proc/vmcore: fix possible deadlock on concurrent mmap and read
Lockdep noticed that there is chance for a deadlock if we have concurrent
mmap, concurrent read, and the addition/removal of a callback.
As nicely explained by Boqun:
"
Lockdep warned about the above sequences because rw_semaphore is a fair
read-write lock, and the following can cause a deadlock:
TASK 1 TASK 2 TASK 3
====== ====== ======
down_write(mmap_lock);
down_read(vmcore_cb_rwsem)
down_write(vmcore_cb_rwsem); // blocked
down_read(vmcore_cb_rwsem); // cannot get the lock because of the fairness
down_read(mmap_lock); // blocked
IOW, a reader can block another read if there is a writer queued by the
second reader and the lock is fair.
"
To fix, convert to srcu to make this deadlock impossible. We need srcu as
our callbacks can sleep. With this change, I cannot trigger any lockdep
warnings.
[ 6.386519] ======================================================
[ 6.387203] WARNING: possible circular locking dependency detected
[ 6.387965] 5.17.0-0.rc0.20220117git0c947b893d69.68.test.fc36.x86_64 #1 Not tainted
[ 6.388899] ------------------------------------------------------
[ 6.389657] makedumpfile/542 is trying to acquire lock:
[ 6.390308]
ffffffff832d2eb8 (vmcore_cb_rwsem){.+.+}-{3:3}, at: mmap_vmcore+0x340/0x580
[ 6.391290]
[ 6.391290] but task is already holding lock:
[ 6.391978]
ffff8880af226438 (&mm->mmap_lock#2){++++}-{3:3}, at: vm_mmap_pgoff+0x84/0x150
[ 6.392898]
[ 6.392898] which lock already depends on the new lock.
[ 6.392898]
[ 6.393866]
[ 6.393866] the existing dependency chain (in reverse order) is:
[ 6.394762]
[ 6.394762] -> #1 (&mm->mmap_lock#2){++++}-{3:3}:
[ 6.395530] lock_acquire+0xc3/0x1a0
[ 6.396047] __might_fault+0x4e/0x70
[ 6.396562] _copy_to_user+0x1f/0x90
[ 6.397093] __copy_oldmem_page+0x72/0xc0
[ 6.397663] read_from_oldmem+0x77/0x1e0
[ 6.398229] read_vmcore+0x2c2/0x310
[ 6.398742] proc_reg_read+0x47/0xa0
[ 6.399265] vfs_read+0x101/0x340
[ 6.399751] __x64_sys_pread64+0x5d/0xa0
[ 6.400314] do_syscall_64+0x43/0x90
[ 6.400778] entry_SYSCALL_64_after_hwframe+0x44/0xae
[ 6.401390]
[ 6.401390] -> #0 (vmcore_cb_rwsem){.+.+}-{3:3}:
[ 6.402063] validate_chain+0x9f4/0x2670
[ 6.402560] __lock_acquire+0x8f7/0xbc0
[ 6.403054] lock_acquire+0xc3/0x1a0
[ 6.403509] down_read+0x4a/0x140
[ 6.403948] mmap_vmcore+0x340/0x580
[ 6.404403] proc_reg_mmap+0x3e/0x90
[ 6.404866] mmap_region+0x504/0x880
[ 6.405322] do_mmap+0x38a/0x520
[ 6.405744] vm_mmap_pgoff+0xc1/0x150
[ 6.406258] ksys_mmap_pgoff+0x178/0x200
[ 6.406823] do_syscall_64+0x43/0x90
[ 6.407339] entry_SYSCALL_64_after_hwframe+0x44/0xae
[ 6.407975]
[ 6.407975] other info that might help us debug this:
[ 6.407975]
[ 6.408945] Possible unsafe locking scenario:
[ 6.408945]
[ 6.409684] CPU0 CPU1
[ 6.410196] ---- ----
[ 6.410703] lock(&mm->mmap_lock#2);
[ 6.411121] lock(vmcore_cb_rwsem);
[ 6.411792] lock(&mm->mmap_lock#2);
[ 6.412465] lock(vmcore_cb_rwsem);
[ 6.412873]
[ 6.412873] *** DEADLOCK ***
[ 6.412873]
[ 6.413522] 1 lock held by makedumpfile/542:
[ 6.414006] #0:
ffff8880af226438 (&mm->mmap_lock#2){++++}-{3:3}, at: vm_mmap_pgoff+0x84/0x150
[ 6.414944]
[ 6.414944] stack backtrace:
[ 6.415432] CPU: 0 PID: 542 Comm: makedumpfile Not tainted 5.17.0-0.rc0.20220117git0c947b893d69.68.test.fc36.x86_64 #1
[ 6.416581] Hardware name: Red Hat KVM, BIOS 0.5.1 01/01/2011
[ 6.417272] Call Trace:
[ 6.417593] <TASK>
[ 6.417882] dump_stack_lvl+0x5d/0x78
[ 6.418346] print_circular_bug+0x5d7/0x5f0
[ 6.418821] ? stack_trace_save+0x3a/0x50
[ 6.419273] ? save_trace+0x3d/0x330
[ 6.419681] check_noncircular+0xd1/0xe0
[ 6.420217] validate_chain+0x9f4/0x2670
[ 6.420715] ? __lock_acquire+0x8f7/0xbc0
[ 6.421234] ? __lock_acquire+0x8f7/0xbc0
[ 6.421685] __lock_acquire+0x8f7/0xbc0
[ 6.422127] lock_acquire+0xc3/0x1a0
[ 6.422535] ? mmap_vmcore+0x340/0x580
[ 6.422965] ? lock_is_held_type+0xe2/0x140
[ 6.423432] ? mmap_vmcore+0x340/0x580
[ 6.423893] down_read+0x4a/0x140
[ 6.424321] ? mmap_vmcore+0x340/0x580
[ 6.424800] mmap_vmcore+0x340/0x580
[ 6.425237] ? vm_area_alloc+0x1c/0x60
[ 6.425661] ? trace_kmem_cache_alloc+0x30/0xe0
[ 6.426174] ? kmem_cache_alloc+0x1e0/0x2f0
[ 6.426641] proc_reg_mmap+0x3e/0x90
[ 6.427052] mmap_region+0x504/0x880
[ 6.427462] do_mmap+0x38a/0x520
[ 6.427842] vm_mmap_pgoff+0xc1/0x150
[ 6.428260] ksys_mmap_pgoff+0x178/0x200
[ 6.428701] do_syscall_64+0x43/0x90
[ 6.429126] entry_SYSCALL_64_after_hwframe+0x44/0xae
[ 6.429745] RIP: 0033:0x7fc7359b8fc7
[ 6.430157] Code: 00 00 00 89 ef e8 69 b3 ff ff eb e4 e8 c2 64 01 00 66 90 f3 0f 1e fa 41 89 ca 41 f7 c1 ff 0f 00 00 75 10 b8 09 00 00 00 0f 05 <48> 3d 00 f0 ff ff 77 21 c3 48 8b 05 21 7e 0e 00 64 c7 00 16 00 00
[ 6.432147] RSP: 002b:
00007fff35b4c208 EFLAGS:
00000246 ORIG_RAX:
0000000000000009
[ 6.432970] RAX:
ffffffffffffffda RBX:
0000000000000001 RCX:
00007fc7359b8fc7
[ 6.433746] RDX:
0000000000000001 RSI:
0000000000400000 RDI:
0000000000000000
[ 6.434529] RBP:
000055a1125ecf10 R08:
0000000000000003 R09:
0000000000002000
[ 6.435310] R10:
0000000000000002 R11:
0000000000000246 R12:
0000000000002000
[ 6.436093] R13:
0000000000400000 R14:
000055a1124269e2 R15:
0000000000000000
[ 6.436887] </TASK>
Link: https://lkml.kernel.org/r/20220119193417.100385-1-david@redhat.com
Fixes:
cc5f2704c934 ("proc/vmcore: convert oldmem_pfn_is_ram callback to more generic vmcore callbacks")
Signed-off-by: David Hildenbrand <david@redhat.com>
Reported-by: Baoquan He <bhe@redhat.com>
Acked-by: Baoquan He <bhe@redhat.com>
Cc: Vivek Goyal <vgoyal@redhat.com>
Cc: Dave Young <dyoung@redhat.com>
Cc: "Paul E. McKenney" <paulmck@kernel.org>
Cc: Josh Triplett <josh@joshtriplett.org>
Cc: Peter Zijlstra <peterz@infradead.org>
Cc: Boqun Feng <boqun.feng@gmail.com>
Cc: <stable@vger.kernel.org>
Signed-off-by: Andrew Morton <akpm@linux-foundation.org>
Signed-off-by: Stephen Rothwell <sfr@canb.auug.org.au>
projects / linux-2.6-block.git / commit