	KASSERT(!vm_page_busied(m),
> > -	    ("vm_page_alloc_init: page %p is busy", m));
> > -	KASSERT(m->dirty == 0,
> > -	    ("vm_page_alloc_init: page %p is dirty", m));
> > +	    ("page %p has unexpected queue %d", m, m->queue));
> > +	KASSERT(m->wire_count == 0, ("page %p is wired", m));
> > +	KASSERT(m->hold_count == 0, ("page %p is held", m));
> > +	KASSERT(!vm_page_busied(m), ("page %p is busy", m));
> > +	KASSERT(m->dirty == 0, ("page %p is dirty", m));
> >  	KASSERT(pmap_page_get_memattr(m) == VM_MEMATTR_DEFAULT,
> > -	    ("vm_page_alloc_init: page %p has unexpected memattr %d",
> > +	    ("page %p has unexpected memattr %d",
> >  	    m, pmap_page_get_memattr(m)));
> > -	mtx_assert(&vm_page_queue_free_mtx, MA_OWNED);
> > -	drop = NULL;
> > -	if ((m->flags & PG_CACHED) != 0) {
> > -		KASSERT((m->flags & PG_ZERO) == 0,
> > -		    ("vm_page_alloc_init: cached page %p is PG_ZERO", m));
> > -		m->valid = 0;
> > -		m_object = m->object;
> > -		vm_page_cache_remove(m);
> > -		if (m_object->type == OBJT_VNODE &&
> > -		    vm_object_cache_is_empty(m_object))
> > -			drop = m_object->handle;
> > -	} else {
> > -		KASSERT(m->valid == 0,
> > -		    ("vm_page_alloc_init: free page %p is valid", m));
> > -		vm_phys_freecnt_adj(m, -1);
> > -	}
> > -	return (drop);
> > +	KASSERT(m->valid == 0, ("free page %p is valid", m));
> >  }
> >  
> >  /*
> > @@ -2077,7 +1826,6 @@ vm_page_alloc_init(vm_page_t m)
> >  vm_page_t
> >  vm_page_alloc_freelist(int flind, int req)
> >  {
> > -	struct vnode *drop;
> >  	vm_page_t m;
> >  	u_int flags;
> >  	int req_class;
> > @@ -2111,8 +1859,9 @@ vm_page_alloc_freelist(int flind, int re
> >  		mtx_unlock(&vm_page_queue_free_mtx);
> >  		return (NULL);
> >  	}
> > -	drop = vm_page_alloc_init(m);
> > +	vm_phys_freecnt_adj(m, -1);
> >  	mtx_unlock(&vm_page_queue_free_mtx);
> > +	vm_page_alloc_check(m);
> >  
> >  	/*
> >  	 * Initialize the page.  Only the PG_ZERO flag is inherited.
> > @@ -2132,8 +1881,6 @@ vm_page_alloc_freelist(int flind, int re
> >  	}
> >  	/* Unmanaged pages don't use "act_count". */
> >  	m->oflags = VPO_UNMANAGED;
> > -	if (drop != NULL)
> > -		vdrop(drop);
> >  	if (vm_paging_needed())
> >  		pagedaemon_wakeup();
> >  	return (m);
> > @@ -2259,38 +2006,8 @@ retry:
> >  			/* Don't care: PG_NODUMP, PG_ZERO. */
> >  			if (object->type != OBJT_DEFAULT &&
> >  			    object->type != OBJT_SWAP &&
> > -			    object->type != OBJT_VNODE)
> > +			    object->type != OBJT_VNODE) {
> >  				run_ext = 0;
> > -			else if ((m->flags & PG_CACHED) != 0 ||
> > -			    m != vm_page_lookup(object, m->pindex)) {
> > -				/*
> > -				 * The page is cached or recently converted
> > -				 * from cached to free.
> > -				 */
> > -#if VM_NRESERVLEVEL > 0
> > -				if (level >= 0) {
> > -					/*
> > -					 * The page is reserved.  Extend the
> > -					 * current run by one page.
> > -					 */
> > -					run_ext = 1;
> > -				} else
> > -#endif
> > -				if ((order = m->order) < VM_NFREEORDER) {
> > -					/*
> > -					 * The page is enqueued in the
> > -					 * physical memory allocator's cache/
> > -					 * free page queues.  Moreover, it is
> > -					 * the first page in a power-of-two-
> > -					 * sized run of contiguous cache/free
> > -					 * pages.  Add these pages to the end
> > -					 * of the current run, and jump
> > -					 * ahead.
> > -					 */
> > -					run_ext = 1 << order;
> > -					m_inc = 1 << order;
> > -				} else
> > -					run_ext = 0;
> >  #if VM_NRESERVLEVEL > 0
> >  			} else if ((options & VPSC_NOSUPER) != 0 &&
> >  			    (level = vm_reserv_level_iffullpop(m)) >= 0) {
> > @@ -2457,15 +2174,7 @@ retry:
> >  			    object->type != OBJT_SWAP &&
> >  			    object->type != OBJT_VNODE)
> >  				error = EINVAL;
> > -			else if ((m->flags & PG_CACHED) != 0 ||
> > -			    m != vm_page_lookup(object, m->pindex)) {
> > -				/*
> > -				 * The page is cached or recently converted
> > -				 * from cached to free.
> > -				 */
> > -				VM_OBJECT_WUNLOCK(object);
> > -				goto cached;
> > -			} else if (object->memattr != VM_MEMATTR_DEFAULT)
> > +			else if (object->memattr != VM_MEMATTR_DEFAULT)
> >  				error = EINVAL;
> >  			else if (m->queue != PQ_NONE && !vm_page_busied(m)) {
> >  				KASSERT(pmap_page_get_memattr(m) ==
> > @@ -2566,7 +2275,6 @@ retry:
> >  unlock:
> >  			VM_OBJECT_WUNLOCK(object);
> >  		} else {
> > -cached:
> >  			mtx_lock(&vm_page_queue_free_mtx);
> >  			order = m->order;
> >  			if (order < VM_NFREEORDER) {
> > @@ -2964,27 +2672,6 @@ vm_page_free_wakeup(void)
> >  }
> >  
> >  /*
> > - *	Turn a cached page into a free page, by changing its attributes.
> > - *	Keep the statistics up-to-date.
> > - *
> > - *	The free page queue must be locked.
> > - */
> > -static void
> > -vm_page_cache_turn_free(vm_page_t m)
> > -{
> > -
> > -	mtx_assert(&vm_page_queue_free_mtx, MA_OWNED);
> > -
> > -	m->object = NULL;
> > -	m->valid = 0;
> > -	KASSERT((m->flags & PG_CACHED) != 0,
> > -	    ("vm_page_cache_turn_free: page %p is not cached", m));
> > -	m->flags &= ~PG_CACHED;
> > -	vm_cnt.v_cache_count--;
> > -	vm_phys_freecnt_adj(m, 1);
> > -}
> > -
> > -/*
> >   *	vm_page_free_toq:
> >   *
> >   *	Returns the given page to the free list,
> > @@ -3383,8 +3070,7 @@ retrylookup:
> >  		VM_WAIT;
> >  		VM_OBJECT_WLOCK(object);
> >  		goto retrylookup;
> > -	} else if (m->valid != 0)
> > -		return (m);
> > +	}
> >  	if (allocflags & VM_ALLOC_ZERO && (m->flags & PG_ZERO) == 0)
> >  		pmap_zero_page(m);
> >  	return (m);
> > 
> > Modified: head/sys/vm/vm_page.h
> > ==============================================================================
> > --- head/sys/vm/vm_page.h	Tue Nov 15 17:01:48 2016	(r308690)
> > +++ head/sys/vm/vm_page.h	Tue Nov 15 18:22:50 2016	(r308691)
> > @@ -326,7 +326,6 @@ extern struct mtx_padalign pa_lock[];
> >   * Page flags.  If changed at any other time than page allocation or
> >   * freeing, the modification must be protected by the vm_page lock.
> >   */
> > -#define	PG_CACHED	0x0001		/* page is cached */
> >  #define	PG_FICTITIOUS	0x0004		/* physical page doesn't exist */
> >  #define	PG_ZERO		0x0008		/* page is zeroed */
> >  #define	PG_MARKER	0x0010		/* special queue marker page */
> > @@ -409,8 +408,6 @@ vm_page_t PHYS_TO_VM_PAGE(vm_paddr_t pa)
> >  #define	VM_ALLOC_ZERO		0x0040	/* (acfg) Try to obtain a zeroed page */
> >  #define	VM_ALLOC_NOOBJ		0x0100	/* (acg) No associated object */
> >  #define	VM_ALLOC_NOBUSY		0x0200	/* (acg) Do not busy the page */
> > -#define	VM_ALLOC_IFCACHED	0x0400	/* (ag) Fail if page is not cached */
> > -#define	VM_ALLOC_IFNOTCACHED	0x0800	/* (ag) Fail if page is cached */
> >  #define	VM_ALLOC_IGN_SBUSY	0x1000	/* (g) Ignore shared busy flag */
> >  #define	VM_ALLOC_NODUMP		0x2000	/* (ag) don't include in dump */
> >  #define	VM_ALLOC_SBUSY		0x4000	/* (acg) Shared busy the page */
> > @@ -453,8 +450,6 @@ vm_page_t vm_page_alloc_contig(vm_object
> >      vm_paddr_t boundary, vm_memattr_t memattr);
> >  vm_page_t vm_page_alloc_freelist(int, int);
> >  vm_page_t vm_page_grab (vm_object_t, vm_pindex_t, int);
> > -void vm_page_cache_free(vm_object_t, vm_pindex_t, vm_pindex_t);
> > -void vm_page_cache_transfer(vm_object_t, vm_pindex_t, vm_object_t);
> >  int vm_page_try_to_free (vm_page_t);
> >  void vm_page_deactivate (vm_page_t);
> >  void vm_page_deactivate_noreuse(vm_page_t);
> > @@ -464,7 +459,6 @@ vm_page_t vm_page_find_least(vm_object_t
> >  vm_page_t vm_page_getfake(vm_paddr_t paddr, vm_memattr_t memattr);
> >  void vm_page_initfake(vm_page_t m, vm_paddr_t paddr, vm_memattr_t memattr);
> >  int vm_page_insert (vm_page_t, vm_object_t, vm_pindex_t);
> > -boolean_t vm_page_is_cached(vm_object_t object, vm_pindex_t pindex);
> >  void vm_page_launder(vm_page_t m);
> >  vm_page_t vm_page_lookup (vm_object_t, vm_pindex_t);
> >  vm_page_t vm_page_next(vm_page_t m);
> > 
> > Modified: head/sys/vm/vm_reserv.c
> > ==============================================================================
> > --- head/sys/vm/vm_reserv.c	Tue Nov 15 17:01:48 2016	(r308690)
> > +++ head/sys/vm/vm_reserv.c	Tue Nov 15 18:22:50 2016	(r308691)
> > @@ -908,45 +908,6 @@ vm_reserv_level_iffullpop(vm_page_t m)
> >  }
> >  
> >  /*
> > - * Prepare for the reactivation of a cached page.
> > - *
> > - * First, suppose that the given page "m" was allocated individually, i.e., not
> > - * as part of a reservation, and cached.  Then, suppose a reservation
> > - * containing "m" is allocated by the same object.  Although "m" and the
> > - * reservation belong to the same object, "m"'s pindex may not match the
> > - * reservation's.
> > - *
> > - * The free page queue must be locked.
> > - */
> > -boolean_t
> > -vm_reserv_reactivate_page(vm_page_t m)
> > -{
> > -	vm_reserv_t rv;
> > -	int index;
> > -
> > -	mtx_assert(&vm_page_queue_free_mtx, MA_OWNED);
> > -	rv = vm_reserv_from_page(m);
> > -	if (rv->object == NULL)
> > -		return (FALSE);
> > -	KASSERT((m->flags & PG_CACHED) != 0,
> > -	    ("vm_reserv_reactivate_page: page %p is not cached", m));
> > -	if (m->object == rv->object &&
> > -	    m->pindex - rv->pindex == (index = VM_RESERV_INDEX(m->object,
> > -	    m->pindex)))
> > -		vm_reserv_populate(rv, index);
> > -	else {
> > -		KASSERT(rv->inpartpopq,
> > -	    ("vm_reserv_reactivate_page: reserv %p's inpartpopq is FALSE",
> > -		    rv));
> > -		TAILQ_REMOVE(&vm_rvq_partpop, rv, partpopq);
> > -		rv->inpartpopq = FALSE;
> > -		/* Don't release "m" to the physical memory allocator. */
> > -		vm_reserv_break(rv, m);
> > -	}
> > -	return (TRUE);
> > -}
> > -
> > -/*
> >   * Breaks the given partially-populated reservation, releasing its cached and
> >   * free pages to the physical memory allocator.
> >   *
> > 
> > Modified: head/sys/vm/vm_reserv.h
> > ==============================================================================
> > --- head/sys/vm/vm_reserv.h	Tue Nov 15 17:01:48 2016	(r308690)
> > +++ head/sys/vm/vm_reserv.h	Tue Nov 15 18:22:50 2016	(r308691)
> > @@ -56,7 +56,6 @@ void		vm_reserv_init(void);
> >  bool		vm_reserv_is_page_free(vm_page_t m);
> >  int		vm_reserv_level(vm_page_t m);
> >  int		vm_reserv_level_iffullpop(vm_page_t m);
> > -boolean_t	vm_reserv_reactivate_page(vm_page_t m);
> >  boolean_t	vm_reserv_reclaim_contig(u_long npages, vm_paddr_t low,
> >  		    vm_paddr_t high, u_long alignment, vm_paddr_t boundary);
> >  boolean_t	vm_reserv_reclaim_inactive(void);
> > 
> > Modified: head/sys/vm/vnode_pager.c
> > ==============================================================================
> > --- head/sys/vm/vnode_pager.c	Tue Nov 15 17:01:48 2016	(r308690)
> > +++ head/sys/vm/vnode_pager.c	Tue Nov 15 18:22:50 2016	(r308691)
> > @@ -466,10 +466,6 @@ vnode_pager_setsize(struct vnode *vp, vm
> >  			 * replacement from working properly.
> >  			 */
> >  			vm_page_clear_dirty(m, base, PAGE_SIZE - base);
> > -		} else if ((nsize & PAGE_MASK) &&
> > -		    vm_page_is_cached(object, OFF_TO_IDX(nsize))) {
> > -			vm_page_cache_free(object, OFF_TO_IDX(nsize),
> > -			    nobjsize);
> >  		}
> >  	}
> >  	object->un_pager.vnp.vnp_size = nsize;
> > @@ -894,8 +890,7 @@ vnode_pager_generic_getpages(struct vnod
> >  		for (tpindex = m[0]->pindex - 1;
> >  		    tpindex >= startpindex && tpindex < m[0]->pindex;
> >  		    tpindex--, i++) {
> > -			p = vm_page_alloc(object, tpindex, VM_ALLOC_NORMAL |
> > -			    VM_ALLOC_IFNOTCACHED);
> > +			p = vm_page_alloc(object, tpindex, VM_ALLOC_NORMAL);
> >  			if (p == NULL) {
> >  				/* Shift the array. */
> >  				for (int j = 0; j < i; j++)
> > @@ -932,8 +927,7 @@ vnode_pager_generic_getpages(struct vnod
> >  
> >  		for (tpindex = m[count - 1]->pindex + 1;
> >  		    tpindex < endpindex; i++, tpindex++) {
> > -			p = vm_page_alloc(object, tpindex, VM_ALLOC_NORMAL |
> > -			    VM_ALLOC_IFNOTCACHED);
> > +			p = vm_page_alloc(object, tpindex, VM_ALLOC_NORMAL);
> >  			if (p == NULL)
> >  				break;
> >  			bp->b_pages[i] = p;
> > 



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20161116165343.GX54029>