From 4143503b486afa8cac2a2c0550fd240593a08834 Mon Sep 17 00:00:00 2001 From: Christian Borntraeger Date: Mon, 30 May 2022 11:27:06 +0200 Subject: [PATCH 01/12] s390/mm: use non-quiescing sske for KVM switch to keyed guest commit 3ae11dbcfac906a8c3a480e98660a823130dc16a upstream. The switch to a keyed guest does not require a classic sske as the other guest CPUs are not accessing the key before the switch is complete. By using the NQ SSKE things are faster especially with multiple guests. Signed-off-by: Christian Borntraeger Suggested-by: Janis Schoetterl-Glausch Reviewed-by: Claudio Imbrenda Link: https://lore.kernel.org/r/20220530092706.11637-3-borntraeger@linux.ibm.com Signed-off-by: Christian Borntraeger Signed-off-by: Heiko Carstens Signed-off-by: Greg Kroah-Hartman --- arch/s390/mm/pgtable.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/arch/s390/mm/pgtable.c b/arch/s390/mm/pgtable.c index 4438c00acb65..28ca07360e97 100644 --- a/arch/s390/mm/pgtable.c +++ b/arch/s390/mm/pgtable.c @@ -716,7 +716,7 @@ void ptep_zap_key(struct mm_struct *mm, unsigned long addr, pte_t *ptep) pgste_val(pgste) |= PGSTE_GR_BIT | PGSTE_GC_BIT; ptev = pte_val(*ptep); if (!(ptev & _PAGE_INVALID) && (ptev & _PAGE_WRITE)) - page_set_storage_key(ptev & PAGE_MASK, PAGE_DEFAULT_KEY, 1); + page_set_storage_key(ptev & PAGE_MASK, PAGE_DEFAULT_KEY, 0); pgste_set_unlock(ptep, pgste); preempt_enable(); } From 2577d67a9a8aa5ee5e3963379f63e427a916de08 Mon Sep 17 00:00:00 2001 From: Mike Snitzer Date: Tue, 21 Jun 2022 12:35:04 -0400 Subject: [PATCH 02/12] dm: remove special-casing of bio-based immutable singleton target on NVMe Commit 9c37de297f6590937f95a28bec1b7ac68a38618f upstream. There is no benefit to DM special-casing NVMe. Remove all code used to establish DM_TYPE_NVME_BIO_BASED. Also, remove 3 'struct mapped_device *md' variables in __map_bio() which masked the same variable that is available within __map_bio()'s scope. Tested-by: Guenter Roeck Signed-off-by: Mike Snitzer Signed-off-by: Greg Kroah-Hartman --- drivers/md/dm-table.c | 32 +-------------- drivers/md/dm.c | 73 ++++------------------------------- include/linux/device-mapper.h | 1 - 3 files changed, 9 insertions(+), 97 deletions(-) diff --git a/drivers/md/dm-table.c b/drivers/md/dm-table.c index 06b382304d92..81bc36a43b32 100644 --- a/drivers/md/dm-table.c +++ b/drivers/md/dm-table.c @@ -872,8 +872,7 @@ EXPORT_SYMBOL(dm_consume_args); static bool __table_type_bio_based(enum dm_queue_mode table_type) { return (table_type == DM_TYPE_BIO_BASED || - table_type == DM_TYPE_DAX_BIO_BASED || - table_type == DM_TYPE_NVME_BIO_BASED); + table_type == DM_TYPE_DAX_BIO_BASED); } static bool __table_type_request_based(enum dm_queue_mode table_type) @@ -929,8 +928,6 @@ bool dm_table_supports_dax(struct dm_table *t, return true; } -static bool dm_table_does_not_support_partial_completion(struct dm_table *t); - static int device_is_rq_stackable(struct dm_target *ti, struct dm_dev *dev, sector_t start, sector_t len, void *data) { @@ -960,7 +957,6 @@ static int dm_table_determine_type(struct dm_table *t) goto verify_bio_based; } BUG_ON(t->type == DM_TYPE_DAX_BIO_BASED); - BUG_ON(t->type == DM_TYPE_NVME_BIO_BASED); goto verify_rq_based; } @@ -999,15 +995,6 @@ verify_bio_based: if (dm_table_supports_dax(t, device_not_dax_capable, &page_size) || (list_empty(devices) && live_md_type == DM_TYPE_DAX_BIO_BASED)) { t->type = DM_TYPE_DAX_BIO_BASED; - } else { - /* Check if upgrading to NVMe bio-based is valid or required */ - tgt = dm_table_get_immutable_target(t); - if (tgt && !tgt->max_io_len && dm_table_does_not_support_partial_completion(t)) { - t->type = DM_TYPE_NVME_BIO_BASED; - goto verify_rq_based; /* must be stacked directly on NVMe (blk-mq) */ - } else if (list_empty(devices) && live_md_type == DM_TYPE_NVME_BIO_BASED) { - t->type = DM_TYPE_NVME_BIO_BASED; - } } return 0; } @@ -1024,8 +1011,7 @@ verify_rq_based: * (e.g. request completion process for partial completion.) */ if (t->num_targets > 1) { - DMERR("%s DM doesn't support multiple targets", - t->type == DM_TYPE_NVME_BIO_BASED ? "nvme bio-based" : "request-based"); + DMERR("request-based DM doesn't support multiple targets"); return -EINVAL; } @@ -1714,20 +1700,6 @@ static int device_is_not_random(struct dm_target *ti, struct dm_dev *dev, return q && !blk_queue_add_random(q); } -static int device_is_partial_completion(struct dm_target *ti, struct dm_dev *dev, - sector_t start, sector_t len, void *data) -{ - char b[BDEVNAME_SIZE]; - - /* For now, NVMe devices are the only devices of this class */ - return (strncmp(bdevname(dev->bdev, b), "nvme", 4) != 0); -} - -static bool dm_table_does_not_support_partial_completion(struct dm_table *t) -{ - return !dm_table_any_dev_attr(t, device_is_partial_completion, NULL); -} - static int device_not_write_same_capable(struct dm_target *ti, struct dm_dev *dev, sector_t start, sector_t len, void *data) { diff --git a/drivers/md/dm.c b/drivers/md/dm.c index 37b8bb4d80f0..77e28f77c59f 100644 --- a/drivers/md/dm.c +++ b/drivers/md/dm.c @@ -1000,7 +1000,7 @@ static void clone_endio(struct bio *bio) struct mapped_device *md = tio->io->md; dm_endio_fn endio = tio->ti->type->end_io; - if (unlikely(error == BLK_STS_TARGET) && md->type != DM_TYPE_NVME_BIO_BASED) { + if (unlikely(error == BLK_STS_TARGET)) { if (bio_op(bio) == REQ_OP_DISCARD && !bio->bi_disk->queue->limits.max_discard_sectors) disable_discard(md); @@ -1325,7 +1325,6 @@ static blk_qc_t __map_bio(struct dm_target_io *tio) sector = clone->bi_iter.bi_sector; if (unlikely(swap_bios_limit(ti, clone))) { - struct mapped_device *md = io->md; int latch = get_swap_bios(); if (unlikely(latch != md->swap_bios)) __set_swap_bios_limit(md, latch); @@ -1340,24 +1339,17 @@ static blk_qc_t __map_bio(struct dm_target_io *tio) /* the bio has been remapped so dispatch it */ trace_block_bio_remap(clone->bi_disk->queue, clone, bio_dev(io->orig_bio), sector); - if (md->type == DM_TYPE_NVME_BIO_BASED) - ret = direct_make_request(clone); - else - ret = generic_make_request(clone); + ret = generic_make_request(clone); break; case DM_MAPIO_KILL: - if (unlikely(swap_bios_limit(ti, clone))) { - struct mapped_device *md = io->md; + if (unlikely(swap_bios_limit(ti, clone))) up(&md->swap_bios_semaphore); - } free_tio(tio); dec_pending(io, BLK_STS_IOERR); break; case DM_MAPIO_REQUEUE: - if (unlikely(swap_bios_limit(ti, clone))) { - struct mapped_device *md = io->md; + if (unlikely(swap_bios_limit(ti, clone))) up(&md->swap_bios_semaphore); - } free_tio(tio); dec_pending(io, BLK_STS_DM_REQUEUE); break; @@ -1732,51 +1724,6 @@ static blk_qc_t __split_and_process_bio(struct mapped_device *md, return ret; } -/* - * Optimized variant of __split_and_process_bio that leverages the - * fact that targets that use it do _not_ have a need to split bios. - */ -static blk_qc_t __process_bio(struct mapped_device *md, struct dm_table *map, - struct bio *bio, struct dm_target *ti) -{ - struct clone_info ci; - blk_qc_t ret = BLK_QC_T_NONE; - int error = 0; - - init_clone_info(&ci, md, map, bio); - - if (bio->bi_opf & REQ_PREFLUSH) { - struct bio flush_bio; - - /* - * Use an on-stack bio for this, it's safe since we don't - * need to reference it after submit. It's just used as - * the basis for the clone(s). - */ - bio_init(&flush_bio, NULL, 0); - flush_bio.bi_opf = REQ_OP_WRITE | REQ_PREFLUSH | REQ_SYNC; - ci.bio = &flush_bio; - ci.sector_count = 0; - error = __send_empty_flush(&ci); - bio_uninit(ci.bio); - /* dec_pending submits any data associated with flush */ - } else { - struct dm_target_io *tio; - - ci.bio = bio; - ci.sector_count = bio_sectors(bio); - if (__process_abnormal_io(&ci, ti, &error)) - goto out; - - tio = alloc_tio(&ci, ti, 0, GFP_NOIO); - ret = __clone_and_map_simple_bio(&ci, tio, NULL); - } -out: - /* drop the extra reference count */ - dec_pending(ci.io, errno_to_blk_status(error)); - return ret; -} - static blk_qc_t dm_process_bio(struct mapped_device *md, struct dm_table *map, struct bio *bio) { @@ -1807,8 +1754,6 @@ static blk_qc_t dm_process_bio(struct mapped_device *md, /* regular IO is split by __split_and_process_bio */ } - if (dm_get_md_type(md) == DM_TYPE_NVME_BIO_BASED) - return __process_bio(md, map, bio, ti); return __split_and_process_bio(md, map, bio); } @@ -2200,12 +2145,10 @@ static struct dm_table *__bind(struct mapped_device *md, struct dm_table *t, if (request_based) dm_stop_queue(q); - if (request_based || md->type == DM_TYPE_NVME_BIO_BASED) { + if (request_based) { /* - * Leverage the fact that request-based DM targets and - * NVMe bio based targets are immutable singletons - * - used to optimize both dm_request_fn and dm_mq_queue_rq; - * and __process_bio. + * Leverage the fact that request-based DM targets are + * immutable singletons - used to optimize dm_mq_queue_rq. */ md->immutable_target = dm_table_get_immutable_target(t); } @@ -2334,7 +2277,6 @@ int dm_setup_md_queue(struct mapped_device *md, struct dm_table *t) break; case DM_TYPE_BIO_BASED: case DM_TYPE_DAX_BIO_BASED: - case DM_TYPE_NVME_BIO_BASED: dm_init_congested_fn(md); break; case DM_TYPE_NONE: @@ -3070,7 +3012,6 @@ struct dm_md_mempools *dm_alloc_md_mempools(struct mapped_device *md, enum dm_qu switch (type) { case DM_TYPE_BIO_BASED: case DM_TYPE_DAX_BIO_BASED: - case DM_TYPE_NVME_BIO_BASED: pool_size = max(dm_get_reserved_bio_based_ios(), min_pool_size); front_pad = roundup(per_io_data_size, __alignof__(struct dm_target_io)) + offsetof(struct dm_target_io, clone); io_front_pad = roundup(front_pad, __alignof__(struct dm_io)) + offsetof(struct dm_io, tio); diff --git a/include/linux/device-mapper.h b/include/linux/device-mapper.h index a53d7d2c2d95..60631f3abddb 100644 --- a/include/linux/device-mapper.h +++ b/include/linux/device-mapper.h @@ -28,7 +28,6 @@ enum dm_queue_mode { DM_TYPE_BIO_BASED = 1, DM_TYPE_REQUEST_BASED = 2, DM_TYPE_DAX_BIO_BASED = 3, - DM_TYPE_NVME_BIO_BASED = 4, }; typedef enum { STATUSTYPE_INFO, STATUSTYPE_TABLE } status_type_t; From bdcbf2602febf5cf19065d4fa091873b1f482779 Mon Sep 17 00:00:00 2001 From: Marian Postevca Date: Fri, 3 Jun 2022 18:34:59 +0300 Subject: [PATCH 03/12] usb: gadget: u_ether: fix regression in setting fixed MAC address commit b337af3a4d6147000b7ca6b3438bf5c820849b37 upstream. In systemd systems setting a fixed MAC address through the "dev_addr" module argument fails systematically. When checking the MAC address after the interface is created it always has the same but different MAC address to the one supplied as argument. This is partially caused by systemd which by default will set an internally generated permanent MAC address for interfaces that are marked as having a randomly generated address. Commit 890d5b40908bfd1a ("usb: gadget: u_ether: fix race in setting MAC address in setup phase") didn't take into account the fact that the interface must be marked as having a set MAC address when it's set as module argument. Fixed by marking the interface with NET_ADDR_SET when the "dev_addr" module argument is supplied. Fixes: 890d5b40908bfd1a ("usb: gadget: u_ether: fix race in setting MAC address in setup phase") Cc: stable@vger.kernel.org Signed-off-by: Marian Postevca Link: https://lore.kernel.org/r/20220603153459.32722-1-posteuca@mutex.one Signed-off-by: Greg Kroah-Hartman --- drivers/usb/gadget/function/u_ether.c | 11 +++++++++-- 1 file changed, 9 insertions(+), 2 deletions(-) diff --git a/drivers/usb/gadget/function/u_ether.c b/drivers/usb/gadget/function/u_ether.c index 271bd08f4a25..3f053b11e2ce 100644 --- a/drivers/usb/gadget/function/u_ether.c +++ b/drivers/usb/gadget/function/u_ether.c @@ -772,9 +772,13 @@ struct eth_dev *gether_setup_name(struct usb_gadget *g, dev->qmult = qmult; snprintf(net->name, sizeof(net->name), "%s%%d", netname); - if (get_ether_addr(dev_addr, net->dev_addr)) + if (get_ether_addr(dev_addr, net->dev_addr)) { + net->addr_assign_type = NET_ADDR_RANDOM; dev_warn(&g->dev, "using random %s ethernet address\n", "self"); + } else { + net->addr_assign_type = NET_ADDR_SET; + } if (get_ether_addr(host_addr, dev->host_mac)) dev_warn(&g->dev, "using random %s ethernet address\n", "host"); @@ -831,6 +835,9 @@ struct net_device *gether_setup_name_default(const char *netname) INIT_LIST_HEAD(&dev->tx_reqs); INIT_LIST_HEAD(&dev->rx_reqs); + /* by default we always have a random MAC address */ + net->addr_assign_type = NET_ADDR_RANDOM; + skb_queue_head_init(&dev->rx_frames); /* network device setup */ @@ -868,7 +875,6 @@ int gether_register_netdev(struct net_device *net) g = dev->gadget; memcpy(net->dev_addr, dev->dev_mac, ETH_ALEN); - net->addr_assign_type = NET_ADDR_RANDOM; status = register_netdev(net); if (status < 0) { @@ -908,6 +914,7 @@ int gether_set_dev_addr(struct net_device *net, const char *dev_addr) if (get_ether_addr(dev_addr, new_addr)) return -EINVAL; memcpy(dev->dev_mac, new_addr, ETH_ALEN); + net->addr_assign_type = NET_ADDR_SET; return 0; } EXPORT_SYMBOL_GPL(gether_set_dev_addr); From 95921a3bab76872e8b5c6ba07903cbdbcc352cc6 Mon Sep 17 00:00:00 2001 From: Eric Dumazet Date: Tue, 9 Feb 2021 11:20:28 -0800 Subject: [PATCH 04/12] tcp: add some entropy in __inet_hash_connect() commit c579bd1b4021c42ae247108f1e6f73dd3f08600c upstream. Even when implementing RFC 6056 3.3.4 (Algorithm 4: Double-Hash Port Selection Algorithm), a patient attacker could still be able to collect enough state from an otherwise idle host. Idea of this patch is to inject some noise, in the cases __inet_hash_connect() found a candidate in the first attempt. This noise should not significantly reduce the collision avoidance, and should be zero if connection table is already well used. Note that this is not implementing RFC 6056 3.3.5 because we think Algorithm 5 could hurt typical workloads. Signed-off-by: Eric Dumazet Cc: David Dworken Cc: Willem de Bruijn Signed-off-by: David S. Miller Cc: Ben Hutchings Signed-off-by: Greg Kroah-Hartman --- net/ipv4/inet_hashtables.c | 5 +++++ 1 file changed, 5 insertions(+) diff --git a/net/ipv4/inet_hashtables.c b/net/ipv4/inet_hashtables.c index 959f4f0c8546..ba1aa49b91ed 100644 --- a/net/ipv4/inet_hashtables.c +++ b/net/ipv4/inet_hashtables.c @@ -782,6 +782,11 @@ next_port: return -EADDRNOTAVAIL; ok: + /* If our first attempt found a candidate, skip next candidate + * in 1/16 of cases to add some noise. + */ + if (!i && !(prandom_u32() % 16)) + i = 2; WRITE_ONCE(table_perturb[index], READ_ONCE(table_perturb[index]) + i + 2); /* Head lock still held and bh's disabled */ From 53c5de3092ade55f82ed4f5373d8a8ca115df818 Mon Sep 17 00:00:00 2001 From: Willy Tarreau Date: Mon, 2 May 2022 10:46:09 +0200 Subject: [PATCH 05/12] tcp: use different parts of the port_offset for index and offset commit 9e9b70ae923baf2b5e8a0ea4fd0c8451801ac526 upstream. Amit Klein suggests that we use different parts of port_offset for the table's index and the port offset so that there is no direct relation between them. Cc: Jason A. Donenfeld Cc: Moshe Kol Cc: Yossi Gilad Cc: Amit Klein Reviewed-by: Eric Dumazet Signed-off-by: Willy Tarreau Signed-off-by: Jakub Kicinski Cc: Ben Hutchings Signed-off-by: Greg Kroah-Hartman --- net/ipv4/inet_hashtables.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/net/ipv4/inet_hashtables.c b/net/ipv4/inet_hashtables.c index ba1aa49b91ed..df3eec00fbc1 100644 --- a/net/ipv4/inet_hashtables.c +++ b/net/ipv4/inet_hashtables.c @@ -726,7 +726,7 @@ int __inet_hash_connect(struct inet_timewait_death_row *death_row, net_get_random_once(table_perturb, sizeof(table_perturb)); index = hash_32(port_offset, INET_TABLE_PERTURB_SHIFT); - offset = READ_ONCE(table_perturb[index]) + port_offset; + offset = READ_ONCE(table_perturb[index]) + (port_offset >> 32); offset %= remaining; /* In first pass we try ports of @low parity. From 7c0a777b7dbdcd39eb45996afe6df7770f7926ac Mon Sep 17 00:00:00 2001 From: Willy Tarreau Date: Mon, 2 May 2022 10:46:11 +0200 Subject: [PATCH 06/12] tcp: add small random increments to the source port commit ca7af0402550f9a0b3316d5f1c30904e42ed257d upstream. Here we're randomly adding between 0 and 7 random increments to the selected source port in order to add some noise in the source port selection that will make the next port less predictable. With the default port range of 32768-60999 this means a worst case reuse scenario of 14116/8=1764 connections between two consecutive uses of the same port, with an average of 14116/4.5=3137. This code was stressed at more than 800000 connections per second to a fixed target with all connections closed by the client using RSTs (worst condition) and only 2 connections failed among 13 billion, despite the hash being reseeded every 10 seconds, indicating a perfectly safe situation. Cc: Moshe Kol Cc: Yossi Gilad Cc: Amit Klein Reviewed-by: Eric Dumazet Signed-off-by: Willy Tarreau Signed-off-by: Jakub Kicinski Cc: Ben Hutchings Signed-off-by: Greg Kroah-Hartman --- net/ipv4/inet_hashtables.c | 9 +++++---- 1 file changed, 5 insertions(+), 4 deletions(-) diff --git a/net/ipv4/inet_hashtables.c b/net/ipv4/inet_hashtables.c index df3eec00fbc1..e88746ed74bf 100644 --- a/net/ipv4/inet_hashtables.c +++ b/net/ipv4/inet_hashtables.c @@ -782,11 +782,12 @@ next_port: return -EADDRNOTAVAIL; ok: - /* If our first attempt found a candidate, skip next candidate - * in 1/16 of cases to add some noise. + /* Here we want to add a little bit of randomness to the next source + * port that will be chosen. We use a max() with a random here so that + * on low contention the randomness is maximal and on high contention + * it may be inexistent. */ - if (!i && !(prandom_u32() % 16)) - i = 2; + i = max_t(int, i, (prandom_u32() & 7) * 2); WRITE_ONCE(table_perturb[index], READ_ONCE(table_perturb[index]) + i + 2); /* Head lock still held and bh's disabled */ From 77d29f3b18c466a6b88bdfceccec3085961a7d0e Mon Sep 17 00:00:00 2001 From: Willy Tarreau Date: Mon, 2 May 2022 10:46:12 +0200 Subject: [PATCH 07/12] tcp: dynamically allocate the perturb table used by source ports commit e9261476184be1abd486c9434164b2acbe0ed6c2 upstream. We'll need to further increase the size of this table and it's likely that at some point its size will not be suitable anymore for a static table. Let's allocate it on boot from inet_hashinfo2_init(), which is called from tcp_init(). Cc: Moshe Kol Cc: Yossi Gilad Cc: Amit Klein Reviewed-by: Eric Dumazet Signed-off-by: Willy Tarreau Signed-off-by: Jakub Kicinski Cc: Ben Hutchings Signed-off-by: Greg Kroah-Hartman --- net/ipv4/inet_hashtables.c | 12 ++++++++++-- 1 file changed, 10 insertions(+), 2 deletions(-) diff --git a/net/ipv4/inet_hashtables.c b/net/ipv4/inet_hashtables.c index e88746ed74bf..8c958f5ac870 100644 --- a/net/ipv4/inet_hashtables.c +++ b/net/ipv4/inet_hashtables.c @@ -680,7 +680,8 @@ EXPORT_SYMBOL_GPL(inet_unhash); * privacy, this only consumes 1 KB of kernel memory. */ #define INET_TABLE_PERTURB_SHIFT 8 -static u32 table_perturb[1 << INET_TABLE_PERTURB_SHIFT]; +#define INET_TABLE_PERTURB_SIZE (1 << INET_TABLE_PERTURB_SHIFT) +static u32 *table_perturb; int __inet_hash_connect(struct inet_timewait_death_row *death_row, struct sock *sk, u64 port_offset, @@ -723,7 +724,8 @@ int __inet_hash_connect(struct inet_timewait_death_row *death_row, if (likely(remaining > 1)) remaining &= ~1U; - net_get_random_once(table_perturb, sizeof(table_perturb)); + net_get_random_once(table_perturb, + INET_TABLE_PERTURB_SIZE * sizeof(*table_perturb)); index = hash_32(port_offset, INET_TABLE_PERTURB_SHIFT); offset = READ_ONCE(table_perturb[index]) + (port_offset >> 32); @@ -861,6 +863,12 @@ void __init inet_hashinfo2_init(struct inet_hashinfo *h, const char *name, low_limit, high_limit); init_hashinfo_lhash2(h); + + /* this one is used for source ports of outgoing connections */ + table_perturb = kmalloc_array(INET_TABLE_PERTURB_SIZE, + sizeof(*table_perturb), GFP_KERNEL); + if (!table_perturb) + panic("TCP: failed to alloc table_perturb"); } int inet_hashinfo2_init_mod(struct inet_hashinfo *h) From c26e1addf15763ae404f4bbf131719a724e768ab Mon Sep 17 00:00:00 2001 From: Willy Tarreau Date: Mon, 2 May 2022 10:46:13 +0200 Subject: [PATCH 08/12] tcp: increase source port perturb table to 2^16 commit 4c2c8f03a5ab7cb04ec64724d7d176d00bcc91e5 upstream. Moshe Kol, Amit Klein, and Yossi Gilad reported being able to accurately identify a client by forcing it to emit only 40 times more connections than there are entries in the table_perturb[] table. The previous two improvements consisting in resalting the secret every 10s and adding randomness to each port selection only slightly improved the situation, and the current value of 2^8 was too small as it's not very difficult to make a client emit 10k connections in less than 10 seconds. Thus we're increasing the perturb table from 2^8 to 2^16 so that the same precision now requires 2.6M connections, which is more difficult in this time frame and harder to hide as a background activity. The impact is that the table now uses 256 kB instead of 1 kB, which could mostly affect devices making frequent outgoing connections. However such components usually target a small set of destinations (load balancers, database clients, perf assessment tools), and in practice only a few entries will be visited, like before. A live test at 1 million connections per second showed no performance difference from the previous value. Reported-by: Moshe Kol Reported-by: Yossi Gilad Reported-by: Amit Klein Reviewed-by: Eric Dumazet Signed-off-by: Willy Tarreau Signed-off-by: Jakub Kicinski Cc: Ben Hutchings Signed-off-by: Greg Kroah-Hartman --- net/ipv4/inet_hashtables.c | 9 +++++---- 1 file changed, 5 insertions(+), 4 deletions(-) diff --git a/net/ipv4/inet_hashtables.c b/net/ipv4/inet_hashtables.c index 8c958f5ac870..ffee136ed8c7 100644 --- a/net/ipv4/inet_hashtables.c +++ b/net/ipv4/inet_hashtables.c @@ -675,11 +675,12 @@ EXPORT_SYMBOL_GPL(inet_unhash); * Note that we use 32bit integers (vs RFC 'short integers') * because 2^16 is not a multiple of num_ephemeral and this * property might be used by clever attacker. - * RFC claims using TABLE_LENGTH=10 buckets gives an improvement, - * we use 256 instead to really give more isolation and - * privacy, this only consumes 1 KB of kernel memory. + * RFC claims using TABLE_LENGTH=10 buckets gives an improvement, though + * attacks were since demonstrated, thus we use 65536 instead to really + * give more isolation and privacy, at the expense of 256kB of kernel + * memory. */ -#define INET_TABLE_PERTURB_SHIFT 8 +#define INET_TABLE_PERTURB_SHIFT 16 #define INET_TABLE_PERTURB_SIZE (1 << INET_TABLE_PERTURB_SHIFT) static u32 *table_perturb; From 2e1591c27b954f1f60ef8ce5d214e8fee1b4d304 Mon Sep 17 00:00:00 2001 From: Willy Tarreau Date: Mon, 2 May 2022 10:46:14 +0200 Subject: [PATCH 09/12] tcp: drop the hash_32() part from the index calculation commit e8161345ddbb66e449abde10d2fdce93f867eba9 upstream. In commit 190cc82489f4 ("tcp: change source port randomizarion at connect() time"), the table_perturb[] array was introduced and an index was taken from the port_offset via hash_32(). But it turns out that hash_32() performs a multiplication while the input here comes from the output of SipHash in secure_seq, that is well distributed enough to avoid the need for yet another hash. Suggested-by: Amit Klein Reviewed-by: Eric Dumazet Signed-off-by: Willy Tarreau Signed-off-by: Jakub Kicinski Cc: Ben Hutchings Signed-off-by: Greg Kroah-Hartman --- net/ipv4/inet_hashtables.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/net/ipv4/inet_hashtables.c b/net/ipv4/inet_hashtables.c index ffee136ed8c7..d9bee15e36a5 100644 --- a/net/ipv4/inet_hashtables.c +++ b/net/ipv4/inet_hashtables.c @@ -727,7 +727,7 @@ int __inet_hash_connect(struct inet_timewait_death_row *death_row, net_get_random_once(table_perturb, INET_TABLE_PERTURB_SIZE * sizeof(*table_perturb)); - index = hash_32(port_offset, INET_TABLE_PERTURB_SHIFT); + index = port_offset & (INET_TABLE_PERTURB_SIZE - 1); offset = READ_ONCE(table_perturb[index]) + (port_offset >> 32); offset %= remaining; From 7b9c3bfbad2510478ac29221d2c40a09184ce6a1 Mon Sep 17 00:00:00 2001 From: Will Deacon Date: Fri, 10 Jun 2022 16:12:27 +0100 Subject: [PATCH 10/12] arm64: mm: Don't invalidate FROM_DEVICE buffers at start of DMA transfer commit c50f11c6196f45c92ca48b16a5071615d4ae0572 upstream. Invalidating the buffer memory in arch_sync_dma_for_device() for FROM_DEVICE transfers When using the streaming DMA API to map a buffer prior to inbound non-coherent DMA (i.e. DMA_FROM_DEVICE), we invalidate any dirty CPU cachelines so that they will not be written back during the transfer and corrupt the buffer contents written by the DMA. This, however, poses two potential problems: (1) If the DMA transfer does not write to every byte in the buffer, then the unwritten bytes will contain stale data once the transfer has completed. (2) If the buffer has a virtual alias in userspace, then stale data may be visible via this alias during the period between performing the cache invalidation and the DMA writes landing in memory. Address both of these issues by cleaning (aka writing-back) the dirty lines in arch_sync_dma_for_device(DMA_FROM_DEVICE) instead of discarding them using invalidation. Cc: Ard Biesheuvel Cc: Christoph Hellwig Cc: Robin Murphy Cc: Russell King Cc: Link: https://lore.kernel.org/r/20220606152150.GA31568@willie-the-truck Signed-off-by: Will Deacon Reviewed-by: Ard Biesheuvel Link: https://lore.kernel.org/r/20220610151228.4562-2-will@kernel.org Signed-off-by: Catalin Marinas Signed-off-by: Greg Kroah-Hartman --- arch/arm64/mm/cache.S | 2 -- 1 file changed, 2 deletions(-) diff --git a/arch/arm64/mm/cache.S b/arch/arm64/mm/cache.S index db767b072601..7b054c67acd8 100644 --- a/arch/arm64/mm/cache.S +++ b/arch/arm64/mm/cache.S @@ -228,8 +228,6 @@ ENDPIPROC(__dma_flush_area) * - dir - DMA direction */ ENTRY(__dma_map_area) - cmp w2, #DMA_FROM_DEVICE - b.eq __dma_inv_area b __dma_clean_area ENDPIPROC(__dma_map_area) From 3994d2ee55e28a08a4af4ef2ad5ced8d194d8b78 Mon Sep 17 00:00:00 2001 From: Greg Kroah-Hartman Date: Thu, 23 Jun 2022 18:19:35 +0200 Subject: [PATCH 11/12] Revert "hwmon: Make chip parameter for with_info API mandatory" This reverts commit 1ec0bc72f5dab3ab367ae5230cf6f212d805a225 which is commit ddaefa209c4ac791c1262e97c9b2d0440c8ef1d5 upstream. It should not have been applied to the stable trees. Link: https://lore.kernel.org/r/20220622154454.GA1864037@roeck-us.net Reported-by: Julian Haller Reported-by: Guenter Roeck Cc: Sasha Levin Signed-off-by: Greg Kroah-Hartman --- Documentation/hwmon/hwmon-kernel-api.rst | 2 +- drivers/hwmon/hwmon.c | 16 +++++++++------- 2 files changed, 10 insertions(+), 8 deletions(-) diff --git a/Documentation/hwmon/hwmon-kernel-api.rst b/Documentation/hwmon/hwmon-kernel-api.rst index 23f27fe78e37..c41eb6108103 100644 --- a/Documentation/hwmon/hwmon-kernel-api.rst +++ b/Documentation/hwmon/hwmon-kernel-api.rst @@ -72,7 +72,7 @@ hwmon_device_register_with_info is the most comprehensive and preferred means to register a hardware monitoring device. It creates the standard sysfs attributes in the hardware monitoring core, letting the driver focus on reading from and writing to the chip instead of having to bother with sysfs attributes. -The parent device parameter as well as the chip parameter must not be NULL. Its +The parent device parameter cannot be NULL with non-NULL chip info. Its parameters are described in more detail below. devm_hwmon_device_register_with_info is similar to diff --git a/drivers/hwmon/hwmon.c b/drivers/hwmon/hwmon.c index c73b93b9bb87..a2175394cd25 100644 --- a/drivers/hwmon/hwmon.c +++ b/drivers/hwmon/hwmon.c @@ -715,12 +715,11 @@ EXPORT_SYMBOL_GPL(hwmon_device_register_with_groups); /** * hwmon_device_register_with_info - register w/ hwmon - * @dev: the parent device (mandatory) - * @name: hwmon name attribute (mandatory) - * @drvdata: driver data to attach to created device (optional) - * @chip: pointer to hwmon chip information (mandatory) + * @dev: the parent device + * @name: hwmon name attribute + * @drvdata: driver data to attach to created device + * @chip: pointer to hwmon chip information * @extra_groups: pointer to list of additional non-standard attribute groups - * (optional) * * hwmon_device_unregister() must be called when the device is no * longer needed. @@ -733,10 +732,13 @@ hwmon_device_register_with_info(struct device *dev, const char *name, const struct hwmon_chip_info *chip, const struct attribute_group **extra_groups) { - if (!dev || !name || !chip) + if (!name) return ERR_PTR(-EINVAL); - if (!chip->ops || !chip->ops->is_visible || !chip->info) + if (chip && (!chip->ops || !chip->ops->is_visible || !chip->info)) + return ERR_PTR(-EINVAL); + + if (chip && !dev) return ERR_PTR(-EINVAL); return __hwmon_device_register(dev, name, drvdata, chip, extra_groups); From 23db944f754e99abf814a79a2273b0191d35e4ff Mon Sep 17 00:00:00 2001 From: Greg Kroah-Hartman Date: Sat, 25 Jun 2022 12:44:36 +0200 Subject: [PATCH 12/12] Linux 5.4.201 Link: https://lore.kernel.org/r/20220623164321.195163701@linuxfoundation.org Tested-by: Florian Fainelli Tested-by: Shuah Khan Tested-by: Hulk Robot Tested-by: Jon Hunter Tested-by: Sudip Mukherjee Tested-by: Guenter Roeck Signed-off-by: Greg Kroah-Hartman --- Makefile | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-) diff --git a/Makefile b/Makefile index 32da9117e9d7..75be5870cc55 100644 --- a/Makefile +++ b/Makefile @@ -1,7 +1,7 @@ # SPDX-License-Identifier: GPL-2.0 VERSION = 5 PATCHLEVEL = 4 -SUBLEVEL = 200 +SUBLEVEL = 201 EXTRAVERSION = NAME = Kleptomaniac Octopus