Re: Add Morton,Peter and David for discussion//Re: [PATCH -next] uprobes: fix two zero old_folio bugs in __replace_page()

[Tong Tiangen <tongtiangen@huawei.com>]

在 2025/2/18 16:23, David Hildenbrand 写道:
> On 18.02.25 03:47, Tong Tiangen wrote:
>>
>>
>> 在 2025/2/17 20:38, Tong Tiangen 写道:
>>> We triggered the following error logs in syzkaller test:
>>>
>>>     BUG: Bad page state in process syz.7.38  pfn:1eff3
>>>     page: refcount:0 mapcount:0 mapping:0000000000000000 index:0x0 
>>> pfn:0x1eff3
>>>     flags: 
>>> 0x3fffff00004004(referenced|reserved|node=0|zone=1|lastcpupid=0x1fffff)
>>>     raw: 003fffff00004004 ffffe6c6c07bfcc8 ffffe6c6c07bfcc8 
>>> 0000000000000000
>>>     raw: 0000000000000000 0000000000000000 00000000fffffffe 
>>> 0000000000000000
>>>     page dumped because: PAGE_FLAGS_CHECK_AT_FREE flag(s) set
>>>     Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS 
>>> 1.13.0-1ubuntu1.1 04/01/2014
>>>     Call Trace:
>>>      <TASK>
>>>      dump_stack_lvl+0x32/0x50
>>>      bad_page+0x69/0xf0
>>>      free_unref_page_prepare+0x401/0x500
>>>      free_unref_page+0x6d/0x1b0
>>>      uprobe_write_opcode+0x460/0x8e0
>>>      install_breakpoint.part.0+0x51/0x80
>>>      register_for_each_vma+0x1d9/0x2b0
>>>      __uprobe_register+0x245/0x300
>>>      bpf_uprobe_multi_link_attach+0x29b/0x4f0
>>>      link_create+0x1e2/0x280
>>>      __sys_bpf+0x75f/0xac0
>>>      __x64_sys_bpf+0x1a/0x30
>>>      do_syscall_64+0x56/0x100
>>>      entry_SYSCALL_64_after_hwframe+0x78/0xe2
>>>
>>>      BUG: Bad rss-counter state mm:00000000452453e0 type:MM_FILEPAGES 
>>> val:-1
>>>
>>> The following syzkaller test case can be used to reproduce:
>>>
>>>     r2 = creat(&(0x7f0000000000)='./file0\x00', 0x8)
>>>     write$nbd(r2, &(0x7f0000000580)=ANY=[], 0x10)
>>>     r4 = openat(0xffffffffffffff9c, &(0x7f0000000040)='./file0\x00', 
>>> 0x42, 0x0)
>>>     mmap$IORING_OFF_SQ_RING(&(0x7f0000ffd000/0x3000)=nil, 0x3000, 
>>> 0x0, 0x12, r4, 0x0)
>>>     r5 = userfaultfd(0x80801)
>>>     ioctl$UFFDIO_API(r5, 0xc018aa3f, &(0x7f0000000040)={0xaa, 0x20})
>>>     r6 = userfaultfd(0x80801)
>>>     ioctl$UFFDIO_API(r6, 0xc018aa3f, &(0x7f0000000140))
>>>     ioctl$UFFDIO_REGISTER(r6, 0xc020aa00, 
>>> &(0x7f0000000100)={{&(0x7f0000ffc000/0x4000)=nil, 0x4000}, 0x2})
>>>     ioctl$UFFDIO_ZEROPAGE(r5, 0xc020aa04, 
>>> &(0x7f0000000000)={{&(0x7f0000ffd000/0x1000)=nil, 0x1000}})
>>>     r7 = bpf$PROG_LOAD(0x5, &(0x7f0000000140)={0x2, 0x3, 
>>> &(0x7f0000000200)=ANY=[@ANYBLOB="1800000000120000000000000000000095"], &(0x7f0000000000)='GPL\x00', 
>>> 0x7, 0x0, 0x0, 0x0, 0x0, '\x00', 0x0, @fallback=0x30, 
>>> 0xffffffffffffffff, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 0x0, 
>>> 0x0, 0x10, 0x0, @void, @value}, 0x94)
>>>     bpf$BPF_LINK_CREATE_XDP(0x1c, &(0x7f0000000040)={r7, 0x0, 0x30, 
>>> 0x1e, @val=@uprobe_multi={&(0x7f0000000080)='./file0\x00', 
>>> &(0x7f0000000100)=[0x2], 0x0, 0x0, 0x1}}, 0x40)
>>>
>>> The cause is that zero pfn is set to the pte without increasing the rss
>>> count in mfill_atomic_pte_zeropage() and the refcount of zero folio does
>>> not increase accordingly. Then, the operation on the same pfn is 
>>> performed
>>> in uprobe_write_opcode()->__replace_page() to unconditional decrease the
>>> rss count and old_folio's refcount.
>>>
>>> Therefore, two bugs are introduced:
>>> 1. The rss count is incorrect, when process exit, the check_mm() report
>>>      error "Bad rss-count".
>>> 2. The reserved folio (zero folio) is freed when folio->refcount is 
>>> zero,
>>>      then free_pages_prepare->free_page_is_bad() report error "Bad 
>>> page state".
>>>
>>> To fix it, add zero folio check before rss counter and refcount 
>>> decrease.
>>>
>>> Fixes: 7396fa818d62 ("uprobes/core: Make background page replacement 
>>> logic account for rss_stat counters")
>>> Fixes: 2b1444983508 ("uprobes, mm, x86: Add the ability to install 
>>> and remove uprobes breakpoints")
>>> Signed-off-by: Tong Tiangen <tongtiangen@huawei.com>
>>> ---
>>>    kernel/events/uprobes.c | 6 ++++--
>>>    1 file changed, 4 insertions(+), 2 deletions(-)
>>>
>>> diff --git a/kernel/events/uprobes.c b/kernel/events/uprobes.c
>>> index 46ddf3a2334d..ff5694acfa68 100644
>>> --- a/kernel/events/uprobes.c
>>> +++ b/kernel/events/uprobes.c
>>> @@ -213,7 +213,8 @@ static int __replace_page(struct vm_area_struct 
>>> *vma, unsigned long addr,
>>>            dec_mm_counter(mm, MM_ANONPAGES);
>>>        if (!folio_test_anon(old_folio)) {
>>> -        dec_mm_counter(mm, mm_counter_file(old_folio));
>>> +        if (!is_zero_folio(old_folio))
>>> +            dec_mm_counter(mm, mm_counter_file(old_folio));
>>>            inc_mm_counter(mm, MM_ANONPAGES);
>>>        }
>>> @@ -227,7 +228,8 @@ static int __replace_page(struct vm_area_struct 
>>> *vma, unsigned long addr,
>>>        if (!folio_mapped(old_folio))
>>>            folio_free_swap(old_folio);
>>>        page_vma_mapped_walk_done(&pvmw);
>>> -    folio_put(old_folio);
>>> +    if (!is_zero_folio(old_folio))
>>> +        folio_put(old_folio);
>  >>    >>        err = 0;
>>>     unlock:
>>
> 
> The whole "manually replace pages" logic is fragile. I tried to rewrite 
> it a while back:
> 
> https://lkml.kernel.org/r/20240604122548.359952-1-david@redhat.com
> 
> But didn't get to follow-up yet.
> 
> I'm not sure if the page_vma_mapped_walk() really does what we would 
> expect here.
> 
> The folio_remove_rmap_pte(old_folio, old_page, vma); is certainly wrong 
> as well for ero folios.
> 
> 
> I don't think there is a sane use case right now where we would hit the 
> shared zeropage.
> 
> So for the time being, I think we should just reject them immediately 
> after get_user_page_vma_remote().
> 
> At some point I'll follow up with my rewrite that will clean this 
> nastiness here up a bit.

OK, Before your rewrite last merged, How about i change the solution to
just reject them immediately after get_user_page_vma_remote()？

Thanks,
Tong.

>