linux-stable.git, branch v5.4.9

Linux 5.4.9

2020-01-09T09:20:08+00:00

mm/hugetlb: defer freeing of huge pages if in non-task context

2020-01-09T09:20:07+00:00

[ Upstream commit c77c0a8ac4c522638a8242fcb9de9496e3cdbb2d ]

The following lockdep splat was observed when a certain hugetlbfs test
was run:

  ================================
  WARNING: inconsistent lock state
  4.18.0-159.el8.x86_64+debug #1 Tainted: G        W --------- -  -
  --------------------------------
  inconsistent {SOFTIRQ-ON-W} -> {IN-SOFTIRQ-W} usage.
  swapper/30/0 [HC0[0]:SC1[1]:HE1:SE0] takes:
  ffffffff9acdc038 (hugetlb_lock){+.?.}, at: free_huge_page+0x36f/0xaa0
  {SOFTIRQ-ON-W} state was registered at:
    lock_acquire+0x14f/0x3b0
    _raw_spin_lock+0x30/0x70
    __nr_hugepages_store_common+0x11b/0xb30
    hugetlb_sysctl_handler_common+0x209/0x2d0
    proc_sys_call_handler+0x37f/0x450
    vfs_write+0x157/0x460
    ksys_write+0xb8/0x170
    do_syscall_64+0xa5/0x4d0
    entry_SYSCALL_64_after_hwframe+0x6a/0xdf
  irq event stamp: 691296
  hardirqs last  enabled at (691296): [] _raw_spin_unlock_irqrestore+0x4b/0x60
  hardirqs last disabled at (691295): [] _raw_spin_lock_irqsave+0x22/0x81
  softirqs last  enabled at (691284): [] irq_enter+0xc3/0xe0
  softirqs last disabled at (691285): [] irq_exit+0x23e/0x2b0

  other info that might help us debug this:
   Possible unsafe locking scenario:

         CPU0
         ----
    lock(hugetlb_lock);
    
      lock(hugetlb_lock);

   *** DEADLOCK ***
      :
  Call Trace:
   
   __lock_acquire+0x146b/0x48c0
   lock_acquire+0x14f/0x3b0
   _raw_spin_lock+0x30/0x70
   free_huge_page+0x36f/0xaa0
   bio_check_pages_dirty+0x2fc/0x5c0
   clone_endio+0x17f/0x670 [dm_mod]
   blk_update_request+0x276/0xe50
   scsi_end_request+0x7b/0x6a0
   scsi_io_completion+0x1c6/0x1570
   blk_done_softirq+0x22e/0x350
   __do_softirq+0x23d/0xad8
   irq_exit+0x23e/0x2b0
   do_IRQ+0x11a/0x200
   common_interrupt+0xf/0xf
   

Both the hugetbl_lock and the subpool lock can be acquired in
free_huge_page().  One way to solve the problem is to make both locks
irq-safe.  However, Mike Kravetz had learned that the hugetlb_lock is
held for a linear scan of ALL hugetlb pages during a cgroup reparentling
operation.  So it is just too long to have irq disabled unless we can
break hugetbl_lock down into finer-grained locks with shorter lock hold
times.

Another alternative is to defer the freeing to a workqueue job.  This
patch implements the deferred freeing by adding a free_hpage_workfn()
work function to do the actual freeing.  The free_huge_page() call in a
non-task context saves the page to be freed in the hpage_freelist linked
list in a lockless manner using the llist APIs.

The generic workqueue is used to process the work, but a dedicated
workqueue can be used instead if it is desirable to have the huge page
freed ASAP.

Thanks to Kirill Tkhai  for suggesting the use of
llist APIs which simplfy the code.

Link: http://lkml.kernel.org/r/20191217170331.30893-1-longman@redhat.com
Signed-off-by: Waiman Long 
Reviewed-by: Mike Kravetz 
Acked-by: Davidlohr Bueso 
Acked-by: Michal Hocko 
Reviewed-by: Kirill Tkhai 
Cc: Aneesh Kumar K.V 
Cc: Matthew Wilcox 
Cc: Andi Kleen 
Signed-off-by: Andrew Morton 
Signed-off-by: Linus Torvalds 
Signed-off-by: Sasha Levin

hsr: fix a race condition in node list insertion and deletion

2020-01-09T09:20:07+00:00

[ Upstream commit 92a35678ec075100ce666a2fb6969151affb0e5d ]

hsr nodes are protected by RCU and there is no write side lock.
But node insertions and deletions could be being operated concurrently.
So write side locking is needed.

Test commands:
    ip netns add nst
    ip link add veth0 type veth peer name veth1
    ip link add veth2 type veth peer name veth3
    ip link set veth1 netns nst
    ip link set veth3 netns nst
    ip link set veth0 up
    ip link set veth2 up
    ip link add hsr0 type hsr slave1 veth0 slave2 veth2
    ip a a 192.168.100.1/24 dev hsr0
    ip link set hsr0 up
    ip netns exec nst ip link set veth1 up
    ip netns exec nst ip link set veth3 up
    ip netns exec nst ip link add hsr1 type hsr slave1 veth1 slave2 veth3
    ip netns exec nst ip a a 192.168.100.2/24 dev hsr1
    ip netns exec nst ip link set hsr1 up

    for i in {0..9}
    do
        for j in {0..9}
	do
	    for k in {0..9}
	    do
	        for l in {0..9}
		do
	        arping 192.168.100.2 -I hsr0 -s 00:01:3$i:4$j:5$k:6$l -c1 &
		done
	    done
	done
    done

Splat looks like:
[  236.066091][ T3286] list_add corruption. next->prev should be prev (ffff8880a5940300), but was ffff8880a5940d0.
[  236.069617][ T3286] ------------[ cut here ]------------
[  236.070545][ T3286] kernel BUG at lib/list_debug.c:25!
[  236.071391][ T3286] invalid opcode: 0000 [#1] SMP DEBUG_PAGEALLOC KASAN PTI
[  236.072343][ T3286] CPU: 0 PID: 3286 Comm: arping Tainted: G        W         5.5.0-rc1+ #209
[  236.073463][ T3286] Hardware name: innotek GmbH VirtualBox/VirtualBox, BIOS VirtualBox 12/01/2006
[  236.074695][ T3286] RIP: 0010:__list_add_valid+0x74/0xd0
[  236.075499][ T3286] Code: 48 39 da 75 27 48 39 f5 74 36 48 39 dd 74 31 48 83 c4 08 b8 01 00 00 00 5b 5d c3 48 b
[  236.078277][ T3286] RSP: 0018:ffff8880aaa97648 EFLAGS: 00010286
[  236.086991][ T3286] RAX: 0000000000000075 RBX: ffff8880d4624c20 RCX: 0000000000000000
[  236.088000][ T3286] RDX: 0000000000000075 RSI: 0000000000000008 RDI: ffffed1015552ebf
[  236.098897][ T3286] RBP: ffff88809b53d200 R08: ffffed101b3c04f9 R09: ffffed101b3c04f9
[  236.099960][ T3286] R10: 00000000308769a1 R11: ffffed101b3c04f8 R12: ffff8880d4624c28
[  236.100974][ T3286] R13: ffff8880d4624c20 R14: 0000000040310100 R15: ffff8880ce17ee02
[  236.138967][ T3286] FS:  00007f23479fa680(0000) GS:ffff8880d9c00000(0000) knlGS:0000000000000000
[  236.144852][ T3286] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[  236.145720][ T3286] CR2: 00007f4a14bab210 CR3: 00000000a61c6001 CR4: 00000000000606f0
[  236.146776][ T3286] Call Trace:
[  236.147222][ T3286]  hsr_add_node+0x314/0x490 [hsr]
[  236.153633][ T3286]  hsr_forward_skb+0x2b6/0x1bc0 [hsr]
[  236.154362][ T3286]  ? rcu_read_lock_sched_held+0x90/0xc0
[  236.155091][ T3286]  ? rcu_read_lock_bh_held+0xa0/0xa0
[  236.156607][ T3286]  hsr_dev_xmit+0x70/0xd0 [hsr]
[  236.157254][ T3286]  dev_hard_start_xmit+0x160/0x740
[  236.157941][ T3286]  __dev_queue_xmit+0x1961/0x2e10
[  236.158565][ T3286]  ? netdev_core_pick_tx+0x2e0/0x2e0
[ ... ]

Reported-by: syzbot+3924327f9ad5f4d2b343@syzkaller.appspotmail.com
Fixes: f421436a591d ("net/hsr: Add support for the High-availability Seamless Redundancy protocol (HSRv0)")
Signed-off-by: Taehee Yoo 
Signed-off-by: David S. Miller 
Signed-off-by: Sasha Levin

hsr: fix error handling routine in hsr_dev_finalize()

2020-01-09T09:20:07+00:00

[ Upstream commit 1d19e2d53e8ed9e4c98fc95e0067492cda7288b0 ]

hsr_dev_finalize() is called to create new hsr interface.
There are some wrong error handling codes.

1. wrong checking return value of debugfs_create_{dir/file}.
These function doesn't return NULL. If error occurs in there,
it returns error pointer.
So, it should check error pointer instead of NULL.

2. It doesn't unregister interface if it fails to setup hsr interface.
If it fails to initialize hsr interface after register_netdevice(),
it should call unregister_netdevice().

3. Ignore failure of creation of debugfs
If creating of debugfs dir and file is failed, creating hsr interface
will be failed. But debugfs doesn't affect actual logic of hsr module.
So, ignoring this is more correct and this behavior is more general.

Fixes: c5a759117210 ("net/hsr: Use list_head (and rcu) instead of array for slave devices.")
Signed-off-by: Taehee Yoo 
Signed-off-by: David S. Miller 
Signed-off-by: Sasha Levin

hsr: avoid debugfs warning message when module is remove

2020-01-09T09:20:07+00:00

[ Upstream commit 84bb59d773853bc2dda2ac1ef8474c40eb33a3c6 ]

When hsr module is being removed, debugfs_remove() is called to remove
both debugfs directory and file.

When module is being removed, module state is changed to
MODULE_STATE_GOING then exit() is called.
At this moment, module couldn't be held so try_module_get()
will be failed.

debugfs's open() callback tries to hold the module if .owner is existing.
If it fails, warning message is printed.

CPU0				CPU1
delete_module()
    try_stop_module()
    hsr_exit()			open() <-- WARNING
        debugfs_remove()

In order to avoid the warning message, this patch makes hsr module does
not set .owner. Unsetting .owner is safe because these are protected by
inode_lock().

Test commands:
    #SHELL1
    ip link add dummy0 type dummy
    ip link add dummy1 type dummy
    while :
    do
        ip link add hsr0 type hsr slave1 dummy0 slave2 dummy1
	modprobe -rv hsr
    done

    #SHELL2
    while :
    do
        cat /sys/kernel/debug/hsr0/node_table
    done

Splat looks like:
[  101.223783][ T1271] ------------[ cut here ]------------
[  101.230309][ T1271] debugfs file owner did not clean up at exit: node_table
[  101.230380][ T1271] WARNING: CPU: 3 PID: 1271 at fs/debugfs/file.c:309 full_proxy_open+0x10f/0x650
[  101.233153][ T1271] Modules linked in: hsr(-) dummy veth openvswitch nsh nf_conncount nf_nat nf_conntrack nf_d]
[  101.237112][ T1271] CPU: 3 PID: 1271 Comm: cat Tainted: G        W         5.5.0-rc1+ #204
[  101.238270][ T1271] Hardware name: innotek GmbH VirtualBox/VirtualBox, BIOS VirtualBox 12/01/2006
[  101.240379][ T1271] RIP: 0010:full_proxy_open+0x10f/0x650
[  101.241166][ T1271] Code: 48 c1 ea 03 80 3c 02 00 0f 85 c1 04 00 00 49 8b 3c 24 e8 04 86 7e ff 84 c0 75 2d 4c 8
[  101.251985][ T1271] RSP: 0018:ffff8880ca22fa38 EFLAGS: 00010286
[  101.273355][ T1271] RAX: dffffc0000000008 RBX: ffff8880cc6e6200 RCX: 0000000000000000
[  101.274466][ T1271] RDX: 0000000000000000 RSI: 0000000000000006 RDI: ffff8880c4dd5c14
[  101.275581][ T1271] RBP: 0000000000000000 R08: fffffbfff2922f5d R09: 0000000000000000
[  101.276733][ T1271] R10: 0000000000000001 R11: 0000000000000000 R12: ffffffffc0551bc0
[  101.277853][ T1271] R13: ffff8880c4059a48 R14: ffff8880be50a5e0 R15: ffffffff941adaa0
[  101.278956][ T1271] FS:  00007f8871cda540(0000) GS:ffff8880da800000(0000) knlGS:0000000000000000
[  101.280216][ T1271] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
[  101.282832][ T1271] CR2: 00007f88717cfd10 CR3: 00000000b9440005 CR4: 00000000000606e0
[  101.283974][ T1271] Call Trace:
[  101.285328][ T1271]  do_dentry_open+0x63c/0xf50
[  101.286077][ T1271]  ? open_proxy_open+0x270/0x270
[  101.288271][ T1271]  ? __x64_sys_fchdir+0x180/0x180
[  101.288987][ T1271]  ? inode_permission+0x65/0x390
[  101.289682][ T1271]  path_openat+0x701/0x2810
[  101.290294][ T1271]  ? path_lookupat+0x880/0x880
[  101.290957][ T1271]  ? check_chain_key+0x236/0x5d0
[  101.291676][ T1271]  ? __lock_acquire+0xdfe/0x3de0
[  101.292358][ T1271]  ? sched_clock+0x5/0x10
[  101.292962][ T1271]  ? sched_clock_cpu+0x18/0x170
[  101.293644][ T1271]  ? find_held_lock+0x39/0x1d0
[  101.305616][ T1271]  do_filp_open+0x17a/0x270
[  101.306061][ T1271]  ? may_open_dev+0xc0/0xc0
[ ... ]

Fixes: fc4ecaeebd26 ("net: hsr: add debugfs support for display node list")
Signed-off-by: Taehee Yoo 
Signed-off-by: David S. Miller 
Signed-off-by: Sasha Levin

net: annotate lockless accesses to sk->sk_pacing_shift

2020-01-09T09:20:07+00:00

[ Upstream commit 7c68fa2bddda6d942bd387c9ba5b4300737fd991 ]

sk->sk_pacing_shift can be read and written without lock
synchronization. This patch adds annotations to
document this fact and avoid future syzbot complains.

This might also avoid unexpected false sharing
in sk_pacing_shift_update(), as the compiler
could remove the conditional check and always
write over sk->sk_pacing_shift :

if (sk->sk_pacing_shift != val)
	sk->sk_pacing_shift = val;

Signed-off-by: Eric Dumazet 
Signed-off-by: David S. Miller 
Signed-off-by: Sasha Levin

perf/x86/intel/bts: Fix the use of page_private()

2020-01-09T09:20:07+00:00

[ Upstream commit ff61541cc6c1962957758ba433c574b76f588d23 ]

Commit

  8062382c8dbe2 ("perf/x86/intel/bts: Add BTS PMU driver")

brought in a warning with the BTS buffer initialization
that is easily tripped with (assuming KPTI is disabled):

instantly throwing:

> ------------[ cut here ]------------
> WARNING: CPU: 2 PID: 326 at arch/x86/events/intel/bts.c:86 bts_buffer_setup_aux+0x117/0x3d0
> Modules linked in:
> CPU: 2 PID: 326 Comm: perf Not tainted 5.4.0-rc8-00291-gceb9e77324fa #904
> RIP: 0010:bts_buffer_setup_aux+0x117/0x3d0
> Call Trace:
>  rb_alloc_aux+0x339/0x550
>  perf_mmap+0x607/0xc70
>  mmap_region+0x76b/0xbd0
...

It appears to assume (for lost raisins) that PagePrivate() is set,
while later it actually tests for PagePrivate() before using
page_private().

Make it consistent and always check PagePrivate() before using
page_private().

Fixes: 8062382c8dbe2 ("perf/x86/intel/bts: Add BTS PMU driver")
Signed-off-by: Alexander Shishkin 
Signed-off-by: Peter Zijlstra (Intel) 
Cc: Jiri Olsa 
Cc: Vince Weaver 
Cc: Ingo Molnar 
Cc: Arnaldo Carvalho de Melo 
Link: https://lkml.kernel.org/r/20191205142853.28894-2-alexander.shishkin@linux.intel.com
Signed-off-by: Sasha Levin

efi: Don't attempt to map RCI2 config table if it doesn't exist

2020-01-09T09:20:07+00:00

[ Upstream commit a470552ee8965da0fe6fd4df0aa39c4cda652c7c ]

Commit:

  1c5fecb61255aa12 ("efi: Export Runtime Configuration Interface table to sysfs")

... added support for a Dell specific UEFI configuration table, but
failed to take into account that mapping the table should not be
attempted unless the table actually exists. If it doesn't exist,
the code usually fails silently unless pr_debug() prints are
enabled. However, on 32-bit PAE x86, the splat below is produced due
to the attempt to map the placeholder value EFI_INVALID_TABLE_ADDR
which we use for non-existing UEFI configuration tables, and which
equals ULONG_MAX.

   memremap attempted on mixed range 0x00000000ffffffff size: 0x1e
   WARNING: CPU: 1 PID: 1 at kernel/iomem.c:81 memremap+0x1a3/0x1c0
   Modules linked in:
   CPU: 1 PID: 1 Comm: swapper/0 Not tainted 5.4.2-smp-mine #1
   Hardware name: Hewlett-Packard HP Z400 Workstation/0B4Ch, BIOS 786G3 v03.61 03/05/2018
   EIP: memremap+0x1a3/0x1c0
  ...
   Call Trace:
    ? map_properties+0x473/0x473
    ? efi_rci2_sysfs_init+0x2c/0x154
    ? map_properties+0x473/0x473
    ? do_one_initcall+0x49/0x1d4
    ? parse_args+0x1e8/0x2a0
    ? do_early_param+0x7a/0x7a
    ? kernel_init_freeable+0x139/0x1c2
    ? rest_init+0x8e/0x8e
    ? kernel_init+0xd/0xf2
    ? ret_from_fork+0x2e/0x38

Fix this by checking whether the table exists before attempting to map it.

Reported-by: Richard Narron 
Tested-by: Richard Narron 
Signed-off-by: Ard Biesheuvel 
Cc: linux-efi@vger.kernel.org
Fixes: 1c5fecb61255aa12 ("efi: Export Runtime Configuration Interface table to sysfs")
Link: https://lkml.kernel.org/r/20191210090945.11501-2-ardb@kernel.org
Signed-off-by: Ingo Molnar 
Signed-off-by: Sasha Levin

lib/ubsan: don't serialize UBSAN report

2020-01-09T09:20:07+00:00

[ Upstream commit ce5c31db3645b649a31044a4d8b6057f6c723702 ]

At the moment, UBSAN report will be serialized using a spin_lock().  On
RT-systems, spinlocks are turned to rt_spin_lock and may sleep.  This
will result to the following splat if the undefined behavior is in a
context that can sleep:

  BUG: sleeping function called from invalid context at /src/linux/kernel/locking/rtmutex.c:968
  in_atomic(): 1, irqs_disabled(): 128, pid: 3447, name: make
  1 lock held by make/3447:
   #0: 000000009a966332 (&mm->mmap_sem){++++}, at: do_page_fault+0x140/0x4f8
  irq event stamp: 6284
  hardirqs last  enabled at (6283): [] _raw_spin_unlock_irqrestore+0x90/0xa0
  hardirqs last disabled at (6284): [] _raw_spin_lock_irqsave+0x30/0x78
  softirqs last  enabled at (2430): [] fpsimd_restore_current_state+0x60/0xe8
  softirqs last disabled at (2427): [] fpsimd_restore_current_state+0x28/0xe8
  Preemption disabled at:
  [] rt_mutex_futex_unlock+0x4c/0xb0
  CPU: 3 PID: 3447 Comm: make Tainted: G        W         5.2.14-rt7-01890-ge6e057589653 #911
  Call trace:
    dump_backtrace+0x0/0x148
    show_stack+0x14/0x20
    dump_stack+0xbc/0x104
    ___might_sleep+0x154/0x210
    rt_spin_lock+0x68/0xa0
    ubsan_prologue+0x30/0x68
    handle_overflow+0x64/0xe0
    __ubsan_handle_add_overflow+0x10/0x18
    __lock_acquire+0x1c28/0x2a28
    lock_acquire+0xf0/0x370
    _raw_spin_lock_irqsave+0x58/0x78
    rt_mutex_futex_unlock+0x4c/0xb0
    rt_spin_unlock+0x28/0x70
    get_page_from_freelist+0x428/0x2b60
    __alloc_pages_nodemask+0x174/0x1708
    alloc_pages_vma+0x1ac/0x238
    __handle_mm_fault+0x4ac/0x10b0
    handle_mm_fault+0x1d8/0x3b0
    do_page_fault+0x1c8/0x4f8
    do_translation_fault+0xb8/0xe0
    do_mem_abort+0x3c/0x98
    el0_da+0x20/0x24

The spin_lock() will protect against multiple CPUs to output a report
together, I guess to prevent them from being interleaved.  However, they
can still interleave with other messages (and even splat from
__might_sleep).

So the lock usefulness seems pretty limited.  Rather than trying to
accomodate RT-system by switching to a raw_spin_lock(), the lock is now
completely dropped.

Link: http://lkml.kernel.org/r/20190920100835.14999-1-julien.grall@arm.com
Signed-off-by: Julien Grall 
Reported-by: Andre Przywara 
Acked-by: Andrey Ryabinin 
Cc: Thomas Gleixner 
Cc: Sebastian Andrzej Siewior 
Cc: Steven Rostedt 
Signed-off-by: Andrew Morton 
Signed-off-by: Linus Torvalds 
Signed-off-by: Sasha Levin

xen/blkback: Avoid unmapping unmapped grant pages

2020-01-09T09:20:07+00:00

[ Upstream commit f9bd84a8a845d82f9b5a081a7ae68c98a11d2e84 ]

For each I/O request, blkback first maps the foreign pages for the
request to its local pages.  If an allocation of a local page for the
mapping fails, it should unmap every mapping already made for the
request.

However, blkback's handling mechanism for the allocation failure does
not mark the remaining foreign pages as unmapped.  Therefore, the unmap
function merely tries to unmap every valid grant page for the request,
including the pages not mapped due to the allocation failure.  On a
system that fails the allocation frequently, this problem leads to
following kernel crash.

  [  372.012538] BUG: unable to handle kernel NULL pointer dereference at 0000000000000001
  [  372.012546] IP: [] gnttab_unmap_refs.part.7+0x1c/0x40
  [  372.012557] PGD 16f3e9067 PUD 16426e067 PMD 0
  [  372.012562] Oops: 0002 [#1] SMP
  [  372.012566] Modules linked in: act_police sch_ingress cls_u32
  ...
  [  372.012746] Call Trace:
  [  372.012752]  [] gnttab_unmap_refs+0x34/0x40
  [  372.012759]  [] xen_blkbk_unmap+0x83/0x150 [xen_blkback]
  ...
  [  372.012802]  [] dispatch_rw_block_io+0x970/0x980 [xen_blkback]
  ...
  Decompressing Linux... Parsing ELF... done.
  Booting the kernel.
  [    0.000000] Initializing cgroup subsys cpuset

This commit fixes this problem by marking the grant pages of the given
request that didn't mapped due to the allocation failure as invalid.

Fixes: c6cc142dac52 ("xen-blkback: use balloon pages for all mappings")

Reviewed-by: David Woodhouse 
Reviewed-by: Maximilian Heyne 
Reviewed-by: Paul Durrant 
Reviewed-by: Roger Pau Monné 
Signed-off-by: SeongJae Park 
Signed-off-by: Jens Axboe 
Signed-off-by: Sasha Levin