Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 2620:137:e000::1:20 as permitted sender) client-ip=2620:137:e000::1:20;
Date:   Wed, 14 Dec 2022 10:19:53 -0500
From:   Peter Xu <peterx@redhat.com>
To:     Pengfei Xu <pengfei.xu@intel.com>
Cc:     linux-kernel@vger.kernel.org, heng.su@intel.com, linux-mm@kvack.org
Subject: Re: [Syzkaller & bisect] There is "copy_page_range WARNING" in
 v6.1-rc8 and v6.1
Message-ID: <Y5npmXzUx4FD12Rp@x1n>
References: <Y5l871w6bwDnJ9HD@xpf.sh.intel.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
In-Reply-To: <Y5l871w6bwDnJ9HD@xpf.sh.intel.com>
Precedence: bulk

On Wed, Dec 14, 2022 at 03:36:15PM +0800, Pengfei Xu wrote:
> Hi Peter and memory expert,

Pengfei,

> 
> It's a soft remind.
> There is "copy_page_range WARNING" in v6.1-rc8 and v6.1.
> [   28.138268] ------------[ cut here ]------------
> [   28.138277] WARNING: CPU: 1 PID: 568 at mm/memory.c:874 copy_page_range+0x1858/0x2960
> [   28.139002] Modules linked in:
> [   28.139219] CPU: 1 PID: 568 Comm: repro Not tainted 5.19.0-rc1-6ff40f00e5a4+ #1
> [   28.139706] Hardware name: QEMU Standard PC (i440FX + PIIX, 1996), BIOS rel-1.14.0-0-g155821a1990b-prebuilt.qemu.org 04/01/2014
> [   28.140306] RIP: 0010:copy_page_range+0x1858/0x2960
> [   28.140584] Code: 5b e2 ff 41 83 fc 1e 0f 85 b1 fe ff ff e8 00 5a e2 ff 31 ff 4c 89 ee e8 26 5b e2 ff 4d 85 ed 0f 85 b6 fe ff ff e8 e8 59 e2 ff <0f> 0b e9 aa 5
> [   28.141550] RSP: 0018:ffffc90000c9baf0 EFLAGS: 00010246
> [   28.141838] RAX: 0000000000000000 RBX: f7fffffffffffc00 RCX: ffffffff81415aba
> [   28.142229] RDX: 0000000000000000 RSI: ffff888007e28000 RDI: 0000000000000002
> [   28.142612] RBP: ffffc90000c9bc98 R08: 0000000000000040 R09: 0000000000000000
> [   28.142996] R10: 0000000000000001 R11: ffffea00002ab580 R12: 000000000000001e
> [   28.143405] R13: 0000000000000000 R14: 0000000000000004 R15: 0000000000000000
> [   28.143905] FS:  00007f5b23733740(0000) GS:ffff88807dd00000(0000) knlGS:0000000000000000
> [   28.144335] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
> [   28.144648] CR2: 00000000200000c0 CR3: 0000000007d5c005 CR4: 0000000000770ee0
> [   28.145032] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 0000000000000000
> [   28.145413] DR3: 0000000000000000 DR6: 00000000ffff07f0 DR7: 0000000000000400
> [   28.145791] PKRU: 55555554
> [   28.145946] Call Trace:
> [   28.146097]  <TASK>
> [   28.146242]  dup_mm+0x689/0x900
> [   28.146436]  copy_process+0x1ade/0x23e0
> [   28.146656]  ? __sanitizer_cov_trace_pc+0x25/0x50
> [   28.146928]  kernel_clone+0x90/0x670
> [   28.147152]  ? userfaultfd_ioctl+0xd2/0x23c0
> [   28.147491]  __do_sys_clone+0xa2/0xd0
> [   28.147779]  __x64_sys_clone+0x2f/0x40
> [   28.148023]  do_syscall_64+0x3b/0x90
> [   28.148243]  entry_SYSCALL_64_after_hwframe+0x46/0xb0
> [   28.148538] RIP: 0033:0x7f5b2385859d
> [   28.148749] Code: 00 c3 66 2e 0f 1f 84 00 00 00 00 00 90 f3 0f 1e fa 48 89 f8 48 89 f7 48 89 d6 48 89 ca 4d 89 c2 4d 89 c8 4c 8b 4c 24 08 0f 05 <48> 3d 01 f0 8
> [   28.149722] RSP: 002b:00007ffc5a2e6b58 EFLAGS: 00000202 ORIG_RAX: 0000000000000038
> [   28.150133] RAX: ffffffffffffffda RBX: 0000000000000000 RCX: 00007f5b2385859d
> [   28.150555] RDX: 0000000000000000 RSI: 0000000000000000 RDI: 0000000000001000
> [   28.150938] RBP: 00007ffc5a2e6ba0 R08: 0000000000000000 R09: 0000000000000000
> [   28.151340] R10: 0000000000000000 R11: 0000000000000202 R12: 0000000000401060
> [   28.151846] R13: 00007ffc5a2e6ca0 R14: 0000000000000000 R15: 0000000000000000
> [   28.152230]  </TASK>
> [   28.152363] ---[ end trace 0000000000000000 ]---
> 
> This issue could be reproduced in v6.1 kernel in guest also.
> 
> Bisected and found that bad commit was:
> "
> 81e0f15f2ef6dad7ccb9c03d8e61ef7ded836b38
> mm: enable PTE markers by default
> "
> 
> After reverted above commit on top of v6.1-rc8 kernel, this issue was gone.
> 
> All detailed bisect info, reproduced code and kconfig are in link:
> https://github.com/xupengfe/syzkaller_logs/tree/main/221208_115556_copy_page_range
> 
> 
> If the info was helpful and it's fixed, please added the Reported tag from me.

I think the warning was improper indeed as when without UFFD_EVENT_FORK
registered dst_vma will not inherit VM_UFFD_WP.

This also reminded me that this can also trigger after the swapin error
replacement using pte markers when there's the swapin error markers and if
we fork() we may hit the same thing.

I'll post a fix soon.  Thanks for the report.

-- 
Peter Xu