aboutsummaryrefslogtreecommitdiffstats
diff options
context:
space:
mode:
authorIngo Molnar <mingo@elte.hu>2011-06-04 06:28:05 -0400
committerIngo Molnar <mingo@elte.hu>2011-06-04 06:28:05 -0400
commit3ce2a0bc9dfb6423491afe0afc9f099e24b8cba4 (patch)
tree58fbef582846fef0e777b1a552aca12e21a071b1
parentaef29bf20bd79c73992ab23d5067e9f0448b466e (diff)
parentaa4a221875873d2a1f9656cb7fd7e545e952b4fa (diff)
Merge branch 'perf/urgent' into perf/core
Conflicts: tools/perf/util/python.c Merge reason: resolve the conflict with perf/urgent. Signed-off-by: Ingo Molnar <mingo@elte.hu>
Diffstat
-rw-r--r--Documentation/kernel-parameters.txt5
-rw-r--r--Documentation/virtual/lguest/Makefile2
-rw-r--r--Documentation/virtual/lguest/lguest.c22
-rw-r--r--arch/ia64/include/asm/unistd.h3
-rw-r--r--arch/ia64/kernel/entry.S1
-rw-r--r--arch/powerpc/platforms/powermac/pic.c3
-rw-r--r--arch/x86/kernel/Makefile2
-rw-r--r--arch/x86/kernel/process.c2
-rw-r--r--arch/x86/kernel/smpboot.c2
-rw-r--r--arch/x86/lguest/boot.c1
-rw-r--r--drivers/block/paride/pcd.c1
-rw-r--r--drivers/block/virtio_blk.c91
-rw-r--r--drivers/cdrom/viocd.c1
-rw-r--r--drivers/char/virtio_console.c5
-rw-r--r--drivers/ide/ide-cd.c1
-rw-r--r--drivers/misc/kgdbts.c5
-rw-r--r--drivers/net/virtio_net.c2
-rw-r--r--drivers/pci/dmar.c7
-rw-r--r--drivers/pci/intel-iommu.c240
-rw-r--r--drivers/pci/iova.c12
-rw-r--r--drivers/vhost/net.c12
-rw-r--r--drivers/vhost/test.c6
-rw-r--r--drivers/vhost/vhost.c138
-rw-r--r--drivers/vhost/vhost.h21
-rw-r--r--drivers/virtio/virtio_balloon.c21
-rw-r--r--drivers/virtio/virtio_ring.c53
-rw-r--r--fs/autofs4/root.c2
-rw-r--r--fs/namei.c3
-rw-r--r--include/linux/dma_remapping.h4
-rw-r--r--include/linux/mtd/physmap.h1
-rw-r--r--include/linux/virtio.h9
-rw-r--r--include/linux/virtio_9p.h25
-rw-r--r--include/linux/virtio_balloon.h25
-rw-r--r--include/linux/virtio_blk.h25
-rw-r--r--include/linux/virtio_config.h25
-rw-r--r--include/linux/virtio_console.h26
-rw-r--r--include/linux/virtio_ids.h24
-rw-r--r--include/linux/virtio_net.h25
-rw-r--r--include/linux/virtio_pci.h23
-rw-r--r--include/linux/virtio_ring.h52
-rw-r--r--kernel/events/core.c22
-rw-r--r--kernel/rcutree.c54
-rw-r--r--kernel/rcutree_plugin.h11
-rw-r--r--kernel/sysctl.c6
-rw-r--r--mm/page_alloc.c4
-rw-r--r--security/apparmor/lsm.c3
-rw-r--r--tools/virtio/virtio_test.c19
47 files changed, 844 insertions, 203 deletions
diff --git a/Documentation/kernel-parameters.txt b/Documentation/kernel-parameters.txt
index 5438a2d7907f..d9a203b058f1 100644
--- a/Documentation/kernel-parameters.txt
+++ b/Documentation/kernel-parameters.txt
@@ -999,7 +999,10 @@ bytes respectively. Such letter suffixes can also be entirely omitted.
999 With this option on every unmap_single operation will 999 With this option on every unmap_single operation will
1000 result in a hardware IOTLB flush operation as opposed 1000 result in a hardware IOTLB flush operation as opposed
1001 to batching them for performance. 1001 to batching them for performance.
1002 1002 sp_off [Default Off]
1003 By default, super page will be supported if Intel IOMMU
1004 has the capability. With this option, super page will
1005 not be supported.
1003 intremap= [X86-64, Intel-IOMMU] 1006 intremap= [X86-64, Intel-IOMMU]
1004 Format: { on (default) | off | nosid } 1007 Format: { on (default) | off | nosid }
1005 on enable Interrupt Remapping (default) 1008 on enable Interrupt Remapping (default)
diff --git a/Documentation/virtual/lguest/Makefile b/Documentation/virtual/lguest/Makefile
index bebac6b4f332..0ac34206f7a7 100644
--- a/Documentation/virtual/lguest/Makefile
+++ b/Documentation/virtual/lguest/Makefile
@@ -1,5 +1,5 @@
1# This creates the demonstration utility "lguest" which runs a Linux guest. 1# This creates the demonstration utility "lguest" which runs a Linux guest.
2# Missing headers? Add "-I../../include -I../../arch/x86/include" 2# Missing headers? Add "-I../../../include -I../../../arch/x86/include"
3CFLAGS:=-m32 -Wall -Wmissing-declarations -Wmissing-prototypes -O3 -U_FORTIFY_SOURCE 3CFLAGS:=-m32 -Wall -Wmissing-declarations -Wmissing-prototypes -O3 -U_FORTIFY_SOURCE
4 4
5all: lguest 5all: lguest
diff --git a/Documentation/virtual/lguest/lguest.c b/Documentation/virtual/lguest/lguest.c
index d9da7e148538..cd9d6af61d07 100644
--- a/Documentation/virtual/lguest/lguest.c
+++ b/Documentation/virtual/lguest/lguest.c
@@ -49,7 +49,7 @@
49#include <linux/virtio_rng.h> 49#include <linux/virtio_rng.h>
50#include <linux/virtio_ring.h> 50#include <linux/virtio_ring.h>
51#include <asm/bootparam.h> 51#include <asm/bootparam.h>
52#include "../../include/linux/lguest_launcher.h" 52#include "../../../include/linux/lguest_launcher.h"
53/*L:110 53/*L:110
54 * We can ignore the 42 include files we need for this program, but I do want 54 * We can ignore the 42 include files we need for this program, but I do want
55 * to draw attention to the use of kernel-style types. 55 * to draw attention to the use of kernel-style types.
@@ -135,9 +135,6 @@ struct device {
135 /* Is it operational */ 135 /* Is it operational */
136 bool running; 136 bool running;
137 137
138 /* Does Guest want an intrrupt on empty? */
139 bool irq_on_empty;
140
141 /* Device-specific data. */ 138 /* Device-specific data. */
142 void *priv; 139 void *priv;
143}; 140};
@@ -637,10 +634,7 @@ static void trigger_irq(struct virtqueue *vq)
637 634
638 /* If they don't want an interrupt, don't send one... */ 635 /* If they don't want an interrupt, don't send one... */
639 if (vq->vring.avail->flags & VRING_AVAIL_F_NO_INTERRUPT) { 636 if (vq->vring.avail->flags & VRING_AVAIL_F_NO_INTERRUPT) {
640 /* ... unless they've asked us to force one on empty. */ 637 return;
641 if (!vq->dev->irq_on_empty
642 || lg_last_avail(vq) != vq->vring.avail->idx)
643 return;
644 } 638 }
645 639
646 /* Send the Guest an interrupt tell them we used something up. */ 640 /* Send the Guest an interrupt tell them we used something up. */
@@ -1057,15 +1051,6 @@ static void create_thread(struct virtqueue *vq)
1057 close(vq->eventfd); 1051 close(vq->eventfd);
1058} 1052}
1059 1053
1060static bool accepted_feature(struct device *dev, unsigned int bit)
1061{
1062 const u8 *features = get_feature_bits(dev) + dev->feature_len;
1063
1064 if (dev->feature_len < bit / CHAR_BIT)
1065 return false;
1066 return features[bit / CHAR_BIT] & (1 << (bit % CHAR_BIT));
1067}
1068
1069static void start_device(struct device *dev) 1054static void start_device(struct device *dev)
1070{ 1055{
1071 unsigned int i; 1056 unsigned int i;
@@ -1079,8 +1064,6 @@ static void start_device(struct device *dev)
1079 verbose(" %02x", get_feature_bits(dev) 1064 verbose(" %02x", get_feature_bits(dev)
1080 [dev->feature_len+i]); 1065 [dev->feature_len+i]);
1081 1066
1082 dev->irq_on_empty = accepted_feature(dev, VIRTIO_F_NOTIFY_ON_EMPTY);
1083
1084 for (vq = dev->vq; vq; vq = vq->next) { 1067 for (vq = dev->vq; vq; vq = vq->next) {
1085 if (vq->service) 1068 if (vq->service)
1086 create_thread(vq); 1069 create_thread(vq);
@@ -1564,7 +1547,6 @@ static void setup_tun_net(char *arg)
1564 /* Set up the tun device. */ 1547 /* Set up the tun device. */
1565 configure_device(ipfd, tapif, ip); 1548 configure_device(ipfd, tapif, ip);
1566 1549
1567 add_feature(dev, VIRTIO_F_NOTIFY_ON_EMPTY);
1568 /* Expect Guest to handle everything except UFO */ 1550 /* Expect Guest to handle everything except UFO */
1569 add_feature(dev, VIRTIO_NET_F_CSUM); 1551 add_feature(dev, VIRTIO_NET_F_CSUM);
1570 add_feature(dev, VIRTIO_NET_F_GUEST_CSUM); 1552 add_feature(dev, VIRTIO_NET_F_GUEST_CSUM);
diff --git a/arch/ia64/include/asm/unistd.h b/arch/ia64/include/asm/unistd.h
index 1cf0f496f744..7c928da35b17 100644
--- a/arch/ia64/include/asm/unistd.h
+++ b/arch/ia64/include/asm/unistd.h
@@ -320,11 +320,12 @@
320#define __NR_clock_adjtime 1328 320#define __NR_clock_adjtime 1328
321#define __NR_syncfs 1329 321#define __NR_syncfs 1329
322#define __NR_setns 1330 322#define __NR_setns 1330
323#define __NR_sendmmsg 1331
323 324
324#ifdef __KERNEL__ 325#ifdef __KERNEL__
325 326
326 327
327#define NR_syscalls 307 /* length of syscall table */ 328#define NR_syscalls 308 /* length of syscall table */
328 329
329/* 330/*
330 * The following defines stop scripts/checksyscalls.sh from complaining about 331 * The following defines stop scripts/checksyscalls.sh from complaining about
diff --git a/arch/ia64/kernel/entry.S b/arch/ia64/kernel/entry.S
index 9ca80193cd4e..97dd2abdeb1a 100644
--- a/arch/ia64/kernel/entry.S
+++ b/arch/ia64/kernel/entry.S
@@ -1776,6 +1776,7 @@ sys_call_table:
1776 data8 sys_clock_adjtime 1776 data8 sys_clock_adjtime
1777 data8 sys_syncfs 1777 data8 sys_syncfs
1778 data8 sys_setns // 1330 1778 data8 sys_setns // 1330
1779 data8 sys_sendmmsg
1779 1780
1780 .org sys_call_table + 8*NR_syscalls // guard against failures to increase NR_syscalls 1781 .org sys_call_table + 8*NR_syscalls // guard against failures to increase NR_syscalls
1781#endif /* __IA64_ASM_PARAVIRTUALIZED_NATIVE */ 1782#endif /* __IA64_ASM_PARAVIRTUALIZED_NATIVE */
diff --git a/arch/powerpc/platforms/powermac/pic.c b/arch/powerpc/platforms/powermac/pic.c
index 9089b0421191..7667db448aa7 100644
--- a/arch/powerpc/platforms/powermac/pic.c
+++ b/arch/powerpc/platforms/powermac/pic.c
@@ -715,7 +715,8 @@ static struct syscore_ops pmacpic_syscore_ops = {
715 715
716static int __init init_pmacpic_syscore(void) 716static int __init init_pmacpic_syscore(void)
717{ 717{
718 register_syscore_ops(&pmacpic_syscore_ops); 718 if (pmac_irq_hw[0])
719 register_syscore_ops(&pmacpic_syscore_ops);
719 return 0; 720 return 0;
720} 721}
721 722
diff --git a/arch/x86/kernel/Makefile b/arch/x86/kernel/Makefile
index f5abe3a245b8..90b06d4daee2 100644
--- a/arch/x86/kernel/Makefile
+++ b/arch/x86/kernel/Makefile
@@ -8,6 +8,7 @@ CPPFLAGS_vmlinux.lds += -U$(UTS_MACHINE)
8 8
9ifdef CONFIG_FUNCTION_TRACER 9ifdef CONFIG_FUNCTION_TRACER
10# Do not profile debug and lowlevel utilities 10# Do not profile debug and lowlevel utilities
11CFLAGS_REMOVE_tsc.o = -pg
11CFLAGS_REMOVE_rtc.o = -pg 12CFLAGS_REMOVE_rtc.o = -pg
12CFLAGS_REMOVE_paravirt-spinlocks.o = -pg 13CFLAGS_REMOVE_paravirt-spinlocks.o = -pg
13CFLAGS_REMOVE_pvclock.o = -pg 14CFLAGS_REMOVE_pvclock.o = -pg
@@ -28,6 +29,7 @@ CFLAGS_paravirt.o := $(nostackp)
28GCOV_PROFILE_vsyscall_64.o := n 29GCOV_PROFILE_vsyscall_64.o := n
29GCOV_PROFILE_hpet.o := n 30GCOV_PROFILE_hpet.o := n
30GCOV_PROFILE_tsc.o := n 31GCOV_PROFILE_tsc.o := n
32GCOV_PROFILE_vread_tsc_64.o := n
31GCOV_PROFILE_paravirt.o := n 33GCOV_PROFILE_paravirt.o := n
32 34
33# vread_tsc_64 is hot and should be fully optimized: 35# vread_tsc_64 is hot and should be fully optimized:
diff --git a/arch/x86/kernel/process.c b/arch/x86/kernel/process.c
index 426a5b66f7e4..2e4928d45a2d 100644
--- a/arch/x86/kernel/process.c
+++ b/arch/x86/kernel/process.c
@@ -642,7 +642,7 @@ static int __init idle_setup(char *str)
642 boot_option_idle_override = IDLE_POLL; 642 boot_option_idle_override = IDLE_POLL;
643 } else if (!strcmp(str, "mwait")) { 643 } else if (!strcmp(str, "mwait")) {
644 boot_option_idle_override = IDLE_FORCE_MWAIT; 644 boot_option_idle_override = IDLE_FORCE_MWAIT;
645 WARN_ONCE(1, "\idle=mwait\" will be removed in 2012\"\n"); 645 WARN_ONCE(1, "\"idle=mwait\" will be removed in 2012\n");
646 } else if (!strcmp(str, "halt")) { 646 } else if (!strcmp(str, "halt")) {
647 /* 647 /*
648 * When the boot option of idle=halt is added, halt is 648 * When the boot option of idle=halt is added, halt is
diff --git a/arch/x86/kernel/smpboot.c b/arch/x86/kernel/smpboot.c
index eefd96765e79..33a0c11797de 100644
--- a/arch/x86/kernel/smpboot.c
+++ b/arch/x86/kernel/smpboot.c
@@ -1332,7 +1332,7 @@ static inline void mwait_play_dead(void)
1332 void *mwait_ptr; 1332 void *mwait_ptr;
1333 struct cpuinfo_x86 *c = __this_cpu_ptr(&cpu_info); 1333 struct cpuinfo_x86 *c = __this_cpu_ptr(&cpu_info);
1334 1334
1335 if (!this_cpu_has(X86_FEATURE_MWAIT) && mwait_usable(c)) 1335 if (!(this_cpu_has(X86_FEATURE_MWAIT) && mwait_usable(c)))
1336 return; 1336 return;
1337 if (!this_cpu_has(X86_FEATURE_CLFLSH)) 1337 if (!this_cpu_has(X86_FEATURE_CLFLSH))
1338 return; 1338 return;
diff --git a/arch/x86/lguest/boot.c b/arch/x86/lguest/boot.c
index e191c096ab90..db832fd65ecb 100644
--- a/arch/x86/lguest/boot.c
+++ b/arch/x86/lguest/boot.c
@@ -993,6 +993,7 @@ static void lguest_time_irq(unsigned int irq, struct irq_desc *desc)
993static void lguest_time_init(void) 993static void lguest_time_init(void)
994{ 994{
995 /* Set up the timer interrupt (0) to go to our simple timer routine */ 995 /* Set up the timer interrupt (0) to go to our simple timer routine */
996 lguest_setup_irq(0);
996 irq_set_handler(0, lguest_time_irq); 997 irq_set_handler(0, lguest_time_irq);
997 998
998 clocksource_register_hz(&lguest_clock, NSEC_PER_SEC); 999 clocksource_register_hz(&lguest_clock, NSEC_PER_SEC);
diff --git a/drivers/block/paride/pcd.c b/drivers/block/paride/pcd.c
index a0aabd904a51..46b8136c31bb 100644
--- a/drivers/block/paride/pcd.c
+++ b/drivers/block/paride/pcd.c
@@ -321,7 +321,6 @@ static void pcd_init_units(void)
321 strcpy(disk->disk_name, cd->name); /* umm... */ 321 strcpy(disk->disk_name, cd->name); /* umm... */
322 disk->fops = &pcd_bdops; 322 disk->fops = &pcd_bdops;
323 disk->flags = GENHD_FL_BLOCK_EVENTS_ON_EXCL_WRITE; 323 disk->flags = GENHD_FL_BLOCK_EVENTS_ON_EXCL_WRITE;
324 disk->events = DISK_EVENT_MEDIA_CHANGE;
325 } 324 }
326} 325}
327 326
diff --git a/drivers/block/virtio_blk.c b/drivers/block/virtio_blk.c
index 6ecf89cdf006..079c08808d8a 100644
--- a/drivers/block/virtio_blk.c
+++ b/drivers/block/virtio_blk.c
@@ -6,10 +6,13 @@
6#include <linux/virtio.h> 6#include <linux/virtio.h>
7#include <linux/virtio_blk.h> 7#include <linux/virtio_blk.h>
8#include <linux/scatterlist.h> 8#include <linux/scatterlist.h>
9#include <linux/string_helpers.h>
10#include <scsi/scsi_cmnd.h>
9 11
10#define PART_BITS 4 12#define PART_BITS 4
11 13
12static int major, index; 14static int major, index;
15struct workqueue_struct *virtblk_wq;
13 16
14struct virtio_blk 17struct virtio_blk
15{ 18{
@@ -26,6 +29,9 @@ struct virtio_blk
26 29
27 mempool_t *pool; 30 mempool_t *pool;
28 31
32 /* Process context for config space updates */
33 struct work_struct config_work;
34
29 /* What host tells us, plus 2 for header & tailer. */ 35 /* What host tells us, plus 2 for header & tailer. */
30 unsigned int sg_elems; 36 unsigned int sg_elems;
31 37
@@ -141,7 +147,7 @@ static bool do_req(struct request_queue *q, struct virtio_blk *vblk,
141 num = blk_rq_map_sg(q, vbr->req, vblk->sg + out); 147 num = blk_rq_map_sg(q, vbr->req, vblk->sg + out);
142 148
143 if (vbr->req->cmd_type == REQ_TYPE_BLOCK_PC) { 149 if (vbr->req->cmd_type == REQ_TYPE_BLOCK_PC) {
144 sg_set_buf(&vblk->sg[num + out + in++], vbr->req->sense, 96); 150 sg_set_buf(&vblk->sg[num + out + in++], vbr->req->sense, SCSI_SENSE_BUFFERSIZE);
145 sg_set_buf(&vblk->sg[num + out + in++], &vbr->in_hdr, 151 sg_set_buf(&vblk->sg[num + out + in++], &vbr->in_hdr,
146 sizeof(vbr->in_hdr)); 152 sizeof(vbr->in_hdr));
147 } 153 }
@@ -291,6 +297,46 @@ static ssize_t virtblk_serial_show(struct device *dev,
291} 297}
292DEVICE_ATTR(serial, S_IRUGO, virtblk_serial_show, NULL); 298DEVICE_ATTR(serial, S_IRUGO, virtblk_serial_show, NULL);
293 299
300static void virtblk_config_changed_work(struct work_struct *work)
301{
302 struct virtio_blk *vblk =
303 container_of(work, struct virtio_blk, config_work);
304 struct virtio_device *vdev = vblk->vdev;
305 struct request_queue *q = vblk->disk->queue;
306 char cap_str_2[10], cap_str_10[10];
307 u64 capacity, size;
308
309 /* Host must always specify the capacity. */
310 vdev->config->get(vdev, offsetof(struct virtio_blk_config, capacity),
311 &capacity, sizeof(capacity));
312
313 /* If capacity is too big, truncate with warning. */
314 if ((sector_t)capacity != capacity) {
315 dev_warn(&vdev->dev, "Capacity %llu too large: truncating\n",
316 (unsigned long long)capacity);
317 capacity = (sector_t)-1;
318 }
319
320 size = capacity * queue_logical_block_size(q);
321 string_get_size(size, STRING_UNITS_2, cap_str_2, sizeof(cap_str_2));
322 string_get_size(size, STRING_UNITS_10, cap_str_10, sizeof(cap_str_10));
323
324 dev_notice(&vdev->dev,
325 "new size: %llu %d-byte logical blocks (%s/%s)\n",
326 (unsigned long long)capacity,
327 queue_logical_block_size(q),
328 cap_str_10, cap_str_2);
329
330 set_capacity(vblk->disk, capacity);
331}
332
333static void virtblk_config_changed(struct virtio_device *vdev)
334{
335 struct virtio_blk *vblk = vdev->priv;
336
337 queue_work(virtblk_wq, &vblk->config_work);
338}
339
294static int __devinit virtblk_probe(struct virtio_device *vdev) 340static int __devinit virtblk_probe(struct virtio_device *vdev)
295{ 341{
296 struct virtio_blk *vblk; 342 struct virtio_blk *vblk;
@@ -327,6 +373,7 @@ static int __devinit virtblk_probe(struct virtio_device *vdev)
327 vblk->vdev = vdev; 373 vblk->vdev = vdev;
328 vblk->sg_elems = sg_elems; 374 vblk->sg_elems = sg_elems;
329 sg_init_table(vblk->sg, vblk->sg_elems); 375 sg_init_table(vblk->sg, vblk->sg_elems);
376 INIT_WORK(&vblk->config_work, virtblk_config_changed_work);
330 377
331 /* We expect one virtqueue, for output. */ 378 /* We expect one virtqueue, for output. */
332 vblk->vq = virtio_find_single_vq(vdev, blk_done, "requests"); 379 vblk->vq = virtio_find_single_vq(vdev, blk_done, "requests");
@@ -477,6 +524,8 @@ static void __devexit virtblk_remove(struct virtio_device *vdev)
477{ 524{
478 struct virtio_blk *vblk = vdev->priv; 525 struct virtio_blk *vblk = vdev->priv;
479 526
527 flush_work(&vblk->config_work);
528
480 /* Nothing should be pending. */ 529 /* Nothing should be pending. */
481 BUG_ON(!list_empty(&vblk->reqs)); 530 BUG_ON(!list_empty(&vblk->reqs));
482 531
@@ -508,27 +557,47 @@ static unsigned int features[] = {
508 * Use __refdata to avoid this warning. 557 * Use __refdata to avoid this warning.
509 */ 558 */
510static struct virtio_driver __refdata virtio_blk = { 559static struct virtio_driver __refdata virtio_blk = {
511 .feature_table = features, 560 .feature_table = features,
512 .feature_table_size = ARRAY_SIZE(features), 561 .feature_table_size = ARRAY_SIZE(features),
513 .driver.name = KBUILD_MODNAME, 562 .driver.name = KBUILD_MODNAME,
514 .driver.owner = THIS_MODULE, 563 .driver.owner = THIS_MODULE,
515 .id_table = id_table, 564 .id_table = id_table,
516 .probe = virtblk_probe, 565 .probe = virtblk_probe,
517 .remove = __devexit_p(virtblk_remove), 566 .remove = __devexit_p(virtblk_remove),
567 .config_changed = virtblk_config_changed,
518}; 568};
519 569
520static int __init init(void) 570static int __init init(void)
521{ 571{
572 int error;
573
574 virtblk_wq = alloc_workqueue("virtio-blk", 0, 0);
575 if (!virtblk_wq)
576 return -ENOMEM;
577
522 major = register_blkdev(0, "virtblk"); 578 major = register_blkdev(0, "virtblk");
523 if (major < 0) 579 if (major < 0) {
524 return major; 580 error = major;
525 return register_virtio_driver(&virtio_blk); 581 goto out_destroy_workqueue;
582 }
583
584 error = register_virtio_driver(&virtio_blk);
585 if (error)
586 goto out_unregister_blkdev;
587 return 0;
588
589out_unregister_blkdev:
590 unregister_blkdev(major, "virtblk");
591out_destroy_workqueue:
592 destroy_workqueue(virtblk_wq);
593 return error;
526} 594}
527 595
528static void __exit fini(void) 596static void __exit fini(void)
529{ 597{
530 unregister_blkdev(major, "virtblk"); 598 unregister_blkdev(major, "virtblk");
531 unregister_virtio_driver(&virtio_blk); 599 unregister_virtio_driver(&virtio_blk);
600 destroy_workqueue(virtblk_wq);
532} 601}
533module_init(init); 602module_init(init);
534module_exit(fini); 603module_exit(fini);
diff --git a/drivers/cdrom/viocd.c b/drivers/cdrom/viocd.c
index ae15a4ddaa9b..7878da89d29e 100644
--- a/drivers/cdrom/viocd.c
+++ b/drivers/cdrom/viocd.c
@@ -627,7 +627,6 @@ static int viocd_probe(struct vio_dev *vdev, const struct vio_device_id *id)
627 gendisk->fops = &viocd_fops; 627 gendisk->fops = &viocd_fops;
628 gendisk->flags = GENHD_FL_CD | GENHD_FL_REMOVABLE | 628 gendisk->flags = GENHD_FL_CD | GENHD_FL_REMOVABLE |
629 GENHD_FL_BLOCK_EVENTS_ON_EXCL_WRITE; 629 GENHD_FL_BLOCK_EVENTS_ON_EXCL_WRITE;
630 gendisk->events = DISK_EVENT_MEDIA_CHANGE;
631 set_capacity(gendisk, 0); 630 set_capacity(gendisk, 0);
632 gendisk->private_data = d; 631 gendisk->private_data = d;
633 d->viocd_disk = gendisk; 632 d->viocd_disk = gendisk;
diff --git a/drivers/char/virtio_console.c b/drivers/char/virtio_console.c
index 838568a7dbf5..fb68b1295373 100644
--- a/drivers/char/virtio_console.c
+++ b/drivers/char/virtio_console.c
@@ -1677,17 +1677,12 @@ static int __devinit virtcons_probe(struct virtio_device *vdev)
1677 portdev->config.max_nr_ports = 1; 1677 portdev->config.max_nr_ports = 1;
1678 if (virtio_has_feature(vdev, VIRTIO_CONSOLE_F_MULTIPORT)) { 1678 if (virtio_has_feature(vdev, VIRTIO_CONSOLE_F_MULTIPORT)) {
1679 multiport = true; 1679 multiport = true;
1680 vdev->features[0] |= 1 << VIRTIO_CONSOLE_F_MULTIPORT;
1681
1682 vdev->config->get(vdev, offsetof(struct virtio_console_config, 1680 vdev->config->get(vdev, offsetof(struct virtio_console_config,
1683 max_nr_ports), 1681 max_nr_ports),
1684 &portdev->config.max_nr_ports, 1682 &portdev->config.max_nr_ports,
1685 sizeof(portdev->config.max_nr_ports)); 1683 sizeof(portdev->config.max_nr_ports));
1686 } 1684 }
1687 1685
1688 /* Let the Host know we support multiple ports.*/
1689 vdev->config->finalize_features(vdev);
1690
1691 err = init_vqs(portdev); 1686 err = init_vqs(portdev);
1692 if (err < 0) { 1687 if (err < 0) {
1693 dev_err(&vdev->dev, "Error %d initializing vqs\n", err); 1688 dev_err(&vdev->dev, "Error %d initializing vqs\n", err);
diff --git a/drivers/ide/ide-cd.c b/drivers/ide/ide-cd.c
index 6e5123b1d341..144d27261e43 100644
--- a/drivers/ide/ide-cd.c
+++ b/drivers/ide/ide-cd.c
@@ -1782,7 +1782,6 @@ static int ide_cd_probe(ide_drive_t *drive)
1782 ide_cd_read_toc(drive, &sense); 1782 ide_cd_read_toc(drive, &sense);
1783 g->fops = &idecd_ops; 1783 g->fops = &idecd_ops;
1784 g->flags |= GENHD_FL_REMOVABLE | GENHD_FL_BLOCK_EVENTS_ON_EXCL_WRITE; 1784 g->flags |= GENHD_FL_REMOVABLE | GENHD_FL_BLOCK_EVENTS_ON_EXCL_WRITE;
1785 g->events = DISK_EVENT_MEDIA_CHANGE;
1786 add_disk(g); 1785 add_disk(g);
1787 return 0; 1786 return 0;
1788 1787
diff --git a/drivers/misc/kgdbts.c b/drivers/misc/kgdbts.c
index b0c56313dbbb..8cebec5e85ee 100644
--- a/drivers/misc/kgdbts.c
+++ b/drivers/misc/kgdbts.c
@@ -304,7 +304,10 @@ static int check_and_rewind_pc(char *put_str, char *arg)
304 return 1; 304 return 1;
305 } 305 }
306 /* Readjust the instruction pointer if needed */ 306 /* Readjust the instruction pointer if needed */
307 instruction_pointer_set(&kgdbts_regs, ip + offset); 307 ip += offset;
308#ifdef GDB_ADJUSTS_BREAK_OFFSET
309 instruction_pointer_set(&kgdbts_regs, ip);
310#endif
308 return 0; 311 return 0;
309} 312}
310 313
diff --git a/drivers/net/virtio_net.c b/drivers/net/virtio_net.c
index 0cb0b0632672..f6853247a620 100644
--- a/drivers/net/virtio_net.c
+++ b/drivers/net/virtio_net.c
@@ -609,7 +609,7 @@ static netdev_tx_t start_xmit(struct sk_buff *skb, struct net_device *dev)
609 * before it gets out of hand. Naturally, this wastes entries. */ 609 * before it gets out of hand. Naturally, this wastes entries. */
610 if (capacity < 2+MAX_SKB_FRAGS) { 610 if (capacity < 2+MAX_SKB_FRAGS) {
611 netif_stop_queue(dev); 611 netif_stop_queue(dev);
612 if (unlikely(!virtqueue_enable_cb(vi->svq))) { 612 if (unlikely(!virtqueue_enable_cb_delayed(vi->svq))) {
613 /* More just got used, free them then recheck. */ 613 /* More just got used, free them then recheck. */
614 capacity += free_old_xmit_skbs(vi); 614 capacity += free_old_xmit_skbs(vi);
615 if (capacity >= 2+MAX_SKB_FRAGS) { 615 if (capacity >= 2+MAX_SKB_FRAGS) {
diff --git a/drivers/pci/dmar.c b/drivers/pci/dmar.c
index 12e02bf92c4a..3dc9befa5aec 100644
--- a/drivers/pci/dmar.c
+++ b/drivers/pci/dmar.c
@@ -698,12 +698,7 @@ int __init detect_intel_iommu(void)
698 { 698 {
699#ifdef CONFIG_INTR_REMAP 699#ifdef CONFIG_INTR_REMAP
700 struct acpi_table_dmar *dmar; 700 struct acpi_table_dmar *dmar;
701 /* 701
702 * for now we will disable dma-remapping when interrupt
703 * remapping is enabled.
704 * When support for queued invalidation for IOTLB invalidation
705 * is added, we will not need this any more.
706 */
707 dmar = (struct acpi_table_dmar *) dmar_tbl; 702 dmar = (struct acpi_table_dmar *) dmar_tbl;
708 if (ret && cpu_has_x2apic && dmar->flags & 0x1) 703 if (ret && cpu_has_x2apic && dmar->flags & 0x1)
709 printk(KERN_INFO 704 printk(KERN_INFO
diff --git a/drivers/pci/intel-iommu.c b/drivers/pci/intel-iommu.c
index 6af6b628175b..59f17acf7f68 100644
--- a/drivers/pci/intel-iommu.c
+++ b/drivers/pci/intel-iommu.c
@@ -47,6 +47,8 @@
47#define ROOT_SIZE VTD_PAGE_SIZE 47#define ROOT_SIZE VTD_PAGE_SIZE
48#define CONTEXT_SIZE VTD_PAGE_SIZE 48#define CONTEXT_SIZE VTD_PAGE_SIZE
49 49
50#define IS_BRIDGE_HOST_DEVICE(pdev) \
51 ((pdev->class >> 8) == PCI_CLASS_BRIDGE_HOST)
50#define IS_GFX_DEVICE(pdev) ((pdev->class >> 16) == PCI_BASE_CLASS_DISPLAY) 52#define IS_GFX_DEVICE(pdev) ((pdev->class >> 16) == PCI_BASE_CLASS_DISPLAY)
51#define IS_ISA_DEVICE(pdev) ((pdev->class >> 8) == PCI_CLASS_BRIDGE_ISA) 53#define IS_ISA_DEVICE(pdev) ((pdev->class >> 8) == PCI_CLASS_BRIDGE_ISA)
52#define IS_AZALIA(pdev) ((pdev)->vendor == 0x8086 && (pdev)->device == 0x3a3e) 54#define IS_AZALIA(pdev) ((pdev)->vendor == 0x8086 && (pdev)->device == 0x3a3e)
@@ -116,6 +118,11 @@ static inline unsigned long align_to_level(unsigned long pfn, int level)
116 return (pfn + level_size(level) - 1) & level_mask(level); 118 return (pfn + level_size(level) - 1) & level_mask(level);
117} 119}
118 120
121static inline unsigned long lvl_to_nr_pages(unsigned int lvl)
122{
123 return 1 << ((lvl - 1) * LEVEL_STRIDE);
124}
125
119/* VT-d pages must always be _smaller_ than MM pages. Otherwise things 126/* VT-d pages must always be _smaller_ than MM pages. Otherwise things
120 are never going to work. */ 127 are never going to work. */
121static inline unsigned long dma_to_mm_pfn(unsigned long dma_pfn) 128static inline unsigned long dma_to_mm_pfn(unsigned long dma_pfn)
@@ -143,6 +150,12 @@ static void __init check_tylersburg_isoch(void);
143static int rwbf_quirk; 150static int rwbf_quirk;
144 151
145/* 152/*
153 * set to 1 to panic kernel if can't successfully enable VT-d
154 * (used when kernel is launched w/ TXT)
155 */
156static int force_on = 0;
157
158/*
146 * 0: Present 159 * 0: Present
147 * 1-11: Reserved 160 * 1-11: Reserved
148 * 12-63: Context Ptr (12 - (haw-1)) 161 * 12-63: Context Ptr (12 - (haw-1))
@@ -338,6 +351,9 @@ struct dmar_domain {
338 int iommu_coherency;/* indicate coherency of iommu access */ 351 int iommu_coherency;/* indicate coherency of iommu access */
339 int iommu_snooping; /* indicate snooping control feature*/ 352 int iommu_snooping; /* indicate snooping control feature*/
340 int iommu_count; /* reference count of iommu */ 353 int iommu_count; /* reference count of iommu */
354 int iommu_superpage;/* Level of superpages supported:
355 0 == 4KiB (no superpages), 1 == 2MiB,
356 2 == 1GiB, 3 == 512GiB, 4 == 1TiB */
341 spinlock_t iommu_lock; /* protect iommu set in domain */ 357 spinlock_t iommu_lock; /* protect iommu set in domain */
342 u64 max_addr; /* maximum mapped address */ 358 u64 max_addr; /* maximum mapped address */
343}; 359};
@@ -387,6 +403,7 @@ int dmar_disabled = 1;
387static int dmar_map_gfx = 1; 403static int dmar_map_gfx = 1;
388static int dmar_forcedac; 404static int dmar_forcedac;
389static int intel_iommu_strict; 405static int intel_iommu_strict;
406static int intel_iommu_superpage = 1;
390 407
391#define DUMMY_DEVICE_DOMAIN_INFO ((struct device_domain_info *)(-1)) 408#define DUMMY_DEVICE_DOMAIN_INFO ((struct device_domain_info *)(-1))
392static DEFINE_SPINLOCK(device_domain_lock); 409static DEFINE_SPINLOCK(device_domain_lock);
@@ -417,6 +434,10 @@ static int __init intel_iommu_setup(char *str)
417 printk(KERN_INFO 434 printk(KERN_INFO
418 "Intel-IOMMU: disable batched IOTLB flush\n"); 435 "Intel-IOMMU: disable batched IOTLB flush\n");
419 intel_iommu_strict = 1; 436 intel_iommu_strict = 1;
437 } else if (!strncmp(str, "sp_off", 6)) {
438 printk(KERN_INFO
439 "Intel-IOMMU: disable supported super page\n");
440 intel_iommu_superpage = 0;
420 } 441 }
421 442
422 str += strcspn(str, ","); 443 str += strcspn(str, ",");
@@ -555,11 +576,32 @@ static void domain_update_iommu_snooping(struct dmar_domain *domain)
555 } 576 }
556} 577}
557 578
579static void domain_update_iommu_superpage(struct dmar_domain *domain)
580{
581 int i, mask = 0xf;
582
583 if (!intel_iommu_superpage) {
584 domain->iommu_superpage = 0;
585 return;
586 }
587
588 domain->iommu_superpage = 4; /* 1TiB */
589
590 for_each_set_bit(i, &domain->iommu_bmp, g_num_of_iommus) {
591 mask |= cap_super_page_val(g_iommus[i]->cap);
592 if (!mask) {
593 break;
594 }
595 }
596 domain->iommu_superpage = fls(mask);
597}
598
558/* Some capabilities may be different across iommus */ 599/* Some capabilities may be different across iommus */
559static void domain_update_iommu_cap(struct dmar_domain *domain) 600static void domain_update_iommu_cap(struct dmar_domain *domain)
560{ 601{
561 domain_update_iommu_coherency(domain); 602 domain_update_iommu_coherency(domain);
562 domain_update_iommu_snooping(domain); 603 domain_update_iommu_snooping(domain);
604 domain_update_iommu_superpage(domain);
563} 605}
564 606
565static struct intel_iommu *device_to_iommu(int segment, u8 bus, u8 devfn) 607static struct intel_iommu *device_to_iommu(int segment, u8 bus, u8 devfn)
@@ -689,23 +731,31 @@ out:
689} 731}
690 732
691static struct dma_pte *pfn_to_dma_pte(struct dmar_domain *domain, 733static struct dma_pte *pfn_to_dma_pte(struct dmar_domain *domain,
692 unsigned long pfn) 734 unsigned long pfn, int large_level)
693{ 735{
694 int addr_width = agaw_to_width(domain->agaw) - VTD_PAGE_SHIFT; 736 int addr_width = agaw_to_width(domain->agaw) - VTD_PAGE_SHIFT;
695 struct dma_pte *parent, *pte = NULL; 737 struct dma_pte *parent, *pte = NULL;
696 int level = agaw_to_level(domain->agaw); 738 int level = agaw_to_level(domain->agaw);
697 int offset; 739 int offset, target_level;
698 740
699 BUG_ON(!domain->pgd); 741 BUG_ON(!domain->pgd);
700 BUG_ON(addr_width < BITS_PER_LONG && pfn >> addr_width); 742 BUG_ON(addr_width < BITS_PER_LONG && pfn >> addr_width);
701 parent = domain->pgd; 743 parent = domain->pgd;
702 744
745 /* Search pte */
746 if (!large_level)
747 target_level = 1;
748 else
749 target_level = large_level;
750
703 while (level > 0) { 751 while (level > 0) {
704 void *tmp_page; 752 void *tmp_page;
705 753
706 offset = pfn_level_offset(pfn, level); 754 offset = pfn_level_offset(pfn, level);
707 pte = &parent[offset]; 755 pte = &parent[offset];
708 if (level == 1) 756 if (!large_level && (pte->val & DMA_PTE_LARGE_PAGE))
757 break;
758 if (level == target_level)
709 break; 759 break;
710 760
711 if (!dma_pte_present(pte)) { 761 if (!dma_pte_present(pte)) {
@@ -733,10 +783,11 @@ static struct dma_pte *pfn_to_dma_pte(struct dmar_domain *domain,
733 return pte; 783 return pte;
734} 784}
735 785
786
736/* return address's pte at specific level */ 787/* return address's pte at specific level */
737static struct dma_pte *dma_pfn_level_pte(struct dmar_domain *domain, 788static struct dma_pte *dma_pfn_level_pte(struct dmar_domain *domain,
738 unsigned long pfn, 789 unsigned long pfn,
739 int level) 790 int level, int *large_page)
740{ 791{
741 struct dma_pte *parent, *pte = NULL; 792 struct dma_pte *parent, *pte = NULL;
742 int total = agaw_to_level(domain->agaw); 793 int total = agaw_to_level(domain->agaw);
@@ -749,8 +800,16 @@ static struct dma_pte *dma_pfn_level_pte(struct dmar_domain *domain,
749 if (level == total) 800 if (level == total)
750 return pte; 801 return pte;
751 802
752 if (!dma_pte_present(pte)) 803 if (!dma_pte_present(pte)) {
804 *large_page = total;
753 break; 805 break;
806 }
807
808 if (pte->val & DMA_PTE_LARGE_PAGE) {
809 *large_page = total;
810 return pte;
811 }
812
754 parent = phys_to_virt(dma_pte_addr(pte)); 813 parent = phys_to_virt(dma_pte_addr(pte));
755 total--; 814 total--;
756 } 815 }
@@ -763,6 +822,7 @@ static void dma_pte_clear_range(struct dmar_domain *domain,
763 unsigned long last_pfn) 822 unsigned long last_pfn)
764{ 823{
765 int addr_width = agaw_to_width(domain->agaw) - VTD_PAGE_SHIFT; 824 int addr_width = agaw_to_width(domain->agaw) - VTD_PAGE_SHIFT;
825 unsigned int large_page = 1;
766 struct dma_pte *first_pte, *pte; 826 struct dma_pte *first_pte, *pte;
767 827
768 BUG_ON(addr_width < BITS_PER_LONG && start_pfn >> addr_width); 828 BUG_ON(addr_width < BITS_PER_LONG && start_pfn >> addr_width);
@@ -771,14 +831,15 @@ static void dma_pte_clear_range(struct dmar_domain *domain,
771 831
772 /* we don't need lock here; nobody else touches the iova range */ 832 /* we don't need lock here; nobody else touches the iova range */
773 do { 833 do {
774 first_pte = pte = dma_pfn_level_pte(domain, start_pfn, 1); 834 large_page = 1;
835 first_pte = pte = dma_pfn_level_pte(domain, start_pfn, 1, &large_page);
775 if (!pte) { 836 if (!pte) {
776 start_pfn = align_to_level(start_pfn + 1, 2); 837 start_pfn = align_to_level(start_pfn + 1, large_page + 1);
777 continue; 838 continue;
778 } 839 }
779 do { 840 do {
780 dma_clear_pte(pte); 841 dma_clear_pte(pte);
781 start_pfn++; 842 start_pfn += lvl_to_nr_pages(large_page);
782 pte++; 843 pte++;
783 } while (start_pfn <= last_pfn && !first_pte_in_page(pte)); 844 } while (start_pfn <= last_pfn && !first_pte_in_page(pte));
784 845
@@ -798,6 +859,7 @@ static void dma_pte_free_pagetable(struct dmar_domain *domain,
798 int total = agaw_to_level(domain->agaw); 859 int total = agaw_to_level(domain->agaw);
799 int level; 860 int level;
800 unsigned long tmp; 861 unsigned long tmp;
862 int large_page = 2;
801 863
802 BUG_ON(addr_width < BITS_PER_LONG && start_pfn >> addr_width); 864 BUG_ON(addr_width < BITS_PER_LONG && start_pfn >> addr_width);
803 BUG_ON(addr_width < BITS_PER_LONG && last_pfn >> addr_width); 865 BUG_ON(addr_width < BITS_PER_LONG && last_pfn >> addr_width);
@@ -813,7 +875,10 @@ static void dma_pte_free_pagetable(struct dmar_domain *domain,
813 return; 875 return;
814 876
815 do { 877 do {
816 first_pte = pte = dma_pfn_level_pte(domain, tmp, level); 878 large_page = level;
879 first_pte = pte = dma_pfn_level_pte(domain, tmp, level, &large_page);
880 if (large_page > level)
881 level = large_page + 1;
817 if (!pte) { 882 if (!pte) {
818 tmp = align_to_level(tmp + 1, level + 1); 883 tmp = align_to_level(tmp + 1, level + 1);
819 continue; 884 continue;
@@ -1397,6 +1462,7 @@ static int domain_init(struct dmar_domain *domain, int guest_width)
1397 else 1462 else
1398 domain->iommu_snooping = 0; 1463 domain->iommu_snooping = 0;
1399 1464
1465 domain->iommu_superpage = fls(cap_super_page_val(iommu->cap));
1400 domain->iommu_count = 1; 1466 domain->iommu_count = 1;
1401 domain->nid = iommu->node; 1467 domain->nid = iommu->node;
1402 1468
@@ -1417,6 +1483,10 @@ static void domain_exit(struct dmar_domain *domain)
1417 if (!domain) 1483 if (!domain)
1418 return; 1484 return;
1419 1485
1486 /* Flush any lazy unmaps that may reference this domain */
1487 if (!intel_iommu_strict)
1488 flush_unmaps_timeout(0);
1489
1420 domain_remove_dev_info(domain); 1490 domain_remove_dev_info(domain);
1421 /* destroy iovas */ 1491 /* destroy iovas */
1422 put_iova_domain(&domain->iovad); 1492 put_iova_domain(&domain->iovad);
@@ -1648,6 +1718,34 @@ static inline unsigned long aligned_nrpages(unsigned long host_addr,
1648 return PAGE_ALIGN(host_addr + size) >> VTD_PAGE_SHIFT; 1718 return PAGE_ALIGN(host_addr + size) >> VTD_PAGE_SHIFT;
1649} 1719}
1650 1720
1721/* Return largest possible superpage level for a given mapping */
1722static inline int hardware_largepage_caps(struct dmar_domain *domain,
1723 unsigned long iov_pfn,
1724 unsigned long phy_pfn,
1725 unsigned long pages)
1726{
1727 int support, level = 1;
1728 unsigned long pfnmerge;
1729
1730 support = domain->iommu_superpage;
1731
1732 /* To use a large page, the virtual *and* physical addresses
1733 must be aligned to 2MiB/1GiB/etc. Lower bits set in either
1734 of them will mean we have to use smaller pages. So just
1735 merge them and check both at once. */
1736 pfnmerge = iov_pfn | phy_pfn;
1737
1738 while (support && !(pfnmerge & ~VTD_STRIDE_MASK)) {
1739 pages >>= VTD_STRIDE_SHIFT;
1740 if (!pages)
1741 break;
1742 pfnmerge >>= VTD_STRIDE_SHIFT;
1743 level++;
1744 support--;
1745 }
1746 return level;
1747}
1748
1651static int __domain_mapping(struct dmar_domain *domain, unsigned long iov_pfn, 1749static int __domain_mapping(struct dmar_domain *domain, unsigned long iov_pfn,
1652 struct scatterlist *sg, unsigned long phys_pfn, 1750 struct scatterlist *sg, unsigned long phys_pfn,
1653 unsigned long nr_pages, int prot) 1751 unsigned long nr_pages, int prot)
@@ -1656,6 +1754,8 @@ static int __domain_mapping(struct dmar_domain *domain, unsigned long iov_pfn,
1656 phys_addr_t uninitialized_var(pteval); 1754 phys_addr_t uninitialized_var(pteval);
1657 int addr_width = agaw_to_width(domain->agaw) - VTD_PAGE_SHIFT; 1755 int addr_width = agaw_to_width(domain->agaw) - VTD_PAGE_SHIFT;
1658 unsigned long sg_res; 1756 unsigned long sg_res;
1757 unsigned int largepage_lvl = 0;
1758 unsigned long lvl_pages = 0;
1659 1759
1660 BUG_ON(addr_width < BITS_PER_LONG && (iov_pfn + nr_pages - 1) >> addr_width); 1760 BUG_ON(addr_width < BITS_PER_LONG && (iov_pfn + nr_pages - 1) >> addr_width);
1661 1761
@@ -1671,7 +1771,7 @@ static int __domain_mapping(struct dmar_domain *domain, unsigned long iov_pfn,
1671 pteval = ((phys_addr_t)phys_pfn << VTD_PAGE_SHIFT) | prot; 1771 pteval = ((phys_addr_t)phys_pfn << VTD_PAGE_SHIFT) | prot;
1672 } 1772 }
1673 1773
1674 while (nr_pages--) { 1774 while (nr_pages > 0) {
1675 uint64_t tmp; 1775 uint64_t tmp;
1676 1776
1677 if (!sg_res) { 1777 if (!sg_res) {
@@ -1679,11 +1779,21 @@ static int __domain_mapping(struct dmar_domain *domain, unsigned long iov_pfn,
1679 sg->dma_address = ((dma_addr_t)iov_pfn << VTD_PAGE_SHIFT) + sg->offset; 1779 sg->dma_address = ((dma_addr_t)iov_pfn << VTD_PAGE_SHIFT) + sg->offset;
1680 sg->dma_length = sg->length; 1780 sg->dma_length = sg->length;
1681 pteval = page_to_phys(sg_page(sg)) | prot; 1781 pteval = page_to_phys(sg_page(sg)) | prot;
1782 phys_pfn = pteval >> VTD_PAGE_SHIFT;
1682 } 1783 }
1784
1683 if (!pte) { 1785 if (!pte) {
1684 first_pte = pte = pfn_to_dma_pte(domain, iov_pfn); 1786 largepage_lvl = hardware_largepage_caps(domain, iov_pfn, phys_pfn, sg_res);
1787
1788 first_pte = pte = pfn_to_dma_pte(domain, iov_pfn, largepage_lvl);
1685 if (!pte) 1789 if (!pte)
1686 return -ENOMEM; 1790 return -ENOMEM;
1791 /* It is large page*/
1792 if (largepage_lvl > 1)
1793 pteval |= DMA_PTE_LARGE_PAGE;
1794 else
1795 pteval &= ~(uint64_t)DMA_PTE_LARGE_PAGE;
1796
1687 } 1797 }
1688 /* We don't need lock here, nobody else 1798 /* We don't need lock here, nobody else
1689 * touches the iova range 1799 * touches the iova range
@@ -1699,16 +1809,38 @@ static int __domain_mapping(struct dmar_domain *domain, unsigned long iov_pfn,
1699 } 1809 }
1700 WARN_ON(1); 1810 WARN_ON(1);
1701 } 1811 }
1812
1813 lvl_pages = lvl_to_nr_pages(largepage_lvl);
1814
1815 BUG_ON(nr_pages < lvl_pages);
1816 BUG_ON(sg_res < lvl_pages);
1817
1818 nr_pages -= lvl_pages;
1819 iov_pfn += lvl_pages;
1820 phys_pfn += lvl_pages;
1821 pteval += lvl_pages * VTD_PAGE_SIZE;
1822 sg_res -= lvl_pages;
1823
1824 /* If the next PTE would be the first in a new page, then we
1825 need to flush the cache on the entries we've just written.
1826 And then we'll need to recalculate 'pte', so clear it and
1827 let it get set again in the if (!pte) block above.
1828
1829 If we're done (!nr_pages) we need to flush the cache too.
1830
1831 Also if we've been setting superpages, we may need to
1832 recalculate 'pte' and switch back to smaller pages for the
1833 end of the mapping, if the trailing size is not enough to
1834 use another superpage (i.e. sg_res < lvl_pages). */
1702 pte++; 1835 pte++;
1703 if (!nr_pages || first_pte_in_page(pte)) { 1836 if (!nr_pages || first_pte_in_page(pte) ||
1837 (largepage_lvl > 1 && sg_res < lvl_pages)) {
1704 domain_flush_cache(domain, first_pte, 1838 domain_flush_cache(domain, first_pte,
1705 (void *)pte - (void *)first_pte); 1839 (void *)pte - (void *)first_pte);
1706 pte = NULL; 1840 pte = NULL;
1707 } 1841 }
1708 iov_pfn++; 1842
1709 pteval += VTD_PAGE_SIZE; 1843 if (!sg_res && nr_pages)
1710 sg_res--;
1711 if (!sg_res)
1712 sg = sg_next(sg); 1844 sg = sg_next(sg);
1713 } 1845 }
1714 return 0; 1846 return 0;
@@ -2016,7 +2148,7 @@ static inline int iommu_prepare_rmrr_dev(struct dmar_rmrr_unit *rmrr,
2016 if (pdev->dev.archdata.iommu == DUMMY_DEVICE_DOMAIN_INFO) 2148 if (pdev->dev.archdata.iommu == DUMMY_DEVICE_DOMAIN_INFO)
2017 return 0; 2149 return 0;
2018 return iommu_prepare_identity_map(pdev, rmrr->base_address, 2150 return iommu_prepare_identity_map(pdev, rmrr->base_address,
2019 rmrr->end_address + 1); 2151 rmrr->end_address);
2020} 2152}
2021 2153
2022#ifdef CONFIG_DMAR_FLOPPY_WA 2154#ifdef CONFIG_DMAR_FLOPPY_WA
@@ -2030,7 +2162,7 @@ static inline void iommu_prepare_isa(void)
2030 return; 2162 return;
2031 2163
2032 printk(KERN_INFO "IOMMU: Prepare 0-16MiB unity mapping for LPC\n"); 2164 printk(KERN_INFO "IOMMU: Prepare 0-16MiB unity mapping for LPC\n");
2033 ret = iommu_prepare_identity_map(pdev, 0, 16*1024*1024); 2165 ret = iommu_prepare_identity_map(pdev, 0, 16*1024*1024 - 1);
2034 2166
2035 if (ret) 2167 if (ret)
2036 printk(KERN_ERR "IOMMU: Failed to create 0-16MiB identity map; " 2168 printk(KERN_ERR "IOMMU: Failed to create 0-16MiB identity map; "
@@ -2106,10 +2238,10 @@ static int identity_mapping(struct pci_dev *pdev)
2106 if (likely(!iommu_identity_mapping)) 2238 if (likely(!iommu_identity_mapping))
2107 return 0; 2239 return 0;
2108 2240
2241 info = pdev->dev.archdata.iommu;
2242 if (info && info != DUMMY_DEVICE_DOMAIN_INFO)
2243 return (info->domain == si_domain);
2109 2244
2110 list_for_each_entry(info, &si_domain->devices, link)
2111 if (info->dev == pdev)
2112 return 1;
2113 return 0; 2245 return 0;
2114} 2246}
2115 2247
@@ -2187,8 +2319,19 @@ static int iommu_should_identity_map(struct pci_dev *pdev, int startup)
2187 * Assume that they will -- if they turn out not to be, then we can 2319 * Assume that they will -- if they turn out not to be, then we can
2188 * take them out of the 1:1 domain later. 2320 * take them out of the 1:1 domain later.
2189 */ 2321 */
2190 if (!startup) 2322 if (!startup) {
2191 return pdev->dma_mask > DMA_BIT_MASK(32); 2323 /*
2324 * If the device's dma_mask is less than the system's memory
2325 * size then this is not a candidate for identity mapping.
2326 */
2327 u64 dma_mask = pdev->dma_mask;
2328
2329 if (pdev->dev.coherent_dma_mask &&
2330 pdev->dev.coherent_dma_mask < dma_mask)
2331 dma_mask = pdev->dev.coherent_dma_mask;
2332
2333 return dma_mask >= dma_get_required_mask(&pdev->dev);
2334 }
2192 2335
2193 return 1; 2336 return 1;
2194} 2337}
@@ -2203,6 +2346,9 @@ static int __init iommu_prepare_static_identity_mapping(int hw)
2203 return -EFAULT; 2346 return -EFAULT;
2204 2347
2205 for_each_pci_dev(pdev) { 2348 for_each_pci_dev(pdev) {
2349 /* Skip Host/PCI Bridge devices */
2350 if (IS_BRIDGE_HOST_DEVICE(pdev))
2351 continue;
2206 if (iommu_should_identity_map(pdev, 1)) { 2352 if (iommu_should_identity_map(pdev, 1)) {
2207 printk(KERN_INFO "IOMMU: %s identity mapping for device %s\n", 2353 printk(KERN_INFO "IOMMU: %s identity mapping for device %s\n",
2208 hw ? "hardware" : "software", pci_name(pdev)); 2354 hw ? "hardware" : "software", pci_name(pdev));
@@ -2218,7 +2364,7 @@ static int __init iommu_prepare_static_identity_mapping(int hw)
2218 return 0; 2364 return 0;
2219} 2365}
2220 2366
2221static int __init init_dmars(int force_on) 2367static int __init init_dmars(void)
2222{ 2368{
2223 struct dmar_drhd_unit *drhd; 2369 struct dmar_drhd_unit *drhd;
2224 struct dmar_rmrr_unit *rmrr; 2370 struct dmar_rmrr_unit *rmrr;
@@ -2592,8 +2738,7 @@ static dma_addr_t __intel_map_single(struct device *hwdev, phys_addr_t paddr,
2592 iommu = domain_get_iommu(domain); 2738 iommu = domain_get_iommu(domain);
2593 size = aligned_nrpages(paddr, size); 2739 size = aligned_nrpages(paddr, size);
2594 2740
2595 iova = intel_alloc_iova(hwdev, domain, dma_to_mm_pfn(size), 2741 iova = intel_alloc_iova(hwdev, domain, dma_to_mm_pfn(size), dma_mask);
2596 pdev->dma_mask);
2597 if (!iova) 2742 if (!iova)
2598 goto error; 2743 goto error;
2599 2744
@@ -3118,7 +3263,17 @@ static int init_iommu_hw(void)
3118 if (iommu->qi) 3263 if (iommu->qi)
3119 dmar_reenable_qi(iommu); 3264 dmar_reenable_qi(iommu);
3120 3265
3121 for_each_active_iommu(iommu, drhd) { 3266 for_each_iommu(iommu, drhd) {
3267 if (drhd->ignored) {
3268 /*
3269 * we always have to disable PMRs or DMA may fail on
3270 * this device
3271 */
3272 if (force_on)
3273 iommu_disable_protect_mem_regions(iommu);
3274 continue;
3275 }
3276
3122 iommu_flush_write_buffer(iommu); 3277 iommu_flush_write_buffer(iommu);
3123 3278
3124 iommu_set_root_entry(iommu); 3279 iommu_set_root_entry(iommu);
@@ -3127,7 +3282,8 @@ static int init_iommu_hw(void)
3127 DMA_CCMD_GLOBAL_INVL); 3282 DMA_CCMD_GLOBAL_INVL);
3128 iommu->flush.flush_iotlb(iommu, 0, 0, 0, 3283 iommu->flush.flush_iotlb(iommu, 0, 0, 0,
3129 DMA_TLB_GLOBAL_FLUSH); 3284 DMA_TLB_GLOBAL_FLUSH);
3130 iommu_enable_translation(iommu); 3285 if (iommu_enable_translation(iommu))
3286 return 1;
3131 iommu_disable_protect_mem_regions(iommu); 3287 iommu_disable_protect_mem_regions(iommu);
3132 } 3288 }
3133 3289
@@ -3194,7 +3350,10 @@ static void iommu_resume(void)
3194 unsigned long flag; 3350 unsigned long flag;
3195 3351
3196 if (init_iommu_hw()) { 3352 if (init_iommu_hw()) {
3197 WARN(1, "IOMMU setup failed, DMAR can not resume!\n"); 3353 if (force_on)
3354 panic("tboot: IOMMU setup failed, DMAR can not resume!\n");
3355 else
3356 WARN(1, "IOMMU setup failed, DMAR can not resume!\n");
3198 return; 3357 return;
3199 } 3358 }
3200 3359
@@ -3271,7 +3430,6 @@ static struct notifier_block device_nb = {
3271int __init intel_iommu_init(void) 3430int __init intel_iommu_init(void)
3272{ 3431{
3273 int ret = 0; 3432 int ret = 0;
3274 int force_on = 0;
3275 3433
3276 /* VT-d is required for a TXT/tboot launch, so enforce that */ 3434 /* VT-d is required for a TXT/tboot launch, so enforce that */
3277 force_on = tboot_force_iommu(); 3435 force_on = tboot_force_iommu();
@@ -3309,7 +3467,7 @@ int __init intel_iommu_init(void)
3309 3467
3310 init_no_remapping_devices(); 3468 init_no_remapping_devices();
3311 3469
3312 ret = init_dmars(force_on); 3470 ret = init_dmars();
3313 if (ret) { 3471 if (ret) {
3314 if (force_on) 3472 if (force_on)
3315 panic("tboot: Failed to initialize DMARs\n"); 3473 panic("tboot: Failed to initialize DMARs\n");
@@ -3380,8 +3538,8 @@ static void domain_remove_one_dev_info(struct dmar_domain *domain,
3380 spin_lock_irqsave(&device_domain_lock, flags); 3538 spin_lock_irqsave(&device_domain_lock, flags);
3381 list_for_each_safe(entry, tmp, &domain->devices) { 3539 list_for_each_safe(entry, tmp, &domain->devices) {
3382 info = list_entry(entry, struct device_domain_info, link); 3540 info = list_entry(entry, struct device_domain_info, link);
3383 /* No need to compare PCI domain; it has to be the same */ 3541 if (info->segment == pci_domain_nr(pdev->bus) &&
3384 if (info->bus == pdev->bus->number && 3542 info->bus == pdev->bus->number &&
3385 info->devfn == pdev->devfn) { 3543 info->devfn == pdev->devfn) {
3386 list_del(&info->link); 3544 list_del(&info->link);
3387 list_del(&info->global); 3545 list_del(&info->global);
@@ -3419,10 +3577,13 @@ static void domain_remove_one_dev_info(struct dmar_domain *domain,
3419 domain_update_iommu_cap(domain); 3577 domain_update_iommu_cap(domain);
3420 spin_unlock_irqrestore(&domain->iommu_lock, tmp_flags); 3578 spin_unlock_irqrestore(&domain->iommu_lock, tmp_flags);
3421 3579
3422 spin_lock_irqsave(&iommu->lock, tmp_flags); 3580 if (!(domain->flags & DOMAIN_FLAG_VIRTUAL_MACHINE) &&
3423 clear_bit(domain->id, iommu->domain_ids); 3581 !(domain->flags & DOMAIN_FLAG_STATIC_IDENTITY)) {
3424 iommu->domains[domain->id] = NULL; 3582 spin_lock_irqsave(&iommu->lock, tmp_flags);
3425 spin_unlock_irqrestore(&iommu->lock, tmp_flags); 3583 clear_bit(domain->id, iommu->domain_ids);
3584 iommu->domains[domain->id] = NULL;
3585 spin_unlock_irqrestore(&iommu->lock, tmp_flags);
3586 }
3426 } 3587 }
3427 3588
3428 spin_unlock_irqrestore(&device_domain_lock, flags); 3589 spin_unlock_irqrestore(&device_domain_lock, flags);
@@ -3505,6 +3666,7 @@ static int md_domain_init(struct dmar_domain *domain, int guest_width)
3505 domain->iommu_count = 0; 3666 domain->iommu_count = 0;
3506 domain->iommu_coherency = 0; 3667 domain->iommu_coherency = 0;
3507 domain->iommu_snooping = 0; 3668 domain->iommu_snooping = 0;
3669 domain->iommu_superpage = 0;
3508 domain->max_addr = 0; 3670 domain->max_addr = 0;
3509 domain->nid = -1; 3671 domain->nid = -1;
3510 3672
@@ -3720,7 +3882,7 @@ static phys_addr_t intel_iommu_iova_to_phys(struct iommu_domain *domain,
3720 struct dma_pte *pte; 3882 struct dma_pte *pte;
3721 u64 phys = 0; 3883 u64 phys = 0;
3722 3884
3723 pte = pfn_to_dma_pte(dmar_domain, iova >> VTD_PAGE_SHIFT); 3885 pte = pfn_to_dma_pte(dmar_domain, iova >> VTD_PAGE_SHIFT, 0);
3724 if (pte) 3886 if (pte)
3725 phys = dma_pte_addr(pte); 3887 phys = dma_pte_addr(pte);
3726 3888
diff --git a/drivers/pci/iova.c b/drivers/pci/iova.c
index 9606e599a475..c5c274ab5c5a 100644
--- a/drivers/pci/iova.c
+++ b/drivers/pci/iova.c
@@ -63,8 +63,16 @@ __cached_rbnode_delete_update(struct iova_domain *iovad, struct iova *free)
63 curr = iovad->cached32_node; 63 curr = iovad->cached32_node;
64 cached_iova = container_of(curr, struct iova, node); 64 cached_iova = container_of(curr, struct iova, node);
65 65
66 if (free->pfn_lo >= cached_iova->pfn_lo) 66 if (free->pfn_lo >= cached_iova->pfn_lo) {
67 iovad->cached32_node = rb_next(&free->node); 67 struct rb_node *node = rb_next(&free->node);
68 struct iova *iova = container_of(node, struct iova, node);
69
70 /* only cache if it's below 32bit pfn */
71 if (node && iova->pfn_lo < iovad->dma_32bit_pfn)
72 iovad->cached32_node = node;
73 else
74 iovad->cached32_node = NULL;
75 }
68} 76}
69 77
70/* Computes the padding size required, to make the 78/* Computes the padding size required, to make the
diff --git a/drivers/vhost/net.c b/drivers/vhost/net.c
index 2f7c76a85e53..e224a92baa16 100644
--- a/drivers/vhost/net.c
+++ b/drivers/vhost/net.c
@@ -144,7 +144,7 @@ static void handle_tx(struct vhost_net *net)
144 } 144 }
145 145
146 mutex_lock(&vq->mutex); 146 mutex_lock(&vq->mutex);
147 vhost_disable_notify(vq); 147 vhost_disable_notify(&net->dev, vq);
148 148
149 if (wmem < sock->sk->sk_sndbuf / 2) 149 if (wmem < sock->sk->sk_sndbuf / 2)
150 tx_poll_stop(net); 150 tx_poll_stop(net);
@@ -166,8 +166,8 @@ static void handle_tx(struct vhost_net *net)
166 set_bit(SOCK_ASYNC_NOSPACE, &sock->flags); 166 set_bit(SOCK_ASYNC_NOSPACE, &sock->flags);
167 break; 167 break;
168 } 168 }
169 if (unlikely(vhost_enable_notify(vq))) { 169 if (unlikely(vhost_enable_notify(&net->dev, vq))) {
170 vhost_disable_notify(vq); 170 vhost_disable_notify(&net->dev, vq);
171 continue; 171 continue;
172 } 172 }
173 break; 173 break;
@@ -315,7 +315,7 @@ static void handle_rx(struct vhost_net *net)
315 return; 315 return;
316 316
317 mutex_lock(&vq->mutex); 317 mutex_lock(&vq->mutex);
318 vhost_disable_notify(vq); 318 vhost_disable_notify(&net->dev, vq);
319 vhost_hlen = vq->vhost_hlen; 319 vhost_hlen = vq->vhost_hlen;
320 sock_hlen = vq->sock_hlen; 320 sock_hlen = vq->sock_hlen;
321 321
@@ -334,10 +334,10 @@ static void handle_rx(struct vhost_net *net)
334 break; 334 break;
335 /* OK, now we need to know about added descriptors. */ 335 /* OK, now we need to know about added descriptors. */
336 if (!headcount) { 336 if (!headcount) {
337 if (unlikely(vhost_enable_notify(vq))) { 337 if (unlikely(vhost_enable_notify(&net->dev, vq))) {
338 /* They have slipped one in as we were 338 /* They have slipped one in as we were
339 * doing that: check again. */ 339 * doing that: check again. */
340 vhost_disable_notify(vq); 340 vhost_disable_notify(&net->dev, vq);
341 continue; 341 continue;
342 } 342 }
343 /* Nothing new? Wait for eventfd to tell us 343 /* Nothing new? Wait for eventfd to tell us
diff --git a/drivers/vhost/test.c b/drivers/vhost/test.c
index 099f30230d06..734e1d74ad80 100644
--- a/drivers/vhost/test.c
+++ b/drivers/vhost/test.c
@@ -49,7 +49,7 @@ static void handle_vq(struct vhost_test *n)
49 return; 49 return;
50 50
51 mutex_lock(&vq->mutex); 51 mutex_lock(&vq->mutex);
52 vhost_disable_notify(vq); 52 vhost_disable_notify(&n->dev, vq);
53 53
54 for (;;) { 54 for (;;) {
55 head = vhost_get_vq_desc(&n->dev, vq, vq->iov, 55 head = vhost_get_vq_desc(&n->dev, vq, vq->iov,
@@ -61,8 +61,8 @@ static void handle_vq(struct vhost_test *n)
61 break; 61 break;
62 /* Nothing new? Wait for eventfd to tell us they refilled. */ 62 /* Nothing new? Wait for eventfd to tell us they refilled. */
63 if (head == vq->num) { 63 if (head == vq->num) {
64 if (unlikely(vhost_enable_notify(vq))) { 64 if (unlikely(vhost_enable_notify(&n->dev, vq))) {
65 vhost_disable_notify(vq); 65 vhost_disable_notify(&n->dev, vq);
66 continue; 66 continue;
67 } 67 }
68 break; 68 break;
diff --git a/drivers/vhost/vhost.c b/drivers/vhost/vhost.c
index 7aa4eea930f1..ea966b356352 100644
--- a/drivers/vhost/vhost.c
+++ b/drivers/vhost/vhost.c
@@ -37,6 +37,9 @@ enum {
37 VHOST_MEMORY_F_LOG = 0x1, 37 VHOST_MEMORY_F_LOG = 0x1,
38}; 38};
39 39
40#define vhost_used_event(vq) ((u16 __user *)&vq->avail->ring[vq->num])
41#define vhost_avail_event(vq) ((u16 __user *)&vq->used->ring[vq->num])
42
40static void vhost_poll_func(struct file *file, wait_queue_head_t *wqh, 43static void vhost_poll_func(struct file *file, wait_queue_head_t *wqh,
41 poll_table *pt) 44 poll_table *pt)
42{ 45{
@@ -161,6 +164,8 @@ static void vhost_vq_reset(struct vhost_dev *dev,
161 vq->last_avail_idx = 0; 164 vq->last_avail_idx = 0;
162 vq->avail_idx = 0; 165 vq->avail_idx = 0;
163 vq->last_used_idx = 0; 166 vq->last_used_idx = 0;
167 vq->signalled_used = 0;
168 vq->signalled_used_valid = false;
164 vq->used_flags = 0; 169 vq->used_flags = 0;
165 vq->log_used = false; 170 vq->log_used = false;
166 vq->log_addr = -1ull; 171 vq->log_addr = -1ull;
@@ -489,16 +494,17 @@ static int memory_access_ok(struct vhost_dev *d, struct vhost_memory *mem,
489 return 1; 494 return 1;
490} 495}
491 496
492static int vq_access_ok(unsigned int num, 497static int vq_access_ok(struct vhost_dev *d, unsigned int num,
493 struct vring_desc __user *desc, 498 struct vring_desc __user *desc,
494 struct vring_avail __user *avail, 499 struct vring_avail __user *avail,
495 struct vring_used __user *used) 500 struct vring_used __user *used)
496{ 501{
502 size_t s = vhost_has_feature(d, VIRTIO_RING_F_EVENT_IDX) ? 2 : 0;
497 return access_ok(VERIFY_READ, desc, num * sizeof *desc) && 503 return access_ok(VERIFY_READ, desc, num * sizeof *desc) &&
498 access_ok(VERIFY_READ, avail, 504 access_ok(VERIFY_READ, avail,
499 sizeof *avail + num * sizeof *avail->ring) && 505 sizeof *avail + num * sizeof *avail->ring + s) &&
500 access_ok(VERIFY_WRITE, used, 506 access_ok(VERIFY_WRITE, used,
501 sizeof *used + num * sizeof *used->ring); 507 sizeof *used + num * sizeof *used->ring + s);
502} 508}
503 509
504/* Can we log writes? */ 510/* Can we log writes? */
@@ -514,9 +520,11 @@ int vhost_log_access_ok(struct vhost_dev *dev)
514 520
515/* Verify access for write logging. */ 521/* Verify access for write logging. */
516/* Caller should have vq mutex and device mutex */ 522/* Caller should have vq mutex and device mutex */
517static int vq_log_access_ok(struct vhost_virtqueue *vq, void __user *log_base) 523static int vq_log_access_ok(struct vhost_dev *d, struct vhost_virtqueue *vq,
524 void __user *log_base)
518{ 525{
519 struct vhost_memory *mp; 526 struct vhost_memory *mp;
527 size_t s = vhost_has_feature(d, VIRTIO_RING_F_EVENT_IDX) ? 2 : 0;
520 528
521 mp = rcu_dereference_protected(vq->dev->memory, 529 mp = rcu_dereference_protected(vq->dev->memory,
522 lockdep_is_held(&vq->mutex)); 530 lockdep_is_held(&vq->mutex));
@@ -524,15 +532,15 @@ static int vq_log_access_ok(struct vhost_virtqueue *vq, void __user *log_base)
524 vhost_has_feature(vq->dev, VHOST_F_LOG_ALL)) && 532 vhost_has_feature(vq->dev, VHOST_F_LOG_ALL)) &&
525 (!vq->log_used || log_access_ok(log_base, vq->log_addr, 533 (!vq->log_used || log_access_ok(log_base, vq->log_addr,
526 sizeof *vq->used + 534 sizeof *vq->used +
527 vq->num * sizeof *vq->used->ring)); 535 vq->num * sizeof *vq->used->ring + s));
528} 536}
529 537
530/* Can we start vq? */ 538/* Can we start vq? */
531/* Caller should have vq mutex and device mutex */ 539/* Caller should have vq mutex and device mutex */
532int vhost_vq_access_ok(struct vhost_virtqueue *vq) 540int vhost_vq_access_ok(struct vhost_virtqueue *vq)
533{ 541{
534 return vq_access_ok(vq->num, vq->desc, vq->avail, vq->used) && 542 return vq_access_ok(vq->dev, vq->num, vq->desc, vq->avail, vq->used) &&
535 vq_log_access_ok(vq, vq->log_base); 543 vq_log_access_ok(vq->dev, vq, vq->log_base);
536} 544}
537 545
538static long vhost_set_memory(struct vhost_dev *d, struct vhost_memory __user *m) 546static long vhost_set_memory(struct vhost_dev *d, struct vhost_memory __user *m)
@@ -577,6 +585,7 @@ static int init_used(struct vhost_virtqueue *vq,
577 585
578 if (r) 586 if (r)
579 return r; 587 return r;
588 vq->signalled_used_valid = false;
580 return get_user(vq->last_used_idx, &used->idx); 589 return get_user(vq->last_used_idx, &used->idx);
581} 590}
582 591
@@ -674,7 +683,7 @@ static long vhost_set_vring(struct vhost_dev *d, int ioctl, void __user *argp)
674 * If it is not, we don't as size might not have been setup. 683 * If it is not, we don't as size might not have been setup.
675 * We will verify when backend is configured. */ 684 * We will verify when backend is configured. */
676 if (vq->private_data) { 685 if (vq->private_data) {
677 if (!vq_access_ok(vq->num, 686 if (!vq_access_ok(d, vq->num,
678 (void __user *)(unsigned long)a.desc_user_addr, 687 (void __user *)(unsigned long)a.desc_user_addr,
679 (void __user *)(unsigned long)a.avail_user_addr, 688 (void __user *)(unsigned long)a.avail_user_addr,
680 (void __user *)(unsigned long)a.used_user_addr)) { 689 (void __user *)(unsigned long)a.used_user_addr)) {
@@ -818,7 +827,7 @@ long vhost_dev_ioctl(struct vhost_dev *d, unsigned int ioctl, unsigned long arg)
818 vq = d->vqs + i; 827 vq = d->vqs + i;
819 mutex_lock(&vq->mutex); 828 mutex_lock(&vq->mutex);
820 /* If ring is inactive, will check when it's enabled. */ 829 /* If ring is inactive, will check when it's enabled. */
821 if (vq->private_data && !vq_log_access_ok(vq, base)) 830 if (vq->private_data && !vq_log_access_ok(d, vq, base))
822 r = -EFAULT; 831 r = -EFAULT;
823 else 832 else
824 vq->log_base = base; 833 vq->log_base = base;
@@ -1219,6 +1228,10 @@ int vhost_get_vq_desc(struct vhost_dev *dev, struct vhost_virtqueue *vq,
1219 1228
1220 /* On success, increment avail index. */ 1229 /* On success, increment avail index. */
1221 vq->last_avail_idx++; 1230 vq->last_avail_idx++;
1231
1232 /* Assume notifications from guest are disabled at this point,
1233 * if they aren't we would need to update avail_event index. */
1234 BUG_ON(!(vq->used_flags & VRING_USED_F_NO_NOTIFY));
1222 return head; 1235 return head;
1223} 1236}
1224 1237
@@ -1267,6 +1280,12 @@ int vhost_add_used(struct vhost_virtqueue *vq, unsigned int head, int len)
1267 eventfd_signal(vq->log_ctx, 1); 1280 eventfd_signal(vq->log_ctx, 1);
1268 } 1281 }
1269 vq->last_used_idx++; 1282 vq->last_used_idx++;
1283 /* If the driver never bothers to signal in a very long while,
1284 * used index might wrap around. If that happens, invalidate
1285 * signalled_used index we stored. TODO: make sure driver
1286 * signals at least once in 2^16 and remove this. */
1287 if (unlikely(vq->last_used_idx == vq->signalled_used))
1288 vq->signalled_used_valid = false;
1270 return 0; 1289 return 0;
1271} 1290}
1272 1291
@@ -1275,6 +1294,7 @@ static int __vhost_add_used_n(struct vhost_virtqueue *vq,
1275 unsigned count) 1294 unsigned count)
1276{ 1295{
1277 struct vring_used_elem __user *used; 1296 struct vring_used_elem __user *used;
1297 u16 old, new;
1278 int start; 1298 int start;
1279 1299
1280 start = vq->last_used_idx % vq->num; 1300 start = vq->last_used_idx % vq->num;
@@ -1292,7 +1312,14 @@ static int __vhost_add_used_n(struct vhost_virtqueue *vq,
1292 ((void __user *)used - (void __user *)vq->used), 1312 ((void __user *)used - (void __user *)vq->used),
1293 count * sizeof *used); 1313 count * sizeof *used);
1294 } 1314 }
1295 vq->last_used_idx += count; 1315 old = vq->last_used_idx;
1316 new = (vq->last_used_idx += count);
1317 /* If the driver never bothers to signal in a very long while,
1318 * used index might wrap around. If that happens, invalidate
1319 * signalled_used index we stored. TODO: make sure driver
1320 * signals at least once in 2^16 and remove this. */
1321 if (unlikely((u16)(new - vq->signalled_used) < (u16)(new - old)))
1322 vq->signalled_used_valid = false;
1296 return 0; 1323 return 0;
1297} 1324}
1298 1325
@@ -1331,29 +1358,47 @@ int vhost_add_used_n(struct vhost_virtqueue *vq, struct vring_used_elem *heads,
1331 return r; 1358 return r;
1332} 1359}
1333 1360
1334/* This actually signals the guest, using eventfd. */ 1361static bool vhost_notify(struct vhost_dev *dev, struct vhost_virtqueue *vq)
1335void vhost_signal(struct vhost_dev *dev, struct vhost_virtqueue *vq)
1336{ 1362{
1337 __u16 flags; 1363 __u16 old, new, event;
1338 1364 bool v;
1339 /* Flush out used index updates. This is paired 1365 /* Flush out used index updates. This is paired
1340 * with the barrier that the Guest executes when enabling 1366 * with the barrier that the Guest executes when enabling
1341 * interrupts. */ 1367 * interrupts. */
1342 smp_mb(); 1368 smp_mb();
1343 1369
1344 if (__get_user(flags, &vq->avail->flags)) { 1370 if (vhost_has_feature(dev, VIRTIO_F_NOTIFY_ON_EMPTY) &&
1345 vq_err(vq, "Failed to get flags"); 1371 unlikely(vq->avail_idx == vq->last_avail_idx))
1346 return; 1372 return true;
1373
1374 if (!vhost_has_feature(dev, VIRTIO_RING_F_EVENT_IDX)) {
1375 __u16 flags;
1376 if (__get_user(flags, &vq->avail->flags)) {
1377 vq_err(vq, "Failed to get flags");
1378 return true;
1379 }
1380 return !(flags & VRING_AVAIL_F_NO_INTERRUPT);
1347 } 1381 }
1382 old = vq->signalled_used;
1383 v = vq->signalled_used_valid;
1384 new = vq->signalled_used = vq->last_used_idx;
1385 vq->signalled_used_valid = true;
1348 1386
1349 /* If they don't want an interrupt, don't signal, unless empty. */ 1387 if (unlikely(!v))
1350 if ((flags & VRING_AVAIL_F_NO_INTERRUPT) && 1388 return true;
1351 (vq->avail_idx != vq->last_avail_idx ||
1352 !vhost_has_feature(dev, VIRTIO_F_NOTIFY_ON_EMPTY)))
1353 return;
1354 1389
1390 if (get_user(event, vhost_used_event(vq))) {
1391 vq_err(vq, "Failed to get used event idx");
1392 return true;
1393 }
1394 return vring_need_event(event, new, old);
1395}
1396
1397/* This actually signals the guest, using eventfd. */
1398void vhost_signal(struct vhost_dev *dev, struct vhost_virtqueue *vq)
1399{
1355 /* Signal the Guest tell them we used something up. */ 1400 /* Signal the Guest tell them we used something up. */
1356 if (vq->call_ctx) 1401 if (vq->call_ctx && vhost_notify(dev, vq))
1357 eventfd_signal(vq->call_ctx, 1); 1402 eventfd_signal(vq->call_ctx, 1);
1358} 1403}
1359 1404
@@ -1376,7 +1421,7 @@ void vhost_add_used_and_signal_n(struct vhost_dev *dev,
1376} 1421}
1377 1422
1378/* OK, now we need to know about added descriptors. */ 1423/* OK, now we need to know about added descriptors. */
1379bool vhost_enable_notify(struct vhost_virtqueue *vq) 1424bool vhost_enable_notify(struct vhost_dev *dev, struct vhost_virtqueue *vq)
1380{ 1425{
1381 u16 avail_idx; 1426 u16 avail_idx;
1382 int r; 1427 int r;
@@ -1384,11 +1429,34 @@ bool vhost_enable_notify(struct vhost_virtqueue *vq)
1384 if (!(vq->used_flags & VRING_USED_F_NO_NOTIFY)) 1429 if (!(vq->used_flags & VRING_USED_F_NO_NOTIFY))
1385 return false; 1430 return false;
1386 vq->used_flags &= ~VRING_USED_F_NO_NOTIFY; 1431 vq->used_flags &= ~VRING_USED_F_NO_NOTIFY;
1387 r = put_user(vq->used_flags, &vq->used->flags); 1432 if (!vhost_has_feature(dev, VIRTIO_RING_F_EVENT_IDX)) {
1388 if (r) { 1433 r = put_user(vq->used_flags, &vq->used->flags);
1389 vq_err(vq, "Failed to enable notification at %p: %d\n", 1434 if (r) {
1390 &vq->used->flags, r); 1435 vq_err(vq, "Failed to enable notification at %p: %d\n",
1391 return false; 1436 &vq->used->flags, r);
1437 return false;
1438 }
1439 } else {
1440 r = put_user(vq->avail_idx, vhost_avail_event(vq));
1441 if (r) {
1442 vq_err(vq, "Failed to update avail event index at %p: %d\n",
1443 vhost_avail_event(vq), r);
1444 return false;
1445 }
1446 }
1447 if (unlikely(vq->log_used)) {
1448 void __user *used;
1449 /* Make sure data is seen before log. */
1450 smp_wmb();
1451 used = vhost_has_feature(dev, VIRTIO_RING_F_EVENT_IDX) ?
1452 &vq->used->flags : vhost_avail_event(vq);
1453 /* Log used flags or event index entry write. Both are 16 bit
1454 * fields. */
1455 log_write(vq->log_base, vq->log_addr +
1456 (used - (void __user *)vq->used),
1457 sizeof(u16));
1458 if (vq->log_ctx)
1459 eventfd_signal(vq->log_ctx, 1);
1392 } 1460 }
1393 /* They could have slipped one in as we were doing that: make 1461 /* They could have slipped one in as we were doing that: make
1394 * sure it's written, then check again. */ 1462 * sure it's written, then check again. */
@@ -1404,15 +1472,17 @@ bool vhost_enable_notify(struct vhost_virtqueue *vq)
1404} 1472}
1405 1473
1406/* We don't need to be notified again. */ 1474/* We don't need to be notified again. */
1407void vhost_disable_notify(struct vhost_virtqueue *vq) 1475void vhost_disable_notify(struct vhost_dev *dev, struct vhost_virtqueue *vq)
1408{ 1476{
1409 int r; 1477 int r;
1410 1478
1411 if (vq->used_flags & VRING_USED_F_NO_NOTIFY) 1479 if (vq->used_flags & VRING_USED_F_NO_NOTIFY)
1412 return; 1480 return;
1413 vq->used_flags |= VRING_USED_F_NO_NOTIFY; 1481 vq->used_flags |= VRING_USED_F_NO_NOTIFY;
1414 r = put_user(vq->used_flags, &vq->used->flags); 1482 if (!vhost_has_feature(dev, VIRTIO_RING_F_EVENT_IDX)) {
1415 if (r) 1483 r = put_user(vq->used_flags, &vq->used->flags);
1416 vq_err(vq, "Failed to enable notification at %p: %d\n", 1484 if (r)
1417 &vq->used->flags, r); 1485 vq_err(vq, "Failed to enable notification at %p: %d\n",
1486 &vq->used->flags, r);
1487 }
1418} 1488}
diff --git a/drivers/vhost/vhost.h b/drivers/vhost/vhost.h
index b3363ae38518..8e03379dd30f 100644
--- a/drivers/vhost/vhost.h
+++ b/drivers/vhost/vhost.h
@@ -84,6 +84,12 @@ struct vhost_virtqueue {
84 /* Used flags */ 84 /* Used flags */
85 u16 used_flags; 85 u16 used_flags;
86 86
87 /* Last used index value we have signalled on */
88 u16 signalled_used;
89
90 /* Last used index value we have signalled on */
91 bool signalled_used_valid;
92
87 /* Log writes to used structure. */ 93 /* Log writes to used structure. */
88 bool log_used; 94 bool log_used;
89 u64 log_addr; 95 u64 log_addr;
@@ -149,8 +155,8 @@ void vhost_add_used_and_signal(struct vhost_dev *, struct vhost_virtqueue *,
149void vhost_add_used_and_signal_n(struct vhost_dev *, struct vhost_virtqueue *, 155void vhost_add_used_and_signal_n(struct vhost_dev *, struct vhost_virtqueue *,
150 struct vring_used_elem *heads, unsigned count); 156 struct vring_used_elem *heads, unsigned count);
151void vhost_signal(struct vhost_dev *, struct vhost_virtqueue *); 157void vhost_signal(struct vhost_dev *, struct vhost_virtqueue *);
152void vhost_disable_notify(struct vhost_virtqueue *); 158void vhost_disable_notify(struct vhost_dev *, struct vhost_virtqueue *);
153bool vhost_enable_notify(struct vhost_virtqueue *); 159bool vhost_enable_notify(struct vhost_dev *, struct vhost_virtqueue *);
154 160
155int vhost_log_write(struct vhost_virtqueue *vq, struct vhost_log *log, 161int vhost_log_write(struct vhost_virtqueue *vq, struct vhost_log *log,
156 unsigned int log_num, u64 len); 162 unsigned int log_num, u64 len);
@@ -162,11 +168,12 @@ int vhost_log_write(struct vhost_virtqueue *vq, struct vhost_log *log,
162 } while (0) 168 } while (0)
163 169
164enum { 170enum {
165 VHOST_FEATURES = (1 << VIRTIO_F_NOTIFY_ON_EMPTY) | 171 VHOST_FEATURES = (1ULL << VIRTIO_F_NOTIFY_ON_EMPTY) |
166 (1 << VIRTIO_RING_F_INDIRECT_DESC) | 172 (1ULL << VIRTIO_RING_F_INDIRECT_DESC) |
167 (1 << VHOST_F_LOG_ALL) | 173 (1ULL << VIRTIO_RING_F_EVENT_IDX) |
168 (1 << VHOST_NET_F_VIRTIO_NET_HDR) | 174 (1ULL << VHOST_F_LOG_ALL) |
169 (1 << VIRTIO_NET_F_MRG_RXBUF), 175 (1ULL << VHOST_NET_F_VIRTIO_NET_HDR) |
176 (1ULL << VIRTIO_NET_F_MRG_RXBUF),
170}; 177};
171 178
172static inline int vhost_has_feature(struct vhost_dev *dev, int bit) 179static inline int vhost_has_feature(struct vhost_dev *dev, int bit)
diff --git a/drivers/virtio/virtio_balloon.c b/drivers/virtio/virtio_balloon.c
index 0f1da45ba47d..e058ace2a4ad 100644
--- a/drivers/virtio/virtio_balloon.c
+++ b/drivers/virtio/virtio_balloon.c
@@ -40,9 +40,6 @@ struct virtio_balloon
40 /* Waiting for host to ack the pages we released. */ 40 /* Waiting for host to ack the pages we released. */
41 struct completion acked; 41 struct completion acked;
42 42
43 /* Do we have to tell Host *before* we reuse pages? */
44 bool tell_host_first;
45
46 /* The pages we've told the Host we're not using. */ 43 /* The pages we've told the Host we're not using. */
47 unsigned int num_pages; 44 unsigned int num_pages;
48 struct list_head pages; 45 struct list_head pages;
@@ -151,13 +148,14 @@ static void leak_balloon(struct virtio_balloon *vb, size_t num)
151 vb->num_pages--; 148 vb->num_pages--;
152 } 149 }
153 150
154 if (vb->tell_host_first) { 151
155 tell_host(vb, vb->deflate_vq); 152 /*
156 release_pages_by_pfn(vb->pfns, vb->num_pfns); 153 * Note that if
157 } else { 154 * virtio_has_feature(vdev, VIRTIO_BALLOON_F_MUST_TELL_HOST);
158 release_pages_by_pfn(vb->pfns, vb->num_pfns); 155 * is true, we *have* to do it in this order
159 tell_host(vb, vb->deflate_vq); 156 */
160 } 157 tell_host(vb, vb->deflate_vq);
158 release_pages_by_pfn(vb->pfns, vb->num_pfns);
161} 159}
162 160
163static inline void update_stat(struct virtio_balloon *vb, int idx, 161static inline void update_stat(struct virtio_balloon *vb, int idx,
@@ -325,9 +323,6 @@ static int virtballoon_probe(struct virtio_device *vdev)
325 goto out_del_vqs; 323 goto out_del_vqs;
326 } 324 }
327 325
328 vb->tell_host_first
329 = virtio_has_feature(vdev, VIRTIO_BALLOON_F_MUST_TELL_HOST);
330
331 return 0; 326 return 0;
332 327
333out_del_vqs: 328out_del_vqs:
diff --git a/drivers/virtio/virtio_ring.c b/drivers/virtio/virtio_ring.c
index b0043fb26a4d..68b9136847af 100644
--- a/drivers/virtio/virtio_ring.c
+++ b/drivers/virtio/virtio_ring.c
@@ -82,6 +82,9 @@ struct vring_virtqueue
82 /* Host supports indirect buffers */ 82 /* Host supports indirect buffers */
83 bool indirect; 83 bool indirect;
84 84
85 /* Host publishes avail event idx */
86 bool event;
87
85 /* Number of free buffers */ 88 /* Number of free buffers */
86 unsigned int num_free; 89 unsigned int num_free;
87 /* Head of free buffer list. */ 90 /* Head of free buffer list. */
@@ -237,18 +240,22 @@ EXPORT_SYMBOL_GPL(virtqueue_add_buf_gfp);
237void virtqueue_kick(struct virtqueue *_vq) 240void virtqueue_kick(struct virtqueue *_vq)
238{ 241{
239 struct vring_virtqueue *vq = to_vvq(_vq); 242 struct vring_virtqueue *vq = to_vvq(_vq);
243 u16 new, old;
240 START_USE(vq); 244 START_USE(vq);
241 /* Descriptors and available array need to be set before we expose the 245 /* Descriptors and available array need to be set before we expose the
242 * new available array entries. */ 246 * new available array entries. */
243 virtio_wmb(); 247 virtio_wmb();
244 248
245 vq->vring.avail->idx += vq->num_added; 249 old = vq->vring.avail->idx;
250 new = vq->vring.avail->idx = old + vq->num_added;
246 vq->num_added = 0; 251 vq->num_added = 0;
247 252
248 /* Need to update avail index before checking if we should notify */ 253 /* Need to update avail index before checking if we should notify */
249 virtio_mb(); 254 virtio_mb();
250 255
251 if (!(vq->vring.used->flags & VRING_USED_F_NO_NOTIFY)) 256 if (vq->event ?
257 vring_need_event(vring_avail_event(&vq->vring), new, old) :
258 !(vq->vring.used->flags & VRING_USED_F_NO_NOTIFY))
252 /* Prod other side to tell it about changes. */ 259 /* Prod other side to tell it about changes. */
253 vq->notify(&vq->vq); 260 vq->notify(&vq->vq);
254 261
@@ -324,6 +331,14 @@ void *virtqueue_get_buf(struct virtqueue *_vq, unsigned int *len)
324 ret = vq->data[i]; 331 ret = vq->data[i];
325 detach_buf(vq, i); 332 detach_buf(vq, i);
326 vq->last_used_idx++; 333 vq->last_used_idx++;
334 /* If we expect an interrupt for the next entry, tell host
335 * by writing event index and flush out the write before
336 * the read in the next get_buf call. */
337 if (!(vq->vring.avail->flags & VRING_AVAIL_F_NO_INTERRUPT)) {
338 vring_used_event(&vq->vring) = vq->last_used_idx;
339 virtio_mb();
340 }
341
327 END_USE(vq); 342 END_USE(vq);
328 return ret; 343 return ret;
329} 344}
@@ -345,7 +360,11 @@ bool virtqueue_enable_cb(struct virtqueue *_vq)
345 360
346 /* We optimistically turn back on interrupts, then check if there was 361 /* We optimistically turn back on interrupts, then check if there was
347 * more to do. */ 362 * more to do. */
363 /* Depending on the VIRTIO_RING_F_EVENT_IDX feature, we need to
364 * either clear the flags bit or point the event index at the next
365 * entry. Always do both to keep code simple. */
348 vq->vring.avail->flags &= ~VRING_AVAIL_F_NO_INTERRUPT; 366 vq->vring.avail->flags &= ~VRING_AVAIL_F_NO_INTERRUPT;
367 vring_used_event(&vq->vring) = vq->last_used_idx;
349 virtio_mb(); 368 virtio_mb();
350 if (unlikely(more_used(vq))) { 369 if (unlikely(more_used(vq))) {
351 END_USE(vq); 370 END_USE(vq);
@@ -357,6 +376,33 @@ bool virtqueue_enable_cb(struct virtqueue *_vq)
357} 376}
358EXPORT_SYMBOL_GPL(virtqueue_enable_cb); 377EXPORT_SYMBOL_GPL(virtqueue_enable_cb);
359 378
379bool virtqueue_enable_cb_delayed(struct virtqueue *_vq)
380{
381 struct vring_virtqueue *vq = to_vvq(_vq);
382 u16 bufs;
383
384 START_USE(vq);
385
386 /* We optimistically turn back on interrupts, then check if there was
387 * more to do. */
388 /* Depending on the VIRTIO_RING_F_USED_EVENT_IDX feature, we need to
389 * either clear the flags bit or point the event index at the next
390 * entry. Always do both to keep code simple. */
391 vq->vring.avail->flags &= ~VRING_AVAIL_F_NO_INTERRUPT;
392 /* TODO: tune this threshold */
393 bufs = (u16)(vq->vring.avail->idx - vq->last_used_idx) * 3 / 4;
394 vring_used_event(&vq->vring) = vq->last_used_idx + bufs;
395 virtio_mb();
396 if (unlikely((u16)(vq->vring.used->idx - vq->last_used_idx) > bufs)) {
397 END_USE(vq);
398 return false;
399 }
400
401 END_USE(vq);
402 return true;
403}
404EXPORT_SYMBOL_GPL(virtqueue_enable_cb_delayed);
405
360void *virtqueue_detach_unused_buf(struct virtqueue *_vq) 406void *virtqueue_detach_unused_buf(struct virtqueue *_vq)
361{ 407{
362 struct vring_virtqueue *vq = to_vvq(_vq); 408 struct vring_virtqueue *vq = to_vvq(_vq);
@@ -438,6 +484,7 @@ struct virtqueue *vring_new_virtqueue(unsigned int num,
438#endif 484#endif
439 485
440 vq->indirect = virtio_has_feature(vdev, VIRTIO_RING_F_INDIRECT_DESC); 486 vq->indirect = virtio_has_feature(vdev, VIRTIO_RING_F_INDIRECT_DESC);
487 vq->event = virtio_has_feature(vdev, VIRTIO_RING_F_EVENT_IDX);
441 488
442 /* No callback? Tell other side not to bother us. */ 489 /* No callback? Tell other side not to bother us. */
443 if (!callback) 490 if (!callback)
@@ -472,6 +519,8 @@ void vring_transport_features(struct virtio_device *vdev)
472 switch (i) { 519 switch (i) {
473 case VIRTIO_RING_F_INDIRECT_DESC: 520 case VIRTIO_RING_F_INDIRECT_DESC:
474 break; 521 break;
522 case VIRTIO_RING_F_EVENT_IDX:
523 break;
475 default: 524 default:
476 /* We don't understand this bit. */ 525 /* We don't understand this bit. */
477 clear_bit(i, vdev->features); 526 clear_bit(i, vdev->features);
diff --git a/fs/autofs4/root.c b/fs/autofs4/root.c
index 87d95a8cddbc..f55ae23b137e 100644
--- a/fs/autofs4/root.c
+++ b/fs/autofs4/root.c
@@ -583,8 +583,6 @@ static int autofs4_dir_unlink(struct inode *dir, struct dentry *dentry)
583 if (!autofs4_oz_mode(sbi) && !capable(CAP_SYS_ADMIN)) 583 if (!autofs4_oz_mode(sbi) && !capable(CAP_SYS_ADMIN))
584 return -EACCES; 584 return -EACCES;
585 585
586 dentry_unhash(dentry);
587
588 if (atomic_dec_and_test(&ino->count)) { 586 if (atomic_dec_and_test(&ino->count)) {
589 p_ino = autofs4_dentry_ino(dentry->d_parent); 587 p_ino = autofs4_dentry_ino(dentry->d_parent);
590 if (p_ino && dentry->d_parent != dentry) 588 if (p_ino && dentry->d_parent != dentry)
diff --git a/fs/namei.c b/fs/namei.c
index 1ab641f2e78e..e2e4e8d032ee 100644
--- a/fs/namei.c
+++ b/fs/namei.c
@@ -2579,6 +2579,7 @@ int vfs_rmdir(struct inode *dir, struct dentry *dentry)
2579 if (error) 2579 if (error)
2580 goto out; 2580 goto out;
2581 2581
2582 shrink_dcache_parent(dentry);
2582 error = dir->i_op->rmdir(dir, dentry); 2583 error = dir->i_op->rmdir(dir, dentry);
2583 if (error) 2584 if (error)
2584 goto out; 2585 goto out;
@@ -2993,6 +2994,8 @@ static int vfs_rename_dir(struct inode *old_dir, struct dentry *old_dentry,
2993 if (d_mountpoint(old_dentry) || d_mountpoint(new_dentry)) 2994 if (d_mountpoint(old_dentry) || d_mountpoint(new_dentry))
2994 goto out; 2995 goto out;
2995 2996
2997 if (target)
2998 shrink_dcache_parent(new_dentry);
2996 error = old_dir->i_op->rename(old_dir, old_dentry, new_dir, new_dentry); 2999 error = old_dir->i_op->rename(old_dir, old_dentry, new_dir, new_dentry);
2997 if (error) 3000 if (error)
2998 goto out; 3001 goto out;
diff --git a/include/linux/dma_remapping.h b/include/linux/dma_remapping.h
index 5619f8522738..bbd8661b3473 100644
--- a/include/linux/dma_remapping.h
+++ b/include/linux/dma_remapping.h
@@ -9,8 +9,12 @@
9#define VTD_PAGE_MASK (((u64)-1) << VTD_PAGE_SHIFT) 9#define VTD_PAGE_MASK (((u64)-1) << VTD_PAGE_SHIFT)
10#define VTD_PAGE_ALIGN(addr) (((addr) + VTD_PAGE_SIZE - 1) & VTD_PAGE_MASK) 10#define VTD_PAGE_ALIGN(addr) (((addr) + VTD_PAGE_SIZE - 1) & VTD_PAGE_MASK)
11 11
12#define VTD_STRIDE_SHIFT (9)
13#define VTD_STRIDE_MASK (((u64)-1) << VTD_STRIDE_SHIFT)
14
12#define DMA_PTE_READ (1) 15#define DMA_PTE_READ (1)
13#define DMA_PTE_WRITE (2) 16#define DMA_PTE_WRITE (2)
17#define DMA_PTE_LARGE_PAGE (1 << 7)
14#define DMA_PTE_SNP (1 << 11) 18#define DMA_PTE_SNP (1 << 11)
15 19
16#define CONTEXT_TT_MULTI_LEVEL 0 20#define CONTEXT_TT_MULTI_LEVEL 0
diff --git a/include/linux/mtd/physmap.h b/include/linux/mtd/physmap.h
index d40bfa1d9c91..e5f21d293c70 100644
--- a/include/linux/mtd/physmap.h
+++ b/include/linux/mtd/physmap.h
@@ -19,6 +19,7 @@
19#include <linux/mtd/partitions.h> 19#include <linux/mtd/partitions.h>
20 20
21struct map_info; 21struct map_info;
22struct platform_device;
22 23
23struct physmap_flash_data { 24struct physmap_flash_data {
24 unsigned int width; 25 unsigned int width;
diff --git a/include/linux/virtio.h b/include/linux/virtio.h
index aff5b4f74041..710885749605 100644
--- a/include/linux/virtio.h
+++ b/include/linux/virtio.h
@@ -51,6 +51,13 @@ struct virtqueue {
51 * This re-enables callbacks; it returns "false" if there are pending 51 * This re-enables callbacks; it returns "false" if there are pending
52 * buffers in the queue, to detect a possible race between the driver 52 * buffers in the queue, to detect a possible race between the driver
53 * checking for more work, and enabling callbacks. 53 * checking for more work, and enabling callbacks.
54 * virtqueue_enable_cb_delayed: restart callbacks after disable_cb.
55 * vq: the struct virtqueue we're talking about.
56 * This re-enables callbacks but hints to the other side to delay
57 * interrupts until most of the available buffers have been processed;
58 * it returns "false" if there are many pending buffers in the queue,
59 * to detect a possible race between the driver checking for more work,
60 * and enabling callbacks.
54 * virtqueue_detach_unused_buf: detach first unused buffer 61 * virtqueue_detach_unused_buf: detach first unused buffer
55 * vq: the struct virtqueue we're talking about. 62 * vq: the struct virtqueue we're talking about.
56 * Returns NULL or the "data" token handed to add_buf 63 * Returns NULL or the "data" token handed to add_buf
@@ -86,6 +93,8 @@ void virtqueue_disable_cb(struct virtqueue *vq);
86 93
87bool virtqueue_enable_cb(struct virtqueue *vq); 94bool virtqueue_enable_cb(struct virtqueue *vq);
88 95
96bool virtqueue_enable_cb_delayed(struct virtqueue *vq);
97
89void *virtqueue_detach_unused_buf(struct virtqueue *vq); 98void *virtqueue_detach_unused_buf(struct virtqueue *vq);
90 99
91/** 100/**
diff --git a/include/linux/virtio_9p.h b/include/linux/virtio_9p.h
index e68b439b2860..277c4ad44e84 100644
--- a/include/linux/virtio_9p.h
+++ b/include/linux/virtio_9p.h
@@ -1,7 +1,30 @@
1#ifndef _LINUX_VIRTIO_9P_H 1#ifndef _LINUX_VIRTIO_9P_H
2#define _LINUX_VIRTIO_9P_H 2#define _LINUX_VIRTIO_9P_H
3/* This header is BSD licensed so anyone can use the definitions to implement 3/* This header is BSD licensed so anyone can use the definitions to implement
4 * compatible drivers/servers. */ 4 * compatible drivers/servers.
5 *
6 * Redistribution and use in source and binary forms, with or without
7 * modification, are permitted provided that the following conditions
8 * are met:
9 * 1. Redistributions of source code must retain the above copyright
10 * notice, this list of conditions and the following disclaimer.
11 * 2. Redistributions in binary form must reproduce the above copyright
12 * notice, this list of conditions and the following disclaimer in the
13 * documentation and/or other materials provided with the distribution.
14 * 3. Neither the name of IBM nor the names of its contributors
15 * may be used to endorse or promote products derived from this software
16 * without specific prior written permission.
17 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS ``AS IS'' AND
18 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
19 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
20 * ARE DISCLAIMED. IN NO EVENT SHALL IBM OR CONTRIBUTORS BE LIABLE
21 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
22 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
23 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
24 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
25 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
26 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
27 * SUCH DAMAGE. */
5#include <linux/types.h> 28#include <linux/types.h>
6#include <linux/virtio_ids.h> 29#include <linux/virtio_ids.h>
7#include <linux/virtio_config.h> 30#include <linux/virtio_config.h>
diff --git a/include/linux/virtio_balloon.h b/include/linux/virtio_balloon.h
index a50ecd1b81a2..652dc8bea921 100644
--- a/include/linux/virtio_balloon.h
+++ b/include/linux/virtio_balloon.h
@@ -1,7 +1,30 @@
1#ifndef _LINUX_VIRTIO_BALLOON_H 1#ifndef _LINUX_VIRTIO_BALLOON_H
2#define _LINUX_VIRTIO_BALLOON_H 2#define _LINUX_VIRTIO_BALLOON_H
3/* This header is BSD licensed so anyone can use the definitions to implement 3/* This header is BSD licensed so anyone can use the definitions to implement
4 * compatible drivers/servers. */ 4 * compatible drivers/servers.
5 *
6 * Redistribution and use in source and binary forms, with or without
7 * modification, are permitted provided that the following conditions
8 * are met:
9 * 1. Redistributions of source code must retain the above copyright
10 * notice, this list of conditions and the following disclaimer.
11 * 2. Redistributions in binary form must reproduce the above copyright
12 * notice, this list of conditions and the following disclaimer in the
13 * documentation and/or other materials provided with the distribution.
14 * 3. Neither the name of IBM nor the names of its contributors
15 * may be used to endorse or promote products derived from this software
16 * without specific prior written permission.
17 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS ``AS IS'' AND
18 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
19 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
20 * ARE DISCLAIMED. IN NO EVENT SHALL IBM OR CONTRIBUTORS BE LIABLE
21 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
22 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
23 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
24 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
25 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
26 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
27 * SUCH DAMAGE. */
5#include <linux/virtio_ids.h> 28#include <linux/virtio_ids.h>
6#include <linux/virtio_config.h> 29#include <linux/virtio_config.h>
7 30
diff --git a/include/linux/virtio_blk.h b/include/linux/virtio_blk.h
index 167720d695ed..e0edb40ca7aa 100644
--- a/include/linux/virtio_blk.h
+++ b/include/linux/virtio_blk.h
@@ -1,7 +1,30 @@
1#ifndef _LINUX_VIRTIO_BLK_H 1#ifndef _LINUX_VIRTIO_BLK_H
2#define _LINUX_VIRTIO_BLK_H 2#define _LINUX_VIRTIO_BLK_H
3/* This header is BSD licensed so anyone can use the definitions to implement 3/* This header is BSD licensed so anyone can use the definitions to implement
4 * compatible drivers/servers. */ 4 * compatible drivers/servers.
5 *
6 * Redistribution and use in source and binary forms, with or without
7 * modification, are permitted provided that the following conditions
8 * are met:
9 * 1. Redistributions of source code must retain the above copyright
10 * notice, this list of conditions and the following disclaimer.
11 * 2. Redistributions in binary form must reproduce the above copyright
12 * notice, this list of conditions and the following disclaimer in the
13 * documentation and/or other materials provided with the distribution.
14 * 3. Neither the name of IBM nor the names of its contributors
15 * may be used to endorse or promote products derived from this software
16 * without specific prior written permission.
17 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS ``AS IS'' AND
18 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
19 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
20 * ARE DISCLAIMED. IN NO EVENT SHALL IBM OR CONTRIBUTORS BE LIABLE
21 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
22 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
23 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
24 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
25 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
26 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
27 * SUCH DAMAGE. */
5#include <linux/types.h> 28#include <linux/types.h>
6#include <linux/virtio_ids.h> 29#include <linux/virtio_ids.h>
7#include <linux/virtio_config.h> 30#include <linux/virtio_config.h>
diff --git a/include/linux/virtio_config.h b/include/linux/virtio_config.h
index 800617b4ddd5..39c88c5ad19d 100644
--- a/include/linux/virtio_config.h
+++ b/include/linux/virtio_config.h
@@ -1,7 +1,30 @@
1#ifndef _LINUX_VIRTIO_CONFIG_H 1#ifndef _LINUX_VIRTIO_CONFIG_H
2#define _LINUX_VIRTIO_CONFIG_H 2#define _LINUX_VIRTIO_CONFIG_H
3/* This header, excluding the #ifdef __KERNEL__ part, is BSD licensed so 3/* This header, excluding the #ifdef __KERNEL__ part, is BSD licensed so
4 * anyone can use the definitions to implement compatible drivers/servers. */ 4 * anyone can use the definitions to implement compatible drivers/servers.
5 *
6 * Redistribution and use in source and binary forms, with or without
7 * modification, are permitted provided that the following conditions
8 * are met:
9 * 1. Redistributions of source code must retain the above copyright
10 * notice, this list of conditions and the following disclaimer.
11 * 2. Redistributions in binary form must reproduce the above copyright
12 * notice, this list of conditions and the following disclaimer in the
13 * documentation and/or other materials provided with the distribution.
14 * 3. Neither the name of IBM nor the names of its contributors
15 * may be used to endorse or promote products derived from this software
16 * without specific prior written permission.
17 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS ``AS IS'' AND
18 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
19 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
20 * ARE DISCLAIMED. IN NO EVENT SHALL IBM OR CONTRIBUTORS BE LIABLE
21 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
22 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
23 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
24 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
25 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
26 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
27 * SUCH DAMAGE. */
5 28
6/* Virtio devices use a standardized configuration space to define their 29/* Virtio devices use a standardized configuration space to define their
7 * features and pass configuration information, but each implementation can 30 * features and pass configuration information, but each implementation can
diff --git a/include/linux/virtio_console.h b/include/linux/virtio_console.h
index e4d333543a33..bdf4b0034739 100644
--- a/include/linux/virtio_console.h
+++ b/include/linux/virtio_console.h
@@ -5,7 +5,31 @@
5#include <linux/virtio_config.h> 5#include <linux/virtio_config.h>
6/* 6/*
7 * This header, excluding the #ifdef __KERNEL__ part, is BSD licensed so 7 * This header, excluding the #ifdef __KERNEL__ part, is BSD licensed so
8 * anyone can use the definitions to implement compatible drivers/servers. 8 * anyone can use the definitions to implement compatible drivers/servers:
9 *
10 *
11 * Redistribution and use in source and binary forms, with or without
12 * modification, are permitted provided that the following conditions
13 * are met:
14 * 1. Redistributions of source code must retain the above copyright
15 * notice, this list of conditions and the following disclaimer.
16 * 2. Redistributions in binary form must reproduce the above copyright
17 * notice, this list of conditions and the following disclaimer in the
18 * documentation and/or other materials provided with the distribution.
19 * 3. Neither the name of IBM nor the names of its contributors
20 * may be used to endorse or promote products derived from this software
21 * without specific prior written permission.
22 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS ``AS IS'' AND
23 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
24 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
25 * ARE DISCLAIMED. IN NO EVENT SHALL IBM OR CONTRIBUTORS BE LIABLE
26 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
27 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
28 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
29 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
30 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
31 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
32 * SUCH DAMAGE.
9 * 33 *
10 * Copyright (C) Red Hat, Inc., 2009, 2010, 2011 34 * Copyright (C) Red Hat, Inc., 2009, 2010, 2011
11 * Copyright (C) Amit Shah <amit.shah@redhat.com>, 2009, 2010, 2011 35 * Copyright (C) Amit Shah <amit.shah@redhat.com>, 2009, 2010, 2011
diff --git a/include/linux/virtio_ids.h b/include/linux/virtio_ids.h
index 06660c0a78d7..85bb0bb66ffc 100644
--- a/include/linux/virtio_ids.h
+++ b/include/linux/virtio_ids.h
@@ -5,7 +5,29 @@
5 * 5 *
6 * This header is BSD licensed so anyone can use the definitions to implement 6 * This header is BSD licensed so anyone can use the definitions to implement
7 * compatible drivers/servers. 7 * compatible drivers/servers.
8 */ 8 *
9 * Redistribution and use in source and binary forms, with or without
10 * modification, are permitted provided that the following conditions
11 * are met:
12 * 1. Redistributions of source code must retain the above copyright
13 * notice, this list of conditions and the following disclaimer.
14 * 2. Redistributions in binary form must reproduce the above copyright
15 * notice, this list of conditions and the following disclaimer in the
16 * documentation and/or other materials provided with the distribution.
17 * 3. Neither the name of IBM nor the names of its contributors
18 * may be used to endorse or promote products derived from this software
19 * without specific prior written permission.
20 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS ``AS IS'' AND
21 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
22 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
23 * ARE DISCLAIMED. IN NO EVENT SHALL IBM OR CONTRIBUTORS BE LIABLE
24 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
25 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
26 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
27 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
28 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
29 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
30 * SUCH DAMAGE. */
9 31
10#define VIRTIO_ID_NET 1 /* virtio net */ 32#define VIRTIO_ID_NET 1 /* virtio net */
11#define VIRTIO_ID_BLOCK 2 /* virtio block */ 33#define VIRTIO_ID_BLOCK 2 /* virtio block */
diff --git a/include/linux/virtio_net.h b/include/linux/virtio_net.h
index 085e42298ce5..136040bba3e3 100644
--- a/include/linux/virtio_net.h
+++ b/include/linux/virtio_net.h
@@ -1,7 +1,30 @@
1#ifndef _LINUX_VIRTIO_NET_H 1#ifndef _LINUX_VIRTIO_NET_H
2#define _LINUX_VIRTIO_NET_H 2#define _LINUX_VIRTIO_NET_H
3/* This header is BSD licensed so anyone can use the definitions to implement 3/* This header is BSD licensed so anyone can use the definitions to implement
4 * compatible drivers/servers. */ 4 * compatible drivers/servers.
5 *
6 * Redistribution and use in source and binary forms, with or without
7 * modification, are permitted provided that the following conditions
8 * are met:
9 * 1. Redistributions of source code must retain the above copyright
10 * notice, this list of conditions and the following disclaimer.
11 * 2. Redistributions in binary form must reproduce the above copyright
12 * notice, this list of conditions and the following disclaimer in the
13 * documentation and/or other materials provided with the distribution.
14 * 3. Neither the name of IBM nor the names of its contributors
15 * may be used to endorse or promote products derived from this software
16 * without specific prior written permission.
17 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS ``AS IS'' AND
18 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
19 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
20 * ARE DISCLAIMED. IN NO EVENT SHALL IBM OR CONTRIBUTORS BE LIABLE
21 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
22 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
23 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
24 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
25 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
26 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
27 * SUCH DAMAGE. */
5#include <linux/types.h> 28#include <linux/types.h>
6#include <linux/virtio_ids.h> 29#include <linux/virtio_ids.h>
7#include <linux/virtio_config.h> 30#include <linux/virtio_config.h>
diff --git a/include/linux/virtio_pci.h b/include/linux/virtio_pci.h
index 9a3d7c48c622..ea66f3f60d63 100644
--- a/include/linux/virtio_pci.h
+++ b/include/linux/virtio_pci.h
@@ -11,6 +11,29 @@
11 * 11 *
12 * This header is BSD licensed so anyone can use the definitions to implement 12 * This header is BSD licensed so anyone can use the definitions to implement
13 * compatible drivers/servers. 13 * compatible drivers/servers.
14 *
15 * Redistribution and use in source and binary forms, with or without
16 * modification, are permitted provided that the following conditions
17 * are met:
18 * 1. Redistributions of source code must retain the above copyright
19 * notice, this list of conditions and the following disclaimer.
20 * 2. Redistributions in binary form must reproduce the above copyright
21 * notice, this list of conditions and the following disclaimer in the
22 * documentation and/or other materials provided with the distribution.
23 * 3. Neither the name of IBM nor the names of its contributors
24 * may be used to endorse or promote products derived from this software
25 * without specific prior written permission.
26 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS ``AS IS'' AND
27 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
28 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
29 * ARE DISCLAIMED. IN NO EVENT SHALL IBM OR CONTRIBUTORS BE LIABLE
30 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
31 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
32 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
33 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
34 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
35 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
36 * SUCH DAMAGE.
14 */ 37 */
15 38
16#ifndef _LINUX_VIRTIO_PCI_H 39#ifndef _LINUX_VIRTIO_PCI_H
diff --git a/include/linux/virtio_ring.h b/include/linux/virtio_ring.h
index e4d144b132b5..4a32cb6da425 100644
--- a/include/linux/virtio_ring.h
+++ b/include/linux/virtio_ring.h
@@ -7,6 +7,29 @@
7 * This header is BSD licensed so anyone can use the definitions to implement 7 * This header is BSD licensed so anyone can use the definitions to implement
8 * compatible drivers/servers. 8 * compatible drivers/servers.
9 * 9 *
10 * Redistribution and use in source and binary forms, with or without
11 * modification, are permitted provided that the following conditions
12 * are met:
13 * 1. Redistributions of source code must retain the above copyright
14 * notice, this list of conditions and the following disclaimer.
15 * 2. Redistributions in binary form must reproduce the above copyright
16 * notice, this list of conditions and the following disclaimer in the
17 * documentation and/or other materials provided with the distribution.
18 * 3. Neither the name of IBM nor the names of its contributors
19 * may be used to endorse or promote products derived from this software
20 * without specific prior written permission.
21 * THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS ``AS IS'' AND
22 * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
23 * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
24 * ARE DISCLAIMED. IN NO EVENT SHALL IBM OR CONTRIBUTORS BE LIABLE
25 * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
26 * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
27 * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
28 * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
29 * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
30 * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
31 * SUCH DAMAGE.
32 *
10 * Copyright Rusty Russell IBM Corporation 2007. */ 33 * Copyright Rusty Russell IBM Corporation 2007. */
11#include <linux/types.h> 34#include <linux/types.h>
12 35
@@ -29,6 +52,12 @@
29/* We support indirect buffer descriptors */ 52/* We support indirect buffer descriptors */
30#define VIRTIO_RING_F_INDIRECT_DESC 28 53#define VIRTIO_RING_F_INDIRECT_DESC 28
31 54
55/* The Guest publishes the used index for which it expects an interrupt
56 * at the end of the avail ring. Host should ignore the avail->flags field. */
57/* The Host publishes the avail index for which it expects a kick
58 * at the end of the used ring. Guest should ignore the used->flags field. */
59#define VIRTIO_RING_F_EVENT_IDX 29
60
32/* Virtio ring descriptors: 16 bytes. These can chain together via "next". */ 61/* Virtio ring descriptors: 16 bytes. These can chain together via "next". */
33struct vring_desc { 62struct vring_desc {
34 /* Address (guest-physical). */ 63 /* Address (guest-physical). */
@@ -83,6 +112,7 @@ struct vring {
83 * __u16 avail_flags; 112 * __u16 avail_flags;
84 * __u16 avail_idx; 113 * __u16 avail_idx;
85 * __u16 available[num]; 114 * __u16 available[num];
115 * __u16 used_event_idx;
86 * 116 *
87 * // Padding to the next align boundary. 117 * // Padding to the next align boundary.
88 * char pad[]; 118 * char pad[];
@@ -91,8 +121,14 @@ struct vring {
91 * __u16 used_flags; 121 * __u16 used_flags;
92 * __u16 used_idx; 122 * __u16 used_idx;
93 * struct vring_used_elem used[num]; 123 * struct vring_used_elem used[num];
124 * __u16 avail_event_idx;
94 * }; 125 * };
95 */ 126 */
127/* We publish the used event index at the end of the available ring, and vice
128 * versa. They are at the end for backwards compatibility. */
129#define vring_used_event(vr) ((vr)->avail->ring[(vr)->num])
130#define vring_avail_event(vr) (*(__u16 *)&(vr)->used->ring[(vr)->num])
131
96static inline void vring_init(struct vring *vr, unsigned int num, void *p, 132static inline void vring_init(struct vring *vr, unsigned int num, void *p,
97 unsigned long align) 133 unsigned long align)
98{ 134{
@@ -107,7 +143,21 @@ static inline unsigned vring_size(unsigned int num, unsigned long align)
107{ 143{
108 return ((sizeof(struct vring_desc) * num + sizeof(__u16) * (2 + num) 144 return ((sizeof(struct vring_desc) * num + sizeof(__u16) * (2 + num)
109 + align - 1) & ~(align - 1)) 145 + align - 1) & ~(align - 1))
110 + sizeof(__u16) * 2 + sizeof(struct vring_used_elem) * num; 146 + sizeof(__u16) * 3 + sizeof(struct vring_used_elem) * num;
147}
148
149/* The following is used with USED_EVENT_IDX and AVAIL_EVENT_IDX */
150/* Assuming a given event_idx value from the other size, if
151 * we have just incremented index from old to new_idx,
152 * should we trigger an event? */
153static inline int vring_need_event(__u16 event_idx, __u16 new_idx, __u16 old)
154{
155 /* Note: Xen has similar logic for notification hold-off
156 * in include/xen/interface/io/ring.h with req_event and req_prod
157 * corresponding to event_idx + 1 and new_idx respectively.
158 * Note also that req_event and req_prod in Xen start at 1,
159 * event indexes in virtio start at 0. */
160 return (__u16)(new_idx - event_idx - 1) < (__u16)(new_idx - old);
111} 161}
112 162
113#ifdef __KERNEL__ 163#ifdef __KERNEL__
diff --git a/kernel/events/core.c b/kernel/events/core.c
index cc5d57d1d0b6..ba89f40abe6a 100644
--- a/kernel/events/core.c
+++ b/kernel/events/core.c
@@ -7388,26 +7388,12 @@ static int __perf_cgroup_move(void *info)
7388 return 0; 7388 return 0;
7389} 7389}
7390 7390
7391static void perf_cgroup_move(struct task_struct *task) 7391static void
7392perf_cgroup_attach_task(struct cgroup *cgrp, struct task_struct *task)
7392{ 7393{
7393 task_function_call(task, __perf_cgroup_move, task); 7394 task_function_call(task, __perf_cgroup_move, task);
7394} 7395}
7395 7396
7396static void perf_cgroup_attach(struct cgroup_subsys *ss, struct cgroup *cgrp,
7397 struct cgroup *old_cgrp, struct task_struct *task,
7398 bool threadgroup)
7399{
7400 perf_cgroup_move(task);
7401 if (threadgroup) {
7402 struct task_struct *c;
7403 rcu_read_lock();
7404 list_for_each_entry_rcu(c, &task->thread_group, thread_group) {
7405 perf_cgroup_move(c);
7406 }
7407 rcu_read_unlock();
7408 }
7409}
7410
7411static void perf_cgroup_exit(struct cgroup_subsys *ss, struct cgroup *cgrp, 7397static void perf_cgroup_exit(struct cgroup_subsys *ss, struct cgroup *cgrp,
7412 struct cgroup *old_cgrp, struct task_struct *task) 7398 struct cgroup *old_cgrp, struct task_struct *task)
7413{ 7399{
@@ -7419,7 +7405,7 @@ static void perf_cgroup_exit(struct cgroup_subsys *ss, struct cgroup *cgrp,
7419 if (!(task->flags & PF_EXITING)) 7405 if (!(task->flags & PF_EXITING))
7420 return; 7406 return;
7421 7407
7422 perf_cgroup_move(task); 7408 perf_cgroup_attach_task(cgrp, task);
7423} 7409}
7424 7410
7425struct cgroup_subsys perf_subsys = { 7411struct cgroup_subsys perf_subsys = {
@@ -7428,6 +7414,6 @@ struct cgroup_subsys perf_subsys = {
7428 .create = perf_cgroup_create, 7414 .create = perf_cgroup_create,
7429 .destroy = perf_cgroup_destroy, 7415 .destroy = perf_cgroup_destroy,
7430 .exit = perf_cgroup_exit, 7416 .exit = perf_cgroup_exit,
7431 .attach = perf_cgroup_attach, 7417 .attach_task = perf_cgroup_attach_task,
7432}; 7418};
7433#endif /* CONFIG_CGROUP_PERF */ 7419#endif /* CONFIG_CGROUP_PERF */
diff --git a/kernel/rcutree.c b/kernel/rcutree.c
index 77a7671dd147..89419ff92e99 100644
--- a/kernel/rcutree.c
+++ b/kernel/rcutree.c
@@ -1648,7 +1648,6 @@ static int __cpuinit rcu_spawn_one_cpu_kthread(int cpu)
1648 if (IS_ERR(t)) 1648 if (IS_ERR(t))
1649 return PTR_ERR(t); 1649 return PTR_ERR(t);
1650 kthread_bind(t, cpu); 1650 kthread_bind(t, cpu);
1651 set_task_state(t, TASK_INTERRUPTIBLE);
1652 per_cpu(rcu_cpu_kthread_cpu, cpu) = cpu; 1651 per_cpu(rcu_cpu_kthread_cpu, cpu) = cpu;
1653 WARN_ON_ONCE(per_cpu(rcu_cpu_kthread_task, cpu) != NULL); 1652 WARN_ON_ONCE(per_cpu(rcu_cpu_kthread_task, cpu) != NULL);
1654 per_cpu(rcu_cpu_kthread_task, cpu) = t; 1653 per_cpu(rcu_cpu_kthread_task, cpu) = t;
@@ -1756,7 +1755,6 @@ static int __cpuinit rcu_spawn_one_node_kthread(struct rcu_state *rsp,
1756 if (IS_ERR(t)) 1755 if (IS_ERR(t))
1757 return PTR_ERR(t); 1756 return PTR_ERR(t);
1758 raw_spin_lock_irqsave(&rnp->lock, flags); 1757 raw_spin_lock_irqsave(&rnp->lock, flags);
1759 set_task_state(t, TASK_INTERRUPTIBLE);
1760 rnp->node_kthread_task = t; 1758 rnp->node_kthread_task = t;
1761 raw_spin_unlock_irqrestore(&rnp->lock, flags); 1759 raw_spin_unlock_irqrestore(&rnp->lock, flags);
1762 sp.sched_priority = 99; 1760 sp.sched_priority = 99;
@@ -1765,6 +1763,8 @@ static int __cpuinit rcu_spawn_one_node_kthread(struct rcu_state *rsp,
1765 return rcu_spawn_one_boost_kthread(rsp, rnp, rnp_index); 1763 return rcu_spawn_one_boost_kthread(rsp, rnp, rnp_index);
1766} 1764}
1767 1765
1766static void rcu_wake_one_boost_kthread(struct rcu_node *rnp);
1767
1768/* 1768/*
1769 * Spawn all kthreads -- called as soon as the scheduler is running. 1769 * Spawn all kthreads -- called as soon as the scheduler is running.
1770 */ 1770 */
@@ -1772,18 +1772,30 @@ static int __init rcu_spawn_kthreads(void)
1772{ 1772{
1773 int cpu; 1773 int cpu;
1774 struct rcu_node *rnp; 1774 struct rcu_node *rnp;
1775 struct task_struct *t;
1775 1776
1776 rcu_kthreads_spawnable = 1; 1777 rcu_kthreads_spawnable = 1;
1777 for_each_possible_cpu(cpu) { 1778 for_each_possible_cpu(cpu) {
1778 per_cpu(rcu_cpu_has_work, cpu) = 0; 1779 per_cpu(rcu_cpu_has_work, cpu) = 0;
1779 if (cpu_online(cpu)) 1780 if (cpu_online(cpu)) {
1780 (void)rcu_spawn_one_cpu_kthread(cpu); 1781 (void)rcu_spawn_one_cpu_kthread(cpu);
1782 t = per_cpu(rcu_cpu_kthread_task, cpu);
1783 if (t)
1784 wake_up_process(t);
1785 }
1781 } 1786 }
1782 rnp = rcu_get_root(rcu_state); 1787 rnp = rcu_get_root(rcu_state);
1783 (void)rcu_spawn_one_node_kthread(rcu_state, rnp); 1788 (void)rcu_spawn_one_node_kthread(rcu_state, rnp);
1789 if (rnp->node_kthread_task)
1790 wake_up_process(rnp->node_kthread_task);
1784 if (NUM_RCU_NODES > 1) { 1791 if (NUM_RCU_NODES > 1) {
1785 rcu_for_each_leaf_node(rcu_state, rnp) 1792 rcu_for_each_leaf_node(rcu_state, rnp) {
1786 (void)rcu_spawn_one_node_kthread(rcu_state, rnp); 1793 (void)rcu_spawn_one_node_kthread(rcu_state, rnp);
1794 t = rnp->node_kthread_task;
1795 if (t)
1796 wake_up_process(t);
1797 rcu_wake_one_boost_kthread(rnp);
1798 }
1787 } 1799 }
1788 return 0; 1800 return 0;
1789} 1801}
@@ -2188,14 +2200,14 @@ rcu_init_percpu_data(int cpu, struct rcu_state *rsp, int preemptible)
2188 raw_spin_unlock_irqrestore(&rsp->onofflock, flags); 2200 raw_spin_unlock_irqrestore(&rsp->onofflock, flags);
2189} 2201}
2190 2202
2191static void __cpuinit rcu_online_cpu(int cpu) 2203static void __cpuinit rcu_prepare_cpu(int cpu)
2192{ 2204{
2193 rcu_init_percpu_data(cpu, &rcu_sched_state, 0); 2205 rcu_init_percpu_data(cpu, &rcu_sched_state, 0);
2194 rcu_init_percpu_data(cpu, &rcu_bh_state, 0); 2206 rcu_init_percpu_data(cpu, &rcu_bh_state, 0);
2195 rcu_preempt_init_percpu_data(cpu); 2207 rcu_preempt_init_percpu_data(cpu);
2196} 2208}
2197 2209
2198static void __cpuinit rcu_online_kthreads(int cpu) 2210static void __cpuinit rcu_prepare_kthreads(int cpu)
2199{ 2211{
2200 struct rcu_data *rdp = per_cpu_ptr(rcu_state->rda, cpu); 2212 struct rcu_data *rdp = per_cpu_ptr(rcu_state->rda, cpu);
2201 struct rcu_node *rnp = rdp->mynode; 2213 struct rcu_node *rnp = rdp->mynode;
@@ -2209,6 +2221,31 @@ static void __cpuinit rcu_online_kthreads(int cpu)
2209} 2221}
2210 2222
2211/* 2223/*
2224 * kthread_create() creates threads in TASK_UNINTERRUPTIBLE state,
2225 * but the RCU threads are woken on demand, and if demand is low this
2226 * could be a while triggering the hung task watchdog.
2227 *
2228 * In order to avoid this, poke all tasks once the CPU is fully
2229 * up and running.
2230 */
2231static void __cpuinit rcu_online_kthreads(int cpu)
2232{
2233 struct rcu_data *rdp = per_cpu_ptr(rcu_state->rda, cpu);
2234 struct rcu_node *rnp = rdp->mynode;
2235 struct task_struct *t;
2236
2237 t = per_cpu(rcu_cpu_kthread_task, cpu);
2238 if (t)
2239 wake_up_process(t);
2240
2241 t = rnp->node_kthread_task;
2242 if (t)
2243 wake_up_process(t);
2244
2245 rcu_wake_one_boost_kthread(rnp);
2246}
2247
2248/*
2212 * Handle CPU online/offline notification events. 2249 * Handle CPU online/offline notification events.
2213 */ 2250 */
2214static int __cpuinit rcu_cpu_notify(struct notifier_block *self, 2251static int __cpuinit rcu_cpu_notify(struct notifier_block *self,
@@ -2221,10 +2258,11 @@ static int __cpuinit rcu_cpu_notify(struct notifier_block *self,
2221 switch (action) { 2258 switch (action) {
2222 case CPU_UP_PREPARE: 2259 case CPU_UP_PREPARE:
2223 case CPU_UP_PREPARE_FROZEN: 2260 case CPU_UP_PREPARE_FROZEN:
2224 rcu_online_cpu(cpu); 2261 rcu_prepare_cpu(cpu);
2225 rcu_online_kthreads(cpu); 2262 rcu_prepare_kthreads(cpu);
2226 break; 2263 break;
2227 case CPU_ONLINE: 2264 case CPU_ONLINE:
2265 rcu_online_kthreads(cpu);
2228 case CPU_DOWN_FAILED: 2266 case CPU_DOWN_FAILED:
2229 rcu_node_kthread_setaffinity(rnp, -1); 2267 rcu_node_kthread_setaffinity(rnp, -1);
2230 rcu_cpu_kthread_setrt(cpu, 1); 2268 rcu_cpu_kthread_setrt(cpu, 1);
diff --git a/kernel/rcutree_plugin.h b/kernel/rcutree_plugin.h
index a767b7dac365..c8bff3099a89 100644
--- a/kernel/rcutree_plugin.h
+++ b/kernel/rcutree_plugin.h
@@ -1295,7 +1295,6 @@ static int __cpuinit rcu_spawn_one_boost_kthread(struct rcu_state *rsp,
1295 if (IS_ERR(t)) 1295 if (IS_ERR(t))
1296 return PTR_ERR(t); 1296 return PTR_ERR(t);
1297 raw_spin_lock_irqsave(&rnp->lock, flags); 1297 raw_spin_lock_irqsave(&rnp->lock, flags);
1298 set_task_state(t, TASK_INTERRUPTIBLE);
1299 rnp->boost_kthread_task = t; 1298 rnp->boost_kthread_task = t;
1300 raw_spin_unlock_irqrestore(&rnp->lock, flags); 1299 raw_spin_unlock_irqrestore(&rnp->lock, flags);
1301 sp.sched_priority = RCU_KTHREAD_PRIO; 1300 sp.sched_priority = RCU_KTHREAD_PRIO;
@@ -1303,6 +1302,12 @@ static int __cpuinit rcu_spawn_one_boost_kthread(struct rcu_state *rsp,
1303 return 0; 1302 return 0;
1304} 1303}
1305 1304
1305static void __cpuinit rcu_wake_one_boost_kthread(struct rcu_node *rnp)
1306{
1307 if (rnp->boost_kthread_task)
1308 wake_up_process(rnp->boost_kthread_task);
1309}
1310
1306#else /* #ifdef CONFIG_RCU_BOOST */ 1311#else /* #ifdef CONFIG_RCU_BOOST */
1307 1312
1308static void rcu_initiate_boost(struct rcu_node *rnp, unsigned long flags) 1313static void rcu_initiate_boost(struct rcu_node *rnp, unsigned long flags)
@@ -1326,6 +1331,10 @@ static int __cpuinit rcu_spawn_one_boost_kthread(struct rcu_state *rsp,
1326 return 0; 1331 return 0;
1327} 1332}
1328 1333
1334static void __cpuinit rcu_wake_one_boost_kthread(struct rcu_node *rnp)
1335{
1336}
1337
1329#endif /* #else #ifdef CONFIG_RCU_BOOST */ 1338#endif /* #else #ifdef CONFIG_RCU_BOOST */
1330 1339
1331#ifndef CONFIG_SMP 1340#ifndef CONFIG_SMP
diff --git a/kernel/sysctl.c b/kernel/sysctl.c
index 4fc92445a29c..f175d98bd355 100644
--- a/kernel/sysctl.c
+++ b/kernel/sysctl.c
@@ -938,6 +938,12 @@ static struct ctl_table kern_table[] = {
938 }, 938 },
939#endif 939#endif
940#ifdef CONFIG_PERF_EVENTS 940#ifdef CONFIG_PERF_EVENTS
941 /*
942 * User-space scripts rely on the existence of this file
943 * as a feature check for perf_events being enabled.
944 *
945 * So it's an ABI, do not remove!
946 */
941 { 947 {
942 .procname = "perf_event_paranoid", 948 .procname = "perf_event_paranoid",
943 .data = &sysctl_perf_event_paranoid, 949 .data = &sysctl_perf_event_paranoid,
diff --git a/mm/page_alloc.c b/mm/page_alloc.c
index a4e1db3f1981..4e8985acdab8 100644
--- a/mm/page_alloc.c
+++ b/mm/page_alloc.c
@@ -2247,10 +2247,6 @@ __alloc_pages_nodemask(gfp_t gfp_mask, unsigned int order,
2247 2247
2248 if (should_fail_alloc_page(gfp_mask, order)) 2248 if (should_fail_alloc_page(gfp_mask, order))
2249 return NULL; 2249 return NULL;
2250#ifndef CONFIG_ZONE_DMA
2251 if (WARN_ON_ONCE(gfp_mask & __GFP_DMA))
2252 return NULL;
2253#endif
2254 2250
2255 /* 2251 /*
2256 * Check the zones suitable for the gfp_mask contain at least one 2252 * Check the zones suitable for the gfp_mask contain at least one
diff --git a/security/apparmor/lsm.c b/security/apparmor/lsm.c
index ae3a698415e6..ec1bcecf2cda 100644
--- a/security/apparmor/lsm.c
+++ b/security/apparmor/lsm.c
@@ -593,7 +593,8 @@ static int apparmor_setprocattr(struct task_struct *task, char *name,
593 sa.aad.op = OP_SETPROCATTR; 593 sa.aad.op = OP_SETPROCATTR;
594 sa.aad.info = name; 594 sa.aad.info = name;
595 sa.aad.error = -EINVAL; 595 sa.aad.error = -EINVAL;
596 return aa_audit(AUDIT_APPARMOR_DENIED, NULL, GFP_KERNEL, 596 return aa_audit(AUDIT_APPARMOR_DENIED,
597 __aa_current_profile(), GFP_KERNEL,
597 &sa, NULL); 598 &sa, NULL);
598 } 599 }
599 } else if (strcmp(name, "exec") == 0) { 600 } else if (strcmp(name, "exec") == 0) {
diff --git a/tools/virtio/virtio_test.c b/tools/virtio/virtio_test.c
index df0c6d2c3860..74d3331bdaf9 100644
--- a/tools/virtio/virtio_test.c
+++ b/tools/virtio/virtio_test.c
@@ -198,6 +198,14 @@ const struct option longopts[] = {
198 .val = 'h', 198 .val = 'h',
199 }, 199 },
200 { 200 {
201 .name = "event-idx",
202 .val = 'E',
203 },
204 {
205 .name = "no-event-idx",
206 .val = 'e',
207 },
208 {
201 .name = "indirect", 209 .name = "indirect",
202 .val = 'I', 210 .val = 'I',
203 }, 211 },
@@ -211,13 +219,17 @@ const struct option longopts[] = {
211 219
212static void help() 220static void help()
213{ 221{
214 fprintf(stderr, "Usage: virtio_test [--help] [--no-indirect]\n"); 222 fprintf(stderr, "Usage: virtio_test [--help]"
223 " [--no-indirect]"
224 " [--no-event-idx]"
225 "\n");
215} 226}
216 227
217int main(int argc, char **argv) 228int main(int argc, char **argv)
218{ 229{
219 struct vdev_info dev; 230 struct vdev_info dev;
220 unsigned long long features = 1ULL << VIRTIO_RING_F_INDIRECT_DESC; 231 unsigned long long features = (1ULL << VIRTIO_RING_F_INDIRECT_DESC) |
232 (1ULL << VIRTIO_RING_F_EVENT_IDX);
221 int o; 233 int o;
222 234
223 for (;;) { 235 for (;;) {
@@ -228,6 +240,9 @@ int main(int argc, char **argv)
228 case '?': 240 case '?':
229 help(); 241 help();
230 exit(2); 242 exit(2);
243 case 'e':
244 features &= ~(1ULL << VIRTIO_RING_F_EVENT_IDX);
245 break;
231 case 'h': 246 case 'h':
232 help(); 247 help();
233 goto done; 248 goto done;
generated by cgit v1.2.2 (git 2.25.0) at 2026-07-22 13:10:30 -0400