Re: [syzbot] general protection fault in try_grab_compound_head

Re: [syzbot] general protection fault in try_grab_compound_head

[syzbot]

syzbot has bisected this issue to:

commit 997acaf6b4b59c6a9c259740312a69ea549cc684
Author: Mark Rutland <mark.rutland@arm.com>
Date:   Mon Jan 11 15:37:07 2021 +0000

    lockdep: report broken irq restoration

bisection log:  https://syzkaller.appspot.com/x/bisect.txt?x=16fbcec4300000
start commit:   3dbdb38e Merge branch 'for-5.14' of git://git.kernel.org/p..
git tree:       upstream
final oops:     https://syzkaller.appspot.com/x/report.txt?x=15fbcec4300000
console output: https://syzkaller.appspot.com/x/log.txt?x=11fbcec4300000
kernel config:  https://syzkaller.appspot.com/x/.config?x=a1fcf15a09815757
dashboard link: https://syzkaller.appspot.com/bug?extid=a3fcd59df1b372066f5a
syz repro:      https://syzkaller.appspot.com/x/repro.syz?x=11a856c4300000
C reproducer:   https://syzkaller.appspot.com/x/repro.c?x=1582c9d8300000

Reported-by: syzbot+a3fcd59df1b372066f5a@syzkaller.appspotmail.com
Fixes: 997acaf6b4b5 ("lockdep: report broken irq restoration")

For information about bisection process see: https://goo.gl/tpsmEJ#bisection

Re: [syzbot] general protection fault in try_grab_compound_head

[Thomas Gleixner]

On Sat, Jul 03 2021 at 13:24, syzbot wrote:
> syzbot has bisected this issue to:
>
> commit 997acaf6b4b59c6a9c259740312a69ea549cc684
> Author: Mark Rutland <mark.rutland@arm.com>
> Date:   Mon Jan 11 15:37:07 2021 +0000
>
>     lockdep: report broken irq restoration

That's the commit which makes the underlying problem visible:

       raw_local_irq_restore() called with IRQs enabled

and is triggered by this call chain:

 kvm_wait arch/x86/kernel/kvm.c:860 [inline]
 kvm_wait+0xc3/0xe0 arch/x86/kernel/kvm.c:837
 pv_wait arch/x86/include/asm/paravirt.h:564 [inline]
 pv_wait_head_or_lock kernel/locking/qspinlock_paravirt.h:470 [inline]
 __pv_queued_spin_lock_slowpath+0x8b8/0xb40 kernel/locking/qspinlock.c:508
 pv_queued_spin_lock_slowpath arch/x86/include/asm/paravirt.h:554 [inline]
 queued_spin_lock_slowpath arch/x86/include/asm/qspinlock.h:51 [inline]
 queued_spin_lock include/asm-generic/qspinlock.h:85 [inline]
 do_raw_spin_lock+0x200/0x2b0 kernel/locking/spinlock_debug.c:113
 spin_lock include/linux/spinlock.h:354 [inline]
 alloc_huge_page+0x2b0/0xda0 mm/hugetlb.c:2318
 hugetlb_no_page mm/hugetlb.c:4323 [inline]
 hugetlb_fault+0xc35/0x1cd0 mm/hugetlb.c:4523
 follow_hugetlb_page+0x317/0xda0 mm/hugetlb.c:4836
 __get_user_pages+0x3fa/0xe30 mm/gup.c:1041
 __get_user_pages_locked mm/gup.c:1256 [inline]
 __gup_longterm_locked+0x15f/0xc80 mm/gup.c:1667
 io_sqe_buffer_register fs/io_uring.c:8462 [inline]
 __io_uring_register fs/io_uring.c:9901 [inline]
 __do_sys_io_uring_register+0xeb1/0x3350 fs/io_uring.c:10000
 do_syscall_64+0x2d/0x70 arch/x86/entry/common.c:46
 entry_SYSCALL_64_after_hwframe+0x44/0xa9

Thanks,

        tglx

Re: [syzbot] general protection fault in try_grab_compound_head

[Sean Christopherson]

On Thu, Jul 08, 2021, Thomas Gleixner wrote:
> On Sat, Jul 03 2021 at 13:24, syzbot wrote:
> > syzbot has bisected this issue to:
> >
> > commit 997acaf6b4b59c6a9c259740312a69ea549cc684
> > Author: Mark Rutland <mark.rutland@arm.com>
> > Date:   Mon Jan 11 15:37:07 2021 +0000
> >
> >     lockdep: report broken irq restoration
> 
> That's the commit which makes the underlying problem visible:
> 
>        raw_local_irq_restore() called with IRQs enabled
> 
> and is triggered by this call chain:
> 
>  kvm_wait arch/x86/kernel/kvm.c:860 [inline]
>  kvm_wait+0xc3/0xe0 arch/x86/kernel/kvm.c:837

And the bug in kvm_wait() was fixed by commit f4e61f0c9add ("x86/kvm: Fix broken
irq restoration in kvm_wait").  The bisection is bad, syzbot happened into the
kvm_wait() WARN and got distracted.  The original #GP looks stable, if someone
from mm land has bandwidth.

Re: [syzbot] general protection fault in try_grab_compound_head

[Joao Martins]

On 7/12/21 10:29 PM, Sean Christopherson wrote:
> On Thu, Jul 08, 2021, Thomas Gleixner wrote:
>> On Sat, Jul 03 2021 at 13:24, syzbot wrote:
>>> syzbot has bisected this issue to:
>>>
>>> commit 997acaf6b4b59c6a9c259740312a69ea549cc684
>>> Author: Mark Rutland <mark.rutland@arm.com>
>>> Date:   Mon Jan 11 15:37:07 2021 +0000
>>>
>>>     lockdep: report broken irq restoration
>>
>> That's the commit which makes the underlying problem visible:
>>
>>        raw_local_irq_restore() called with IRQs enabled
>>
>> and is triggered by this call chain:
>>
>>  kvm_wait arch/x86/kernel/kvm.c:860 [inline]
>>  kvm_wait+0xc3/0xe0 arch/x86/kernel/kvm.c:837
> 
> And the bug in kvm_wait() was fixed by commit f4e61f0c9add ("x86/kvm: Fix broken
> irq restoration in kvm_wait").  The bisection is bad, syzbot happened into the
> kvm_wait() WARN and got distracted.  The original #GP looks stable, if someone
> from mm land has bandwidth.
> 

I've bisected this to (my) recent commit 82e5d378b0e47 ("mm/hugetlb: refactor subpage
recording").

I have this fix below and should formally submit tomorrow after more testing.
My apologies for the trouble.

	Joao

------>8------

Subject: mm/hugetlb: fix refs calculation from unaligned @vaddr

commit 82e5d378b0e47 ("mm/hugetlb: refactor subpage recording")
refactored the count of subpages but missed an edge case when @vaddr is
less than a PAGE_SIZE close to vma->vm_end. It would errousnly set @refs
to 0 and record_subpages_vmas() wouldn't set the pages array element to
its value, consequently causing the reported #GP by syzbot.

Fix it by aligning down @vaddr in @refs calculation.

Reported-by: syzbot+a3fcd59df1b372066f5a@syzkaller.appspotmail.com
Fixes: 82e5d378b0e47 ("mm/hugetlb: refactor subpage recording")
Signed-off-by: Joao Martins <joao.m.martins@oracle.com>

diff --git a/mm/hugetlb.c b/mm/hugetlb.c
index a86a58ef132d..cbc448c1a3c8 100644
--- a/mm/hugetlb.c
+++ b/mm/hugetlb.c
@@ -4949,8 +4949,9 @@ long follow_hugetlb_page(struct mm_struct *mm, struct vm_area_struct
*vma,
                        continue;
                }

-               refs = min3(pages_per_huge_page(h) - pfn_offset,
-                           (vma->vm_end - vaddr) >> PAGE_SHIFT, remainder);
+               /* [vaddr .. vm_end] may not be aligned to PAGE_SIZE */
+               refs = min3(pages_per_huge_page(h) - pfn_offset, remainder,
+                   (vma->vm_end - ALIGN_DOWN(vaddr, PAGE_SIZE)) >> PAGE_SHIFT);

                if (pages || vmas)
                        record_subpages_vmas(mem_map_offset(page, pfn_offset),