Received-SPF: pass (google.com: domain of linux-kernel+bounces-81879-linux.lists.archive=gmail.com@vger.kernel.org designates 2604:1380:45d1:ec00::1 as permitted sender) client-ip=2604:1380:45d1:ec00::1;
Date: Mon, 26 Feb 2024 09:14:38 -0700
From: Alex Williamson <alex.williamson@redhat.com>
To: Yisheng Xie <ethan.xys@linux.alibaba.com>
Cc: akpm@linux-foundation.org, kvm@vger.kernel.org,
 linux-kernel@vger.kernel.org, linux-mm@kvack.org
Subject: Re: [PATCH] vfio/type1: unpin PageReserved page
Message-ID: <20240226091438.1fc37957.alex.williamson@redhat.com>
In-Reply-To: <20240226160106.24222-1-ethan.xys@linux.alibaba.com>
References: <20240226160106.24222-1-ethan.xys@linux.alibaba.com>
Precedence: bulk
MIME-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit

On Tue, 27 Feb 2024 00:01:06 +0800
Yisheng Xie <ethan.xys@linux.alibaba.com> wrote:

> We meet a warning as following:
>  WARNING: CPU: 99 PID: 1766859 at mm/gup.c:209 try_grab_page.part.0+0xe8/0x1b0
>  CPU: 99 PID: 1766859 Comm: qemu-kvm Kdump: loaded Tainted: GOE  5.10.134-008.2.x86_64 #1
                                                                   ^^^^^^^^

Does this issue reproduce on mainline?  Thanks,

Alex

>  Hardware name: Foxconn AliServer-Thor-04-12U-v2/Thunder2, BIOS 1.0.PL.FC.P.031.00 05/18/2022
>  RIP: 0010:try_grab_page.part.0+0xe8/0x1b0
>  Code: b9 00 04 00 00 83 e6 01 74 ca 48 8b 32 b9 00 04 00 00 f7 c6 00 00 01 00 74 ba eb 91 8b 57 34 48 89 f8 85 d2 0f 8f 48 ff ff ff <0f> 0b 31 c0 c3 48 89 fa 48 8b 0a f7 c1 00 00 01 00 0f 85 5c ff ff
>  RSP: 0018:ffffc900b1a63b98 EFLAGS: 00010282
>  RAX: ffffea00000e4580 RBX: 0000000000052202 RCX: ffffea00000e4580
>  RDX: 0000000080000001 RSI: 0000000000052202 RDI: ffffea00000e4580
>  RBP: ffff88efa5d3d860 R08: 0000000000000000 R09: 0000000000000002
>  R10: 0000000000000008 R11: ffff89403fff7000 R12: ffff88f589165818
>  R13: 00007f1320600000 R14: ffffea0181296ca8 R15: ffffea00000e4580
>  FS:  00007f1324f93e00(0000) GS:ffff893ebfb80000(0000) knlGS:0000000000000000
>  CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
>  CR2: 00007f1321694070 CR3: 0000006046014004 CR4: 00000000007726e0
>  DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
>  DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 0000000000000400
>  PKRU: 55555554
>  Call Trace:
>   follow_page_pte+0x64b/0x800
>   __get_user_pages+0x228/0x560
>   __gup_longterm_locked+0xa0/0x2f0
>   vaddr_get_pfns+0x67/0x100 [vfio_iommu_type1]
>   vfio_pin_pages_remote+0x30b/0x460 [vfio_iommu_type1]
>   vfio_pin_map_dma+0xd4/0x2e0 [vfio_iommu_type1]
>   vfio_dma_do_map+0x21e/0x340 [vfio_iommu_type1]
>   vfio_iommu_type1_ioctl+0xdd/0x170 [vfio_iommu_type1]
>   ? __fget_files+0x79/0xb0
>   ksys_ioctl+0x7b/0xb0
>   ? ksys_write+0xc4/0xe0
>   __x64_sys_ioctl+0x16/0x20
>   do_syscall_64+0x2d/0x40
>   entry_SYSCALL_64_after_hwframe+0x44/0xa9
> 
> After add dumppage, it shows that it is a PageReserved page(e.g. zero page),
> whoes refcount is just overflow:
>  page:00000000b0504535 refcount:-2147483647 mapcount:0 mapping:0000000000000000 index:0x0 pfn:0x3916
>  flags: 0xffffc000001002(referenced|reserved)
>  raw: 00ffffc000001002 ffffea00000e4588 ffffea00000e4588 0000000000000000
>  raw: 0000000000000000 0000000000000000 80000001ffffffff 0000000000000000
> 
> gup will _pin_ a page which is PageReserved, however, put_pfn in vfio will
> skip unpin page which is PageReserved. So use pfn_valid in put_pfn
> instead of !is_invalid_reserved_pfn to unpin PageReserved page.
> 
> Signed-off-by: Yisheng Xie <ethan.xys@linux.alibaba.com>
> ---
>  drivers/vfio/vfio_iommu_type1.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/drivers/vfio/vfio_iommu_type1.c b/drivers/vfio/vfio_iommu_type1.c
> index b2854d7939ce..12775bab27ee 100644
> --- a/drivers/vfio/vfio_iommu_type1.c
> +++ b/drivers/vfio/vfio_iommu_type1.c
> @@ -461,7 +461,7 @@ static bool is_invalid_reserved_pfn(unsigned long pfn)
>  
>  static int put_pfn(unsigned long pfn, int prot)
>  {
> -	if (!is_invalid_reserved_pfn(pfn)) {
> +	if (pfn_valid(pfn)) {
>  		struct page *page = pfn_to_page(pfn);
>  
>  		unpin_user_pages_dirty_lock(&page, 1, prot & IOMMU_WRITE);