version 1.75, 2002/11/27 04:07:42 |
version 1.83.2.5, 2004/09/18 14:53:04 |
|
|
* 2. Redistributions in binary form must reproduce the above copyright |
* 2. Redistributions in binary form must reproduce the above copyright |
* notice, this list of conditions and the following disclaimer in the |
* notice, this list of conditions and the following disclaimer in the |
* documentation and/or other materials provided with the distribution. |
* documentation and/or other materials provided with the distribution. |
* 3. All advertising materials mentioning features or use of this software |
* 3. Neither the name of the University nor the names of its contributors |
* must display the following acknowledgement: |
|
* This product includes software developed by the University of |
|
* California, Berkeley and its contributors. |
|
* 4. Neither the name of the University nor the names of its contributors |
|
* may be used to endorse or promote products derived from this software |
* may be used to endorse or promote products derived from this software |
* without specific prior written permission. |
* without specific prior written permission. |
* |
* |
Line 76 __KERNEL_RCSID(0, "$NetBSD$"); |
|
Line 72 __KERNEL_RCSID(0, "$NetBSD$"); |
|
|
|
#include "opt_sock_counters.h" |
#include "opt_sock_counters.h" |
#include "opt_sosend_loan.h" |
#include "opt_sosend_loan.h" |
|
#include "opt_mbuftrace.h" |
|
#include "opt_somaxkva.h" |
|
|
#include <sys/param.h> |
#include <sys/param.h> |
#include <sys/systm.h> |
#include <sys/systm.h> |
Line 92 __KERNEL_RCSID(0, "$NetBSD$"); |
|
Line 90 __KERNEL_RCSID(0, "$NetBSD$"); |
|
#include <sys/resourcevar.h> |
#include <sys/resourcevar.h> |
#include <sys/pool.h> |
#include <sys/pool.h> |
#include <sys/event.h> |
#include <sys/event.h> |
|
#include <sys/poll.h> |
|
|
#include <uvm/uvm.h> |
#include <uvm/uvm.h> |
|
|
struct pool socket_pool; |
POOL_INIT(socket_pool, sizeof(struct socket), 0, 0, 0, "sockpl", NULL); |
|
|
|
MALLOC_DEFINE(M_SOOPTS, "soopts", "socket options"); |
|
MALLOC_DEFINE(M_SONAME, "soname", "socket name"); |
|
|
extern int somaxconn; /* patchable (XXX sysctl) */ |
extern int somaxconn; /* patchable (XXX sysctl) */ |
int somaxconn = SOMAXCONN; |
int somaxconn = SOMAXCONN; |
Line 114 struct evcnt sosend_kvalimit = EVCNT_INI |
|
Line 116 struct evcnt sosend_kvalimit = EVCNT_INI |
|
|
|
#define SOSEND_COUNTER_INCR(ev) (ev)->ev_count++ |
#define SOSEND_COUNTER_INCR(ev) (ev)->ev_count++ |
|
|
|
EVCNT_ATTACH_STATIC(sosend_loan_big); |
|
EVCNT_ATTACH_STATIC(sosend_copy_big); |
|
EVCNT_ATTACH_STATIC(sosend_copy_small); |
|
EVCNT_ATTACH_STATIC(sosend_kvalimit); |
#else |
#else |
|
|
#define SOSEND_COUNTER_INCR(ev) /* nothing */ |
#define SOSEND_COUNTER_INCR(ev) /* nothing */ |
|
|
soinit(void) |
soinit(void) |
{ |
{ |
|
|
pool_init(&socket_pool, sizeof(struct socket), 0, 0, 0, |
/* Set the initial adjusted socket buffer size. */ |
"sockpl", NULL); |
if (sb_max_set(sb_max)) |
|
panic("bad initial sb_max value: %lu\n", sb_max); |
|
|
#ifdef SOSEND_COUNTERS |
|
evcnt_attach_static(&sosend_loan_big); |
|
evcnt_attach_static(&sosend_copy_big); |
|
evcnt_attach_static(&sosend_copy_small); |
|
evcnt_attach_static(&sosend_kvalimit); |
|
#endif /* SOSEND_COUNTERS */ |
|
} |
} |
|
|
#ifdef SOSEND_NO_LOAN |
#ifdef SOSEND_NO_LOAN |
Line 141 int use_sosend_loan = 0; |
|
Line 142 int use_sosend_loan = 0; |
|
int use_sosend_loan = 1; |
int use_sosend_loan = 1; |
#endif |
#endif |
|
|
|
struct simplelock so_pendfree_slock = SIMPLELOCK_INITIALIZER; |
struct mbuf *so_pendfree; |
struct mbuf *so_pendfree; |
|
|
int somaxkva = 16 * 1024 * 1024; |
#ifndef SOMAXKVA |
|
#define SOMAXKVA (16 * 1024 * 1024) |
|
#endif |
|
int somaxkva = SOMAXKVA; |
int socurkva; |
int socurkva; |
int sokvawaiters; |
int sokvawaiters; |
|
|
#define SOCK_LOAN_THRESH 4096 |
#define SOCK_LOAN_THRESH 4096 |
#define SOCK_LOAN_CHUNK 65536 |
#define SOCK_LOAN_CHUNK 65536 |
|
|
|
static size_t sodopendfree(struct socket *); |
|
static size_t sodopendfreel(struct socket *); |
|
static __inline vsize_t sokvareserve(struct socket *, vsize_t); |
|
static __inline void sokvaunreserve(vsize_t); |
|
|
|
static __inline vsize_t |
|
sokvareserve(struct socket *so, vsize_t len) |
|
{ |
|
int s; |
|
int error; |
|
|
|
s = splvm(); |
|
simple_lock(&so_pendfree_slock); |
|
while (socurkva + len > somaxkva) { |
|
size_t freed; |
|
|
|
/* |
|
* try to do pendfree. |
|
*/ |
|
|
|
freed = sodopendfreel(so); |
|
|
|
/* |
|
* if some kva was freed, try again. |
|
*/ |
|
|
|
if (freed) |
|
continue; |
|
|
|
SOSEND_COUNTER_INCR(&sosend_kvalimit); |
|
sokvawaiters++; |
|
error = ltsleep(&socurkva, PVM | PCATCH, "sokva", 0, |
|
&so_pendfree_slock); |
|
sokvawaiters--; |
|
if (error) { |
|
len = 0; |
|
break; |
|
} |
|
} |
|
socurkva += len; |
|
simple_unlock(&so_pendfree_slock); |
|
splx(s); |
|
return len; |
|
} |
|
|
|
static __inline void |
|
sokvaunreserve(vsize_t len) |
|
{ |
|
int s; |
|
|
|
s = splvm(); |
|
simple_lock(&so_pendfree_slock); |
|
socurkva -= len; |
|
if (sokvawaiters) |
|
wakeup(&socurkva); |
|
simple_unlock(&so_pendfree_slock); |
|
splx(s); |
|
} |
|
|
|
/* |
|
* sokvaalloc: allocate kva for loan. |
|
*/ |
|
|
|
vaddr_t |
|
sokvaalloc(vsize_t len, struct socket *so) |
|
{ |
|
vaddr_t lva; |
|
|
|
/* |
|
* reserve kva. |
|
*/ |
|
|
|
if (sokvareserve(so, len) == 0) |
|
return 0; |
|
|
|
/* |
|
* allocate kva. |
|
*/ |
|
|
|
lva = uvm_km_valloc_wait(kernel_map, len); |
|
if (lva == 0) { |
|
sokvaunreserve(len); |
|
return (0); |
|
} |
|
|
|
return lva; |
|
} |
|
|
|
/* |
|
* sokvafree: free kva for loan. |
|
*/ |
|
|
|
void |
|
sokvafree(vaddr_t sva, vsize_t len) |
|
{ |
|
|
|
/* |
|
* free kva. |
|
*/ |
|
|
|
uvm_km_free(kernel_map, sva, len); |
|
|
|
/* |
|
* unreserve kva. |
|
*/ |
|
|
|
sokvaunreserve(len); |
|
} |
|
|
static void |
static void |
sodoloanfree(caddr_t buf, u_int size) |
sodoloanfree(struct vm_page **pgs, caddr_t buf, size_t size) |
{ |
{ |
struct vm_page **pgs; |
|
vaddr_t va, sva, eva; |
vaddr_t va, sva, eva; |
vsize_t len; |
vsize_t len; |
paddr_t pa; |
paddr_t pa; |
Line 164 sodoloanfree(caddr_t buf, u_int size) |
|
Line 277 sodoloanfree(caddr_t buf, u_int size) |
|
len = eva - sva; |
len = eva - sva; |
npgs = len >> PAGE_SHIFT; |
npgs = len >> PAGE_SHIFT; |
|
|
pgs = alloca(npgs * sizeof(*pgs)); |
if (__predict_false(pgs == NULL)) { |
|
pgs = alloca(npgs * sizeof(*pgs)); |
|
|
for (i = 0, va = sva; va < eva; i++, va += PAGE_SIZE) { |
for (i = 0, va = sva; va < eva; i++, va += PAGE_SIZE) { |
if (pmap_extract(pmap_kernel(), va, &pa) == FALSE) |
if (pmap_extract(pmap_kernel(), va, &pa) == FALSE) |
panic("sodoloanfree: va 0x%lx not mapped", va); |
panic("sodoloanfree: va 0x%lx not mapped", va); |
pgs[i] = PHYS_TO_VM_PAGE(pa); |
pgs[i] = PHYS_TO_VM_PAGE(pa); |
|
} |
} |
} |
|
|
pmap_kremove(sva, len); |
pmap_kremove(sva, len); |
pmap_update(pmap_kernel()); |
pmap_update(pmap_kernel()); |
uvm_unloan(pgs, npgs, UVM_LOAN_TOPAGE); |
uvm_unloan(pgs, npgs, UVM_LOAN_TOPAGE); |
uvm_km_free(kernel_map, sva, len); |
sokvafree(sva, len); |
socurkva -= len; |
|
if (sokvawaiters) |
|
wakeup(&socurkva); |
|
} |
} |
|
|
static size_t |
static size_t |
sodopendfree(struct socket *so) |
sodopendfree(struct socket *so) |
{ |
{ |
struct mbuf *m; |
|
size_t rv = 0; |
|
int s; |
int s; |
|
size_t rv; |
|
|
s = splvm(); |
s = splvm(); |
|
simple_lock(&so_pendfree_slock); |
|
rv = sodopendfreel(so); |
|
simple_unlock(&so_pendfree_slock); |
|
splx(s); |
|
|
for (;;) { |
return rv; |
m = so_pendfree; |
} |
if (m == NULL) |
|
break; |
|
so_pendfree = m->m_next; |
|
splx(s); |
|
|
|
rv += m->m_ext.ext_size; |
/* |
sodoloanfree(m->m_ext.ext_buf, m->m_ext.ext_size); |
* sodopendfreel: free mbufs on "pendfree" list. |
s = splvm(); |
* unlock and relock so_pendfree_slock when freeing mbufs. |
pool_cache_put(&mbpool_cache, m); |
* |
} |
* => called with so_pendfree_slock held. |
|
* => called at splvm. |
|
*/ |
|
|
|
static size_t |
|
sodopendfreel(struct socket *so) |
|
{ |
|
size_t rv = 0; |
|
|
|
LOCK_ASSERT(simple_lock_held(&so_pendfree_slock)); |
|
|
for (;;) { |
for (;;) { |
m = so->so_pendfree; |
struct mbuf *m; |
|
struct mbuf *next; |
|
|
|
m = so_pendfree; |
if (m == NULL) |
if (m == NULL) |
break; |
break; |
so->so_pendfree = m->m_next; |
so_pendfree = NULL; |
splx(s); |
simple_unlock(&so_pendfree_slock); |
|
/* XXX splx */ |
|
|
|
for (; m != NULL; m = next) { |
|
next = m->m_next; |
|
|
|
rv += m->m_ext.ext_size; |
|
sodoloanfree((m->m_flags & M_EXT_PAGES) ? |
|
m->m_ext.ext_pgs : NULL, m->m_ext.ext_buf, |
|
m->m_ext.ext_size); |
|
pool_cache_put(&mbpool_cache, m); |
|
} |
|
|
rv += m->m_ext.ext_size; |
/* XXX splvm */ |
sodoloanfree(m->m_ext.ext_buf, m->m_ext.ext_size); |
simple_lock(&so_pendfree_slock); |
s = splvm(); |
|
pool_cache_put(&mbpool_cache, m); |
|
} |
} |
|
|
splx(s); |
|
return (rv); |
return (rv); |
} |
} |
|
|
static void |
void |
soloanfree(struct mbuf *m, caddr_t buf, u_int size, void *arg) |
soloanfree(struct mbuf *m, caddr_t buf, size_t size, void *arg) |
{ |
{ |
struct socket *so = arg; |
|
int s; |
int s; |
|
|
if (m == NULL) { |
if (m == NULL) { |
sodoloanfree(buf, size); |
|
|
/* |
|
* called from MEXTREMOVE. |
|
*/ |
|
|
|
sodoloanfree(NULL, buf, size); |
return; |
return; |
} |
} |
|
|
|
/* |
|
* postpone freeing mbuf. |
|
* |
|
* we can't do it in interrupt context |
|
* because we need to put kva back to kernel_map. |
|
*/ |
|
|
s = splvm(); |
s = splvm(); |
m->m_next = so->so_pendfree; |
simple_lock(&so_pendfree_slock); |
so->so_pendfree = m; |
m->m_next = so_pendfree; |
splx(s); |
so_pendfree = m; |
if (sokvawaiters) |
if (sokvawaiters) |
wakeup(&socurkva); |
wakeup(&socurkva); |
|
simple_unlock(&so_pendfree_slock); |
|
splx(s); |
} |
} |
|
|
static long |
static long |
Line 245 sosend_loan(struct socket *so, struct ui |
|
Line 389 sosend_loan(struct socket *so, struct ui |
|
struct iovec *iov = uio->uio_iov; |
struct iovec *iov = uio->uio_iov; |
vaddr_t sva, eva; |
vaddr_t sva, eva; |
vsize_t len; |
vsize_t len; |
struct vm_page **pgs; |
|
vaddr_t lva, va; |
vaddr_t lva, va; |
int npgs, s, i, error; |
int npgs, i, error; |
|
|
if (uio->uio_segflg != UIO_USERSPACE) |
if (uio->uio_segflg != UIO_USERSPACE) |
return (0); |
return (0); |
Line 262 sosend_loan(struct socket *so, struct ui |
|
Line 405 sosend_loan(struct socket *so, struct ui |
|
len = eva - sva; |
len = eva - sva; |
npgs = len >> PAGE_SHIFT; |
npgs = len >> PAGE_SHIFT; |
|
|
while (socurkva + len > somaxkva) { |
/* XXX KDASSERT */ |
if (sodopendfree(so)) |
KASSERT(npgs <= M_EXT_MAXPAGES); |
continue; |
KASSERT(uio->uio_procp != NULL); |
SOSEND_COUNTER_INCR(&sosend_kvalimit); |
|
s = splvm(); |
|
sokvawaiters++; |
|
(void) tsleep(&socurkva, PVM, "sokva", 0); |
|
sokvawaiters--; |
|
splx(s); |
|
} |
|
|
|
lva = uvm_km_valloc_wait(kernel_map, len); |
lva = sokvaalloc(len, so); |
if (lva == 0) |
if (lva == 0) |
return (0); |
return 0; |
socurkva += len; |
|
|
|
pgs = alloca(npgs * sizeof(*pgs)); |
|
|
|
error = uvm_loan(&uio->uio_procp->p_vmspace->vm_map, sva, len, |
error = uvm_loan(&uio->uio_procp->p_vmspace->vm_map, sva, len, |
pgs, UVM_LOAN_TOPAGE); |
m->m_ext.ext_pgs, UVM_LOAN_TOPAGE); |
if (error) { |
if (error) { |
uvm_km_free(kernel_map, lva, len); |
sokvafree(lva, len); |
socurkva -= len; |
|
return (0); |
return (0); |
} |
} |
|
|
for (i = 0, va = lva; i < npgs; i++, va += PAGE_SIZE) |
for (i = 0, va = lva; i < npgs; i++, va += PAGE_SIZE) |
pmap_kenter_pa(va, VM_PAGE_TO_PHYS(pgs[i]), VM_PROT_READ); |
pmap_kenter_pa(va, VM_PAGE_TO_PHYS(m->m_ext.ext_pgs[i]), |
|
VM_PROT_READ); |
pmap_update(pmap_kernel()); |
pmap_update(pmap_kernel()); |
|
|
lva += (vaddr_t) iov->iov_base & PAGE_MASK; |
lva += (vaddr_t) iov->iov_base & PAGE_MASK; |
|
|
MEXTADD(m, (caddr_t) lva, space, M_MBUF, soloanfree, so); |
MEXTADD(m, (caddr_t) lva, space, M_MBUF, soloanfree, so); |
|
m->m_flags |= M_EXT_PAGES | M_EXT_ROMAP; |
|
|
uio->uio_resid -= space; |
uio->uio_resid -= space; |
/* uio_offset not updated, not set/used for write(2) */ |
/* uio_offset not updated, not set/used for write(2) */ |
Line 317 sosend_loan(struct socket *so, struct ui |
|
Line 451 sosend_loan(struct socket *so, struct ui |
|
*/ |
*/ |
/*ARGSUSED*/ |
/*ARGSUSED*/ |
int |
int |
socreate(int dom, struct socket **aso, int type, int proto) |
socreate(int dom, struct socket **aso, int type, int proto, struct proc *p) |
{ |
{ |
struct proc *p; |
const struct protosw *prp; |
struct protosw *prp; |
|
struct socket *so; |
struct socket *so; |
int error, s; |
int error, s; |
|
|
p = curproc; /* XXX */ |
|
if (proto) |
if (proto) |
prp = pffindproto(dom, proto, type); |
prp = pffindproto(dom, proto, type); |
else |
else |
Line 342 socreate(int dom, struct socket **aso, i |
|
Line 474 socreate(int dom, struct socket **aso, i |
|
so->so_proto = prp; |
so->so_proto = prp; |
so->so_send = sosend; |
so->so_send = sosend; |
so->so_receive = soreceive; |
so->so_receive = soreceive; |
|
#ifdef MBUFTRACE |
|
so->so_rcv.sb_mowner = &prp->pr_domain->dom_mowner; |
|
so->so_snd.sb_mowner = &prp->pr_domain->dom_mowner; |
|
so->so_mowner = &prp->pr_domain->dom_mowner; |
|
#endif |
if (p != 0) |
if (p != 0) |
so->so_uid = p->p_ucred->cr_uid; |
so->so_uid = p->p_ucred->cr_uid; |
|
else |
|
so->so_uid = UID_MAX; |
error = (*prp->pr_usrreq)(so, PRU_ATTACH, (struct mbuf *)0, |
error = (*prp->pr_usrreq)(so, PRU_ATTACH, (struct mbuf *)0, |
(struct mbuf *)(long)proto, (struct mbuf *)0, p); |
(struct mbuf *)(long)proto, (struct mbuf *)0, p); |
if (error) { |
if (error) { |
Line 393 solisten(struct socket *so, int backlog) |
|
Line 532 solisten(struct socket *so, int backlog) |
|
void |
void |
sofree(struct socket *so) |
sofree(struct socket *so) |
{ |
{ |
struct mbuf *m; |
|
|
|
if (so->so_pcb || (so->so_state & SS_NOFDREF) == 0) |
if (so->so_pcb || (so->so_state & SS_NOFDREF) == 0) |
return; |
return; |
Line 406 sofree(struct socket *so) |
|
Line 544 sofree(struct socket *so) |
|
if (!soqremque(so, 0)) |
if (!soqremque(so, 0)) |
return; |
return; |
} |
} |
sbrelease(&so->so_snd); |
if (so->so_rcv.sb_hiwat) |
|
(void)chgsbsize(so->so_uid, &so->so_rcv.sb_hiwat, 0, |
|
RLIM_INFINITY); |
|
if (so->so_snd.sb_hiwat) |
|
(void)chgsbsize(so->so_uid, &so->so_snd.sb_hiwat, 0, |
|
RLIM_INFINITY); |
|
sbrelease(&so->so_snd, so); |
sorflush(so); |
sorflush(so); |
while ((m = so->so_pendfree) != NULL) { |
|
so->so_pendfree = m->m_next; |
|
m->m_next = so_pendfree; |
|
so_pendfree = m; |
|
} |
|
pool_put(&socket_pool, so); |
pool_put(&socket_pool, so); |
} |
} |
|
|
Line 510 soaccept(struct socket *so, struct mbuf |
|
Line 649 soaccept(struct socket *so, struct mbuf |
|
} |
} |
|
|
int |
int |
soconnect(struct socket *so, struct mbuf *nam) |
soconnect(struct socket *so, struct mbuf *nam, struct proc *p) |
{ |
{ |
struct proc *p; |
|
int s, error; |
int s, error; |
|
|
p = curproc; /* XXX */ |
|
if (so->so_options & SO_ACCEPTCONN) |
if (so->so_options & SO_ACCEPTCONN) |
return (EOPNOTSUPP); |
return (EOPNOTSUPP); |
s = splsoftnet(); |
s = splsoftnet(); |
Line 592 sodisconnect(struct socket *so) |
|
Line 729 sodisconnect(struct socket *so) |
|
*/ |
*/ |
int |
int |
sosend(struct socket *so, struct mbuf *addr, struct uio *uio, struct mbuf *top, |
sosend(struct socket *so, struct mbuf *addr, struct uio *uio, struct mbuf *top, |
struct mbuf *control, int flags) |
struct mbuf *control, int flags, struct proc *p) |
{ |
{ |
struct proc *p; |
|
struct mbuf **mp, *m; |
struct mbuf **mp, *m; |
long space, len, resid, clen, mlen; |
long space, len, resid, clen, mlen; |
int error, s, dontroute, atomic; |
int error, s, dontroute, atomic; |
|
|
sodopendfree(so); |
sodopendfree(so); |
|
|
p = curproc; /* XXX */ |
|
clen = 0; |
clen = 0; |
atomic = sosendallatonce(so) || top; |
atomic = sosendallatonce(so) || top; |
if (uio) |
if (uio) |
Line 622 sosend(struct socket *so, struct mbuf *a |
|
Line 757 sosend(struct socket *so, struct mbuf *a |
|
dontroute = |
dontroute = |
(flags & MSG_DONTROUTE) && (so->so_options & SO_DONTROUTE) == 0 && |
(flags & MSG_DONTROUTE) && (so->so_options & SO_DONTROUTE) == 0 && |
(so->so_proto->pr_flags & PR_ATOMIC); |
(so->so_proto->pr_flags & PR_ATOMIC); |
p->p_stats->p_ru.ru_msgsnd++; |
if (p) |
|
p->p_stats->p_ru.ru_msgsnd++; |
if (control) |
if (control) |
clen = control->m_len; |
clen = control->m_len; |
#define snderr(errno) { error = errno; splx(s); goto release; } |
#define snderr(errno) { error = errno; splx(s); goto release; } |
Line 654 sosend(struct socket *so, struct mbuf *a |
|
Line 790 sosend(struct socket *so, struct mbuf *a |
|
if ((atomic && resid > so->so_snd.sb_hiwat) || |
if ((atomic && resid > so->so_snd.sb_hiwat) || |
clen > so->so_snd.sb_hiwat) |
clen > so->so_snd.sb_hiwat) |
snderr(EMSGSIZE); |
snderr(EMSGSIZE); |
if (space < resid + clen && uio && |
if (space < resid + clen && |
(atomic || space < so->so_snd.sb_lowat || space < clen)) { |
(atomic || space < so->so_snd.sb_lowat || space < clen)) { |
if (so->so_state & SS_NBIO) |
if (so->so_state & SS_NBIO) |
snderr(EWOULDBLOCK); |
snderr(EWOULDBLOCK); |
Line 678 sosend(struct socket *so, struct mbuf *a |
|
Line 814 sosend(struct socket *so, struct mbuf *a |
|
top->m_flags |= M_EOR; |
top->m_flags |= M_EOR; |
} else do { |
} else do { |
if (top == 0) { |
if (top == 0) { |
MGETHDR(m, M_WAIT, MT_DATA); |
m = m_gethdr(M_WAIT, MT_DATA); |
mlen = MHLEN; |
mlen = MHLEN; |
m->m_pkthdr.len = 0; |
m->m_pkthdr.len = 0; |
m->m_pkthdr.rcvif = (struct ifnet *)0; |
m->m_pkthdr.rcvif = (struct ifnet *)0; |
} else { |
} else { |
MGET(m, M_WAIT, MT_DATA); |
m = m_get(M_WAIT, MT_DATA); |
mlen = MLEN; |
mlen = MLEN; |
} |
} |
|
MCLAIM(m, so->so_snd.sb_mowner); |
if (use_sosend_loan && |
if (use_sosend_loan && |
uio->uio_iov->iov_len >= SOCK_LOAN_THRESH && |
uio->uio_iov->iov_len >= SOCK_LOAN_THRESH && |
space >= SOCK_LOAN_THRESH && |
space >= SOCK_LOAN_THRESH && |
Line 697 sosend(struct socket *so, struct mbuf *a |
|
Line 834 sosend(struct socket *so, struct mbuf *a |
|
} |
} |
if (resid >= MINCLSIZE && space >= MCLBYTES) { |
if (resid >= MINCLSIZE && space >= MCLBYTES) { |
SOSEND_COUNTER_INCR(&sosend_copy_big); |
SOSEND_COUNTER_INCR(&sosend_copy_big); |
MCLGET(m, M_WAIT); |
m_clget(m, M_WAIT); |
if ((m->m_flags & M_EXT) == 0) |
if ((m->m_flags & M_EXT) == 0) |
goto nopages; |
goto nopages; |
mlen = MCLBYTES; |
mlen = MCLBYTES; |
|
|
soreceive(struct socket *so, struct mbuf **paddr, struct uio *uio, |
soreceive(struct socket *so, struct mbuf **paddr, struct uio *uio, |
struct mbuf **mp0, struct mbuf **controlp, int *flagsp) |
struct mbuf **mp0, struct mbuf **controlp, int *flagsp) |
{ |
{ |
|
struct proc * p; |
struct mbuf *m, **mp; |
struct mbuf *m, **mp; |
int flags, len, error, s, offset, moff, type, orig_resid; |
int flags, len, error, s, offset, moff, type, orig_resid; |
struct protosw *pr; |
const struct protosw *pr; |
struct mbuf *nextrecord; |
struct mbuf *nextrecord; |
int mbuf_removed = 0; |
int mbuf_removed = 0; |
|
|
Line 804 soreceive(struct socket *so, struct mbuf |
|
Line 942 soreceive(struct socket *so, struct mbuf |
|
mp = mp0; |
mp = mp0; |
type = 0; |
type = 0; |
orig_resid = uio->uio_resid; |
orig_resid = uio->uio_resid; |
|
p = uio->uio_procp; |
|
|
if (paddr) |
if (paddr) |
*paddr = 0; |
*paddr = 0; |
if (controlp) |
if (controlp) |
Line 819 soreceive(struct socket *so, struct mbuf |
|
Line 959 soreceive(struct socket *so, struct mbuf |
|
if (flags & MSG_OOB) { |
if (flags & MSG_OOB) { |
m = m_get(M_WAIT, MT_DATA); |
m = m_get(M_WAIT, MT_DATA); |
error = (*pr->pr_usrreq)(so, PRU_RCVOOB, m, |
error = (*pr->pr_usrreq)(so, PRU_RCVOOB, m, |
(struct mbuf *)(long)(flags & MSG_PEEK), (struct mbuf *)0, |
(struct mbuf *)(long)(flags & MSG_PEEK), |
(struct proc *)0); |
(struct mbuf *)0, p); |
if (error) |
if (error) |
goto bad; |
goto bad; |
do { |
do { |
Line 837 soreceive(struct socket *so, struct mbuf |
|
Line 977 soreceive(struct socket *so, struct mbuf |
|
*mp = (struct mbuf *)0; |
*mp = (struct mbuf *)0; |
if (so->so_state & SS_ISCONFIRMING && uio->uio_resid) |
if (so->so_state & SS_ISCONFIRMING && uio->uio_resid) |
(*pr->pr_usrreq)(so, PRU_RCVD, (struct mbuf *)0, |
(*pr->pr_usrreq)(so, PRU_RCVD, (struct mbuf *)0, |
(struct mbuf *)0, (struct mbuf *)0, (struct proc *)0); |
(struct mbuf *)0, (struct mbuf *)0, p); |
|
|
restart: |
restart: |
if ((error = sblock(&so->so_rcv, SBLOCKWAIT(flags))) != 0) |
if ((error = sblock(&so->so_rcv, SBLOCKWAIT(flags))) != 0) |
Line 910 soreceive(struct socket *so, struct mbuf |
|
Line 1050 soreceive(struct socket *so, struct mbuf |
|
* While we process the initial mbufs containing address and control |
* While we process the initial mbufs containing address and control |
* info, we save a copy of m->m_nextpkt into nextrecord. |
* info, we save a copy of m->m_nextpkt into nextrecord. |
*/ |
*/ |
#ifdef notyet /* XXXX */ |
if (p) |
if (uio->uio_procp) |
p->p_stats->p_ru.ru_msgrcv++; |
uio->uio_procp->p_stats->p_ru.ru_msgrcv++; |
|
#endif |
|
KASSERT(m == so->so_rcv.sb_mb); |
KASSERT(m == so->so_rcv.sb_mb); |
SBLASTRECORDCHK(&so->so_rcv, "soreceive 1"); |
SBLASTRECORDCHK(&so->so_rcv, "soreceive 1"); |
SBLASTMBUFCHK(&so->so_rcv, "soreceive 1"); |
SBLASTMBUFCHK(&so->so_rcv, "soreceive 1"); |
Line 951 soreceive(struct socket *so, struct mbuf |
|
Line 1089 soreceive(struct socket *so, struct mbuf |
|
sbfree(&so->so_rcv, m); |
sbfree(&so->so_rcv, m); |
mbuf_removed = 1; |
mbuf_removed = 1; |
if (controlp) { |
if (controlp) { |
if (pr->pr_domain->dom_externalize && |
struct domain *dom = pr->pr_domain; |
|
if (dom->dom_externalize && p && |
mtod(m, struct cmsghdr *)->cmsg_type == |
mtod(m, struct cmsghdr *)->cmsg_type == |
SCM_RIGHTS) |
SCM_RIGHTS) |
error = (*pr->pr_domain->dom_externalize)(m); |
error = (*dom->dom_externalize)(m, p); |
*controlp = m; |
*controlp = m; |
so->so_rcv.sb_mb = m->m_next; |
so->so_rcv.sb_mb = m->m_next; |
m->m_next = 0; |
m->m_next = 0; |
m = so->so_rcv.sb_mb; |
m = so->so_rcv.sb_mb; |
} else { |
} else { |
|
/* |
|
* Dispose of any SCM_RIGHTS message that went |
|
* through the read path rather than recv. |
|
*/ |
|
if (pr->pr_domain->dom_dispose && |
|
mtod(m, struct cmsghdr *)->cmsg_type == SCM_RIGHTS) |
|
(*pr->pr_domain->dom_dispose)(m); |
MFREE(m, so->so_rcv.sb_mb); |
MFREE(m, so->so_rcv.sb_mb); |
m = so->so_rcv.sb_mb; |
m = so->so_rcv.sb_mb; |
} |
} |
Line 1140 soreceive(struct socket *so, struct mbuf |
|
Line 1286 soreceive(struct socket *so, struct mbuf |
|
(*pr->pr_usrreq)(so, PRU_RCVD, |
(*pr->pr_usrreq)(so, PRU_RCVD, |
(struct mbuf *)0, |
(struct mbuf *)0, |
(struct mbuf *)(long)flags, |
(struct mbuf *)(long)flags, |
(struct mbuf *)0, |
(struct mbuf *)0, p); |
(struct proc *)0); |
|
SBLASTRECORDCHK(&so->so_rcv, "soreceive sbwait 2"); |
SBLASTRECORDCHK(&so->so_rcv, "soreceive sbwait 2"); |
SBLASTMBUFCHK(&so->so_rcv, "soreceive sbwait 2"); |
SBLASTMBUFCHK(&so->so_rcv, "soreceive sbwait 2"); |
error = sbwait(&so->so_rcv); |
error = sbwait(&so->so_rcv); |
Line 1178 soreceive(struct socket *so, struct mbuf |
|
Line 1323 soreceive(struct socket *so, struct mbuf |
|
SBLASTMBUFCHK(&so->so_rcv, "soreceive 4"); |
SBLASTMBUFCHK(&so->so_rcv, "soreceive 4"); |
if (pr->pr_flags & PR_WANTRCVD && so->so_pcb) |
if (pr->pr_flags & PR_WANTRCVD && so->so_pcb) |
(*pr->pr_usrreq)(so, PRU_RCVD, (struct mbuf *)0, |
(*pr->pr_usrreq)(so, PRU_RCVD, (struct mbuf *)0, |
(struct mbuf *)(long)flags, (struct mbuf *)0, |
(struct mbuf *)(long)flags, (struct mbuf *)0, p); |
(struct proc *)0); |
|
} |
} |
if (orig_resid == uio->uio_resid && orig_resid && |
if (orig_resid == uio->uio_resid && orig_resid && |
(flags & MSG_EOR) == 0 && (so->so_state & SS_CANTRCVMORE) == 0) { |
(flags & MSG_EOR) == 0 && (so->so_state & SS_CANTRCVMORE) == 0) { |
Line 1199 soreceive(struct socket *so, struct mbuf |
|
Line 1343 soreceive(struct socket *so, struct mbuf |
|
int |
int |
soshutdown(struct socket *so, int how) |
soshutdown(struct socket *so, int how) |
{ |
{ |
struct protosw *pr; |
const struct protosw *pr; |
|
|
pr = so->so_proto; |
pr = so->so_proto; |
if (!(how == SHUT_RD || how == SHUT_WR || how == SHUT_RDWR)) |
if (!(how == SHUT_RD || how == SHUT_WR || how == SHUT_RDWR)) |
|
|
sorflush(struct socket *so) |
sorflush(struct socket *so) |
{ |
{ |
struct sockbuf *sb, asb; |
struct sockbuf *sb, asb; |
struct protosw *pr; |
const struct protosw *pr; |
int s; |
int s; |
|
|
sb = &so->so_rcv; |
sb = &so->so_rcv; |
Line 1228 sorflush(struct socket *so) |
|
Line 1372 sorflush(struct socket *so) |
|
socantrcvmore(so); |
socantrcvmore(so); |
sbunlock(sb); |
sbunlock(sb); |
asb = *sb; |
asb = *sb; |
memset((caddr_t)sb, 0, sizeof(*sb)); |
/* |
|
* Clear most of the sockbuf structure, but leave some of the |
|
* fields valid. |
|
*/ |
|
memset(&sb->sb_startzero, 0, |
|
sizeof(*sb) - offsetof(struct sockbuf, sb_startzero)); |
splx(s); |
splx(s); |
if (pr->pr_flags & PR_RIGHTS && pr->pr_domain->dom_dispose) |
if (pr->pr_flags & PR_RIGHTS && pr->pr_domain->dom_dispose) |
(*pr->pr_domain->dom_dispose)(asb.sb_mb); |
(*pr->pr_domain->dom_dispose)(asb.sb_mb); |
sbrelease(&asb); |
sbrelease(&asb, so); |
} |
} |
|
|
int |
int |
Line 1306 sosetopt(struct socket *so, int level, i |
|
Line 1455 sosetopt(struct socket *so, int level, i |
|
case SO_RCVBUF: |
case SO_RCVBUF: |
if (sbreserve(optname == SO_SNDBUF ? |
if (sbreserve(optname == SO_SNDBUF ? |
&so->so_snd : &so->so_rcv, |
&so->so_snd : &so->so_rcv, |
(u_long) optval) == 0) { |
(u_long) optval, so) == 0) { |
error = ENOBUFS; |
error = ENOBUFS; |
goto bad; |
goto bad; |
} |
} |
Line 1334 sosetopt(struct socket *so, int level, i |
|
Line 1483 sosetopt(struct socket *so, int level, i |
|
case SO_RCVTIMEO: |
case SO_RCVTIMEO: |
{ |
{ |
struct timeval *tv; |
struct timeval *tv; |
short val; |
int val; |
|
|
if (m == NULL || m->m_len < sizeof(*tv)) { |
if (m == NULL || m->m_len < sizeof(*tv)) { |
error = EINVAL; |
error = EINVAL; |
goto bad; |
goto bad; |
} |
} |
tv = mtod(m, struct timeval *); |
tv = mtod(m, struct timeval *); |
if (tv->tv_sec > (SHRT_MAX - tv->tv_usec / tick) / hz) { |
if (tv->tv_sec > (INT_MAX - tv->tv_usec / tick) / hz) { |
error = EDOM; |
error = EDOM; |
goto bad; |
goto bad; |
} |
} |
Line 1451 sogetopt(struct socket *so, int level, i |
|
Line 1600 sogetopt(struct socket *so, int level, i |
|
break; |
break; |
} |
} |
|
|
|
case SO_OVERFLOWED: |
|
*mtod(m, int *) = so->so_rcv.sb_overflowed; |
|
break; |
|
|
default: |
default: |
(void)m_free(m); |
(void)m_free(m); |
return (ENOPROTOOPT); |
return (ENOPROTOOPT); |
Line 1463 sogetopt(struct socket *so, int level, i |
|
Line 1616 sogetopt(struct socket *so, int level, i |
|
void |
void |
sohasoutofband(struct socket *so) |
sohasoutofband(struct socket *so) |
{ |
{ |
struct proc *p; |
fownsignal(so->so_pgid, SIGURG, POLL_PRI, POLLPRI|POLLRDBAND, so); |
|
|
if (so->so_pgid < 0) |
|
gsignal(-so->so_pgid, SIGURG); |
|
else if (so->so_pgid > 0 && (p = pfind(so->so_pgid)) != 0) |
|
psignal(p, SIGURG); |
|
selwakeup(&so->so_rcv.sb_sel); |
selwakeup(&so->so_rcv.sb_sel); |
} |
} |
|
|
Line 1587 soo_kqfilter(struct file *fp, struct kno |
|
Line 1735 soo_kqfilter(struct file *fp, struct kno |
|
return (0); |
return (0); |
} |
} |
|
|
|
#include <sys/sysctl.h> |
|
|
|
static int sysctl_kern_somaxkva(SYSCTLFN_PROTO); |
|
|
|
/* |
|
* sysctl helper routine for kern.somaxkva. ensures that the given |
|
* value is not too small. |
|
* (XXX should we maybe make sure it's not too large as well?) |
|
*/ |
|
static int |
|
sysctl_kern_somaxkva(SYSCTLFN_ARGS) |
|
{ |
|
int error, new_somaxkva; |
|
struct sysctlnode node; |
|
int s; |
|
|
|
new_somaxkva = somaxkva; |
|
node = *rnode; |
|
node.sysctl_data = &new_somaxkva; |
|
error = sysctl_lookup(SYSCTLFN_CALL(&node)); |
|
if (error || newp == NULL) |
|
return (error); |
|
|
|
if (new_somaxkva < (16 * 1024 * 1024)) /* sanity */ |
|
return (EINVAL); |
|
|
|
s = splvm(); |
|
simple_lock(&so_pendfree_slock); |
|
somaxkva = new_somaxkva; |
|
wakeup(&socurkva); |
|
simple_unlock(&so_pendfree_slock); |
|
splx(s); |
|
|
|
return (error); |
|
} |
|
|
|
SYSCTL_SETUP(sysctl_kern_somaxkva_setup, "sysctl kern.somaxkva setup") |
|
{ |
|
|
|
sysctl_createv(clog, 0, NULL, NULL, |
|
CTLFLAG_PERMANENT, |
|
CTLTYPE_NODE, "kern", NULL, |
|
NULL, 0, NULL, 0, |
|
CTL_KERN, CTL_EOL); |
|
|
|
sysctl_createv(clog, 0, NULL, NULL, |
|
CTLFLAG_PERMANENT|CTLFLAG_READWRITE, |
|
CTLTYPE_INT, "somaxkva", |
|
SYSCTL_DESCR("Maximum amount of kernel memory to be " |
|
"used for socket buffers"), |
|
sysctl_kern_somaxkva, 0, NULL, 0, |
|
CTL_KERN, KERN_SOMAXKVA, CTL_EOL); |
|
} |