summaryrefslogtreecommitdiffstats
path: root/fs/pnode.c
Commit message (Collapse)AuthorAgeFilesLines
* pnode: terminate at peers of sourceChristian Brauner2022-12-211-1/+1
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | The propagate_mnt() function handles mount propagation when creating mounts and propagates the source mount tree @source_mnt to all applicable nodes of the destination propagation mount tree headed by @dest_mnt. Unfortunately it contains a bug where it fails to terminate at peers of @source_mnt when looking up copies of the source mount that become masters for copies of the source mount tree mounted on top of slaves in the destination propagation tree causing a NULL dereference. Once the mechanics of the bug are understood it's easy to trigger. Because of unprivileged user namespaces it is available to unprivileged users. While fixing this bug we've gotten confused multiple times due to unclear terminology or missing concepts. So let's start this with some clarifications: * The terms "master" or "peer" denote a shared mount. A shared mount belongs to a peer group. * A peer group is a set of shared mounts that propagate to each other. They are identified by a peer group id. The peer group id is available in @shared_mnt->mnt_group_id. Shared mounts within the same peer group have the same peer group id. The peers in a peer group can be reached via @shared_mnt->mnt_share. * The terms "slave mount" or "dependent mount" denote a mount that receives propagation from a peer in a peer group. IOW, shared mounts may have slave mounts and slave mounts have shared mounts as their master. Slave mounts of a given peer in a peer group are listed on that peers slave list available at @shared_mnt->mnt_slave_list. * The term "master mount" denotes a mount in a peer group. IOW, it denotes a shared mount or a peer mount in a peer group. The term "master mount" - or "master" for short - is mostly used when talking in the context of slave mounts that receive propagation from a master mount. A master mount of a slave identifies the closest peer group a slave mount receives propagation from. The master mount of a slave can be identified via @slave_mount->mnt_master. Different slaves may point to different masters in the same peer group. * Multiple peers in a peer group can have non-empty ->mnt_slave_lists. Non-empty ->mnt_slave_lists of peers don't intersect. Consequently, to ensure all slave mounts of a peer group are visited the ->mnt_slave_lists of all peers in a peer group have to be walked. * Slave mounts point to a peer in the closest peer group they receive propagation from via @slave_mnt->mnt_master (see above). Together with these peers they form a propagation group (see below). The closest peer group can thus be identified through the peer group id @slave_mnt->mnt_master->mnt_group_id of the peer/master that a slave mount receives propagation from. * A shared-slave mount is a slave mount to a peer group pg1 while also a peer in another peer group pg2. IOW, a peer group may receive propagation from another peer group. If a peer group pg1 is a slave to another peer group pg2 then all peers in peer group pg1 point to the same peer in peer group pg2 via ->mnt_master. IOW, all peers in peer group pg1 appear on the same ->mnt_slave_list. IOW, they cannot be slaves to different peer groups. * A pure slave mount is a slave mount that is a slave to a peer group but is not a peer in another peer group. * A propagation group denotes the set of mounts consisting of a single peer group pg1 and all slave mounts and shared-slave mounts that point to a peer in that peer group via ->mnt_master. IOW, all slave mounts such that @slave_mnt->mnt_master->mnt_group_id is equal to @shared_mnt->mnt_group_id. The concept of a propagation group makes it easier to talk about a single propagation level in a propagation tree. For example, in propagate_mnt() the immediate peers of @dest_mnt and all slaves of @dest_mnt's peer group form a propagation group propg1. So a shared-slave mount that is a slave in propg1 and that is a peer in another peer group pg2 forms another propagation group propg2 together with all slaves that point to that shared-slave mount in their ->mnt_master. * A propagation tree refers to all mounts that receive propagation starting from a specific shared mount. For example, for propagate_mnt() @dest_mnt is the start of a propagation tree. The propagation tree ecompasses all mounts that receive propagation from @dest_mnt's peer group down to the leafs. With that out of the way let's get to the actual algorithm. We know that @dest_mnt is guaranteed to be a pure shared mount or a shared-slave mount. This is guaranteed by a check in attach_recursive_mnt(). So propagate_mnt() will first propagate the source mount tree to all peers in @dest_mnt's peer group: for (n = next_peer(dest_mnt); n != dest_mnt; n = next_peer(n)) { ret = propagate_one(n); if (ret) goto out; } Notice, that the peer propagation loop of propagate_mnt() doesn't propagate @dest_mnt itself. @dest_mnt is mounted directly in attach_recursive_mnt() after we propagated to the destination propagation tree. The mount that will be mounted on top of @dest_mnt is @source_mnt. This copy was created earlier even before we entered attach_recursive_mnt() and doesn't concern us a lot here. It's just important to notice that when propagate_mnt() is called @source_mnt will not yet have been mounted on top of @dest_mnt. Thus, @source_mnt->mnt_parent will either still point to @source_mnt or - in the case @source_mnt is moved and thus already attached - still to its former parent. For each peer @m in @dest_mnt's peer group propagate_one() will create a new copy of the source mount tree and mount that copy @child on @m such that @child->mnt_parent points to @m after propagate_one() returns. propagate_one() will stash the last destination propagation node @m in @last_dest and the last copy it created for the source mount tree in @last_source. Hence, if we call into propagate_one() again for the next destination propagation node @m, @last_dest will point to the previous destination propagation node and @last_source will point to the previous copy of the source mount tree and mounted on @last_dest. Each new copy of the source mount tree is created from the previous copy of the source mount tree. This will become important later. The peer loop in propagate_mnt() is straightforward. We iterate through the peers copying and updating @last_source and @last_dest as we go through them and mount each copy of the source mount tree @child on a peer @m in @dest_mnt's peer group. After propagate_mnt() handled the peers in @dest_mnt's peer group propagate_mnt() will propagate the source mount tree down the propagation tree that @dest_mnt's peer group propagates to: for (m = next_group(dest_mnt, dest_mnt); m; m = next_group(m, dest_mnt)) { /* everything in that slave group */ n = m; do { ret = propagate_one(n); if (ret) goto out; n = next_peer(n); } while (n != m); } The next_group() helper will recursively walk the destination propagation tree, descending into each propagation group of the propagation tree. The important part is that it takes care to propagate the source mount tree to all peers in the peer group of a propagation group before it propagates to the slaves to those peers in the propagation group. IOW, it creates and mounts copies of the source mount tree that become masters before it creates and mounts copies of the source mount tree that become slaves to these masters. It is important to remember that propagating the source mount tree to each mount @m in the destination propagation tree simply means that we create and mount new copies @child of the source mount tree on @m such that @child->mnt_parent points to @m. Since we know that each node @m in the destination propagation tree headed by @dest_mnt's peer group will be overmounted with a copy of the source mount tree and since we know that the propagation properties of each copy of the source mount tree we create and mount at @m will mostly mirror the propagation properties of @m. We can use that information to create and mount the copies of the source mount tree that become masters before their slaves. The easy case is always when @m and @last_dest are peers in a peer group of a given propagation group. In that case we know that we can simply copy @last_source without having to figure out what the master for the new copy @child of the source mount tree needs to be as we've done that in a previous call to propagate_one(). The hard case is when we're dealing with a slave mount or a shared-slave mount @m in a destination propagation group that we need to create and mount a copy of the source mount tree on. For each propagation group in the destination propagation tree we propagate the source mount tree to we want to make sure that the copies @child of the source mount tree we create and mount on slaves @m pick an ealier copy of the source mount tree that we mounted on a master @m of the destination propagation group as their master. This is a mouthful but as far as we can tell that's the core of it all. But, if we keep track of the masters in the destination propagation tree @m we can use the information to find the correct master for each copy of the source mount tree we create and mount at the slaves in the destination propagation tree @m. Let's walk through the base case as that's still fairly easy to grasp. If we're dealing with the first slave in the propagation group that @dest_mnt is in then we don't yet have marked any masters in the destination propagation tree. We know the master for the first slave to @dest_mnt's peer group is simple @dest_mnt. So we expect this algorithm to yield a copy of the source mount tree that was mounted on a peer in @dest_mnt's peer group as the master for the copy of the source mount tree we want to mount at the first slave @m: for (n = m; ; n = p) { p = n->mnt_master; if (p == dest_master || IS_MNT_MARKED(p)) break; } For the first slave we walk the destination propagation tree all the way up to a peer in @dest_mnt's peer group. IOW, the propagation hierarchy can be walked by walking up the @mnt->mnt_master hierarchy of the destination propagation tree @m. We will ultimately find a peer in @dest_mnt's peer group and thus ultimately @dest_mnt->mnt_master. Btw, here the assumption we listed at the beginning becomes important. Namely, that peers in a peer group pg1 that are slaves in another peer group pg2 appear on the same ->mnt_slave_list. IOW, all slaves who are peers in peer group pg1 point to the same peer in peer group pg2 via their ->mnt_master. Otherwise the termination condition in the code above would be wrong and next_group() would be broken too. So the first iteration sets: n = m; p = n->mnt_master; such that @p now points to a peer or @dest_mnt itself. We walk up one more level since we don't have any marked mounts. So we end up with: n = dest_mnt; p = dest_mnt->mnt_master; If @dest_mnt's peer group is not slave to another peer group then @p is now NULL. If @dest_mnt's peer group is a slave to another peer group then @p now points to @dest_mnt->mnt_master points which is a master outside the propagation tree we're dealing with. Now we need to figure out the master for the copy of the source mount tree we're about to create and mount on the first slave of @dest_mnt's peer group: do { struct mount *parent = last_source->mnt_parent; if (last_source == first_source) break; done = parent->mnt_master == p; if (done && peers(n, parent)) break; last_source = last_source->mnt_master; } while (!done); We know that @last_source->mnt_parent points to @last_dest and @last_dest is the last peer in @dest_mnt's peer group we propagated to in the peer loop in propagate_mnt(). Consequently, @last_source is the last copy we created and mount on that last peer in @dest_mnt's peer group. So @last_source is the master we want to pick. We know that @last_source->mnt_parent->mnt_master points to @last_dest->mnt_master. We also know that @last_dest->mnt_master is either NULL or points to a master outside of the destination propagation tree and so does @p. Hence: done = parent->mnt_master == p; is trivially true in the base condition. We also know that for the first slave mount of @dest_mnt's peer group that @last_dest either points @dest_mnt itself because it was initialized to: last_dest = dest_mnt; at the beginning of propagate_mnt() or it will point to a peer of @dest_mnt in its peer group. In both cases it is guaranteed that on the first iteration @n and @parent are peers (Please note the check for peers here as that's important.): if (done && peers(n, parent)) break; So, as we expected, we select @last_source, which referes to the last copy of the source mount tree we mounted on the last peer in @dest_mnt's peer group, as the master of the first slave in @dest_mnt's peer group. The rest is taken care of by clone_mnt(last_source, ...). We'll skip over that part otherwise this becomes a blogpost. At the end of propagate_mnt() we now mark @m->mnt_master as the first master in the destination propagation tree that is distinct from @dest_mnt->mnt_master. IOW, we mark @dest_mnt itself as a master. By marking @dest_mnt or one of it's peers we are able to easily find it again when we later lookup masters for other copies of the source mount tree we mount copies of the source mount tree on slaves @m to @dest_mnt's peer group. This, in turn allows us to find the master we selected for the copies of the source mount tree we mounted on master in the destination propagation tree again. The important part is to realize that the code makes use of the fact that the last copy of the source mount tree stashed in @last_source was mounted on top of the previous destination propagation node @last_dest. What this means is that @last_source allows us to walk the destination propagation hierarchy the same way each destination propagation node @m does. If we take @last_source, which is the copy of @source_mnt we have mounted on @last_dest in the previous iteration of propagate_one(), then we know @last_source->mnt_parent points to @last_dest but we also know that as we walk through the destination propagation tree that @last_source->mnt_master will point to an earlier copy of the source mount tree we mounted one an earlier destination propagation node @m. IOW, @last_source->mnt_parent will be our hook into the destination propagation tree and each consecutive @last_source->mnt_master will lead us to an earlier propagation node @m via @last_source->mnt_master->mnt_parent. Hence, by walking up @last_source->mnt_master, each of which is mounted on a node that is a master @m in the destination propagation tree we can also walk up the destination propagation hierarchy. So, for each new destination propagation node @m we use the previous copy of @last_source and the fact it's mounted on the previous propagation node @last_dest via @last_source->mnt_master->mnt_parent to determine what the master of the new copy of @last_source needs to be. The goal is to find the _closest_ master that the new copy of the source mount tree we are about to create and mount on a slave @m in the destination propagation tree needs to pick. IOW, we want to find a suitable master in the propagation group. As the propagation structure of the source mount propagation tree we create mirrors the propagation structure of the destination propagation tree we can find @m's closest master - i.e., a marked master - which is a peer in the closest peer group that @m receives propagation from. We store that closest master of @m in @p as before and record the slave to that master in @n We then search for this master @p via @last_source by walking up the master hierarchy starting from the last copy of the source mount tree stored in @last_source that we created and mounted on the previous destination propagation node @m. We will try to find the master by walking @last_source->mnt_master and by comparing @last_source->mnt_master->mnt_parent->mnt_master to @p. If we find @p then we can figure out what earlier copy of the source mount tree needs to be the master for the new copy of the source mount tree we're about to create and mount at the current destination propagation node @m. If @last_source->mnt_master->mnt_parent and @n are peers then we know that the closest master they receive propagation from is @last_source->mnt_master->mnt_parent->mnt_master. If not then the closest immediate peer group that they receive propagation from must be one level higher up. This builds on the earlier clarification at the beginning that all peers in a peer group which are slaves of other peer groups all point to the same ->mnt_master, i.e., appear on the same ->mnt_slave_list, of the closest peer group that they receive propagation from. However, terminating the walk has corner cases. If the closest marked master for a given destination node @m cannot be found by walking up the master hierarchy via @last_source->mnt_master then we need to terminate the walk when we encounter @source_mnt again. This isn't an arbitrary termination. It simply means that the new copy of the source mount tree we're about to create has a copy of the source mount tree we created and mounted on a peer in @dest_mnt's peer group as its master. IOW, @source_mnt is the peer in the closest peer group that the new copy of the source mount tree receives propagation from. We absolutely have to stop @source_mnt because @last_source->mnt_master either points outside the propagation hierarchy we're dealing with or it is NULL because @source_mnt isn't a shared-slave. So continuing the walk past @source_mnt would cause a NULL dereference via @last_source->mnt_master->mnt_parent. And so we have to stop the walk when we encounter @source_mnt again. One scenario where this can happen is when we first handled a series of slaves of @dest_mnt's peer group and then encounter peers in a new peer group that is a slave to @dest_mnt's peer group. We handle them and then we encounter another slave mount to @dest_mnt that is a pure slave to @dest_mnt's peer group. That pure slave will have a peer in @dest_mnt's peer group as its master. Consequently, the new copy of the source mount tree will need to have @source_mnt as it's master. So we walk the propagation hierarchy all the way up to @source_mnt based on @last_source->mnt_master. So terminate on @source_mnt, easy peasy. Except, that the check misses something that the rest of the algorithm already handles. If @dest_mnt has peers in it's peer group the peer loop in propagate_mnt(): for (n = next_peer(dest_mnt); n != dest_mnt; n = next_peer(n)) { ret = propagate_one(n); if (ret) goto out; } will consecutively update @last_source with each previous copy of the source mount tree we created and mounted at the previous peer in @dest_mnt's peer group. So after that loop terminates @last_source will point to whatever copy of the source mount tree was created and mounted on the last peer in @dest_mnt's peer group. Furthermore, if there is even a single additional peer in @dest_mnt's peer group then @last_source will __not__ point to @source_mnt anymore. Because, as we mentioned above, @dest_mnt isn't even handled in this loop but directly in attach_recursive_mnt(). So it can't even accidently come last in that peer loop. So the first time we handle a slave mount @m of @dest_mnt's peer group the copy of the source mount tree we create will make the __last copy of the source mount tree we created and mounted on the last peer in @dest_mnt's peer group the master of the new copy of the source mount tree we create and mount on the first slave of @dest_mnt's peer group__. But this means that the termination condition that checks for @source_mnt is wrong. The @source_mnt cannot be found anymore by propagate_one(). Instead it will find the last copy of the source mount tree we created and mounted for the last peer of @dest_mnt's peer group again. And that is a peer of @source_mnt not @source_mnt itself. IOW, we fail to terminate the loop correctly and ultimately dereference @last_source->mnt_master->mnt_parent. When @source_mnt's peer group isn't slave to another peer group then @last_source->mnt_master is NULL causing the splat below. For example, assume @dest_mnt is a pure shared mount and has three peers in its peer group: =================================================================================== mount-id mount-parent-id peer-group-id =================================================================================== (@dest_mnt) mnt_master[216] 309 297 shared:216 \ (@source_mnt) mnt_master[218]: 609 609 shared:218 (1) mnt_master[216]: 607 605 shared:216 \ (P1) mnt_master[218]: 624 607 shared:218 (2) mnt_master[216]: 576 574 shared:216 \ (P2) mnt_master[218]: 625 576 shared:218 (3) mnt_master[216]: 545 543 shared:216 \ (P3) mnt_master[218]: 626 545 shared:218 After this sequence has been processed @last_source will point to (P3), the copy generated for the third peer in @dest_mnt's peer group we handled. So the copy of the source mount tree (P4) we create and mount on the first slave of @dest_mnt's peer group: =================================================================================== mount-id mount-parent-id peer-group-id =================================================================================== mnt_master[216] 309 297 shared:216 / / (S0) mnt_slave 483 481 master:216 \ \ (P3) mnt_master[218] 626 545 shared:218 \ / \/ (P4) mnt_slave 627 483 master:218 will pick the last copy of the source mount tree (P3) as master, not (S0). When walking the propagation hierarchy via @last_source's master hierarchy we encounter (P3) but not (S0), i.e., @source_mnt. We can fix this in multiple ways: (1) By setting @last_source to @source_mnt after we processed the peers in @dest_mnt's peer group right after the peer loop in propagate_mnt(). (2) By changing the termination condition that relies on finding exactly @source_mnt to finding a peer of @source_mnt. (3) By only moving @last_source when we actually venture into a new peer group or some clever variant thereof. The first two options are minimally invasive and what we want as a fix. The third option is more intrusive but something we'd like to explore in the near future. This passes all LTP tests and specifically the mount propagation testsuite part of it. It also holds up against all known reproducers of this issues. Final words. First, this is a clever but __worringly__ underdocumented algorithm. There isn't a single detailed comment to be found in next_group(), propagate_one() or anywhere else in that file for that matter. This has been a giant pain to understand and work through and a bug like this is insanely difficult to fix without a detailed understanding of what's happening. Let's not talk about the amount of time that was sunk into fixing this. Second, all the cool kids with access to unshare --mount --user --map-root --propagation=unchanged are going to have a lot of fun. IOW, triggerable by unprivileged users while namespace_lock() lock is held. [ 115.848393] BUG: kernel NULL pointer dereference, address: 0000000000000010 [ 115.848967] #PF: supervisor read access in kernel mode [ 115.849386] #PF: error_code(0x0000) - not-present page [ 115.849803] PGD 0 P4D 0 [ 115.850012] Oops: 0000 [#1] PREEMPT SMP PTI [ 115.850354] CPU: 0 PID: 15591 Comm: mount Not tainted 6.1.0-rc7 #3 [ 115.850851] Hardware name: innotek GmbH VirtualBox/VirtualBox, BIOS VirtualBox 12/01/2006 [ 115.851510] RIP: 0010:propagate_one.part.0+0x7f/0x1a0 [ 115.851924] Code: 75 eb 4c 8b 05 c2 25 37 02 4c 89 ca 48 8b 4a 10 49 39 d0 74 1e 48 3b 81 e0 00 00 00 74 26 48 8b 92 e0 00 00 00 be 01 00 00 00 <48> 8b 4a 10 49 39 d0 75 e2 40 84 f6 74 38 4c 89 05 84 25 37 02 4d [ 115.853441] RSP: 0018:ffffb8d5443d7d50 EFLAGS: 00010282 [ 115.853865] RAX: ffff8e4d87c41c80 RBX: ffff8e4d88ded780 RCX: ffff8e4da4333a00 [ 115.854458] RDX: 0000000000000000 RSI: 0000000000000001 RDI: ffff8e4d88ded780 [ 115.855044] RBP: ffff8e4d88ded780 R08: ffff8e4da4338000 R09: ffff8e4da43388c0 [ 115.855693] R10: 0000000000000002 R11: ffffb8d540158000 R12: ffffb8d5443d7da8 [ 115.856304] R13: ffff8e4d88ded780 R14: 0000000000000000 R15: 0000000000000000 [ 115.856859] FS: 00007f92c90c9800(0000) GS:ffff8e4dfdc00000(0000) knlGS:0000000000000000 [ 115.857531] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 115.858006] CR2: 0000000000000010 CR3: 0000000022f4c002 CR4: 00000000000706f0 [ 115.858598] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 [ 115.859393] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400 [ 115.860099] Call Trace: [ 115.860358] <TASK> [ 115.860535] propagate_mnt+0x14d/0x190 [ 115.860848] attach_recursive_mnt+0x274/0x3e0 [ 115.861212] path_mount+0x8c8/0xa60 [ 115.861503] __x64_sys_mount+0xf6/0x140 [ 115.861819] do_syscall_64+0x5b/0x80 [ 115.862117] ? do_faccessat+0x123/0x250 [ 115.862435] ? syscall_exit_to_user_mode+0x17/0x40 [ 115.862826] ? do_syscall_64+0x67/0x80 [ 115.863133] ? syscall_exit_to_user_mode+0x17/0x40 [ 115.863527] ? do_syscall_64+0x67/0x80 [ 115.863835] ? do_syscall_64+0x67/0x80 [ 115.864144] ? do_syscall_64+0x67/0x80 [ 115.864452] ? exc_page_fault+0x70/0x170 [ 115.864775] entry_SYSCALL_64_after_hwframe+0x63/0xcd [ 115.865187] RIP: 0033:0x7f92c92b0ebe [ 115.865480] Code: 48 8b 0d 75 4f 0c 00 f7 d8 64 89 01 48 83 c8 ff c3 66 2e 0f 1f 84 00 00 00 00 00 90 f3 0f 1e fa 49 89 ca b8 a5 00 00 00 0f 05 <48> 3d 01 f0 ff ff 73 01 c3 48 8b 0d 42 4f 0c 00 f7 d8 64 89 01 48 [ 115.866984] RSP: 002b:00007fff000aa728 EFLAGS: 00000246 ORIG_RAX: 00000000000000a5 [ 115.867607] RAX: ffffffffffffffda RBX: 000055a77888d6b0 RCX: 00007f92c92b0ebe [ 115.868240] RDX: 000055a77888d8e0 RSI: 000055a77888e6e0 RDI: 000055a77888e620 [ 115.868823] RBP: 0000000000000000 R08: 0000000000000000 R09: 0000000000000001 [ 115.869403] R10: 0000000000001000 R11: 0000000000000246 R12: 000055a77888e620 [ 115.869994] R13: 000055a77888d8e0 R14: 00000000ffffffff R15: 00007f92c93e4076 [ 115.870581] </TASK> [ 115.870763] Modules linked in: nft_fib_inet nft_fib_ipv4 nft_fib_ipv6 nft_fib nft_reject_inet nf_reject_ipv4 nf_reject_ipv6 nft_reject nft_ct nft_chain_nat nf_nat nf_conntrack nf_defrag_ipv6 nf_defrag_ipv4 ip_set rfkill nf_tables nfnetlink qrtr snd_intel8x0 sunrpc snd_ac97_codec ac97_bus snd_pcm snd_timer intel_rapl_msr intel_rapl_common snd vboxguest intel_powerclamp video rapl joydev soundcore i2c_piix4 wmi fuse zram xfs vmwgfx crct10dif_pclmul crc32_pclmul crc32c_intel polyval_clmulni polyval_generic drm_ttm_helper ttm e1000 ghash_clmulni_intel serio_raw ata_generic pata_acpi scsi_dh_rdac scsi_dh_emc scsi_dh_alua dm_multipath [ 115.875288] CR2: 0000000000000010 [ 115.875641] ---[ end trace 0000000000000000 ]--- [ 115.876135] RIP: 0010:propagate_one.part.0+0x7f/0x1a0 [ 115.876551] Code: 75 eb 4c 8b 05 c2 25 37 02 4c 89 ca 48 8b 4a 10 49 39 d0 74 1e 48 3b 81 e0 00 00 00 74 26 48 8b 92 e0 00 00 00 be 01 00 00 00 <48> 8b 4a 10 49 39 d0 75 e2 40 84 f6 74 38 4c 89 05 84 25 37 02 4d [ 115.878086] RSP: 0018:ffffb8d5443d7d50 EFLAGS: 00010282 [ 115.878511] RAX: ffff8e4d87c41c80 RBX: ffff8e4d88ded780 RCX: ffff8e4da4333a00 [ 115.879128] RDX: 0000000000000000 RSI: 0000000000000001 RDI: ffff8e4d88ded780 [ 115.879715] RBP: ffff8e4d88ded780 R08: ffff8e4da4338000 R09: ffff8e4da43388c0 [ 115.880359] R10: 0000000000000002 R11: ffffb8d540158000 R12: ffffb8d5443d7da8 [ 115.880962] R13: ffff8e4d88ded780 R14: 0000000000000000 R15: 0000000000000000 [ 115.881548] FS: 00007f92c90c9800(0000) GS:ffff8e4dfdc00000(0000) knlGS:0000000000000000 [ 115.882234] CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 [ 115.882713] CR2: 0000000000000010 CR3: 0000000022f4c002 CR4: 00000000000706f0 [ 115.883314] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000 [ 115.883966] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400 Fixes: f2ebb3a921c1 ("smarter propagate_mnt()") Fixes: 5ec0811d3037 ("propogate_mnt: Handle the first propogated copy being a slave") Cc: <stable@vger.kernel.org> Reported-by: Ditang Chen <ditang.c@gmail.com> Signed-off-by: Seth Forshee (Digital Ocean) <sforshee@kernel.org> Signed-off-by: Christian Brauner (Microsoft) <brauner@kernel.org> --- If there are no big objections I'll get this to Linus rather sooner than later.
* propagate_one(): mnt_set_mountpoint() needs mount_lockAl Viro2020-04-271-5/+4
| | | | | | | | | | | | ... to protect the modification of mp->m_count done by it. Most of the places that modify that thing also have namespace_lock held, but not all of them can do so, so we really need mount_lock here. Kudos to Piotr Krysiuk <piotras@gmail.com>, who'd spotted a related bug in pivot_root(2) (fixed unnoticed in 5.3); search for other similar turds has caught out this one. Cc: stable@kernel.org Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
* fs/namespace: fix unprivileged mount propagationChristian Brauner2019-06-171-1/+0
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | When propagating mounts across mount namespaces owned by different user namespaces it is not possible anymore to move or umount the mount in the less privileged mount namespace. Here is a reproducer: sudo mount -t tmpfs tmpfs /mnt sudo --make-rshared /mnt # create unprivileged user + mount namespace and preserve propagation unshare -U -m --map-root --propagation=unchanged # now change back to the original mount namespace in another terminal: sudo mkdir /mnt/aaa sudo mount -t tmpfs tmpfs /mnt/aaa # now in the unprivileged user + mount namespace mount --move /mnt/aaa /opt Unfortunately, this is a pretty big deal for userspace since this is e.g. used to inject mounts into running unprivileged containers. So this regression really needs to go away rather quickly. The problem is that a recent change falsely locked the root of the newly added mounts by setting MNT_LOCKED. Fix this by only locking the mounts on copy_mnt_ns() and not when adding a new mount. Fixes: 3bd045cc9c4b ("separate copying and locking mount tree on cross-userns copies") Cc: Linus Torvalds <torvalds@linux-foundation.org> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: <stable@vger.kernel.org> Tested-by: Christian Brauner <christian@brauner.io> Acked-by: Christian Brauner <christian@brauner.io> Signed-off-by: "Eric W. Biederman" <ebiederm@xmission.com> Signed-off-by: Christian Brauner <christian@brauner.io> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
* treewide: Replace GPLv2 boilerplate/reference with SPDX - rule 209Thomas Gleixner2019-05-301-2/+1
| | | | | | | | | | | | | | | | | | | | Based on 1 normalized pattern(s): released under gpl v2 extracted by the scancode license scanner the SPDX license identifier GPL-2.0-only has been chosen to replace the boilerplate/reference in 15 file(s). Signed-off-by: Thomas Gleixner <tglx@linutronix.de> Reviewed-by: Steve Winslow <swinslow@gmail.com> Reviewed-by: Allison Randal <allison@lohutok.net> Reviewed-by: Alexios Zavras <alexios.zavras@intel.com> Cc: linux-spdx@vger.kernel.org Link: https://lkml.kernel.org/r/20190528171438.895196075@linutronix.de Signed-off-by: Greg Kroah-Hartman <gregkh@linuxfoundation.org>
* separate copying and locking mount tree on cross-userns copiesAl Viro2019-01-301-5/+0
| | | | | | | Rather than having propagate_mnt() check doing unprivileged copies, lock them before commit_tree(). Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
* vfs: Suppress MS_* flag defs within the kernel unless explicitly enabledDavid Howells2018-12-201-0/+1
| | | | | | | | | | Only the mount namespace code that implements mount(2) should be using the MS_* flags. Suppress them inside the kernel unless uapi/linux/mount.h is included. Signed-off-by: David Howells <dhowells@redhat.com> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk> Reviewed-by: David Howells <dhowells@redhat.com>
* mnt: Make propagate_umount less slow for overlapping mount propagation treesEric W. Biederman2017-05-231-1/+62
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Andrei Vagin pointed out that time to executue propagate_umount can go non-linear (and take a ludicrious amount of time) when the mount propogation trees of the mounts to be unmunted by a lazy unmount overlap. Make the walk of the mount propagation trees nearly linear by remembering which mounts have already been visited, allowing subsequent walks to detect when walking a mount propgation tree or a subtree of a mount propgation tree would be duplicate work and to skip them entirely. Walk the list of mounts whose propgatation trees need to be traversed from the mount highest in the mount tree to mounts lower in the mount tree so that odds are higher that the code will walk the largest trees first, allowing later tree walks to be skipped entirely. Add cleanup_umount_visitation to remover the code's memory of which mounts have been visited. Add the functions last_slave and skip_propagation_subtree to allow skipping appropriate parts of the mount propagation tree without needing to change the logic of the rest of the code. A script to generate overlapping mount propagation trees: $ cat runs.h set -e mount -t tmpfs zdtm /mnt mkdir -p /mnt/1 /mnt/2 mount -t tmpfs zdtm /mnt/1 mount --make-shared /mnt/1 mkdir /mnt/1/1 iteration=10 if [ -n "$1" ] ; then iteration=$1 fi for i in $(seq $iteration); do mount --bind /mnt/1/1 /mnt/1/1 done mount --rbind /mnt/1 /mnt/2 TIMEFORMAT='%Rs' nr=$(( ( 2 ** ( $iteration + 1 ) ) + 1 )) echo -n "umount -l /mnt/1 -> $nr " time umount -l /mnt/1 nr=$(cat /proc/self/mountinfo | grep zdtm | wc -l ) time umount -l /mnt/2 $ for i in $(seq 9 19); do echo $i; unshare -Urm bash ./run.sh $i; done Here are the performance numbers with and without the patch: mhash | 8192 | 8192 | 1048576 | 1048576 mounts | before | after | before | after ------------------------------------------------ 1025 | 0.040s | 0.016s | 0.038s | 0.019s 2049 | 0.094s | 0.017s | 0.080s | 0.018s 4097 | 0.243s | 0.019s | 0.206s | 0.023s 8193 | 1.202s | 0.028s | 1.562s | 0.032s 16385 | 9.635s | 0.036s | 9.952s | 0.041s 32769 | 60.928s | 0.063s | 44.321s | 0.064s 65537 | | 0.097s | | 0.097s 131073 | | 0.233s | | 0.176s 262145 | | 0.653s | | 0.344s 524289 | | 2.305s | | 0.735s 1048577 | | 7.107s | | 2.603s Andrei Vagin reports fixing the performance problem is part of the work to fix CVE-2016-6213. Cc: stable@vger.kernel.org Fixes: a05964f3917c ("[PATCH] shared mounts handling: umount") Reported-by: Andrei Vagin <avagin@openvz.org> Reviewed-by: Andrei Vagin <avagin@virtuozzo.com> Signed-off-by: "Eric W. Biederman" <ebiederm@xmission.com>
* mnt: In propgate_umount handle visiting mounts in any orderEric W. Biederman2017-05-231-58/+86
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | While investigating some poor umount performance I realized that in the case of overlapping mount trees where some of the mounts are locked the code has been failing to unmount all of the mounts it should have been unmounting. This failure to unmount all of the necessary mounts can be reproduced with: $ cat locked_mounts_test.sh mount -t tmpfs test-base /mnt mount --make-shared /mnt mkdir -p /mnt/b mount -t tmpfs test1 /mnt/b mount --make-shared /mnt/b mkdir -p /mnt/b/10 mount -t tmpfs test2 /mnt/b/10 mount --make-shared /mnt/b/10 mkdir -p /mnt/b/10/20 mount --rbind /mnt/b /mnt/b/10/20 unshare -Urm --propagation unchaged /bin/sh -c 'sleep 5; if [ $(grep test /proc/self/mountinfo | wc -l) -eq 1 ] ; then echo SUCCESS ; else echo FAILURE ; fi' sleep 1 umount -l /mnt/b wait %% $ unshare -Urm ./locked_mounts_test.sh This failure is corrected by removing the prepass that marks mounts that may be umounted. A first pass is added that umounts mounts if possible and if not sets mount mark if they could be unmounted if they weren't locked and adds them to a list to umount possibilities. This first pass reconsiders the mounts parent if it is on the list of umount possibilities, ensuring that information of umoutability will pass from child to mount parent. A second pass then walks through all mounts that are umounted and processes their children unmounting them or marking them for reparenting. A last pass cleans up the state on the mounts that could not be umounted and if applicable reparents them to their first parent that remained mounted. While a bit longer than the old code this code is much more robust as it allows information to flow up from the leaves and down from the trunk making the order in which mounts are encountered in the umount propgation tree irrelevant. Cc: stable@vger.kernel.org Fixes: 0c56fe31420c ("mnt: Don't propagate unmounts to locked mounts") Reviewed-by: Andrei Vagin <avagin@virtuozzo.com> Signed-off-by: "Eric W. Biederman" <ebiederm@xmission.com>
* mnt: In umount propagation reparent in a separate passEric W. Biederman2017-05-231-5/+30
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | It was observed that in some pathlogical cases that the current code does not unmount everything it should. After investigation it was determined that the issue is that mnt_change_mntpoint can can change which mounts are available to be unmounted during mount propagation which is wrong. The trivial reproducer is: $ cat ./pathological.sh mount -t tmpfs test-base /mnt cd /mnt mkdir 1 2 1/1 mount --bind 1 1 mount --make-shared 1 mount --bind 1 2 mount --bind 1/1 1/1 mount --bind 1/1 1/1 echo grep test-base /proc/self/mountinfo umount 1/1 echo grep test-base /proc/self/mountinfo $ unshare -Urm ./pathological.sh The expected output looks like: 46 31 0:25 / /mnt rw,relatime - tmpfs test-base rw,uid=1000,gid=1000 47 46 0:25 /1 /mnt/1 rw,relatime shared:1 - tmpfs test-base rw,uid=1000,gid=1000 48 46 0:25 /1 /mnt/2 rw,relatime shared:1 - tmpfs test-base rw,uid=1000,gid=1000 49 54 0:25 /1/1 /mnt/1/1 rw,relatime shared:1 - tmpfs test-base rw,uid=1000,gid=1000 50 53 0:25 /1/1 /mnt/2/1 rw,relatime shared:1 - tmpfs test-base rw,uid=1000,gid=1000 51 49 0:25 /1/1 /mnt/1/1 rw,relatime shared:1 - tmpfs test-base rw,uid=1000,gid=1000 54 47 0:25 /1/1 /mnt/1/1 rw,relatime shared:1 - tmpfs test-base rw,uid=1000,gid=1000 53 48 0:25 /1/1 /mnt/2/1 rw,relatime shared:1 - tmpfs test-base rw,uid=1000,gid=1000 52 50 0:25 /1/1 /mnt/2/1 rw,relatime shared:1 - tmpfs test-base rw,uid=1000,gid=1000 46 31 0:25 / /mnt rw,relatime - tmpfs test-base rw,uid=1000,gid=1000 47 46 0:25 /1 /mnt/1 rw,relatime shared:1 - tmpfs test-base rw,uid=1000,gid=1000 48 46 0:25 /1 /mnt/2 rw,relatime shared:1 - tmpfs test-base rw,uid=1000,gid=1000 The output without the fix looks like: 46 31 0:25 / /mnt rw,relatime - tmpfs test-base rw,uid=1000,gid=1000 47 46 0:25 /1 /mnt/1 rw,relatime shared:1 - tmpfs test-base rw,uid=1000,gid=1000 48 46 0:25 /1 /mnt/2 rw,relatime shared:1 - tmpfs test-base rw,uid=1000,gid=1000 49 54 0:25 /1/1 /mnt/1/1 rw,relatime shared:1 - tmpfs test-base rw,uid=1000,gid=1000 50 53 0:25 /1/1 /mnt/2/1 rw,relatime shared:1 - tmpfs test-base rw,uid=1000,gid=1000 51 49 0:25 /1/1 /mnt/1/1 rw,relatime shared:1 - tmpfs test-base rw,uid=1000,gid=1000 54 47 0:25 /1/1 /mnt/1/1 rw,relatime shared:1 - tmpfs test-base rw,uid=1000,gid=1000 53 48 0:25 /1/1 /mnt/2/1 rw,relatime shared:1 - tmpfs test-base rw,uid=1000,gid=1000 52 50 0:25 /1/1 /mnt/2/1 rw,relatime shared:1 - tmpfs test-base rw,uid=1000,gid=1000 46 31 0:25 / /mnt rw,relatime - tmpfs test-base rw,uid=1000,gid=1000 47 46 0:25 /1 /mnt/1 rw,relatime shared:1 - tmpfs test-base rw,uid=1000,gid=1000 48 46 0:25 /1 /mnt/2 rw,relatime shared:1 - tmpfs test-base rw,uid=1000,gid=1000 52 48 0:25 /1/1 /mnt/2/1 rw,relatime shared:1 - tmpfs test-base rw,uid=1000,gid=1000 That last mount in the output was in the propgation tree to be unmounted but was missed because the mnt_change_mountpoint changed it's parent before the walk through the mount propagation tree observed it. Cc: stable@vger.kernel.org Fixes: 1064f874abc0 ("mnt: Tuck mounts under others instead of creating shadow/side mounts.") Acked-by: Andrei Vagin <avagin@virtuozzo.com> Reviewed-by: Ram Pai <linuxram@us.ibm.com> Signed-off-by: "Eric W. Biederman" <ebiederm@xmission.com>
* mnt: Tuck mounts under others instead of creating shadow/side mounts.Eric W. Biederman2017-02-041-12/+49
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | Ever since mount propagation was introduced in cases where a mount in propagated to parent mount mountpoint pair that is already in use the code has placed the new mount behind the old mount in the mount hash table. This implementation detail is problematic as it allows creating arbitrary length mount hash chains. Furthermore it invalidates the constraint maintained elsewhere in the mount code that a parent mount and a mountpoint pair will have exactly one mount upon them. Making it hard to deal with and to talk about this special case in the mount code. Modify mount propagation to notice when there is already a mount at the parent mount and mountpoint where a new mount is propagating to and place that preexisting mount on top of the new mount. Modify unmount propagation to notice when a mount that is being unmounted has another mount on top of it (and no other children), and to replace the unmounted mount with the mount on top of it. Move the MNT_UMUONT test from __lookup_mnt_last into __propagate_umount as that is the only call of __lookup_mnt_last where MNT_UMOUNT may be set on any mount visible in the mount hash table. These modifications allow: - __lookup_mnt_last to be removed. - attach_shadows to be renamed __attach_mnt and its shadow handling to be removed. - commit_tree to be simplified - copy_tree to be simplified The result is an easier to understand tree of mounts that does not allow creation of arbitrary length hash chains in the mount hash table. The result is also a very slight userspace visible difference in semantics. The following two cases now behave identically, where before order mattered: case 1: (explicit user action) B is a slave of A mount something on A/a , it will propagate to B/a and than mount something on B/a case 2: (tucked mount) B is a slave of A mount something on B/a and than mount something on A/a Histroically umount A/a would fail in case 1 and succeed in case 2. Now umount A/a succeeds in both configurations. This very small change in semantics appears if anything to be a bug fix to me and my survey of userspace leads me to believe that no programs will notice or care of this subtle semantic change. v2: Updated to mnt_change_mountpoint to not call dput or mntput and instead to decrement the counts directly. It is guaranteed that there will be other references when mnt_change_mountpoint is called so this is safe. v3: Moved put_mountpoint under mount_lock in attach_recursive_mnt As the locking in fs/namespace.c changed between v2 and v3. v4: Reworked the logic in propagate_mount_busy and __propagate_umount that detects when a mount completely covers another mount. v5: Removed unnecessary tests whose result is alwasy true in find_topper and attach_recursive_mnt. v6: Document the user space visible semantic difference. Cc: stable@vger.kernel.org Fixes: b90fa9ae8f51 ("[PATCH] shared mount handling: bind and rbind") Tested-by: Andrei Vagin <avagin@virtuozzo.com> Signed-off-by: "Eric W. Biederman" <ebiederm@xmission.com>
* reorganize do_make_slave()Al Viro2016-12-161-38/+36
| | | | | | | | Make sure that clone_mnt() never returns a mount with MNT_SHARED in flags, but without a valid ->mnt_group_id. That allows to demystify do_make_slave() quite a bit, among other things. Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
* mnt: Add a per mount namespace limit on the number of mountsEric W. Biederman2016-09-301-1/+1
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | CAI Qian <caiqian@redhat.com> pointed out that the semantics of shared subtrees make it possible to create an exponentially increasing number of mounts in a mount namespace. mkdir /tmp/1 /tmp/2 mount --make-rshared / for i in $(seq 1 20) ; do mount --bind /tmp/1 /tmp/2 ; done Will create create 2^20 or 1048576 mounts, which is a practical problem as some people have managed to hit this by accident. As such CVE-2016-6213 was assigned. Ian Kent <raven@themaw.net> described the situation for autofs users as follows: > The number of mounts for direct mount maps is usually not very large because of > the way they are implemented, large direct mount maps can have performance > problems. There can be anywhere from a few (likely case a few hundred) to less > than 10000, plus mounts that have been triggered and not yet expired. > > Indirect mounts have one autofs mount at the root plus the number of mounts that > have been triggered and not yet expired. > > The number of autofs indirect map entries can range from a few to the common > case of several thousand and in rare cases up to between 30000 and 50000. I've > not heard of people with maps larger than 50000 entries. > > The larger the number of map entries the greater the possibility for a large > number of active mounts so it's not hard to expect cases of a 1000 or somewhat > more active mounts. So I am setting the default number of mounts allowed per mount namespace at 100,000. This is more than enough for any use case I know of, but small enough to quickly stop an exponential increase in mounts. Which should be perfect to catch misconfigurations and malfunctioning programs. For anyone who needs a higher limit this can be changed by writing to the new /proc/sys/fs/mount-max sysctl. Tested-by: CAI Qian <caiqian@redhat.com> Signed-off-by: "Eric W. Biederman" <ebiederm@xmission.com>
* propogate_mnt: Handle the first propogated copy being a slaveEric W. Biederman2016-05-051-11/+14
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | When the first propgated copy was a slave the following oops would result: > BUG: unable to handle kernel NULL pointer dereference at 0000000000000010 > IP: [<ffffffff811fba4e>] propagate_one+0xbe/0x1c0 > PGD bacd4067 PUD bac66067 PMD 0 > Oops: 0000 [#1] SMP > Modules linked in: > CPU: 1 PID: 824 Comm: mount Not tainted 4.6.0-rc5userns+ #1523 > Hardware name: Bochs Bochs, BIOS Bochs 01/01/2007 > task: ffff8800bb0a8000 ti: ffff8800bac3c000 task.ti: ffff8800bac3c000 > RIP: 0010:[<ffffffff811fba4e>] [<ffffffff811fba4e>] propagate_one+0xbe/0x1c0 > RSP: 0018:ffff8800bac3fd38 EFLAGS: 00010283 > RAX: 0000000000000000 RBX: ffff8800bb77ec00 RCX: 0000000000000010 > RDX: 0000000000000000 RSI: ffff8800bb58c000 RDI: ffff8800bb58c480 > RBP: ffff8800bac3fd48 R08: 0000000000000001 R09: 0000000000000000 > R10: 0000000000001ca1 R11: 0000000000001c9d R12: 0000000000000000 > R13: ffff8800ba713800 R14: ffff8800bac3fda0 R15: ffff8800bb77ec00 > FS: 00007f3c0cd9b7e0(0000) GS:ffff8800bfb00000(0000) knlGS:0000000000000000 > CS: 0010 DS: 0000 ES: 0000 CR0: 0000000080050033 > CR2: 0000000000000010 CR3: 00000000bb79d000 CR4: 00000000000006e0 > Stack: > ffff8800bb77ec00 0000000000000000 ffff8800bac3fd88 ffffffff811fbf85 > ffff8800bac3fd98 ffff8800bb77f080 ffff8800ba713800 ffff8800bb262b40 > 0000000000000000 0000000000000000 ffff8800bac3fdd8 ffffffff811f1da0 > Call Trace: > [<ffffffff811fbf85>] propagate_mnt+0x105/0x140 > [<ffffffff811f1da0>] attach_recursive_mnt+0x120/0x1e0 > [<ffffffff811f1ec3>] graft_tree+0x63/0x70 > [<ffffffff811f1f6b>] do_add_mount+0x9b/0x100 > [<ffffffff811f2c1a>] do_mount+0x2aa/0xdf0 > [<ffffffff8117efbe>] ? strndup_user+0x4e/0x70 > [<ffffffff811f3a45>] SyS_mount+0x75/0xc0 > [<ffffffff8100242b>] do_syscall_64+0x4b/0xa0 > [<ffffffff81988f3c>] entry_SYSCALL64_slow_path+0x25/0x25 > Code: 00 00 75 ec 48 89 0d 02 22 22 01 8b 89 10 01 00 00 48 89 05 fd 21 22 01 39 8e 10 01 00 00 0f 84 e0 00 00 00 48 8b 80 d8 00 00 00 <48> 8b 50 10 48 89 05 df 21 22 01 48 89 15 d0 21 22 01 8b 53 30 > RIP [<ffffffff811fba4e>] propagate_one+0xbe/0x1c0 > RSP <ffff8800bac3fd38> > CR2: 0000000000000010 > ---[ end trace 2725ecd95164f217 ]--- This oops happens with the namespace_sem held and can be triggered by non-root users. An all around not pleasant experience. To avoid this scenario when finding the appropriate source mount to copy stop the walk up the mnt_master chain when the first source mount is encountered. Further rewrite the walk up the last_source mnt_master chain so that it is clear what is going on. The reason why the first source mount is special is that it it's mnt_parent is not a mount in the dest_mnt propagation tree, and as such termination conditions based up on the dest_mnt mount propgation tree do not make sense. To avoid other kinds of confusion last_dest is not changed when computing last_source. last_dest is only used once in propagate_one and that is above the point of the code being modified, so changing the global variable is meaningless and confusing. Cc: stable@vger.kernel.org fixes: f2ebb3a921c1ca1e2ddd9242e95a1989a50c4c68 ("smarter propagate_mnt()") Reported-by: Tycho Andersen <tycho.andersen@canonical.com> Reviewed-by: Seth Forshee <seth.forshee@canonical.com> Tested-by: Seth Forshee <seth.forshee@canonical.com> Signed-off-by: "Eric W. Biederman" <ebiederm@xmission.com>
* fs/pnode.c: treat zero mnt_group_id-s as unequalMaxim Patlasov2016-02-201-2/+7
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | propagate_one(m) calculates "type" argument for copy_tree() like this: > if (m->mnt_group_id == last_dest->mnt_group_id) { > type = CL_MAKE_SHARED; > } else { > type = CL_SLAVE; > if (IS_MNT_SHARED(m)) > type |= CL_MAKE_SHARED; > } The "type" argument then governs clone_mnt() behavior with respect to flags and mnt_master of new mount. When we iterate through a slave group, it is possible that both current "m" and "last_dest" are not shared (although, both are slaves, i.e. have non-NULL mnt_master-s). Then the comparison above erroneously makes new mount shared and sets its mnt_master to last_source->mnt_master. The patch fixes the problem by handling zero mnt_group_id-s as though they are unequal. The similar problem exists in the implementation of "else" clause above when we have to ascend upward in the master/slave tree by calling: > last_source = last_source->mnt_master; > last_dest = last_source->mnt_parent; proper number of times. The last step is governed by "n->mnt_group_id != last_dest->mnt_group_id" condition that may lie if both are zero. The patch fixes this case in the same way as the former one. [AV: don't open-code an obvious helper...] Signed-off-by: Maxim Patlasov <mpatlasov@virtuozzo.com> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
* mnt: Don't propagate unmounts to locked mountsEric W. Biederman2015-04-021-3/+29
| | | | | | | | | | | | | | | | If the first mount in shared subtree is locked don't unmount the shared subtree. This is ensured by walking through the mounts parents before children and marking a mount as unmountable if it is not locked or it is locked but it's parent is marked. This allows recursive mount detach to propagate through a set of mounts when unmounting them would not reveal what is under any locked mount. Cc: stable@vger.kernel.org Signed-off-by: "Eric W. Biederman" <ebiederm@xmission.com>
* mnt: On an unmount propagate clearing of MNT_LOCKEDEric W. Biederman2015-04-021-0/+20
| | | | | | | | | | | | | | A prerequisite of calling umount_tree is that the point where the tree is mounted at is valid to unmount. If we are propagating the effect of the unmount clear MNT_LOCKED in every instance where the same filesystem is mounted on the same mountpoint in the mount tree, as we know (by virtue of the fact that umount_tree was called) that it is safe to reveal what is at that mountpoint. Cc: stable@vger.kernel.org Signed-off-by: "Eric W. Biederman" <ebiederm@xmission.com>
* mnt: Delay removal from the mount hash.Eric W. Biederman2015-04-021-1/+0
| | | | | | | | | | | | | | | | - Modify __lookup_mnt_hash_last to ignore mounts that have MNT_UMOUNTED set. - Don't remove mounts from the mount hash table in propogate_umount - Don't remove mounts from the mount hash table in umount_tree before the entire list of mounts to be umounted is selected. - Remove mounts from the mount hash table as the last thing that happens in the case where a mount has a parent in umount_tree. Mounts without parents are not hashed (by definition). This paves the way for delaying removal from the mount hash table even farther and fixing the MNT_LOCKED vs MNT_DETACH issue. Cc: stable@vger.kernel.org Signed-off-by: "Eric W. Biederman" <ebiederm@xmission.com>
* mnt: Add MNT_UMOUNT flagEric W. Biederman2015-04-021-0/+1
| | | | | | | | | | | In some instances it is necessary to know if the the unmounting process has begun on a mount. Add MNT_UMOUNT to make that reliably testable. This fix gets used in fixing locked mounts in MNT_DETACH Cc: stable@vger.kernel.org Signed-off-by: "Eric W. Biederman" <ebiederm@xmission.com>
* mnt: In umount_tree reuse mnt_list instead of mnt_hashEric W. Biederman2015-04-021-3/+3
| | | | | | | | | | | | | umount_tree builds a list of mounts that need to be unmounted. Utilize mnt_list for this purpose instead of mnt_hash. This begins to allow keeping a mount on the mnt_hash after it is unmounted, which is necessary for a properly functioning MNT_LOCKED implementation. The fact that mnt_list is an ordinary list makding available list_move is nice bonus. Cc: stable@vger.kernel.org Signed-off-by: "Eric W. Biederman" <ebiederm@xmission.com>
* mnt: Move the clear of MNT_LOCKED from copy_tree to it's callers.Eric W. Biederman2014-12-021-0/+1
| | | | | | | | | | | | | | Clear MNT_LOCKED in the callers of copy_tree except copy_mnt_ns, and collect_mounts. In copy_mnt_ns it is necessary to create an exact copy of a mount tree, so not clearing MNT_LOCKED is important. Similarly collect_mounts is used to take a snapshot of the mount tree for audit logging purposes and auditing using a faithful copy of the tree is important. This becomes particularly significant when we start setting MNT_LOCKED on rootfs to prevent it from being unmounted. Signed-off-by: "Eric W. Biederman" <ebiederm@xmission.com>
* get rid of propagate_umount() mistakenly treating slaves as busy.Al Viro2014-08-301-0/+1
| | | | | | | | | The check in __propagate_umount() ("has somebody explicitly mounted something on that slave?") is done *before* taking the already doomed victims out of the child lists. Cc: stable@vger.kernel.org Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
* smarter propagate_mnt()Al Viro2014-04-011-79/+119
| | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | The current mainline has copies propagated to *all* nodes, then tears down the copies we made for nodes that do not contain counterparts of the desired mountpoint. That sets the right propagation graph for the copies (at teardown time we move the slaves of removed node to a surviving peer or directly to master), but we end up paying a fairly steep price in useless allocations. It's fairly easy to create a situation where N calls of mount(2) create exactly N bindings, with O(N^2) vfsmounts allocated and freed in process. Fortunately, it is possible to avoid those allocations/freeings. The trick is to create copies in the right order and find which one would've eventually become a master with the current algorithm. It turns out to be possible in O(nodes getting propagation) time and with no extra allocations at all. One part is that we need to make sure that eventual master will be created before its slaves, so we need to walk the propagation tree in a different order - by peer groups. And iterate through the peers before dealing with the next group. Another thing is finding the (earlier) copy that will be a master of one we are about to create; to do that we are (temporary) marking the masters of mountpoints we are attaching the copies to. Either we are in a peer of the last mountpoint we'd dealt with, or we have the following situation: we are attaching to mountpoint M, the last copy S_0 had been attached to M_0 and there are sequences S_0...S_n, M_0...M_n such that S_{i+1} is a master of S_{i}, S_{i} mounted on M{i} and we need to create a slave of the first S_{k} such that M is getting propagation from M_{k}. It means that the master of M_{k} will be among the sequence of masters of M. On the other hand, the nearest marked node in that sequence will either be the master of M_{k} or the master of M_{k-1} (the latter - in the case if M_{k-1} is a slave of something M gets propagation from, but in a wrong peer group). So we go through the sequence of masters of M until we find a marked one (P). Let N be the one before it. Then we go through the sequence of masters of S_0 until we find one (say, S) mounted on a node D that has P as master and check if D is a peer of N. If it is, S will be the master of new copy, if not - the master of S will be. That's it for the hard part; the rest is fairly simple. Iterator is in next_group(), handling of one prospective mountpoint is propagate_one(). It seems to survive all tests and gives a noticably better performance than the current mainline for setups that are seriously using shared subtrees. Cc: stable@vger.kernel.org Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
* switch mnt_hash to hlistAl Viro2014-03-301-11/+15
| | | | | | | | | | | | | fixes RCU bug - walking through hlist is safe in face of element moves, since it's self-terminating. Cyclic lists are not - if we end up jumping to another hash chain, we'll loop infinitely without ever hitting the original list head. [fix for dumb braino folded] Spotted by: Max Kellermann <mk@cm4all.com> Cc: stable@vger.kernel.org Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
* split __lookup_mnt() in two functionsAl Viro2013-10-241-3/+3
| | | | | | | | Instead of passing the direction as argument (and checking it on every step through the hash chain), just have separate __lookup_mnt() and __lookup_mnt_last(). And use the standard iterators... Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
* new helpers: lock_mount_hash/unlock_mount_hashAl Viro2013-10-241-2/+2
| | | | | | | aka br_write_{lock,unlock} of vfsmount_lock. Inlines in fs/mount.h, vfsmount_lock extern moved over there as well. Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
* namespace.c: get rid of mnt_ghostsAl Viro2013-10-241-2/+1
| | | | Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
* vfs: Fix invalid ida_remove() callTakashi Iwai2013-05-311-1/+2
| | | | | | | | | | | | | When the group id of a shared mount is not allocated, the umount still tries to call mnt_release_group_id(), which eventually hits a kernel warning at ida_remove() spewing a message like: ida_remove called for id=0 which is not allocated. This patch fixes the bug simply checking the group id in the caller. Reported-by: Cristian Rodríguez <crrodriguez@opensuse.org> Signed-off-by: Takashi Iwai <tiwai@suse.de> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
* Merge branch 'for-linus' of ↵Linus Torvalds2013-05-011-6/+4
|\ | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | | git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs Pull VFS updates from Al Viro, Misc cleanups all over the place, mainly wrt /proc interfaces (switch create_proc_entry to proc_create(), get rid of the deprecated create_proc_read_entry() in favor of using proc_create_data() and seq_file etc). 7kloc removed. * 'for-linus' of git://git.kernel.org/pub/scm/linux/kernel/git/viro/vfs: (204 commits) don't bother with deferred freeing of fdtables proc: Move non-public stuff from linux/proc_fs.h to fs/proc/internal.h proc: Make the PROC_I() and PDE() macros internal to procfs proc: Supply a function to remove a proc entry by PDE take cgroup_open() and cpuset_open() to fs/proc/base.c ppc: Clean up scanlog ppc: Clean up rtas_flash driver somewhat hostap: proc: Use remove_proc_subtree() drm: proc: Use remove_proc_subtree() drm: proc: Use minor->index to label things, not PDE->name drm: Constify drm_proc_list[] zoran: Don't print proc_dir_entry data in debug reiserfs: Don't access the proc_dir_entry in r_open(), r_start() r_show() proc: Supply an accessor for getting the data from a PDE's parent airo: Use remove_proc_subtree() rtl8192u: Don't need to save device proc dir PDE rtl8187se: Use a dir under /proc/net/r8180/ proc: Add proc_mkdir_data() proc: Move some bits from linux/proc_fs.h to linux/{of.h,signal.h,tty.h} proc: Move PDE_NET() to fs/proc/proc_net.c ...
| * switch unlock_mount() to namespace_unlock(), convert all umount_tree() callersAl Viro2013-04-091-3/+1
| | | | | | | | | | | | | | which allows to kill the last argument of umount_tree() and make release_mounts() static. Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
| * get rid of full-hash scan on detaching vfsmountsAl Viro2013-04-091-3/+3
| | | | | | | | Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
* | vfs: Carefully propogate mounts across user namespacesEric W. Biederman2013-03-271-0/+6
|/ | | | | | | | | | | | | | | | | | | | | As a matter of policy MNT_READONLY should not be changable if the original mounter had more privileges than creator of the mount namespace. Add the flag CL_UNPRIVILEGED to note when we are copying a mount from a mount namespace that requires more privileges to a mount namespace that requires fewer privileges. When the CL_UNPRIVILEGED flag is set cause clone_mnt to set MNT_NO_REMOUNT if any of the mnt flags that should never be changed are set. This protects both mount propagation and the initial creation of a less privileged mount namespace. Cc: stable@vger.kernel.org Acked-by: Serge Hallyn <serge.hallyn@canonical.com> Reported-by: Andy Lutomirski <luto@amacapital.net> Signed-off-by: "Eric W. Biederman" <ebiederm@xmission.com>
* VFS: Make clone_mnt()/copy_tree()/collect_mounts() return errorsDavid Howells2012-07-141-2/+3
| | | | | | | | | | | | | | | | | | copy_tree() can theoretically fail in a case other than ENOMEM, but always returns NULL which is interpreted by callers as -ENOMEM. Change it to return an explicit error. Also change clone_mnt() for consistency and because union mounts will add new error cases. Thanks to Andreas Gruenbacher <agruen@suse.de> for a bug fix. [AV: folded braino fix by Dan Carpenter] Original-author: Valerie Aurora <vaurora@redhat.com> Signed-off-by: David Howells <dhowells@redhat.com> Cc: Valerie Aurora <valerie.aurora@gmail.com> Cc: Andreas Gruenbacher <agruen@suse.de> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
* brlocks/lglocks: API cleanupsAndi Kleen2012-05-291-2/+2
| | | | | | | | | | | | | | | | | | | | | lglocks and brlocks are currently generated with some complicated macros in lglock.h. But there's no reason to not just use common utility functions and put all the data into a common data structure. In preparation, this patch changes the API to look more like normal function calls with pointers, not magic macros. The patch is rather large because I move over all users in one go to keep it bisectable. This impacts the VFS somewhat in terms of lines changed. But no actual behaviour change. [akpm@linux-foundation.org: checkpatch fixes] Signed-off-by: Andi Kleen <ak@linux.intel.com> Cc: Al Viro <viro@zeniv.linux.org.uk> Cc: Rusty Russell <rusty@rustcorp.com.au> Signed-off-by: Andrew Morton <akpm@linux-foundation.org> Signed-off-by: Rusty Russell <rusty@rustcorp.com.au> Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
* vfs: switch pnode.h macros to struct mount *Al Viro2012-01-031-3/+3
| | | | Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
* vfs: move the rest of int fields to struct mountAl Viro2012-01-031-1/+1
| | | | Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
* vfs: mnt_id/mnt_group_id movedAl Viro2012-01-031-2/+2
| | | | Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
* vfs: mnt_ns moved to struct mountAl Viro2012-01-031-5/+5
| | | | Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
* vfs: take mnt_share/mnt_slave/mnt_slave_list and mnt_expire to struct mountAl Viro2012-01-031-15/+15
| | | | Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
* vfs: and now we can make ->mnt_master point to struct mountAl Viro2012-01-031-9/+9
| | | | Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
* vfs: take mnt_master to struct mountAl Viro2012-01-031-13/+13
| | | | | | make IS_MNT_SLAVE take struct mount * at the same time Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
* vfs: spread struct mount - remaining argument of mnt_set_mountpoint()Al Viro2012-01-031-1/+1
| | | | Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
* vfs: spread struct mount - propagate_mnt()Al Viro2012-01-031-6/+6
| | | | Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
* vfs: spread struct mount - shared subtree iteratorsAl Viro2012-01-031-50/+47
| | | | Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
* vfs: spread struct mount - get_dominating_id / do_make_slaveAl Viro2012-01-031-29/+29
| | | | | | | next pile of horrors, similar to mnt_parent one; this time it's mnt_master. Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
* vfs: take mnt_child/mnt_mounts to struct mountAl Viro2012-01-031-3/+3
| | | | Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
* vfs: spread struct mount - work with countersAl Viro2012-01-031-1/+1
| | | | Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
* vfs: move mnt_mountpoint to struct mountAl Viro2012-01-031-2/+2
| | | | Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
* vfs: now it can be done - make mnt_parent point to struct mountAl Viro2012-01-031-8/+8
| | | | Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
* vfs: mnt_parent moved to struct mountAl Viro2012-01-031-2/+2
| | | | | | the second victim... Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>
* vfs: spread struct mount - is_path_reachableAl Viro2012-01-031-5/+5
| | | | Signed-off-by: Al Viro <viro@zeniv.linux.org.uk>