linux.git
2 years agonetfilter: nf_tables: unregister flowtable hooks on netns exit
Sasha Levin [Wed, 27 Sep 2023 15:30:06 +0000 (17:30 +0200)]
netfilter: nf_tables: unregister flowtable hooks on netns exit

commit 6069da443bf65f513bb507bb21e2f87cfb1ad0b6 upstream.

Unregister flowtable hooks before they are releases via
nf_tables_flowtable_destroy() otherwise hook core reports UAF.

BUG: KASAN: use-after-free in nf_hook_entries_grow+0x5a7/0x700 net/netfilter/core.c:142 net/netfilter/core.c:142
Read of size 4 at addr ffff8880736f7438 by task syz-executor579/3666

CPU: 0 PID: 3666 Comm: syz-executor579 Not tainted 5.16.0-rc5-syzkaller #0
Hardware name: Google Google Compute Engine/Google Compute Engine, BIOS Google 01/01/2011
Call Trace:
 <TASK>
 __dump_stack lib/dump_stack.c:88 [inline]
 __dump_stack lib/dump_stack.c:88 [inline] lib/dump_stack.c:106
 dump_stack_lvl+0x1dc/0x2d8 lib/dump_stack.c:106 lib/dump_stack.c:106
 print_address_description+0x65/0x380 mm/kasan/report.c:247 mm/kasan/report.c:247
 __kasan_report mm/kasan/report.c:433 [inline]
 __kasan_report mm/kasan/report.c:433 [inline] mm/kasan/report.c:450
 kasan_report+0x19a/0x1f0 mm/kasan/report.c:450 mm/kasan/report.c:450
 nf_hook_entries_grow+0x5a7/0x700 net/netfilter/core.c:142 net/netfilter/core.c:142
 __nf_register_net_hook+0x27e/0x8d0 net/netfilter/core.c:429 net/netfilter/core.c:429
 nf_register_net_hook+0xaa/0x180 net/netfilter/core.c:571 net/netfilter/core.c:571
 nft_register_flowtable_net_hooks+0x3c5/0x730 net/netfilter/nf_tables_api.c:7232 net/netfilter/nf_tables_api.c:7232
 nf_tables_newflowtable+0x2022/0x2cf0 net/netfilter/nf_tables_api.c:7430 net/netfilter/nf_tables_api.c:7430
 nfnetlink_rcv_batch net/netfilter/nfnetlink.c:513 [inline]
 nfnetlink_rcv_skb_batch net/netfilter/nfnetlink.c:634 [inline]
 nfnetlink_rcv_batch net/netfilter/nfnetlink.c:513 [inline] net/netfilter/nfnetlink.c:652
 nfnetlink_rcv_skb_batch net/netfilter/nfnetlink.c:634 [inline] net/netfilter/nfnetlink.c:652
 nfnetlink_rcv+0x10e6/0x2550 net/netfilter/nfnetlink.c:652 net/netfilter/nfnetlink.c:652

__nft_release_hook() calls nft_unregister_flowtable_net_hooks() which
only unregisters the hooks, then after RCU grace period, it is
guaranteed that no packets add new entries to the flowtable (no flow
offload rules and flowtable hooks are reachable from packet path), so it
is safe to call nf_flow_table_free() which cleans up the remaining
entries from the flowtable (both software and hardware) and it unbinds
the flow_block.

Fixes: ff4bf2f42a40 ("netfilter: nf_tables: add nft_unregister_flowtable_hook()")
Reported-by: syzbot+e918523f77e62790d6d9@syzkaller.appspotmail.com
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
Gbp-Pq: Topic bugfix/all
Gbp-Pq: Name netfilter-nf_tables-unregister-flowtable-hooks-on-ne.patch

2 years agonetfilter: ipset: Fix race between IPSET_CMD_CREATE and IPSET_CMD_SWAP
Sasha Levin [Tue, 19 Sep 2023 18:04:45 +0000 (20:04 +0200)]
netfilter: ipset: Fix race between IPSET_CMD_CREATE and IPSET_CMD_SWAP

[ Upstream commit 7433b6d2afd512d04398c73aa984d1e285be125b ]

Kyle Zeng reported that there is a race between IPSET_CMD_ADD and IPSET_CMD_SWAP
in netfilter/ip_set, which can lead to the invocation of `__ip_set_put` on a
wrong `set`, triggering the `BUG_ON(set->ref == 0);` check in it.

The race is caused by using the wrong reference counter, i.e. the ref counter instead
of ref_netlink.

Fixes: 24e227896bbf ("netfilter: ipset: Add schedule point in call_ad().")
Reported-by: Kyle Zeng <zengyhkyle@gmail.com>
Closes: https://lore.kernel.org/netfilter-devel/ZPZqetxOmH+w%2Fmyc@westworld/#r
Tested-by: Kyle Zeng <zengyhkyle@gmail.com>
Signed-off-by: Jozsef Kadlecsik <kadlec@netfilter.org>
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Sasha Levin <sashal@kernel.org>
Gbp-Pq: Topic bugfix/all
Gbp-Pq: Name netfilter-ipset-fix-race-between-ipset_cmd_create-an.patch

2 years agonetfilter: nf_tables: disallow element removal on anonymous sets
Sasha Levin [Sun, 10 Sep 2023 17:04:45 +0000 (19:04 +0200)]
netfilter: nf_tables: disallow element removal on anonymous sets

[ Upstream commit 23a3bfd4ba7acd36abf52b78605f61b21bdac216 ]

Anonymous sets need to be populated once at creation and then they are
bound to rule since 938154b93be8 ("netfilter: nf_tables: reject unbound
anonymous set before commit phase"), otherwise transaction reports
EINVAL.

Userspace does not need to delete elements of anonymous sets that are
not yet bound, reject this with EOPNOTSUPP.

From flush command path, skip anonymous sets, they are expected to be
bound already. Otherwise, EINVAL is hit at the end of this transaction
for unbound sets.

Fixes: 96518518cc41 ("netfilter: add nftables")
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
Gbp-Pq: Topic bugfix/all
Gbp-Pq: Name netfilter-nf_tables-disallow-element-removal-on-anon.patch

2 years agonetfilter: nf_tables: fix memleak when more than 255 elements expired
Sasha Levin [Fri, 22 Sep 2023 17:01:18 +0000 (19:01 +0200)]
netfilter: nf_tables: fix memleak when more than 255 elements expired

commit cf5000a7787cbc10341091d37245a42c119d26c5 upstream.

When more than 255 elements expired we're supposed to switch to a new gc
container structure.

This never happens: u8 type will wrap before reaching the boundary
and nft_trans_gc_space() always returns true.

This means we recycle the initial gc container structure and
lose track of the elements that came before.

While at it, don't deref 'gc' after we've passed it to call_rcu.

Fixes: 5f68718b34a5 ("netfilter: nf_tables: GC transaction API to avoid race with control plane")
Reported-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
Gbp-Pq: Topic bugfix/all
Gbp-Pq: Name netfilter-nf_tables-fix-memleak-when-more-than-255-e.patch

2 years agonetfilter: nft_set_hash: try later when GC hits EAGAIN on iteration
Sasha Levin [Fri, 22 Sep 2023 17:01:17 +0000 (19:01 +0200)]
netfilter: nft_set_hash: try later when GC hits EAGAIN on iteration

commit b079155faae94e9b3ab9337e82100a914ebb4e8d upstream.

Skip GC run if iterator rewinds to the beginning with EAGAIN, otherwise GC
might collect the same element more than once.

Fixes: f6c383b8c31a ("netfilter: nf_tables: adapt set backend to use GC transaction API")
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
Gbp-Pq: Topic bugfix/all
Gbp-Pq: Name netfilter-nft_set_hash-try-later-when-gc-hits-eagain.patch

2 years agonetfilter: nft_set_pipapo: stop GC iteration if GC transaction allocation fails
Sasha Levin [Fri, 22 Sep 2023 17:01:16 +0000 (19:01 +0200)]
netfilter: nft_set_pipapo: stop GC iteration if GC transaction allocation fails

commit 6d365eabce3c018a80f6e0379b17df2abb17405e upstream.

nft_trans_gc_queue_sync() enqueues the GC transaction and it allocates a
new one. If this allocation fails, then stop this GC sync run and retry
later.

Fixes: 5f68718b34a5 ("netfilter: nf_tables: GC transaction API to avoid race with control plane")
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
Gbp-Pq: Topic bugfix/all
Gbp-Pq: Name netfilter-nft_set_pipapo-stop-gc-iteration-if-gc-tra.patch

2 years agonetfilter: nft_set_rbtree: use read spinlock to avoid datapath contention
Sasha Levin [Fri, 22 Sep 2023 17:01:15 +0000 (19:01 +0200)]
netfilter: nft_set_rbtree: use read spinlock to avoid datapath contention

commit 96b33300fba880ec0eafcf3d82486f3463b4b6da upstream.

rbtree GC does not modify the datastructure, instead it collects expired
elements and it enqueues a GC transaction. Use a read spinlock instead
to avoid data contention while GC worker is running.

Fixes: f6c383b8c31a ("netfilter: nf_tables: adapt set backend to use GC transaction API")
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
Gbp-Pq: Topic bugfix/all
Gbp-Pq: Name netfilter-nft_set_rbtree-use-read-spinlock-to-avoid-.patch

2 years agonetfilter: nft_set_rbtree: skip sync GC for new elements in this transaction
Sasha Levin [Fri, 22 Sep 2023 17:01:14 +0000 (19:01 +0200)]
netfilter: nft_set_rbtree: skip sync GC for new elements in this transaction

commit 2ee52ae94baabf7ee09cf2a8d854b990dac5d0e4 upstream.

New elements in this transaction might expired before such transaction
ends. Skip sync GC for such elements otherwise commit path might walk
over an already released object. Once transaction is finished, async GC
will collect such expired element.

Fixes: f6c383b8c31a ("netfilter: nf_tables: adapt set backend to use GC transaction API")
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
Gbp-Pq: Topic bugfix/all
Gbp-Pq: Name netfilter-nft_set_rbtree-skip-sync-gc-for-new-elemen.patch

2 years agonetfilter: nf_tables: defer gc run if previous batch is still pending
Sasha Levin [Fri, 22 Sep 2023 17:01:13 +0000 (19:01 +0200)]
netfilter: nf_tables: defer gc run if previous batch is still pending

commit 8e51830e29e12670b4c10df070a4ea4c9593e961 upstream.

Don't queue more gc work, else we may queue the same elements multiple
times.

If an element is flagged as dead, this can mean that either the previous
gc request was invalidated/discarded by a transaction or that the previous
request is still pending in the system work queue.

The latter will happen if the gc interval is set to a very low value,
e.g. 1ms, and system work queue is backlogged.

The sets refcount is 1 if no previous gc requeusts are queued, so add
a helper for this and skip gc run if old requests are pending.

Add a helper for this and skip the gc run in this case.

Fixes: f6c383b8c31a ("netfilter: nf_tables: adapt set backend to use GC transaction API")
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Sasha Levin <sashal@kernel.org>
Gbp-Pq: Topic bugfix/all
Gbp-Pq: Name netfilter-nf_tables-defer-gc-run-if-previous-batch-i.patch

2 years agonetfilter: nf_tables: use correct lock to protect gc_list
Sasha Levin [Fri, 22 Sep 2023 17:01:12 +0000 (19:01 +0200)]
netfilter: nf_tables: use correct lock to protect gc_list

commit 8357bc946a2abc2a10ca40e5a2105d2b4c57515e upstream.

Use nf_tables_gc_list_lock spinlock, not nf_tables_destroy_list_lock to
protect the gc_list.

Fixes: 5f68718b34a5 ("netfilter: nf_tables: GC transaction API to avoid race with control plane")
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
Gbp-Pq: Topic bugfix/all
Gbp-Pq: Name netfilter-nf_tables-use-correct-lock-to-protect-gc_l.patch

2 years agonetfilter: nf_tables: GC transaction race with abort path
Sasha Levin [Fri, 22 Sep 2023 17:01:11 +0000 (19:01 +0200)]
netfilter: nf_tables: GC transaction race with abort path

commit 720344340fb9be2765bbaab7b292ece0a4570eae upstream.

Abort path is missing a synchronization point with GC transactions. Add
GC sequence number hence any GC transaction losing race will be
discarded.

Fixes: 5f68718b34a5 ("netfilter: nf_tables: GC transaction API to avoid race with control plane")
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
Gbp-Pq: Topic bugfix/all
Gbp-Pq: Name netfilter-nf_tables-gc-transaction-race-with-abort-p.patch

2 years agonetfilter: nf_tables: GC transaction race with netns dismantle
Sasha Levin [Fri, 22 Sep 2023 17:01:10 +0000 (19:01 +0200)]
netfilter: nf_tables: GC transaction race with netns dismantle

commit 02c6c24402bf1c1e986899c14ba22a10b510916b upstream.

Use maybe_get_net() since GC workqueue might race with netns exit path.

Fixes: 5f68718b34a5 ("netfilter: nf_tables: GC transaction API to avoid race with control plane")
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Sasha Levin <sashal@kernel.org>
Gbp-Pq: Topic bugfix/all
Gbp-Pq: Name netfilter-nf_tables-gc-transaction-race-with-netns-d.patch

2 years agonetfilter: nf_tables: fix GC transaction races with netns and netlink event exit...
Sasha Levin [Fri, 22 Sep 2023 17:01:09 +0000 (19:01 +0200)]
netfilter: nf_tables: fix GC transaction races with netns and netlink event exit path

commit 6a33d8b73dfac0a41f3877894b38082bd0c9a5bc upstream.

Netlink event path is missing a synchronization point with GC
transactions. Add GC sequence number update to netns release path and
netlink event path, any GC transaction losing race will be discarded.

Fixes: 5f68718b34a5 ("netfilter: nf_tables: GC transaction API to avoid race with control plane")
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Sasha Levin <sashal@kernel.org>
Gbp-Pq: Topic bugfix/all
Gbp-Pq: Name netfilter-nf_tables-fix-gc-transaction-races-with-ne.patch

2 years agonetfilter: nf_tables: don't fail inserts if duplicate has expired
Sasha Levin [Fri, 22 Sep 2023 17:01:08 +0000 (19:01 +0200)]
netfilter: nf_tables: don't fail inserts if duplicate has expired

commit 7845914f45f066497ac75b30c50dbc735e84e884 upstream.

nftables selftests fail:
run-tests.sh testcases/sets/0044interval_overlap_0
Expected: 0-2 . 0-3, got:
W: [FAILED]     ./testcases/sets/0044interval_overlap_0: got 1

Insertion must ignore duplicate but expired entries.

Moreover, there is a strange asymmetry in nft_pipapo_activate:

It refetches the current element, whereas the other ->activate callbacks
(bitmap, hash, rhash, rbtree) use elem->priv.
Same for .remove: other set implementations take elem->priv,
nft_pipapo_remove fetches elem->priv, then does a relookup,
remove this.

I suspect this was the reason for the change that prompted the
removal of the expired check in pipapo_get() in the first place,
but skipping exired elements there makes no sense to me, this helper
is used for normal get requests, insertions (duplicate check)
and deactivate callback.

In first two cases expired elements must be skipped.

For ->deactivate(), this gets called for DELSETELEM, so it
seems to me that expired elements should be skipped as well, i.e.
delete request should fail with -ENOENT error.

Fixes: 24138933b97b ("netfilter: nf_tables: don't skip expired elements during walk")
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
Gbp-Pq: Topic bugfix/all
Gbp-Pq: Name netfilter-nf_tables-don-t-fail-inserts-if-duplicate-.patch

2 years agonetfilter: nf_tables: remove busy mark and gc batch API
Sasha Levin [Fri, 22 Sep 2023 17:01:07 +0000 (19:01 +0200)]
netfilter: nf_tables: remove busy mark and gc batch API

commit a2dd0233cbc4d8a0abb5f64487487ffc9265beb5 upstream.

Ditch it, it has been replace it by the GC transaction API and it has no
clients anymore.

Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
Gbp-Pq: Topic bugfix/all
Gbp-Pq: Name netfilter-nf_tables-remove-busy-mark-and-gc-batch-ap.patch

2 years agonetfilter: nft_set_hash: mark set element as dead when deleting from packet path
Sasha Levin [Fri, 22 Sep 2023 17:01:06 +0000 (19:01 +0200)]
netfilter: nft_set_hash: mark set element as dead when deleting from packet path

commit c92db3030492b8ad1d0faace7a93bbcf53850d0c upstream.

Set on the NFT_SET_ELEM_DEAD_BIT flag on this element, instead of
performing element removal which might race with an ongoing transaction.
Enable gc when dynamic flag is set on since dynset deletion requires
garbage collection after this patch.

Fixes: d0a8d877da97 ("netfilter: nft_dynset: support for element deletion")
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
Gbp-Pq: Topic bugfix/all
Gbp-Pq: Name netfilter-nft_set_hash-mark-set-element-as-dead-when.patch

2 years agonetfilter: nf_tables: adapt set backend to use GC transaction API
Sasha Levin [Fri, 22 Sep 2023 17:01:05 +0000 (19:01 +0200)]
netfilter: nf_tables: adapt set backend to use GC transaction API

commit f6c383b8c31a93752a52697f8430a71dcbc46adf upstream.

Use the GC transaction API to replace the old and buggy gc API and the
busy mark approach.

No set elements are removed from async garbage collection anymore,
instead the _DEAD bit is set on so the set element is not visible from
lookup path anymore. Async GC enqueues transaction work that might be
aborted and retried later.

rbtree and pipapo set backends does not set on the _DEAD bit from the
sync GC path since this runs in control plane path where mutex is held.
In this case, set elements are deactivated, removed and then released
via RCU callback, sync GC never fails.

Fixes: 3c4287f62044 ("nf_tables: Add set type for arbitrary concatenation of ranges")
Fixes: 8d8540c4f5e0 ("netfilter: nft_set_rbtree: add timeout support")
Fixes: 9d0982927e79 ("netfilter: nft_hash: add support for timeouts")
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
Gbp-Pq: Topic bugfix/all
Gbp-Pq: Name netfilter-nf_tables-adapt-set-backend-to-use-gc-tran.patch

2 years agonetfilter: nf_tables: GC transaction API to avoid race with control plane
Sasha Levin [Fri, 22 Sep 2023 17:01:04 +0000 (19:01 +0200)]
netfilter: nf_tables: GC transaction API to avoid race with control plane

commit 5f68718b34a531a556f2f50300ead2862278da26 upstream.

The set types rhashtable and rbtree use a GC worker to reclaim memory.
From system work queue, in periodic intervals, a scan of the table is
done.

The major caveat here is that the nft transaction mutex is not held.
This causes a race between control plane and GC when they attempt to
delete the same element.

We cannot grab the netlink mutex from the work queue, because the
control plane has to wait for the GC work queue in case the set is to be
removed, so we get following deadlock:

   cpu 1                                cpu2
     GC work                            transaction comes in , lock nft mutex
       `acquire nft mutex // BLOCKS
                                        transaction asks to remove the set
                                        set destruction calls cancel_work_sync()

cancel_work_sync will now block forever, because it is waiting for the
mutex the caller already owns.

This patch adds a new API that deals with garbage collection in two
steps:

1) Lockless GC of expired elements sets on the NFT_SET_ELEM_DEAD_BIT
   so they are not visible via lookup. Annotate current GC sequence in
   the GC transaction. Enqueue GC transaction work as soon as it is
   full. If ruleset is updated, then GC transaction is aborted and
   retried later.

2) GC work grabs the mutex. If GC sequence has changed then this GC
   transaction lost race with control plane, abort it as it contains
   stale references to objects and let GC try again later. If the
   ruleset is intact, then this GC transaction deactivates and removes
   the elements and it uses call_rcu() to destroy elements.

Note that no elements are removed from GC lockless path, the _DEAD bit
is set and pointers are collected. GC catchall does not remove the
elements anymore too. There is a new set->dead flag that is set on to
abort the GC transaction to deal with set->ops->destroy() path which
removes the remaining elements in the set from commit_release, where no
mutex is held.

To deal with GC when mutex is held, which allows safe deactivate and
removal, add sync GC API which releases the set element object via
call_rcu(). This is used by rbtree and pipapo backends which also
perform garbage collection from control plane path.

Since element removal from sets can happen from control plane and
element garbage collection/timeout, it is necessary to keep the set
structure alive until all elements have been deactivated and destroyed.

We cannot do a cancel_work_sync or flush_work in nft_set_destroy because
its called with the transaction mutex held, but the aforementioned async
work queue might be blocked on the very mutex that nft_set_destroy()
callchain is sitting on.

This gives us the choice of ABBA deadlock or UaF.

To avoid both, add set->refs refcount_t member. The GC API can then
increment the set refcount and release it once the elements have been
free'd.

Set backends are adapted to use the GC transaction API in a follow up
patch entitled:

  ("netfilter: nf_tables: use gc transaction API in set backends")

This is joint work with Florian Westphal.

Fixes: cfed7e1b1f8e ("netfilter: nf_tables: add set garbage collection helpers")
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
Gbp-Pq: Topic bugfix/all
Gbp-Pq: Name netfilter-nf_tables-gc-transaction-api-to-avoid-race.patch

2 years agonetfilter: nf_tables: don't skip expired elements during walk
Sasha Levin [Fri, 22 Sep 2023 17:01:03 +0000 (19:01 +0200)]
netfilter: nf_tables: don't skip expired elements during walk

commit 24138933b97b055d486e8064b4a1721702442a9b upstream.

There is an asymmetry between commit/abort and preparation phase if the
following conditions are met:

1. set is a verdict map ("1.2.3.4 : jump foo")
2. timeouts are enabled

In this case, following sequence is problematic:

1. element E in set S refers to chain C
2. userspace requests removal of set S
3. kernel does a set walk to decrement chain->use count for all elements
   from preparation phase
4. kernel does another set walk to remove elements from the commit phase
   (or another walk to do a chain->use increment for all elements from
    abort phase)

If E has already expired in 1), it will be ignored during list walk, so its use count
won't have been changed.

Then, when set is culled, ->destroy callback will zap the element via
nf_tables_set_elem_destroy(), but this function is only safe for
elements that have been deactivated earlier from the preparation phase:
lack of earlier deactivate removes the element but leaks the chain use
count, which results in a WARN splat when the chain gets removed later,
plus a leak of the nft_chain structure.

Update pipapo_get() not to skip expired elements, otherwise flush
command reports bogus ENOENT errors.

Fixes: 3c4287f62044 ("nf_tables: Add set type for arbitrary concatenation of ranges")
Fixes: 8d8540c4f5e0 ("netfilter: nft_set_rbtree: add timeout support")
Fixes: 9d0982927e79 ("netfilter: nft_hash: add support for timeouts")
Signed-off-by: Florian Westphal <fw@strlen.de>
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
Gbp-Pq: Topic bugfix/all
Gbp-Pq: Name netfilter-nf_tables-don-t-skip-expired-elements-duri.patch

2 years agonetfilter: nf_tables: integrate pipapo into commit protocol
Sasha Levin [Fri, 22 Sep 2023 17:01:02 +0000 (19:01 +0200)]
netfilter: nf_tables: integrate pipapo into commit protocol

commit 212ed75dc5fb9d1423b3942c8f872a868cda3466 upstream.

The pipapo set backend follows copy-on-update approach, maintaining one
clone of the existing datastructure that is being updated. The clone
and current datastructures are swapped via rcu from the commit step.

The existing integration with the commit protocol is flawed because
there is no operation to clean up the clone if the transaction is
aborted. Moreover, the datastructure swap happens on set element
activation.

This patch adds two new operations for sets: commit and abort, these new
operations are invoked from the commit and abort steps, after the
transactions have been digested, and it updates the pipapo set backend
to use it.

This patch adds a new ->pending_update field to sets to maintain a list
of sets that require this new commit and abort operations.

Fixes: 3c4287f62044 ("nf_tables: Add set type for arbitrary concatenation of ranges")
Signed-off-by: Pablo Neira Ayuso <pablo@netfilter.org>
Signed-off-by: Sasha Levin <sashal@kernel.org>
Gbp-Pq: Topic bugfix/all
Gbp-Pq: Name netfilter-nf_tables-integrate-pipapo-into-commit-pro.patch

2 years agoovl: fail on invalid uid/gid mapping at copy up
Miklos Szeredi [Tue, 24 Jan 2023 15:41:18 +0000 (16:41 +0100)]
ovl: fail on invalid uid/gid mapping at copy up

Origin: https://git.kernel.org/linus/4f11ada10d0ad3fd53e2bd67806351de63a4f9c3
Bug-Debian-Security: https://security-tracker.debian.org/tracker/CVE-2023-0386

If st_uid/st_gid doesn't have a mapping in the mounter's user_ns, then
copy-up should fail, just like it would fail if the mounter task was doing
the copy using "cp -a".

There's a corner case where the "cp -a" would succeed but copy up fail: if
there's a mapping of the invalid uid/gid (65534 by default) in the user
namespace.  This is because stat(2) will return this value if the mapping
doesn't exist in the current user_ns and "cp -a" will in turn be able to
create a file with this uid/gid.

This behavior would be inconsistent with POSIX ACL's, which return -1 for
invalid uid/gid which result in a failed copy.

For consistency and simplicity fail the copy of the st_uid/st_gid are
invalid.

Fixes: 459c7c565ac3 ("ovl: unprivieged mounts")
Cc: <stable@vger.kernel.org> # v5.11
Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
Reviewed-by: Christian Brauner <brauner@kernel.org>
Reviewed-by: Seth Forshee <sforshee@kernel.org>
Gbp-Pq: Topic bugfix/all
Gbp-Pq: Name ovl-fail-on-invalid-uid-gid-mapping-at-copy-up.patch

2 years agovfs: move cap_convert_nscap() call into vfs_setxattr()
Miklos Szeredi [Mon, 14 Dec 2020 14:26:13 +0000 (15:26 +0100)]
vfs: move cap_convert_nscap() call into vfs_setxattr()

Origin: https://git.kernel.org/linus/7c03e2cda4a584cadc398e8f6641ca9988a39d52
Bug-Debian-Security: https://security-tracker.debian.org/tracker/CVE-2021-3493

cap_convert_nscap() does permission checking as well as conversion of the
xattr value conditionally based on fs's user-ns.

This is needed by overlayfs and probably other layered fs (ecryptfs) and is
what vfs_foo() is supposed to do anyway.

Signed-off-by: Miklos Szeredi <mszeredi@redhat.com>
Acked-by: James Morris <jamorris@linux.microsoft.com>
Gbp-Pq: Topic bugfix/all
Gbp-Pq: Name vfs-move-cap_convert_nscap-call-into-vfs_setxattr.patch

2 years agontfs: mark it as broken
Ben Hutchings [Thu, 25 Apr 2019 14:31:33 +0000 (15:31 +0100)]
ntfs: mark it as broken

NTFS has unfixed issues CVE-2018-12929, CVE-2018-12930, and
CVE-2018-12931.  ntfs-3g is a better supported alternative.

Make sure it can't be enabled even in custom kernels.

Gbp-Pq: Topic debian
Gbp-Pq: Name ntfs-mark-it-as-broken.patch

2 years ago[i386/686-pae] PCI: Set pci=nobios by default
Ben Hutchings [Tue, 16 Feb 2016 02:45:42 +0000 (02:45 +0000)]
[i386/686-pae] PCI: Set pci=nobios by default

Forwarded: not-needed

CONFIG_PCI_GOBIOS results in physical addresses 640KB-1MB being mapped
W+X, which is undesirable for security reasons and will result in a
warning at boot now that we enable CONFIG_DEBUG_WX.

This can be overridden using the kernel parameter "pci=nobios", but we
want to disable W+X by default.  Disable PCI BIOS probing by default;
it can still be enabled using "pci=bios".

Gbp-Pq: Topic debian
Gbp-Pq: Name i386-686-pae-pci-set-pci-nobios-by-default.patch

2 years ago[PATCH] KEYS: Make use of platform keyring for module signature verify
Robert Holmes [Tue, 23 Apr 2019 07:39:29 +0000 (07:39 +0000)]
[PATCH] KEYS: Make use of platform keyring for module signature verify

Bug-Debian: https://bugs.debian.org/935945
Origin: https://src.fedoraproject.org/rpms/kernel/raw/master/f/KEYS-Make-use-of-platform-keyring-for-module-signature.patch

This patch completes commit 278311e417be ("kexec, KEYS: Make use of
platform keyring for signature verify") which, while adding the
platform keyring for bzImage verification, neglected to also add
this keyring for module verification.

As such, kernel modules signed with keys from the MokList variable
were not successfully verified.

Signed-off-by: Robert Holmes <robeholmes@gmail.com>
Signed-off-by: Jeremy Cline <jcline@redhat.com>
Gbp-Pq: Topic features/all/db-mok-keyring
Gbp-Pq: Name KEYS-Make-use-of-platform-keyring-for-module-signature.patch

2 years agoMODSIGN: Make shash allocation failure fatal
Ben Hutchings [Sun, 5 May 2019 12:45:06 +0000 (13:45 +0100)]
MODSIGN: Make shash allocation failure fatal

mod_is_hash_blacklisted() currently returns 0 (suceess) if
crypto_alloc_shash() fails.  This should instead be a fatal error,
so unwrap and pass up the error code.

Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
Gbp-Pq: Topic features/all/db-mok-keyring
Gbp-Pq: Name modsign-make-shash-allocation-failure-fatal.patch

2 years ago[PATCH 4/4] MODSIGN: check the attributes of db and mok
Lee, Chun-Yi [Tue, 13 Mar 2018 10:38:03 +0000 (18:38 +0800)]
[PATCH 4/4] MODSIGN: check the attributes of db and mok

Origin: https://lore.kernel.org/patchwork/patch/933176/

That's better for checking the attributes of db and mok variables
before loading certificates to kernel keyring.

For db and dbx, both of them are authenticated variables. Which
means that they can only be modified by manufacturer's key. So
the kernel should checks EFI_VARIABLE_TIME_BASED_AUTHENTICATED_WRITE_ACCESS
attribute before we trust it.

For mok-rt and mokx-rt, both of them are created by shim boot loader
to forward the mok/mokx content to runtime. They must be runtime-volatile
variables. So kernel should checks that the attributes map did not set
EFI_VARIABLE_NON_VOLATILE bit before we trust it.

Cc: David Howells <dhowells@redhat.com>
Cc: Josh Boyer <jwboyer@fedoraproject.org>
Cc: James Bottomley <James.Bottomley@HansenPartnership.com>
Signed-off-by: "Lee, Chun-Yi" <jlee@suse.com>
[Rebased by Luca Boccassi]
[bwh: Forward-ported to 5.5.9:
 - get_cert_list() takes a pointer to status and returns the cert list
 - Adjust filename, context]
[bwh: Forward-ported to 5.10: MokListRT and MokListXRT are now both
 loaded through a single code path.]

Gbp-Pq: Topic features/all/db-mok-keyring
Gbp-Pq: Name 0004-MODSIGN-check-the-attributes-of-db-and-mok.patch

2 years ago[PATCH 3/4] MODSIGN: checking the blacklisted hash before loading a kernel module
Lee, Chun-Yi [Tue, 13 Mar 2018 10:38:02 +0000 (18:38 +0800)]
[PATCH 3/4] MODSIGN: checking the blacklisted hash before loading a kernel module

Origin: https://lore.kernel.org/patchwork/patch/933175/

This patch adds the logic for checking the kernel module's hash
base on blacklist. The hash must be generated by sha256 and enrolled
to dbx/mokx.

For example:
sha256sum sample.ko
mokutil --mokx --import-hash $HASH_RESULT

Whether the signature on ko file is stripped or not, the hash can be
compared by kernel.

Cc: David Howells <dhowells@redhat.com>
Cc: Josh Boyer <jwboyer@fedoraproject.org>
Cc: James Bottomley <James.Bottomley@HansenPartnership.com>
Signed-off-by: "Lee, Chun-Yi" <jlee@suse.com>
[Rebased by Luca Boccassi]

Gbp-Pq: Topic features/all/db-mok-keyring
Gbp-Pq: Name 0003-MODSIGN-checking-the-blacklisted-hash-before-loading-a-kernel-module.patch

2 years ago[PATCH 1/5] MODSIGN: do not load mok when secure boot disabled
Lee, Chun-Yi [Tue, 13 Mar 2018 10:37:59 +0000 (18:37 +0800)]
[PATCH 1/5] MODSIGN: do not load mok when secure boot disabled

Origin: https://lore.kernel.org/patchwork/patch/933173/

The mok can not be trusted when the secure boot is disabled. Which
means that the kernel embedded certificate is the only trusted key.

Due to db/dbx are authenticated variables, they needs manufacturer's
KEK for update. So db/dbx are secure when secureboot disabled.

Cc: David Howells <dhowells@redhat.com>
Cc: Josh Boyer <jwboyer@fedoraproject.org>
Cc: James Bottomley <James.Bottomley@HansenPartnership.com>
Signed-off-by: "Lee, Chun-Yi" <jlee@suse.com>
[Rebased by Luca Boccassi]
[bwh: Forward-ported to 5.5.9:
 - get_cert_list() takes a pointer to status and returns the cert list
 - Adjust filename]
[Salvatore Bonaccorso: Forward-ported to 5.10: Refresh for changes in
38a1f03aa240 ("integrity: Move import of MokListRT certs to a separate
routine"). Refresh in context for change in ebd9c2ae369a ("integrity: Load mokx
variables into the blacklist keyring")]

Gbp-Pq: Topic features/all/db-mok-keyring
Gbp-Pq: Name 0001-MODSIGN-do-not-load-mok-when-secure-boot-disabled.patch

2 years agoarm64: add kernel config option to lock down when in Secure Boot mode
Linn Crosetto [Tue, 30 Aug 2016 17:54:38 +0000 (11:54 -0600)]
arm64: add kernel config option to lock down when in Secure Boot mode

Bug-Debian: https://bugs.debian.org/831827
Forwarded: no

Add a kernel configuration option to lock down the kernel, to restrict
userspace's ability to modify the running kernel when UEFI Secure Boot is
enabled. Based on the x86 patch by Matthew Garrett.

Determine the state of Secure Boot in the EFI stub and pass this to the
kernel using the FDT.

Signed-off-by: Linn Crosetto <linn@hpe.com>
[bwh: Forward-ported to 4.10: adjust context]
[Lukas Wunner: Forward-ported to 4.11: drop parts applied upstream]
[bwh: Forward-ported to 4.15 and lockdown patch set:
 - Pass result of efi_get_secureboot() in stub through to
   efi_set_secure_boot() in main kernel
 - Use lockdown API and naming]
[bwh: Forward-ported to 4.19.3: adjust context in update_fdt()]
[dannf: Moved init_lockdown() call after uefi_init(), fixing SB detection]
[bwh: Drop call to init_lockdown(), as efi_set_secure_boot() now calls this]
[bwh: Forward-ported to 5.6: efi_get_secureboot() no longer takes a
 sys_table parameter]
[bwh: Forward-ported to 5.7: EFI initialisation from FDT was rewritten, so:
 - Add Secure Boot mode to the parameter enumeration in fdtparams.c
 - Add a parameter to efi_get_fdt_params() to return the Secure Boot mode
 - Since Xen does not have a property name defined for Secure Boot mode,
   change efi_get_fdt_prop() to handle a missing property name by clearing
   the output variable]
[Salvatore Bonaccorso: Forward-ported to 5.10: f30f242fb131 ("efi: Rename
arm-init to efi-init common for all arch") renamed arm-init.c to efi-init.c]

Gbp-Pq: Topic features/all/lockdown
Gbp-Pq: Name arm64-add-kernel-config-option-to-lock-down-when.patch

2 years agomtd: phram,slram: Disable when the kernel is locked down
Ben Hutchings [Fri, 30 Aug 2019 14:54:24 +0000 (15:54 +0100)]
mtd: phram,slram: Disable when the kernel is locked down

Forwarded: https://lore.kernel.org/linux-security-module/20190830154720.eekfjt6c4jzvlbfz@decadent.org.uk/

These drivers allow mapping arbitrary memory ranges as MTD devices.
This should be disabled to preserve the kernel's integrity when it is
locked down.

* Add the HWPARAM flag to the module parameters
* When slram is built-in, it uses __setup() to read kernel parameters,
  so add an explicit check security_locked_down() check

Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
Cc: Matthew Garrett <mjg59@google.com>
Cc: David Howells <dhowells@redhat.com>
Cc: Joern Engel <joern@lazybastard.org>
Cc: linux-mtd@lists.infradead.org
Gbp-Pq: Topic features/all/lockdown
Gbp-Pq: Name mtd-disable-slram-and-phram-when-locked-down.patch

2 years agoefi: Lock down the kernel if booted in secure boot mode
Ben Hutchings [Tue, 10 Sep 2019 10:54:28 +0000 (11:54 +0100)]
efi: Lock down the kernel if booted in secure boot mode

Based on an earlier patch by David Howells, who wrote the following
description:

> UEFI Secure Boot provides a mechanism for ensuring that the firmware will
> only load signed bootloaders and kernels.  Certain use cases may also
> require that all kernel modules also be signed.  Add a configuration option
> that to lock down the kernel - which includes requiring validly signed
> modules - if the kernel is secure-booted.

Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
Gbp-Pq: Topic features/all/lockdown
Gbp-Pq: Name efi-lock-down-the-kernel-if-booted-in-secure-boot-mo.patch

2 years ago[28/30] efi: Add an EFI_SECURE_BOOT flag to indicate secure boot mode
David Howells [Mon, 18 Feb 2019 12:45:03 +0000 (12:45 +0000)]
[28/30] efi: Add an EFI_SECURE_BOOT flag to indicate secure boot mode

Origin: https://git.kernel.org/pub/scm/linux/kernel/git/dhowells/linux-fs.git/commit?id=a5d70c55c603233c192b375f72116a395909da28

UEFI machines can be booted in Secure Boot mode.  Add an EFI_SECURE_BOOT
flag that can be passed to efi_enabled() to find out whether secure boot is
enabled.

Move the switch-statement in x86's setup_arch() that inteprets the
secure_boot boot parameter to generic code and set the bit there.

Suggested-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
Signed-off-by: David Howells <dhowells@redhat.com>
Reviewed-by: Ard Biesheuvel <ard.biesheuvel@linaro.org>
cc: linux-efi@vger.kernel.org
[rperier: Forward-ported to 5.5:
 - Use pr_warn()
 - Adjust context]
[bwh: Forward-ported to 5.6: adjust context]
[bwh: Forward-ported to 5.7:
 - Use the next available bit in efi.flags
 - Adjust context]

Gbp-Pq: Topic features/all/lockdown
Gbp-Pq: Name efi-add-an-efi_secure_boot-flag-to-indicate-secure-b.patch

2 years agowireguard: Clear keys after suspend despite CONFIG_ANDROID=y
Ben Hutchings [Thu, 7 Jul 2022 16:58:43 +0000 (18:58 +0200)]
wireguard: Clear keys after suspend despite CONFIG_ANDROID=y

Forwarded: not-needed

WireGuard assumes that CONFIG_ANDROID implies Android power
management, i.e. user-space suspending the system automatically at
short intervals, and so does not clear keys after a suspend/resume
cycle.  Debian systems don't do that kind of power management but we
do set CONFIG_ANDROID on some architectures as a dependency of Binder.

In 5.20, CONFIG_PM_USERSPACE_AUTOSLEEP will be introduced to tell the
kernel that this kind of power management is in use, and
CONFIG_ANDROID will be removed.  For now, remove this one test that
does the wrong thing for us.

References: https://lwn.net/Articles/899743/

Gbp-Pq: Topic bugfix/all
Gbp-Pq: Name wireguard-ignore-config_android.patch

2 years agoPartially revert "net: socket: implement 64-bit timestamps"
Ben Hutchings [Tue, 20 Aug 2019 17:12:35 +0000 (18:12 +0100)]
Partially revert "net: socket: implement 64-bit timestamps"

The introduction of SIOCGSTAMP{,NS}_OLD and move of SICOGSTAMP{,NS} to
a different header has caused build failures for various user-space
programs including qemu and suricata.  It also causes a test failure
for glibc.

For now, remove the _OLD suffix on the old ioctl numbers and require
programs using 64-bit timestamps to explicitly use SIOCGSTAMP{,NS}_NEW.

References: https://lore.kernel.org/lkml/af0eb47a-5b98-1bd9-3e8d-652e7f28b01f@de.ibm.com/
References: https://bugs.debian.org/934316
References: https://ci.debian.net/data/autopkgtest/testing/amd64/g/glibc/2772289/log.gz
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
Gbp-Pq: Topic bugfix/all
Gbp-Pq: Name partially-revert-net-socket-implement-64-bit-timestamps.patch

2 years agoMakefile: Do not check for libelf when building OOT module
Ben Hutchings [Mon, 7 Sep 2020 02:38:04 +0000 (03:38 +0100)]
Makefile: Do not check for libelf when building OOT module

When building out-of-tree modules, the necessary tools should have
already been built.  We therefore do not need libelf-dev to be
installed.

This effectively reverts commit 9f0c18aec620 "objtool: Fix
CONFIG_STACK_VALIDATION=y warning for out-of-tree modules", and
similarly moves the check introduced by commit 33a57ce0a54d "bpf:
Compile resolve_btfids tool at kernel compilation start".

Gbp-Pq: Topic debian
Gbp-Pq: Name makefile-do-not-check-for-libelf-when-building-oot-module.patch

2 years agofs: Add MODULE_SOFTDEP declarations for hard-coded crypto drivers
Ben Hutchings [Wed, 13 Apr 2016 20:48:06 +0000 (21:48 +0100)]
fs: Add MODULE_SOFTDEP declarations for hard-coded crypto drivers

Bug-Debian: https://bugs.debian.org/819725
Forwarded: http://mid.gmane.org/20160517133631.GF7555@decadent.org.uk

This helps initramfs builders and other tools to find the full
dependencies of a module.

Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
[Lukas Wunner: Forward-ported to 4.11: drop parts applied upstream]

Gbp-Pq: Topic bugfix/all
Gbp-Pq: Name fs-add-module_softdep-declarations-for-hard-coded-cr.patch

2 years agophy/marvell: disable 4-port phys
Ian Campbell [Wed, 20 Nov 2013 08:30:14 +0000 (08:30 +0000)]
phy/marvell: disable 4-port phys

Bug-Debian: https://bugs.debian.org/723177
Forwarded: http://thread.gmane.org/gmane.linux.debian.devel.bugs.general/1107774/

The Marvell PHY was originally disabled because it can cause networking
failures on some systems. According to Lennert Buytenhek this is because some
of the variants added did not share the same register layout. Since the known
cases are all 4-ports disable those variants (indicated by a 4 in the
penultimate position of the model name) until they can be audited for
correctness.

[bwh: Also #if-out the init functions for these PHYs to avoid
 compiler warnings]

Gbp-Pq: Topic bugfix/all
Gbp-Pq: Name disable-some-marvell-phys.patch

2 years agox86: Make x32 syscall support conditional on a kernel parameter
Ben Hutchings [Mon, 12 Feb 2018 23:59:26 +0000 (23:59 +0000)]
x86: Make x32 syscall support conditional on a kernel parameter

Bug-Debian: https://bugs.debian.org/708070
Forwarded: https://lore.kernel.org/lkml/1415245982.3398.53.camel@decadent.org.uk/T/#u

Enabling x32 in the standard amd64 kernel would increase its attack
surface while provide no benefit to the vast majority of its users.
No-one seems interested in regularly checking for vulnerabilities
specific to x32 (at least no-one with a white hat).

Still, adding another flavour just to turn on x32 seems wasteful.  And
the only differences on syscall entry are a few instructions that mask
out the x32 flag and compare the syscall number.

Use a static key to control whether x32 syscalls are really enabled, a
Kconfig parameter to set its default value and a kernel parameter
"syscall.x32" to change it at boot time.

Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
Gbp-Pq: Topic features/x86
Gbp-Pq: Name x86-make-x32-syscall-support-conditional.patch

2 years agox86: memtest: WARN if bad RAM found
Ben Hutchings [Mon, 5 Dec 2011 04:00:58 +0000 (04:00 +0000)]
x86: memtest: WARN if bad RAM found

Bug-Debian: https://bugs.debian.org/613321
Forwarded: http://thread.gmane.org/gmane.linux.kernel/1286471

Since this is not a particularly thorough test, if we find any bad
bits of RAM then there is a fair chance that there are other bad bits
we fail to detect.

Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
Gbp-Pq: Topic features/x86
Gbp-Pq: Name x86-memtest-WARN-if-bad-RAM-found.patch

2 years agoarm64: compat: Implement misalignment fixups for multiword loads
Ard Biesheuvel [Fri, 1 Jul 2022 13:53:22 +0000 (15:53 +0200)]
arm64: compat: Implement misalignment fixups for multiword loads

Origin: https://git.kernel.org/linus/3fc24ef32d3b9368f4c103dcd21d6a3f959b4870

The 32-bit ARM kernel implements fixups on behalf of user space when
using LDM/STM or LDRD/STRD instructions on addresses that are not 32-bit
aligned. This is not something that is supported by the architecture,
but was done anyway to increase compatibility with user space software,
which mostly targeted x86 at the time and did not care about aligned
accesses.

This feature is one of the remaining impediments to being able to switch
to 64-bit kernels on 64-bit capable hardware running 32-bit user space,
so let's implement it for the arm64 compat layer as well.

Note that the intent is to implement the exact same handling of
misaligned multi-word loads and stores as the 32-bit kernel does,
including what appears to be missing support for user space programs
that rely on SETEND to switch to a different byte order and back. Also,
like the 32-bit ARM version, we rely on the faulting address reported by
the CPU to infer the memory address, instead of decoding the instruction
fully to obtain this information.

This implementation is taken from the 32-bit ARM tree, with all pieces
removed that deal with instructions other than LDRD/STRD and LDM/STM, or
that deal with alignment exceptions taken in kernel mode.

Cc: debian-arm@lists.debian.org
Cc: Vagrant Cascadian <vagrant@debian.org>
Cc: Riku Voipio <riku.voipio@iki.fi>
Cc: Steve McIntyre <steve@einval.com>
Signed-off-by: Ard Biesheuvel <ardb@kernel.org>
Reviewed-by: Arnd Bergmann <arnd@arndb.de>
Link: https://lore.kernel.org/r/20220701135322.3025321-1-ardb@kernel.org
[catalin.marinas@arm.com: change the option to 'default n']
Signed-off-by: Catalin Marinas <catalin.marinas@arm.com>
Gbp-Pq: Topic features/arm64
Gbp-Pq: Name arm64-compat-Implement-misalignment-fixups-for-multi.patch

2 years agoarm64: dts: Add support for Raspberry Pi Compute Module 4 IO Board
Cyril Brulebois [Mon, 3 Jan 2022 20:59:36 +0000 (21:59 +0100)]
arm64: dts: Add support for Raspberry Pi Compute Module 4 IO Board

It was introduced in mainline during the v5.16 release cycle. Since
many broadcom includes were reworked since v5.10, adding support would
involve more than cherry-picking a DTS addition that uses a few
includes.

To avoid side effects on other models, introduce a DTS that leverages
some existing includes (bcm2711.dtsi and bcm283x-rpi-usb-host.dtsi)
and describes the rest without re-using parts of the Raspberry Pi 4 B
model.

To avoid phandle rotation (0x16, 0x17, and 0x18) across 3 nodes
(dma@7e007000, i2c@7e205000, and interrupt-controller@7ef00100), and the
related changes in other nodes referencing them, hardcode 0x16 as the
phandle for interrupt-controller@7ef00100. This leads to an empty dtdiff
between this new DTB and the one produced by a v5.16-rc8 build.

Authored-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
Reviewed-by: Cyril Brulebois <cyril@debamax.com>
Gbp-Pq: Topic features/arm64
Gbp-Pq: Name arm64-dts-raspberry-Add-support-for-the-CM4.patch

2 years agoarm64: dts: rockchip: Add support for PCIe on helios64
Uwe Kleine-König [Mon, 10 May 2021 09:09:32 +0000 (11:09 +0200)]
arm64: dts: rockchip: Add support for PCIe on helios64

Origin: https://git.kernel.org/pub/scm/linux/kernel/git/mmind/linux-rockchip.git/patch/?id=5a65adfa2ad1542f856fc7de3999d51f3a35d2e2

This is enough to make the SATA controller visible:

# lspci
00:00.0 PCI bridge: Fuzhou Rockchip Electronics Co., Ltd RK3399 PCI Express Root Port
01:00.0 SATA controller: JMicron Technology Corp. JMB58x AHCI SATA controller

Signed-off-by: Uwe Kleine-König <uwe@kleine-koenig.org>
Link: https://lore.kernel.org/r/20210510090932.970447-1-uwe@kleine-koenig.org
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Gbp-Pq: Topic features/arm64
Gbp-Pq: Name arm64-dts-rockchip-Add-support-for-PCIe-on-helios64.patch

2 years agoarm64: dts: rockchip: Add support for two PWM fans on helios64
Uwe Kleine-König [Mon, 10 May 2021 09:06:07 +0000 (11:06 +0200)]
arm64: dts: rockchip: Add support for two PWM fans on helios64

Origin: https://git.kernel.org/pub/scm/linux/kernel/git/mmind/linux-rockchip.git/patch/?id=271b66414df0b172c936b3cfd1894b7939f84165

On the helios64 board the two connectors P6 and P7 are supposed to
power two fans. Add the corresponding pwm-fan devices.

Signed-off-by: Uwe Kleine-König <uwe@kleine-koenig.org>
Link: https://lore.kernel.org/r/20210510090607.970145-1-uwe@kleine-koenig.org
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Gbp-Pq: Topic features/arm64
Gbp-Pq: Name arm64-dts-rockchip-Add-support-for-two-PWM-fans-on-h.patch

2 years agoarm64: dts: rockchip: kobol-helios64: Add mmc aliases
Uwe Kleine-König [Mon, 29 Mar 2021 08:45:58 +0000 (09:45 +0100)]
arm64: dts: rockchip: kobol-helios64: Add mmc aliases

This patch is part of commit 5dcbe7e3862d ("arm64: dts: rockchip: move mmc
aliases to board dts on rk3399") upstream. It is applied here only for Kobol's
helios64 to simplify conflict resolution for some further patches. It currently
is a noop as the same aliases already exist in rk3399.dtsi.

Link: https://lore.kernel.org/r/20210324122235.1059292-7-heiko@sntech.de
Gbp-Pq: Topic features/arm64
Gbp-Pq: Name arm64-dts-rockchip-kobol-helios64-Add-mmc-aliases.patch

2 years agoarm64: dts: rockchip: Rely on SoC external pull up on pmic-int-l on Helios64
Uwe Kleine-König [Sun, 24 Jan 2021 21:03:28 +0000 (22:03 +0100)]
arm64: dts: rockchip: Rely on SoC external pull up on pmic-int-l on Helios64

Origin: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/patch/?id=1e58ba111421375c5948c3e8145bdd84b06ac095

According to the schematic there is an external pull up, so there is no
need to enable the internal one additionally. Using no pull up matches
the vendor device tree.

Signed-off-by: Uwe Kleine-König <uwe@kleine-koenig.org>
Link: https://lore.kernel.org/r/20210124210328.611707-2-uwe@kleine-koenig.org
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Gbp-Pq: Topic features/arm64
Gbp-Pq: Name arm64-dts-rockchip-Rely-on-SoC-external-pull-up-on-p.patch

2 years agoarm64: dts: rockchip: Add basic support for Kobol's Helios64
Uwe Kleine-König [Wed, 14 Oct 2020 20:00:30 +0000 (22:00 +0200)]
arm64: dts: rockchip: Add basic support for Kobol's Helios64

Origin: https://git.kernel.org/pub/scm/linux/kernel/git/torvalds/linux.git/patch/?id=09e006cfb43e8ec38afe28278b210dab72e6cac8

The hardware is described in detail on Kobol's wiki at
https://wiki.kobol.io/helios64/intro/.

Up to now the following peripherals are working:

 - UART
 - Micro-SD card
 - eMMC
 - ethernet port 1
 - status LED
 - temperature sensor on i2c bus 2

Signed-off-by: Uwe Kleine-König <uwe@kleine-koenig.org>
Link: https://lore.kernel.org/r/20201014200030.845759-3-uwe@kleine-koenig.org
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Gbp-Pq: Topic features/arm64
Gbp-Pq: Name arm64-dts-rockchip-Add-basic-support-for-Kobol-s-Hel.patch

2 years agoplatform/x86: toshiba_haps: Fix missing newline in pr_debug call in toshiba_haps_notify
Hans de Goede [Wed, 19 May 2021 13:56:18 +0000 (15:56 +0200)]
platform/x86: toshiba_haps: Fix missing newline in pr_debug call in toshiba_haps_notify

Origin: https://git.kernel.org/linus/7dc4a18d017ca26abd1cea197e486fb3e5cd7632
Bug-Debian: https://bugs.debian.org/799193

The pr_debug() call in toshiba_haps_notify() is missing a newline at the
end of the string, add this.

BugLink: https://bugs.debian.org/799193
Reported-by: Salvatore Bonaccorso <carnil@debian.org>
Signed-off-by: Hans de Goede <hdegoede@redhat.com>
Link: https://lore.kernel.org/r/20210519135618.139701-1-hdegoede@redhat.com
Gbp-Pq: Topic bugfix/x86
Gbp-Pq: Name platform-x86-toshiba_haps-Fix-missing-newline-in-pr_.patch

2 years agox86-32: Disable 3D-Now in generic config
Ben Hutchings [Tue, 25 Sep 2018 18:44:13 +0000 (19:44 +0100)]
x86-32: Disable 3D-Now in generic config

We want the 686 flavour to run on Geode LX and similar AMD family 5
CPUs as well as family 6 and higher CPUs.  This used to work with
CONFIG_M686=y.  However commit 25d76ac88821 "x86/Kconfig: Explicitly
enumerate i686-class CPUs in Kconfig" in Linux 4.16 has made the
kernel require family 6 or higher.

It looks like a sensible choice would be to enable CONFIG_MGEODE_LX
and CONFIG_X86_GENERIC (for more generic optimisations), but this
currently enables CONFIG_X86_USE_3D_NOW which will cause the kernel to
crash on CPUs without the AMD-specific 3D-Now instructions.

Make CONFIG_X86_USE_3DNOW depend on CONFIG_X86_GENERIC being disabled.

Gbp-Pq: Topic bugfix/x86
Gbp-Pq: Name x86-32-disable-3dnow-in-generic-config.patch

2 years agoarm64/acpi: Add fixup for HPE m400 quirks
Geoff Levand [Wed, 13 Jun 2018 17:56:08 +0000 (10:56 -0700)]
arm64/acpi: Add fixup for HPE m400 quirks

Forwarded: https://patchwork.codeaurora.org/patch/547277/

Adds a new ACPI init routine acpi_fixup_m400_quirks that adds
a work-around for HPE ProLiant m400 APEI firmware problems.

The work-around disables APEI when CONFIG_ACPI_APEI is set and
m400 firmware is detected.  Without this fixup m400 systems
experience errors like these on startup:

  [Hardware Error]: Hardware error from APEI Generic Hardware Error Source: 2
  [Hardware Error]: event severity: fatal
  [Hardware Error]:  Error 0, type: fatal
  [Hardware Error]:   section_type: memory error
  [Hardware Error]:   error_status: 0x0000000000001300
  [Hardware Error]:   error_type: 10, invalid address
  Kernel panic - not syncing: Fatal hardware error!

Signed-off-by: Geoff Levand <geoff@infradead.org>
[bwh: Adjust context to apply to Linux 4.19]

Gbp-Pq: Topic bugfix/arm64
Gbp-Pq: Name arm64-acpi-Add-fixup-for-HPE-m400-quirks.patch

2 years agopowerpc/boot: Fix missing crc32poly.h when building with KERNEL_XZ
Krzysztof Kozlowski [Wed, 29 Aug 2018 07:32:23 +0000 (09:32 +0200)]
powerpc/boot: Fix missing crc32poly.h when building with KERNEL_XZ

Origin: https://patchwork.ozlabs.org/patch/963258/

After commit faa16bc404d7 ("lib: Use existing define with
polynomial") the lib/xz/xz_crc32.c includes a header from include/linux
directory thus any other user of this code should define proper include
path.

This fixes the build error on powerpc with CONFIG_KERNEL_XZ:

    In file included from ../arch/powerpc/boot/../../../lib/decompress_unxz.c:233:0,
                     from ../arch/powerpc/boot/decompress.c:42:
    ../arch/powerpc/boot/../../../lib/xz/xz_crc32.c:18:29: fatal error: linux/crc32poly.h: No such file or directory

Reported-by: Michal Kubecek <mkubecek@suse.cz>
Fixes: faa16bc404d7 ("lib: Use existing define with polynomial")
Signed-off-by: Krzysztof Kozlowski <krzk@kernel.org>
Reported-by: kbuild test robot <lkp@intel.com>
Reported-by: Meelis Roos <mroos@linux.ee>
Tested-by: Michal Kubecek <mkubecek@suse.cz>
Gbp-Pq: Topic bugfix/powerpc
Gbp-Pq: Name powerpc-boot-fix-missing-crc32poly.h-when-building-with-kernel_xz.patch

2 years agoARM: mm: Export __sync_icache_dcache() for xen-privcmd
Ben Hutchings [Wed, 11 Jul 2018 22:40:55 +0000 (23:40 +0100)]
ARM: mm: Export __sync_icache_dcache() for xen-privcmd

Forwarded: https://marc.info/?l=linux-arm-kernel&m=153134944429241

The xen-privcmd driver, which can be modular, calls set_pte_at()
which in turn may call __sync_icache_dcache().

The call to __sync_icache_dcache() may be optimised out because it is
conditional on !pte_special(), and xen-privcmd calls pte_mkspecial().
However, in a non-LPAE configuration there is no "special" bit and the
call is really unconditional.

Fixes: 3ad0876554ca ("xen/privcmd: add IOCTL_PRIVCMD_MMAP_RESOURCE")
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
Gbp-Pq: Topic bugfix/arm
Gbp-Pq: Name arm-mm-export-__sync_icache_dcache-for-xen-privcmd.patch

2 years agosh: Do not use hyphen in exported variable names
Ben Hutchings [Sat, 19 Aug 2017 20:42:09 +0000 (21:42 +0100)]
sh: Do not use hyphen in exported variable names

Forwarded: https://marc.info/?l=linux-sh&m=150317827322995&w=2

arch/sh/Makefile defines and exports ld-bfd to be used by
arch/sh/boot/Makefile and arch/sh/boot/compressed/Makefile.  Similarly
arch/sh/boot/Makefile defines and exports suffix-y to be used by
arch/sh/boot/compressed/Makefile.  However some shells, including
dash, will not pass through environment variables whose name includes
a hyphen.  Usually GNU make does not use a shell to recurse, but if
e.g. $(srctree) contains '~' it will use a shell here.

Rename these variables to ld_bfd and suffix_y.

References: https://buildd.debian.org/status/fetch.php?pkg=linux&arch=sh4&ver=4.13%7Erc5-1%7Eexp1&stamp=1502943967&raw=0
Fixes: ef9b542fce00 ("sh: bzip2/lzma uImage support.")
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
Gbp-Pq: Topic bugfix/sh
Gbp-Pq: Name sh-boot-do-not-use-hyphen-in-exported-variable-name.patch

2 years agoperf tools: Fix unwind build on i386
Ben Hutchings [Sat, 22 Jul 2017 16:37:33 +0000 (17:37 +0100)]
perf tools: Fix unwind build on i386

Forwarded: no

EINVAL may not be defined when building unwind-libunwind.c with
REMOTE_UNWIND_LIBUNWIND, resulting in a compiler error in
LIBUNWIND__ARCH_REG_ID().  Its only caller, access_reg(), only checks
for a negative return value and doesn't care what it is.  So change
-EINVAL to -1.

Fixes: 52ffe0ff02fc ("Support x86(32-bit) cross platform callchain unwind.")
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
Gbp-Pq: Topic bugfix/x86
Gbp-Pq: Name perf-tools-fix-unwind-build-on-i386.patch

2 years agoarm64: dts: rockchip: correct voltage selector on Firefly-RK3399
Heinrich Schuchardt [Mon, 4 Jun 2018 17:15:23 +0000 (19:15 +0200)]
arm64: dts: rockchip: correct voltage selector on Firefly-RK3399

Bug-Debian: https://bugs.debian.org/900799
Origin: https://git.kernel.org/pub/scm/linux/kernel/git/mmind/linux-rockchip.git/patch/?id=710e8c4a54be82ee8a97324e7b4330bf191e08bf

Without this patch the Firefly-RK3399 board boot process hangs after these
lines:

   fan53555-regulator 0-0040: FAN53555 Option[8] Rev[1] Detected!
   fan53555-reg: supplied by vcc_sys
   vcc1v8_s3: supplied by vcc_1v8

Blacklisting driver fan53555 allows booting.

The device tree uses a value of fcs,suspend-voltage-selector different to
any other board.

Changing this setting to the usual value is sufficient to enable booting
and also matches the value used in the vendor kernel.

Fixes: 171582e00db1 ("arm64: dts: rockchip: add support for firefly-rk3399 board")
Cc: stable@vger.kernel.org
Signed-off-by: Heinrich Schuchardt <xypron.glpk@gmx.de>
Signed-off-by: Heiko Stuebner <heiko@sntech.de>
Gbp-Pq: Topic bugfix/arm64
Gbp-Pq: Name dts-rockchip-correct-voltage-selector-firefly-RK3399.patch

2 years agoARM: dts: kirkwood: Fix SATA pinmux-ing for TS419
Ben Hutchings [Fri, 17 Feb 2017 01:30:30 +0000 (01:30 +0000)]
ARM: dts: kirkwood: Fix SATA pinmux-ing for TS419

Forwarded: https://www.spinics.net/lists/arm-kernel/msg563610.html
Bug-Debian: https://bugs.debian.org/855017

The old board code for the TS419 assigns MPP pins 15 and 16 as SATA
activity signals (and none as SATA presence signals).  Currently the
device tree assigns the SoC's default pinmux groups for SATA, which
conflict with the second Ethernet port.

Reported-by: gmbh@gazeta.pl
Tested-by: gmbh@gazeta.pl
References: https://bugs.debian.org/855017
Cc: stable@vger.kernel.org # 3.15+
Fixes: 934b524b3f49 ("ARM: Kirkwood: Add DT description of QNAP 419")
Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
Gbp-Pq: Topic bugfix/arm
Gbp-Pq: Name arm-dts-kirkwood-fix-sata-pinmux-ing-for-ts419.patch

2 years agobtrfs: warn about RAID5/6 being experimental at mount time
Adam Borowski [Tue, 28 Mar 2017 14:55:05 +0000 (16:55 +0200)]
btrfs: warn about RAID5/6 being experimental at mount time

Bug-Debian: https://bugs.debian.org/863290
Origin: https://bugs.debian.org/863290#5

Too many people come complaining about losing their data -- and indeed,
there's no warning outside a wiki and the mailing list tribal knowledge.
Message severity chosen for consistency with XFS -- "alert" makes dmesg
produce nice red background which should get the point across.

Signed-off-by: Adam Borowski <kilobyte@angband.pl>
[bwh: Also add_taint() so this is flagged in bug reports]

Gbp-Pq: Topic debian
Gbp-Pq: Name btrfs-warn-about-raid5-6-being-experimental-at-mount.patch

2 years agofanotify: Taint on use of FANOTIFY_ACCESS_PERMISSIONS
Ben Hutchings [Wed, 13 Jul 2016 00:37:22 +0000 (01:37 +0100)]
fanotify: Taint on use of FANOTIFY_ACCESS_PERMISSIONS

Forwarded: not-needed

Various free and proprietary AV products use this feature and users
apparently want it.  But punting access checks to userland seems like
an easy way to deadlock the system, and there will be nothing we can
do about that.  So warn and taint the kernel if this feature is
actually used.

Gbp-Pq: Topic debian
Gbp-Pq: Name fanotify-taint-on-use-of-fanotify_access_permissions.patch

2 years agofjes: Disable auto-loading
Ben Hutchings [Sat, 18 Mar 2017 20:47:58 +0000 (20:47 +0000)]
fjes: Disable auto-loading

Bug-Debian: https://bugs.debian.org/853976
Forwarded: no

fjes matches a generic ACPI device ID, and relies on its probe
function to distinguish whether that really corresponds to a supported
device.  Very few system will need the driver and it wastes memory on
all the other systems where the same device ID appears, so disable
auto-loading.

Gbp-Pq: Topic debian
Gbp-Pq: Name fjes-disable-autoload.patch

2 years agoviafb: Autoload on OLPC XO 1.5 only
Ben Hutchings [Sat, 20 Apr 2013 14:52:02 +0000 (15:52 +0100)]
viafb: Autoload on OLPC XO 1.5 only

Bug-Debian: https://bugs.debian.org/705788
Forwarded: no

It appears that viafb won't work automatically on all the boards for
which it has a PCI device ID match.  Currently, it is blacklisted by
udev along with most other framebuffer drivers, so this doesn't matter
much.

However, this driver is required for console support on the XO 1.5.
We need to allow it to be autoloaded on this model only, and then
un-blacklist it in udev.

Gbp-Pq: Topic bugfix/x86
Gbp-Pq: Name viafb-autoload-on-olpc-xo1.5-only.patch

2 years agosnd-pcsp: Disable autoload
Ben Hutchings [Wed, 5 Feb 2014 23:01:30 +0000 (23:01 +0000)]
snd-pcsp: Disable autoload

Forwarded: not-needed
Bug-Debian: https://bugs.debian.org/697709

There are two drivers claiming the platform:pcspkr device:
- pcspkr creates an input(!) device that can only beep
- snd-pcsp creates an equivalent input device plus a PCM device that can
  play barely recognisable renditions of sampled sound

snd-pcsp is blacklisted by the alsa-base package, but not everyone
installs that.  On PCs where no sound is wanted at all, both drivers
will still be loaded and one or other will complain that it couldn't
claim the relevant I/O range.

In case anyone finds snd-pcsp useful, we continue to build it.  But
remove the alias, to ensure it's not loaded where it's not wanted.

Gbp-Pq: Topic debian
Gbp-Pq: Name snd-pcsp-disable-autoload.patch

2 years agocdc_ncm,cdc_mbim: Use NCM by default
Ben Hutchings [Sun, 31 Mar 2013 02:58:04 +0000 (03:58 +0100)]
cdc_ncm,cdc_mbim: Use NCM by default

Forwarded: not-needed

Devices that support both NCM and MBIM modes should be kept in NCM
mode unless there is userland support for MBIM.

Set the default value of cdc_ncm.prefer_mbim to false and leave it to
userland (modem-manager) to override this with a modprobe.conf file
once it's ready to speak MBIM.

Gbp-Pq: Topic debian
Gbp-Pq: Name cdc_ncm-cdc_mbim-use-ncm-by-default.patch

2 years agointel-iommu: Add Kconfig option to exclude iGPU by default
Ben Hutchings [Tue, 20 Aug 2019 23:32:16 +0000 (00:32 +0100)]
intel-iommu: Add Kconfig option to exclude iGPU by default

Bug-Debian: https://bugs.debian.org/935270
Bug-Kali: https://bugs.kali.org/view.php?id=5644

There is still laptop firmware that touches the integrated GPU behind
the operating system's back, and doesn't say so in the RMRR table.
Enabling the IOMMU for all devices causes breakage.

Replace CONFIG_INTEL_IOMMU_DEFAULT_ON with a 3-way choice
corresponding to "on", "off", and "on,intgpu_off".

Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
Gbp-Pq: Topic features/x86
Gbp-Pq: Name intel-iommu-add-kconfig-option-to-exclude-igpu-by-default.patch

2 years agointel-iommu: Add option to exclude integrated GPU only
Ben Hutchings [Tue, 20 Aug 2019 23:05:30 +0000 (00:05 +0100)]
intel-iommu: Add option to exclude integrated GPU only

Bug-Debian: https://bugs.debian.org/935270
Bug-Kali: https://bugs.kali.org/view.php?id=5644

There is still laptop firmware that touches the integrated GPU behind
the operating system's back, and doesn't say so in the RMRR table.
Enabling the IOMMU for all devices causes breakage, but turning it off
for all graphics devices seems like a major weakness.

Add an option, intel_iommu=igpu_off, to exclude only integrated GPUs
from remapping.  This is a narrower exclusion than igfx_off: it only
affects Intel devices on the root bus.  Devices attached through an
external port (Thunderbolt or ExpressCard) won't be on the root bus.

Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
Gbp-Pq: Topic features/x86
Gbp-Pq: Name intel-iommu-add-option-to-exclude-integrated-gpu-only.patch

2 years agosecurity,perf: Allow further restriction of perf_event_open
Ben Hutchings [Mon, 11 Jan 2016 15:23:55 +0000 (15:23 +0000)]
security,perf: Allow further restriction of perf_event_open

Forwarded: https://lkml.org/lkml/2016/1/11/587

When kernel.perf_event_open is set to 3 (or greater), disallow all
access to performance events by users without CAP_SYS_ADMIN.
Add a Kconfig symbol CONFIG_SECURITY_PERF_EVENTS_RESTRICT that
makes this value the default.

This is based on a similar feature in grsecurity
(CONFIG_GRKERNSEC_PERF_HARDEN).  This version doesn't include making
the variable read-only.  It also allows enabling further restriction
at run-time regardless of whether the default is changed.

Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
Gbp-Pq: Topic features/all
Gbp-Pq: Name security-perf-allow-further-restriction-of-perf_event_open.patch

2 years agoadd sysctl to disallow unprivileged CLONE_NEWUSER by default
Serge Hallyn [Fri, 31 May 2013 18:12:12 +0000 (19:12 +0100)]
add sysctl to disallow unprivileged CLONE_NEWUSER by default

Origin: http://kernel.ubuntu.com/git?p=serge%2Fubuntu-saucy.git;a=commit;h=5c847404dcb2e3195ad0057877e1422ae90892b8

add sysctl to disallow unprivileged CLONE_NEWUSER by default

This is a short-term patch.  Unprivileged use of CLONE_NEWUSER
is certainly an intended feature of user namespaces.  However
for at least saucy we want to make sure that, if any security
issues are found, we have a fail-safe.

Signed-off-by: Serge Hallyn <serge.hallyn@ubuntu.com>
[bwh: Remove unneeded binary sysctl bits]
[bwh: Keep this sysctl, but change the default to enabled]

Gbp-Pq: Topic debian
Gbp-Pq: Name add-sysctl-to-disallow-unprivileged-CLONE_NEWUSER-by-default.patch

2 years agoyama: Disable by default
Ben Hutchings [Wed, 19 Jun 2013 03:35:28 +0000 (04:35 +0100)]
yama: Disable by default

Bug-Debian: https://bugs.debian.org/712740
Forwarded: not-needed

Gbp-Pq: Topic debian
Gbp-Pq: Name yama-disable-by-default.patch

2 years agosched: Do not enable autogrouping by default
Ben Hutchings [Wed, 16 Mar 2011 03:17:06 +0000 (03:17 +0000)]
sched: Do not enable autogrouping by default

Forwarded: not-needed

We want to provide the option of autogrouping but without enabling
it by default yet.

Gbp-Pq: Topic debian
Gbp-Pq: Name sched-autogroup-disabled.patch

2 years agofs: Enable link security restrictions by default
Ben Hutchings [Fri, 2 Nov 2012 05:32:06 +0000 (05:32 +0000)]
fs: Enable link security restrictions by default

Bug-Debian: https://bugs.debian.org/609455
Forwarded: not-needed

This reverts commit 561ec64ae67ef25cac8d72bb9c4bfc955edfd415
('VFS: don't do protected {sym,hard}links by default').

Gbp-Pq: Topic debian
Gbp-Pq: Name fs-enable-link-security-restrictions-by-default.patch

2 years agohamradio: Disable auto-loading as mitigation against local exploits
Ben Hutchings [Sun, 4 Aug 2019 23:29:11 +0000 (00:29 +0100)]
hamradio: Disable auto-loading as mitigation against local exploits

Forwarded: not-needed

We can mitigate the effect of vulnerabilities in obscure protocols by
preventing unprivileged users from loading the modules, so that they
are only exploitable on systems where the administrator has chosen to
load the protocol.

The 'ham' radio protocols (ax25, netrom, rose) are not actively
maintained or widely used.  Therefore disable auto-loading.

Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
Gbp-Pq: Topic debian
Gbp-Pq: Name hamradio-disable-auto-loading-as-mitigation-against-local-exploits.patch

2 years agodccp: Disable auto-loading as mitigation against local exploits
Ben Hutchings [Thu, 16 Feb 2017 19:09:17 +0000 (19:09 +0000)]
dccp: Disable auto-loading as mitigation against local exploits

Forwarded: not-needed

We can mitigate the effect of vulnerabilities in obscure protocols by
preventing unprivileged users from loading the modules, so that they
are only exploitable on systems where the administrator has chosen to
load the protocol.

The 'dccp' protocol is not actively maintained or widely used.
Therefore disable auto-loading.

Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
Gbp-Pq: Topic debian
Gbp-Pq: Name dccp-disable-auto-loading-as-mitigation-against-local-exploits.patch

2 years ago[PATCH 1/3] rds: Disable auto-loading as mitigation against local exploits
Ben Hutchings [Fri, 19 Nov 2010 02:12:48 +0000 (02:12 +0000)]
[PATCH 1/3] rds: Disable auto-loading as mitigation against local exploits

Forwarded: not-needed

Recent review has revealed several bugs in obscure protocol
implementations that can be exploited by local users for denial of
service or privilege escalation.  We can mitigate the effect of any
remaining vulnerabilities in such protocols by preventing unprivileged
users from loading the modules, so that they are only exploitable on
systems where the administrator has chosen to load the protocol.

The 'rds' protocol is one such protocol that has been found to be
vulnerable, and which was not present in the 'lenny' kernel.
Therefore disable auto-loading.

Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
Gbp-Pq: Topic debian
Gbp-Pq: Name rds-Disable-auto-loading-as-mitigation-against-local.patch

2 years ago[PATCH 2/3] af_802154: Disable auto-loading as mitigation against local exploits
Ben Hutchings [Fri, 19 Nov 2010 02:12:48 +0000 (02:12 +0000)]
[PATCH 2/3] af_802154: Disable auto-loading as mitigation against local exploits

Forwarded: not-needed

Recent review has revealed several bugs in obscure protocol
implementations that can be exploited by local users for denial of
service or privilege escalation.  We can mitigate the effect of any
remaining vulnerabilities in such protocols by preventing unprivileged
users from loading the modules, so that they are only exploitable on
systems where the administrator has chosen to load the protocol.

The 'af_802154' (IEEE 802.15.4) protocol is not widely used, was
not present in the 'lenny' kernel, and seems to receive only sporadic
maintenance.  Therefore disable auto-loading.

Signed-off-by: Ben Hutchings <ben@decadent.org.uk>
Gbp-Pq: Topic debian
Gbp-Pq: Name af_802154-Disable-auto-loading-as-mitigation-against.patch

2 years agofirmware_class: Refer to Debian wiki page when logging missing firmware
Ben Hutchings [Mon, 12 Mar 2018 01:14:03 +0000 (01:14 +0000)]
firmware_class: Refer to Debian wiki page when logging missing firmware

Bug-Debian: https://bugs.debian.org/888405
Forwarded: not-needed

If firmware loading fails due to a missing file, log a second error
message referring to our wiki page about firmware.  This will explain
why some firmware is in non-free, or can't be packaged at all.  Only
do this once per boot.

Do something similar in the radeon and amdgpu drivers, where we have
an early check to avoid failing at a point where we cannot display
anything.

Gbp-Pq: Topic debian
Gbp-Pq: Name firmware_class-refer-to-debian-wiki-firmware-page.patch

2 years agoradeon, amdgpu: Firmware is required for DRM and KMS on R600 onward
Ben Hutchings [Tue, 8 Jan 2013 03:25:52 +0000 (03:25 +0000)]
radeon, amdgpu: Firmware is required for DRM and KMS on R600 onward

Bug-Debian: https://bugs.debian.org/607194
Bug-Debian: https://bugs.debian.org/607471
Bug-Debian: https://bugs.debian.org/610851
Bug-Debian: https://bugs.debian.org/627497
Bug-Debian: https://bugs.debian.org/632212
Bug-Debian: https://bugs.debian.org/637943
Bug-Debian: https://bugs.debian.org/649448
Bug-Debian: https://bugs.debian.org/697229
Forwarded: no

radeon requires firmware/microcode for the GPU in all chips, but for
newer chips (apparently R600 'Evergreen' onward) it also expects
firmware for the memory controller and other sub-blocks.

radeon attempts to gracefully fall back and disable some features if
the firmware is not available, but becomes unstable - the framebuffer
and/or system memory may be corrupted, or the display may stay black.

Therefore, perform a basic check for the existence of
/lib/firmware/{radeon,amdgpu} when a device is probed, and abort if it
is missing, except for the pre-R600 case.

Gbp-Pq: Topic bugfix/all
Gbp-Pq: Name radeon-amdgpu-firmware-is-required-for-drm-and-kms-on-r600-onward.patch

2 years agofirmware: Remove redundant log messages from drivers
Ben Hutchings [Sun, 9 Dec 2012 16:40:31 +0000 (16:40 +0000)]
firmware: Remove redundant log messages from drivers

Forwarded: no

Now that firmware_class logs every success and failure consistently,
many other log messages can be removed from drivers.

This will probably need to be split up into multiple patches prior to
upstream submission.

Gbp-Pq: Topic bugfix/all
Gbp-Pq: Name firmware-remove-redundant-log-messages-from-drivers.patch

2 years agofirmware_class: Log every success and failure against given device
Ben Hutchings [Sun, 9 Dec 2012 16:02:00 +0000 (16:02 +0000)]
firmware_class: Log every success and failure against given device

Forwarded: no

The hundreds of users of request_firmware() have nearly as many
different log formats for reporting failures.  They also have only the
vaguest hint as to what went wrong; only firmware_class really knows
that.  Therefore, add specific log messages for the failure modes that
aren't currently logged.

In case of a driver that tries multiple names, this may result in the
impression that it failed to initialise.  Therefore, also log successes.

This makes many error messages in drivers redundant, which will be
removed in later patches.

This does not cover the case where we fall back to a user-mode helper
(which is no longer enabled in Debian).

NOTE: hw-detect will depend on the "firmware: failed to load %s (%d)\n"
format to detect missing firmware.

Gbp-Pq: Topic bugfix/all
Gbp-Pq: Name firmware_class-log-every-success-and-failure.patch

2 years agoiwlwifi: Do not request unreleased firmware for IWL6000
Ben Hutchings [Fri, 29 Sep 2023 04:25:15 +0000 (05:25 +0100)]
iwlwifi: Do not request unreleased firmware for IWL6000

Bug-Debian: https://bugs.debian.org/689416
Forwarded: not-needed

The iwlwifi driver currently supports firmware API versions 4-6 for
these devices.  It will request the file for the latest supported
version and then fall back to earlier versions.  However, the latest
version that has actually been released is 4, so we expect the
requests for versions 6 and then 5 to fail.

The installer appears to report any failed request, and it is probably
not easy to detect that this particular failure is harmless.  So stop
requesting the unreleased firmware.

Gbp-Pq: Topic debian
Gbp-Pq: Name iwlwifi-do-not-request-unreleased-firmware.patch

2 years agoaf9005: Use request_firmware() to load register init script
Ben Hutchings [Mon, 24 Aug 2009 22:19:58 +0000 (23:19 +0100)]
af9005: Use request_firmware() to load register init script

Forwarded: no

Read the register init script from the Windows driver.  This is sick
but should avoid the potential copyright infringement in distributing
a version of the script which is directly derived from the driver.

Gbp-Pq: Topic features/all
Gbp-Pq: Name drivers-media-dvb-usb-af9005-request_firmware.patch

2 years agokbuild: Abort build if SUBDIRS used
Ben Hutchings [Mon, 26 Apr 2021 16:27:16 +0000 (18:27 +0200)]
kbuild: Abort build if SUBDIRS used

Forwarded: not-needed
Bug-Debian: https://bugs.debian.org/987575

DKMS and module-assistant both build OOT modules as root.  If they
build an old OOT module that still use SUBDIRS this causes Kbuild
to try building a full kernel, which obviously fails but not before
deleting files from the installed headers package.

To avoid such mishaps, detect this situation and abort the build.

The error message is based on that used in commit 0126be38d988
"kbuild: announce removal of SUBDIRS if used".

Gbp-Pq: Topic debian
Gbp-Pq: Name kbuild-abort-build-if-subdirs-used.patch

2 years agokbuild: Look for module.lds under arch directory too
Ben Hutchings [Thu, 10 Dec 2020 16:31:39 +0000 (17:31 +0100)]
kbuild: Look for module.lds under arch directory too

Forwarded: not-needed
Bug-Debian: https://bugs.debian.org/975571

The module.lds linker script is now built under the scripts directory,
where previously it was under arch/$(SRCARCH).

However, we package the scripts directory as linux-kbuild, which is
meant to be able to do support native and cross-builds.  That means it
shouldn't contain files for a specific target architecture without a
wrapper to select between them, and it doesn't appear that linker
scripts are powerful enough to implement such a wrapper.

Building module.lds in a different location would require relatively
large changes.  Moving it in the package build rules can work, but we
need to support custom kernel builds from the same source so we can't
assume it's moved.

Therefore, we move module.lds under the arch build directory in
rules.real and change Makefile.modfinal to look for it in both places.

Gbp-Pq: Topic debian
Gbp-Pq: Name kbuild-look-for-module.lds-under-arch-directory-too.patch

2 years ago[PATCH 2/2] perf/traceevent: Support asciidoctor for documentation
Bastian Blank [Tue, 4 Aug 2020 09:44:37 +0000 (09:44 +0000)]
[PATCH 2/2] perf/traceevent: Support asciidoctor for documentation

From cd02fc78859ef9aefd7c92406f9523622da0b472 Mon Sep 17 00:00:00 2001
Forwarded: not-needed

Gbp-Pq: Topic debian
Gbp-Pq: Name perf-traceevent-support-asciidoctor-for-documentatio.patch

2 years ago[PATCH 1/2] Documentation: Drop sphinx version check
Bastian Blank [Tue, 4 Aug 2020 09:44:19 +0000 (09:44 +0000)]
[PATCH 1/2] Documentation: Drop sphinx version check

From 252aa79fdbd4ac2da09d9b98f81bf11f5e3e1870 Mon Sep 17 00:00:00 2001
Forwarded: not-needed

Gbp-Pq: Topic debian
Gbp-Pq: Name documentation-drop-sphinx-version-check.patch

2 years agoandroid: Enable building ashmem and binder as modules
Ben Hutchings [Fri, 22 Jun 2018 16:27:00 +0000 (17:27 +0100)]
android: Enable building ashmem and binder as modules

Bug-Debian: https://bugs.debian.org/901492

We want to enable use of the Android ashmem and binder drivers to
support Anbox, but they should not be built-in as that would waste
resources and increase security attack surface on systems that don't
need them.

- Add a MODULE_LICENSE declaration to ashmem
- Change the Makefiles to build each driver as an object with the
  "_linux" suffix (which is what Anbox expects)
- Change config symbol types to tristate

Gbp-Pq: Topic debian
Gbp-Pq: Name android-enable-building-ashmem-and-binder-as-modules.patch

2 years agoExport symbols needed by Android drivers
Ben Hutchings [Mon, 7 Sep 2020 01:51:53 +0000 (02:51 +0100)]
Export symbols needed by Android drivers

Bug-Debian: https://bugs.debian.org/901492

We want to enable use of the Android ashmem and binder drivers to
support Anbox, but they should not be built-in as that would waste
resources and increase security attack surface on systems that don't
need them.

Export the currently un-exported symbols they depend on.

Gbp-Pq: Topic debian
Gbp-Pq: Name export-symbols-needed-by-android-drivers.patch

2 years agowireless: Add Debian wireless-regdb certificates
Ben Hutchings [Fri, 13 Apr 2018 19:10:28 +0000 (20:10 +0100)]
wireless: Add Debian wireless-regdb certificates

Forwarded: not-needed

This hex dump is generated using:

{
    for cert in debian/certs/wireless-regdb-*.pem; do
        openssl x509 -in $cert -outform der;
    done
} | hexdump -v -e '1/1 "0x%.2x," "\n"' > net/wireless/certs/debian.hex

Gbp-Pq: Topic debian
Gbp-Pq: Name wireless-add-debian-wireless-regdb-certificates.patch

2 years agoInstall perf scripts non-executable
Bastian Blank [Fri, 7 Oct 2011 20:37:52 +0000 (21:37 +0100)]
Install perf scripts non-executable

Forwarded: no

[bwh: Forward-ported to 4.13]

Gbp-Pq: Topic debian
Gbp-Pq: Name tools-perf-install.patch

2 years agoCreate manpages and binaries including the version
Bastian Blank [Mon, 26 Sep 2011 12:53:12 +0000 (13:53 +0100)]
Create manpages and binaries including the version

Forwarded: no

[bwh: Fix version insertion in perf man page cross-references and perf
man page title.  Install bash_completion script for perf with a
version-dependent name.  And do the same for trace.]

Gbp-Pq: Topic debian
Gbp-Pq: Name tools-perf-version.patch

2 years ago[sh4] Fix uImage build
Nobuhiro Iwamatsu [Fri, 29 Sep 2023 04:25:15 +0000 (05:25 +0100)]
[sh4] Fix uImage build

Bug-Debian: https://bugs.debian.org/569034
Forwarded: not-needed

[bwh: This was added without a description, but I think it is done
 only to avoid a build-dependency on u-boot-tools.]

Gbp-Pq: Topic debian
Gbp-Pq: Name arch-sh4-fix-uimage-build.patch

2 years agoUse RELAXED ieee754 mode for Loongson-3 as 3A 4000 is 2008-only
YunQiang Su [Mon, 16 Nov 2020 01:11:00 +0000 (09:11 +0800)]
Use RELAXED ieee754 mode for Loongson-3 as 3A 4000 is 2008-only

Forwarded: not-needed

There are 2 mode of value of IEEE NaN hardcoded by CPU.
Currently, our mipsel/mips64el port is in so-called lagacy mode.
Loongson 3A 4000 is set as the so-called 2008 mode.

To make Debian workable on Loongson 3A 4000, we need set the kerenl in
RELAXED mode.

https://web.archive.org/web/20180830093617/https://dmz-portal.mips.com/wiki/MIPS_ABI_-_NaN_Interlinking

Gbp-Pq: Topic debian
Gbp-Pq: Name mips-ieee754-relaxed.patch

2 years agoDisable uImage generation for mips generic
YunQiang Su [Mon, 14 May 2018 08:16:18 +0000 (16:16 +0800)]
Disable uImage generation for mips generic

Forwarded: not-needed

MIPS generic trys to generate uImage when build, which then ask for
u-boot-tools.

Gbp-Pq: Topic debian
Gbp-Pq: Name mips-boston-disable-its.patch

2 years ago[PATCH] Partially revert "MIPS: Add -Werror to arch/mips/Kbuild"
Ben Hutchings [Mon, 13 Sep 2010 01:16:18 +0000 (02:16 +0100)]
[PATCH] Partially revert "MIPS: Add -Werror to arch/mips/Kbuild"

Forwarded: not-needed

This reverts commits 66f9ba101f54bda63ab1db97f9e9e94763d0651b and
5373633cc9253ba82547473e899cab141c54133e.

We really don't want to add -Werror anywhere.

Gbp-Pq: Topic debian
Gbp-Pq: Name mips-disable-werror.patch

2 years agoHardcode arch script output
dann frazier [Mon, 26 Mar 2007 22:30:51 +0000 (16:30 -0600)]
Hardcode arch script output

Bug-Debian: https://bugs.debian.org/392592
Forwarded: not-needed

Here's a patch that simply uses hardcoded definitions instead of
doing the dynamic tests that require architecture-specific scripts.

I don't particularly like this approach because it restricts
portability and diverts from upstream. But, it is simpler, and this
really needs to be fixed somehow before etch (along with a rebuild of
linux-modules-extra-2.6), so I'm willing to live with it if my other
patch is deemed unacceptable.

My primary concern is that, in the future, the output of these scripts
will change and we (or our successors) will either not notice or
forget to update the hardcoded values.

Including the scripts in linux-kbuild will avoid this manual step
altogether, and allow for the possibility of other archs to provide
their own scripts in the future.

Gbp-Pq: Topic debian
Gbp-Pq: Name ia64-hardcode-arch-script-output.patch

2 years agokbuild: Make the toolchain variables easily overwritable
Bastian Blank [Sun, 22 Feb 2009 14:39:35 +0000 (15:39 +0100)]
kbuild: Make the toolchain variables easily overwritable

Forwarded: not-needed

Allow make variables to be overridden for each flavour by a file in
the build tree, .kernelvariables.

We currently use this for ARCH, KERNELRELEASE, CC, and in some cases
also CROSS_COMPILE, KCFLAGS.

This file can only be read after we establish the build tree, and all
use of $(ARCH) needs to be moved after this.

[bwh: Updated for 5.3: include .kernelvariables from current directory
 rather than using undefined $(obj).]

Gbp-Pq: Topic debian
Gbp-Pq: Name kernelvariables.patch

2 years agoMake mkcompile_h accept an alternate timestamp string
Ben Hutchings [Tue, 12 May 2015 18:29:22 +0000 (19:29 +0100)]
Make mkcompile_h accept an alternate timestamp string

Forwarded: not-needed

We want to include the Debian version in the utsname::version string
instead of a full timestamp string.  However, we still need to provide
a standard timestamp string for gen_initramfs_list.sh to make the
kernel image reproducible.

Make mkcompile_h use $KBUILD_BUILD_VERSION_TIMESTAMP in preference to
$KBUILD_BUILD_TIMESTAMP.

Gbp-Pq: Topic debian
Gbp-Pq: Name uname-version-timestamp.patch

2 years agoInclude package version along with kernel release in stack traces
Ben Hutchings [Tue, 24 Jul 2012 02:13:10 +0000 (03:13 +0100)]
Include package version along with kernel release in stack traces

Forwarded: not-needed

For distribution binary packages we assume
$DISTRIBUTION_OFFICIAL_BUILD, $DISTRIBUTOR and $DISTRIBUTION_VERSION
are set.

Gbp-Pq: Topic debian
Gbp-Pq: Name version.patch

2 years agoDocumentation: Fix broken link to CIPSO draft
Ben Hutchings [Sat, 24 Aug 2019 18:00:41 +0000 (19:00 +0100)]
Documentation: Fix broken link to CIPSO draft

Forwarded: not-needed

We exclude the CIPSO draft text as its licence is not DFSG compliant.
Link to the IETF's online version instead.

Gbp-Pq: Topic debian/dfsg
Gbp-Pq: Name documentation-fix-broken-link-to-cipso-draft.patch

2 years agovideo: Remove nvidiafb and rivafb
Ben Hutchings [Sat, 2 Jun 2012 18:53:38 +0000 (19:53 +0100)]
video: Remove nvidiafb and rivafb

Bug-Debian: https://bugs.debian.org/383481
Forwarded: no

These drivers contain register programming code provided by the
hardware vendor that appears to have been deliberately obfuscated.
This is arguably not the preferred form for modification.

These drivers are also largely redundant with nouveau.  The RIVA 128
(NV3) is not supported by nouveau but is about 15 years old and
probably discontinued 10 years ago.

Gbp-Pq: Topic debian/dfsg
Gbp-Pq: Name video-remove-nvidiafb-and-rivafb.patch

2 years agoAdd removal patches for: 3c359, smctr, keyspan, cops
Frederik Schüler [Fri, 5 Jan 2007 15:55:24 +0000 (15:55 +0000)]
Add removal patches for: 3c359, smctr, keyspan, cops

Forwarded: not-needed

Gbp-Pq: Topic debian/dfsg
Gbp-Pq: Name drivers-net-appletalk-cops.patch

2 years agovs6624: mark as broken
Ben Hutchings [Sun, 27 May 2012 00:56:58 +0000 (01:56 +0100)]
vs6624: mark as broken

Forwarded: not-needed

Gbp-Pq: Topic debian/dfsg
Gbp-Pq: Name vs6624-disable.patch