mm: another VM_BUG_ON_PAGE(PageTail(page))

[Dmitry Vyukov]

Hello,

The following program triggers VM_BUG_ON_PAGE(PageTail(page)):

// autogenerated by syzkaller (http://github.com/google/syzkaller)
#include <fcntl.h>
#include <numaif.h>
#include <sys/mman.h>
#include <unistd.h>

int main()
{
int fd;

mmap((void*)0x20000000, 4096, PROT_READ|PROT_WRITE,
MAP_PRIVATE|MAP_FIXED|MAP_ANONYMOUS, -1, 0);
fd = open("/dev/sg1", O_RDONLY|O_SYNC|0x100000);
mmap((void*)0x20001000, 0x4000, PROT_READ|PROT_WRITE,
MAP_PRIVATE|MAP_FIXED, fd, 0);
mbind((void*)0x20000000, 0x4000, 0x8002, (void*)0x20002ff8, 3660,
MPOL_MF_STRICT|MPOL_MF_MOVE);
return 0;
}


page:ffffea0000f5da80 count:0 mapcount:1 mapping:dead000000000400
index:0x0 compound_mapcount: 0
flags: 0x1fffc0000000000()
page dumped because: VM_BUG_ON_PAGE(PageTail(page))
------------[ cut here ]------------
kernel BUG at mm/vmscan.c:1446!
invalid opcode: 0000 [#27] SMP DEBUG_PAGEALLOC KASAN
Modules linked in:
CPU: 1 PID: 8185 Comm: a.out Tainted: G D 4.5.0-rc1+ #300
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Bochs 01/01/2011
task: ffff88006146c740 ti: ffff88006c8f0000 task.ti: ffff88006c8f0000
RIP: 0010:[<ffffffff8169309a>] [<ffffffff8169309a>]
isolate_lru_page+0x4ea/0x6d0
RSP: 0018:ffff88006c8f7a50 EFLAGS: 00010282
RAX: ffff88006146c740 RBX: ffffea0000f5da80 RCX: 0000000000000000
RDX: 0000000000000000 RSI: 0000000000000001 RDI: ffffea0000f5dab8
RBP: ffff88006c8f7a88 R08: 0000000000000001 R09: 0000000000000000
R10: 0000000000000002 R11: 1ffffffff134edc5 R12: ffffea0000f5daa0
R13: ffffea0000f5da00 R14: ffffea0000f5da01 R15: 0000000020004000
FS: 00000000022ec880(0063) GS:ffff88003ed00000(0000) knlGS:0000000000000000
CS: 0010 DS: 0000 ES: 0000 CR0: 000000008005003b
CR2: 0000000020003000 CR3: 0000000031d0b000 CR4: 00000000000006e0
Stack:
00000001ffffffff ffffffff816e4159 ffff8800310f6018 0000000020003000
dffffc0000000000 ffffea0000f5da80 0000000020004000 ffff88006c8f7b10
ffffffff8174efdd ffffea0000000001 ffff88006c8f7c70 ffff88006c8f7de8
Call Trace:
[< inline >] migrate_page_add mm/mempolicy.c:966
[<ffffffff8174efdd>] queue_pages_pte_range+0x4ad/0x10b0 mm/mempolicy.c:552
[< inline >] walk_pmd_range mm/pagewalk.c:50
[< inline >] walk_pud_range mm/pagewalk.c:90
[< inline >] walk_pgd_range mm/pagewalk.c:116
[<ffffffff817209c3>] __walk_page_range+0x653/0xcd0 mm/pagewalk.c:204
[<ffffffff81721174>] walk_page_range+0x134/0x300 mm/pagewalk.c:281
[<ffffffff8174cd9b>] queue_pages_range+0xfb/0x130 mm/mempolicy.c:687
[<ffffffff817555e1>] do_mbind+0x2c1/0xdc0 mm/mempolicy.c:1239
[< inline >] SYSC_mbind mm/mempolicy.c:1351
[<ffffffff8175643d>] SyS_mbind+0x13d/0x150 mm/mempolicy.c:1333
[<ffffffff86653236>] entry_SYSCALL_64_fastpath+0x16/0x7a
arch/x86/entry/entry_64.S:185
Code: 89 df e8 5a 62 04 00 0f 0b e8 43 71 ed ff 4d 8d 6e ff e9 73 fb
ff ff e8 35 71 ed ff 48 c7 c6 a0 ab 79 86 48 89 df e8 36 62 04 00 <0f>
0b e8 1f 71 ed ff 4d 8d 6e ff e9 eb fb ff ff c7 45 d0 f0 ff
RIP [<ffffffff8169309a>] isolate_lru_page+0x4ea/0x6d0 mm/vmscan.c:1446
RSP <ffff88006c8f7a50>
---[ end trace 4b07319c03942d44 ]---
BUG: sleeping function called from invalid context at include/linux/sched.h:2805
in_atomic(): 1, irqs_disabled(): 0, pid: 8185, name: a.out
INFO: lockdep is turned off.
CPU: 1 PID: 8185 Comm: a.out Tainted: G D 4.5.0-rc1+ #300
Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Bochs 01/01/2011
00000000ffffffff ffff88006c8f7548 ffffffff82be118d ffff88006146c740
0000000000001ff9 0000000000000000 ffff88006c8f7570 ffffffff813cb8cb
ffff88006146c740 ffffffff867387a0 0000000000000af5 ffff88006c8f75b0
Call Trace:
[< inline >] __dump_stack lib/dump_stack.c:15
[<ffffffff82be118d>] dump_stack+0x6f/0xa2 lib/dump_stack.c:50
[<ffffffff813cb8cb>] ___might_sleep+0x27b/0x3a0 kernel/sched/core.c:7703
[<ffffffff813cba80>] __might_sleep+0x90/0x1a0 kernel/sched/core.c:7665
[< inline >] threadgroup_change_begin include/linux/sched.h:2805
[<ffffffff813830d1>] exit_signals+0x81/0x430 kernel/signal.c:2392
[<ffffffff8135c3dc>] do_exit+0x23c/0x2cb0 kernel/exit.c:701
[<ffffffff811aa28f>] oops_end+0x9f/0xd0 arch/x86/kernel/dumpstack.c:250
[<ffffffff811aa686>] die+0x46/0x60 arch/x86/kernel/dumpstack.c:316
[< inline >] do_trap_no_signal arch/x86/kernel/traps.c:205
[<ffffffff811a3b9f>] do_trap+0x18f/0x380 arch/x86/kernel/traps.c:251
[<ffffffff811a400e>] do_error_trap+0x11e/0x280 arch/x86/kernel/traps.c:290
[<ffffffff811a527b>] do_invalid_op+0x1b/0x20 arch/x86/kernel/traps.c:303
[<ffffffff86654f4e>] invalid_op+0x1e/0x30 arch/x86/entry/entry_64.S:830
[< inline >] migrate_page_add mm/mempolicy.c:966
[<ffffffff8174efdd>] queue_pages_pte_range+0x4ad/0x10b0 mm/mempolicy.c:552
[< inline >] walk_pmd_range mm/pagewalk.c:50
[< inline >] walk_pud_range mm/pagewalk.c:90
[< inline >] walk_pgd_range mm/pagewalk.c:116
[<ffffffff817209c3>] __walk_page_range+0x653/0xcd0 mm/pagewalk.c:204
[<ffffffff81721174>] walk_page_range+0x134/0x300 mm/pagewalk.c:281
[<ffffffff8174cd9b>] queue_pages_range+0xfb/0x130 mm/mempolicy.c:687
[<ffffffff817555e1>] do_mbind+0x2c1/0xdc0 mm/mempolicy.c:1239
[< inline >] SYSC_mbind mm/mempolicy.c:1351
[<ffffffff8175643d>] SyS_mbind+0x13d/0x150 mm/mempolicy.c:1333
[<ffffffff86653236>] entry_SYSCALL_64_fastpath+0x16/0x7a
arch/x86/entry/entry_64.S:185
note: a.out[8185] exited with preempt_count 1


On commit 92e963f50fc74041b5e9e744c330dca48e04f08d with Kirill recent fix:

--- a/drivers/scsi/sg.c
+++ b/drivers/scsi/sg.c
@@ -1261,7 +1261,7 @@ sg_mmap(struct file *filp, struct vm_area_struct *vma)
}

sfp->mmap_called = 1;
- vma->vm_flags |= VM_DONTEXPAND | VM_DONTDUMP;
+ vma->vm_flags |= VM_IO | VM_DONTEXPAND | VM_DONTDUMP;
vma->vm_private_data = sfp;
vma->vm_ops = &sg_mmap_vm_ops;
return 0;

[Kirill A. Shutemov]

On Thu, Jan 28, 2016 at 11:27:11AM +0100, Dmitry Vyukov wrote:
> Hello,
>
> The following program triggers VM_BUG_ON_PAGE(PageTail(page)):
>
> // autogenerated by syzkaller (http://github.com/google/syzkaller)
> #include <fcntl.h>
> #include <numaif.h>
> #include <sys/mman.h>
> #include <unistd.h>
>
> int main()
> {
> int fd;
>
> mmap((void*)0x20000000, 4096, PROT_READ|PROT_WRITE,
> MAP_PRIVATE|MAP_FIXED|MAP_ANONYMOUS, -1, 0);
> fd = open("/dev/sg1", O_RDONLY|O_SYNC|0x100000);
> mmap((void*)0x20001000, 0x4000, PROT_READ|PROT_WRITE,
> MAP_PRIVATE|MAP_FIXED, fd, 0);
> mbind((void*)0x20000000, 0x4000, 0x8002, (void*)0x20002ff8, 3660,
> MPOL_MF_STRICT|MPOL_MF_MOVE);
> return 0;
> }

I don't have sg1 in my VM. I changed it to sg0 and it doesn't trigger an
issue: mbind() returns -EINVAL as it supposed to. Hm..

--
Kirill A. Shutemov

[Dmitry Vyukov]

I've attached my config, and here is how I start qemu:

qemu-system-x86_64 -hda wheezy.img -net
user,host=10.0.2.10,hostfwd=tcp::10022-:22 -net nic -nographic -kernel
arch/x86/boot/bzImage -append "console=ttyS0 root=/dev/sda debug
earlyprintk=serial slub_debug=UZ" -enable-kvm -pidfile vm_pid -m 2G
-numa node,nodeid=0,cpus=0-1 -numa node,nodeid=1,cpus=2-3 -smp
sockets=2,cores=2,threads=1 -usb -usbdevice mouse -usbdevice tablet
-soundhw all

[Kirill A. Shutemov]

Still no luck. :-/

Could you try patch below. I want to see what vm_flags are.

diff --git a/mm/mempolicy.c b/mm/mempolicy.c
index 27d135408a22..93edf181f88a 100644
--- a/mm/mempolicy.c
+++ b/mm/mempolicy.c
@@ -548,8 +548,10 @@ retry:
goto retry;
}

- if (flags & (MPOL_MF_MOVE | MPOL_MF_MOVE_ALL))
+ if (flags & (MPOL_MF_MOVE | MPOL_MF_MOVE_ALL)) {
+ VM_BUG_ON_VMA(PageTail(page), vma);
migrate_page_add(page, qp->pagelist, flags);
+ }
}
pte_unmap_unlock(pte - 1, ptl);
cond_resched();
--
Kirill A. Shutemov

[Dmitry Vyukov]

On Thu, Jan 28, 2016 at 12:40 PM, Kirill A. Shutemov

Humm... now I cannot reproduce it with the original program as well.
But I reproduced it with another program. Please try the one below.
I've updated to 26cd83670f2f5a3d5b5514a1f7d96567cdb9558b and have few
pending fixes to mm (see below) including your VM_BUG_ON_VMA change
above.

This report contains vm_flags:


vma ffff880062082450 start 0000000020001000 end 0000000020005000
next ffff880030460a60 prev ffff880062083910 mm ffff88002f691380
prot 8000000000000025 anon_vma ffff880062a55180 vm_ops ffffffff86d5aac0
pgoff 0 file ffff880060b1f0c0 private_data ffff880061240000
flags: 0x4144073(read|write|mayread|maywrite|mayexec|io|dontexpand|account)
------------[ cut here ]------------
kernel BUG at mm/mempolicy.c:552!
invalid opcode: 0000 [#1] SMP DEBUG_PAGEALLOC KASAN
Modules linked in:
CPU: 3 PID: 11434 Comm: tail Not tainted 4.5.0-rc1+ #301

Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Bochs 01/01/2011

task: ffff88002ecddf00 ti: ffff880033ed0000 task.ti: ffff880033ed0000
RIP: 0010:[<ffffffff817502db>] [<ffffffff817502db>]
queue_pages_pte_range+0x8ab/0x10f0
RSP: 0018:ffff880033ed7a98 EFLAGS: 00010296
RAX: 0000000000000001 RBX: ffff880061e77018 RCX: 0000000000000000
RDX: 0000000000000001 RSI: 0000000000000001 RDI: ffffed00067daf29
RBP: ffff880033ed7b10 R08: 0000000000000001 R09: 0000000000000000
R10: 1ffff1000c41048a R11: 0000000000000001 R12: 0000000020003000
R13: dffffc0000000000 R14: ffffea0001b4c280 R15: 0000000020004000
FS: 00007f670c480700(0000) GS:ffff88006d700000(0000) knlGS:0000000000000000

CS: 0010 DS: 0000 ES: 0000 CR0: 000000008005003b

CR2: 0000000020003000 CR3: 00000000004e7000 CR4: 00000000000006e0
Stack:
ffffea0000000001 ffff880033ed7c70 ffffea0001b4c200 ffff880062020840
ffffed00067daf8e ffffea0001b4c2a0 0000000000000023 ffff880062082450
ffff880033ed7c60 ffff880032747800 ffffffff8174fa30 dffffc0000000000
Call Trace:

[< inline >] walk_pmd_range mm/pagewalk.c:50
[< inline >] walk_pud_range mm/pagewalk.c:90
[< inline >] walk_pgd_range mm/pagewalk.c:116

[<ffffffff817209b3>] __walk_page_range+0x653/0xcd0 mm/pagewalk.c:204
[<ffffffff81721164>] walk_page_range+0x134/0x300 mm/pagewalk.c:281
[<ffffffff8174cd8b>] queue_pages_range+0xfb/0x130 mm/mempolicy.c:689
[<ffffffff81755611>] do_mbind+0x2c1/0xdc0 mm/mempolicy.c:1241
[< inline >] SYSC_mbind mm/mempolicy.c:1353
[<ffffffff8175646d>] SyS_mbind+0x13d/0x150 mm/mempolicy.c:1335
[<ffffffff86653276>] entry_SYSCALL_64_fastpath+0x16/0x7a
arch/x86/entry/entry_64.S:185
Code: 55 98 48 8d 42 ff e9 ce fa ff ff e8 00 9f e1 ff 4c 89 f7 e8 18
fc f2 ff e9 92 fe ff ff e8 ee 9e e1 ff 48 8b 7d c0 e8 85 87 f8 ff <0f>
0b e8 de 9e e1 ff 48 89 df 48 89 f8 0f 1f 40 00 48 89 c3 48
RIP [<ffffffff817502db>] queue_pages_pte_range+0x8ab/0x10f0 mm/mempolicy.c:552
RSP <ffff880033ed7a98>
---[ end trace 3f5635b07e2902a8 ]---

BUG: sleeping function called from invalid context at include/linux/sched.h:2805

in_atomic(): 1, irqs_disabled(): 0, pid: 11434, name: tail

INFO: lockdep is turned off.

CPU: 3 PID: 11434 Comm: tail Tainted: G D 4.5.0-rc1+ #301

Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS Bochs 01/01/2011

00000000ffffffff ffff880033ed7588 ffffffff82be11cd ffff88002ecddf00
0000000000002caa 0000000000000000 ffff880033ed75b0 ffffffff813cb8cb
ffff88002ecddf00 ffffffff867387a0 0000000000000af5 ffff880033ed75f0

Call Trace:
[< inline >] __dump_stack lib/dump_stack.c:15

[<ffffffff82be11cd>] dump_stack+0x6f/0xa2 lib/dump_stack.c:50

[<ffffffff813cb8cb>] ___might_sleep+0x27b/0x3a0 kernel/sched/core.c:7703
[<ffffffff813cba80>] __might_sleep+0x90/0x1a0 kernel/sched/core.c:7665
[< inline >] threadgroup_change_begin include/linux/sched.h:2805
[<ffffffff813830d1>] exit_signals+0x81/0x430 kernel/signal.c:2392
[<ffffffff8135c3dc>] do_exit+0x23c/0x2cb0 kernel/exit.c:701
[<ffffffff811aa28f>] oops_end+0x9f/0xd0 arch/x86/kernel/dumpstack.c:250
[<ffffffff811aa686>] die+0x46/0x60 arch/x86/kernel/dumpstack.c:316
[< inline >] do_trap_no_signal arch/x86/kernel/traps.c:205
[<ffffffff811a3b9f>] do_trap+0x18f/0x380 arch/x86/kernel/traps.c:251
[<ffffffff811a400e>] do_error_trap+0x11e/0x280 arch/x86/kernel/traps.c:290
[<ffffffff811a527b>] do_invalid_op+0x1b/0x20 arch/x86/kernel/traps.c:303

[<ffffffff86654f8e>] invalid_op+0x1e/0x30 arch/x86/entry/entry_64.S:830

[< inline >] walk_pmd_range mm/pagewalk.c:50
[< inline >] walk_pud_range mm/pagewalk.c:90
[< inline >] walk_pgd_range mm/pagewalk.c:116

[<ffffffff817209b3>] __walk_page_range+0x653/0xcd0 mm/pagewalk.c:204
[<ffffffff81721164>] walk_page_range+0x134/0x300 mm/pagewalk.c:281
[<ffffffff8174cd8b>] queue_pages_range+0xfb/0x130 mm/mempolicy.c:689
[<ffffffff81755611>] do_mbind+0x2c1/0xdc0 mm/mempolicy.c:1241
[< inline >] SYSC_mbind mm/mempolicy.c:1353
[<ffffffff8175646d>] SyS_mbind+0x13d/0x150 mm/mempolicy.c:1335
[<ffffffff86653276>] entry_SYSCALL_64_fastpath+0x16/0x7a
arch/x86/entry/entry_64.S:185
note: tail[11434] exited with preempt_count 1

// autogenerated by syzkaller (http://github.com/google/syzkaller)

#include <pthread.h>
#include <stdint.h>
#include <string.h>
#include <sys/syscall.h>
#include <unistd.h>

long r[12];

void* thr(void* arg)
{
switch ((long)arg) {
case 0:
r[0] = syscall(SYS_mmap, 0x20000000ul, 0x1000ul, 0x3ul, 0x32ul,
0xfffffffffffffffful, 0x0ul);
break;
case 1:
r[2] = syscall(SYS_open, "/dev/sg1", 0x101000ul, 0);
break;
case 2:
r[3] = syscall(SYS_mmap, 0x20001000ul, 0x4000ul, 0x3ul, 0x12ul,
r[2], 0x0ul);
break;
case 3:
*(uint64_t*)0x20002ff8 = (uint64_t)0xffffffff;
r[5] = syscall(SYS_mbind, 0x20000000ul, 0x4000ul, 0x8002ul,
0x20002ff8ul, 0xe4cul, 0x3ul);
break;
case 4:
r[6] = syscall(SYS_mmap, 0x20005000ul, 0x1000ul, 0x3ul, 0x32ul,
0xfffffffffffffffful, 0x0ul);
break;
case 5:
memcpy((void*)0x20005ffd, "\x7b\x3a\x00", 3);
memcpy((void*)0x200053e8, "\x70\x6f\x73\x69\x78\x5f\x61\x63\x6c\x5f"
"\x61\x63\x63\x65\x73\x73\x70\x70\x70\x31"
"\x6b\x65\x79\x72\x69\x6e\x67\x73\x65\x6c"
"\x66\x65\x74\x68\x30\x2f\x5c\x00",
38);
memcpy((void*)0x20000ffe, "\x73\x65\x63\x75\x72\x69\x74\x79\x00",
9);
r[10] = syscall(SYS_request_key, 0x20005ffdul, 0x200053e8ul,
0x20000ffeul, 0xfffffffffffffffful, 0, 0);
break;
case 6:
r[11] =
syscall(SYS_keyctl, 0x11ul, r[10], 0x20003fbaul, 0xa9ul, 0, 0);
break;
}
return 0;
}

int main()
{
long i;
pthread_t th[7];

memset(r, -1, sizeof(r));
for (i = 0; i < 7; i++) {
pthread_create(&th[i], 0, thr, (void*)i);
usleep(10000);
}
for (i = 0; i < 7; i++) {
pthread_create(&th[i], 0, thr, (void*)i);
if (i % 2 == 0)
usleep(10000);
}
usleep(100000);
return 0;
}



diff --git a/mm/mempolicy.c b/mm/mempolicy.c
index 27d1354..93edf18 100644

--- a/mm/mempolicy.c
+++ b/mm/mempolicy.c
@@ -548,8 +548,10 @@ retry:
goto retry;
}

- if (flags & (MPOL_MF_MOVE | MPOL_MF_MOVE_ALL))
+ if (flags & (MPOL_MF_MOVE | MPOL_MF_MOVE_ALL)) {
+ VM_BUG_ON_VMA(PageTail(page), vma);
migrate_page_add(page, qp->pagelist, flags);
+ }
}
pte_unmap_unlock(pte - 1, ptl);
cond_resched();

diff --git a/mm/mmap.c b/mm/mmap.c
index 84b1262..082b8a9 100644
--- a/mm/mmap.c
+++ b/mm/mmap.c
@@ -387,8 +387,9 @@ static long vma_compute_subtree_gap(struct
vm_area_struct *vma)
}

#ifdef CONFIG_DEBUG_VM_RB
-static int browse_rb(struct rb_root *root)
+static int browse_rb(struct mm_struct *mm)
{
+ struct rb_root *root = &mm->mm_rb;
int i = 0, j, bug = 0;
struct rb_node *nd, *pn = NULL;
unsigned long prev = 0, pend = 0;
@@ -411,12 +412,14 @@ static int browse_rb(struct rb_root *root)
vma->vm_start, vma->vm_end);
bug = 1;
}
+ spin_lock(&mm->page_table_lock);
if (vma->rb_subtree_gap != vma_compute_subtree_gap(vma)) {
pr_emerg("free gap %lx, correct %lx\n",
vma->rb_subtree_gap,
vma_compute_subtree_gap(vma));
bug = 1;
}
+ spin_unlock(&mm->page_table_lock);
i++;
pn = nd;
prev = vma->vm_start;
@@ -453,12 +456,16 @@ static void validate_mm(struct mm_struct *mm)
struct vm_area_struct *vma = mm->mmap;

while (vma) {
+ struct anon_vma *anon_vma = vma->anon_vma;
struct anon_vma_chain *avc;

- vma_lock_anon_vma(vma);
- list_for_each_entry(avc, &vma->anon_vma_chain, same_vma)
- anon_vma_interval_tree_verify(avc);
- vma_unlock_anon_vma(vma);
+ if (anon_vma) {
+ anon_vma_lock_read(anon_vma);
+ list_for_each_entry(avc, &vma->anon_vma_chain, same_vma)
+ anon_vma_interval_tree_verify(avc);
+ anon_vma_unlock_read(anon_vma);
+ }
+
highest_address = vma->vm_end;
vma = vma->vm_next;
i++;
@@ -472,7 +479,7 @@ static void validate_mm(struct mm_struct *mm)
mm->highest_vm_end, highest_address);
bug = 1;
}
- i = browse_rb(&mm->mm_rb);
+ i = browse_rb(mm);
if (i != mm->map_count) {
if (i != -1)
pr_emerg("map_count %d rb %d\n", mm->map_count, i);

[Kirill A. Shutemov]

From e20cd50a8f612dbc720d31d269e748215607a0b8 Mon Sep 17 00:00:00 2001
From: "Kirill A. Shutemov" <kirill....@linux.intel.com>
Date: Fri, 29 Jan 2016 14:55:40 +0300
Subject: [PATCH 1/2] mm: fix bogus VM_BUG_ON_PAGE() in isolate_lru_page()

We don't care if there's a tail pages which is not on LRU. We are not
going to isolate them anyway.

Testcase:

#include <fcntl.h>
#include <unistd.h>
#include <stdio.h>
#include <sys/mman.h>
#include <numaif.h>

#define SIZE 0x2000

int foo;

int main()
{
int fd;
char *p;
unsigned long mask = 2;

fd = open("/dev/sg0", O_RDWR);
p = mmap(NULL, SIZE, PROT_READ | PROT_WRITE, MAP_PRIVATE, fd, 0);
/* Faultin pages */
foo = p[0] + p[0x1000];
mbind(p, SIZE, MPOL_BIND, &mask, 4, MPOL_MF_MOVE | MPOL_MF_STRICT);
return 0;
}

MPOL_MF_STRICT makes queue_pages_test_walk() ignore !vma_megratable()
and we try to queue such pages for migration. It's good question why we
ignore !vma_megratable() for MPOL_MF_STRICT, but it's subject for a
separate patch.

Signed-off-by: Kirill A. Shutemov <kirill....@linux.intel.com>
Reported-by: Dmitry Vyukov <dvy...@google.com>
Fixes: bb5b8589767a ("mm: make sure isolate_lru_page() is never called for tail page")
---
mm/vmscan.c | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/mm/vmscan.c b/mm/vmscan.c
index eb3dd37ccd7c..492fbe73420b 100644
--- a/mm/vmscan.c
+++ b/mm/vmscan.c
@@ -1443,7 +1443,7 @@ int isolate_lru_page(struct page *page)
int ret = -EBUSY;

VM_BUG_ON_PAGE(!page_count(page), page);
- VM_BUG_ON_PAGE(PageTail(page), page);
+ VM_BUG_ON_PAGE(PageLRU(page) && PageTail(page), page);

if (PageLRU(page)) {
struct zone *zone = page_zone(page);
--
2.7.0.rc3

[Kirill A. Shutemov]

From 691a961bb401c5815ed741dac63591efbc6027e3 Mon Sep 17 00:00:00 2001

From: "Kirill A. Shutemov" <kirill....@linux.intel.com>

Date: Fri, 29 Jan 2016 15:06:17 +0300
Subject: [PATCH 2/2] mempolicy: do not try to queue pages from
!vma_migratable()

Maybe I miss some point, but I don't see a reason why we try to queue
pages from non migratable VMAs.

The only case when we can queue pages from such VMA is MPOL_MF_STRICT
plus MPOL_MF_MOVE or MPOL_MF_MOVE_ALL for VMA which has pages on LRU,
but gfp mask is not sutable for migaration (see mapping_gfp_mask() check
in vma_migratable()). That's looks like a bug to me.

Let's filter out non-migratable vma at start of queue_pages_test_walk()
and go to queue_pages_pte_range() only if MPOL_MF_MOVE or
MPOL_MF_MOVE_ALL flag is set.

Signed-off-by: Kirill A. Shutemov <kirill....@linux.intel.com>

---
mm/mempolicy.c | 14 +++++---------
1 file changed, 5 insertions(+), 9 deletions(-)

diff --git a/mm/mempolicy.c b/mm/mempolicy.c
index 27d135408a22..4c4187c0e1de 100644
--- a/mm/mempolicy.c
+++ b/mm/mempolicy.c
@@ -548,8 +548,7 @@ retry:

goto retry;
}

- if (flags & (MPOL_MF_MOVE | MPOL_MF_MOVE_ALL))

- migrate_page_add(page, qp->pagelist, flags);
+ migrate_page_add(page, qp->pagelist, flags);

}
pte_unmap_unlock(pte - 1, ptl);
cond_resched();

@@ -625,7 +624,7 @@ static int queue_pages_test_walk(unsigned long start, unsigned long end,
unsigned long endvma = vma->vm_end;
unsigned long flags = qp->flags;

- if (vma->vm_flags & VM_PFNMAP)
+ if (!vma_migratable(vma))
return 1;

if (endvma > end)
@@ -644,16 +643,13 @@ static int queue_pages_test_walk(unsigned long start, unsigned long end,

if (flags & MPOL_MF_LAZY) {
/* Similar to task_numa_work, skip inaccessible VMAs */
- if (vma_migratable(vma) &&
- vma->vm_flags & (VM_READ | VM_EXEC | VM_WRITE))
+ if (vma->vm_flags & (VM_READ | VM_EXEC | VM_WRITE))
change_prot_numa(vma, start, endvma);
return 1;
}

- if ((flags & MPOL_MF_STRICT) ||
- ((flags & (MPOL_MF_MOVE | MPOL_MF_MOVE_ALL)) &&
- vma_migratable(vma)))
- /* queue pages from current vma */
+ /* queue pages from current vma */

+ if (flags & (MPOL_MF_MOVE | MPOL_MF_MOVE_ALL))

return 0;
return 1;
}
--
2.7.0.rc3

[Dmitry Vyukov]

On Fri, Jan 29, 2016 at 1:35 PM, Kirill A. Shutemov
<kirill....@linux.intel.com> wrote:
> From 691a961bb401c5815ed741dac63591efbc6027e3 Mon Sep 17 00:00:00 2001
> From: "Kirill A. Shutemov" <kirill....@linux.intel.com>
> Date: Fri, 29 Jan 2016 15:06:17 +0300
> Subject: [PATCH 2/2] mempolicy: do not try to queue pages from
> !vma_migratable()
>
> Maybe I miss some point, but I don't see a reason why we try to queue
> pages from non migratable VMAs.
>
> The only case when we can queue pages from such VMA is MPOL_MF_STRICT
> plus MPOL_MF_MOVE or MPOL_MF_MOVE_ALL for VMA which has pages on LRU,
> but gfp mask is not sutable for migaration (see mapping_gfp_mask() check
> in vma_migratable()). That's looks like a bug to me.
>
> Let's filter out non-migratable vma at start of queue_pages_test_walk()
> and go to queue_pages_pte_range() only if MPOL_MF_MOVE or
> MPOL_MF_MOVE_ALL flag is set.

I've run the fuzzer with these two patches for the weekend and seen no crashes.
I guess we can consider this as fixed.
Thanks!

> --
> You received this message because you are subscribed to the Google Groups "syzkaller" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to syzkaller+...@googlegroups.com.
> For more options, visit https://groups.google.com/d/optout.