Message-ID: <e6b8d928-36d3-d2e5-a773-2f73b8f92bbc@huaweicloud.com>
Date: Fri, 24 Jan 2025 09:30:48 +0800
From: Yu Kuai <yukuai1@...weicloud.com>
To: Harshit Mogalapalli <harshit.m.mogalapalli@...cle.com>,
 yukuai1@...weicloud.com, LKML <linux-kernel@...r.kernel.org>,
 linux-raid@...r.kernel.org, mariusz.tkaczyk@...ux.intel.com,
 song@...nel.org, pmenzel@...gen.mpg.de
Cc: Himanshu Madhani <himanshu.madhani@...cle.com>,
 "regressions@...ts.linux.dev" <regressions@...ts.linux.dev>,
 "stable@...r.kernel.org" <stable@...r.kernel.org>,
 Darren Kenny <darren.kenny@...cle.com>, "yukuai (C)" <yukuai3@...wei.com>
Subject: Re: [REGRESSION] kernel panic at bitmap_get_stats+0x2b/0xa0 since
 6.12

Hi,

在 2025/01/23 5:58, Harshit Mogalapalli 写道:
> Hi all,
> 
> 
> We started seeing panic during boot cycle on 6.12 upstream kernel.
> 
> Data points:
> * This is reproducible on 6.12.9
> * Also reproducible on 6.13 from yesterday.
> * Not reproducible on 6.11
> 
> So I looked at commits between 6.11-> 6.12 , and narrowed it down to a 
> patch series which made changed to md-bitmap.c
> 
> https://lore.kernel.org/all/20240826074452.1490072-1-yukuai1@huaweicloud.com/ 
> 
> 
> After narrowing down further: it is narrowed down to this commit
> 
> ec6bb299c7c3 md/md-bitmap: add 'sync_size' into struct md_bitmap_stats
> 
> 
> #regzbot introduced: ec6bb299c7c3
> 
> 
> Also, the panic points to the middle line below:
> 
>      sb = kmap_local_page(bitmap->storage.sb_page);
> *    stats->sync_size = le64_to_cpu(sb->sync_size);
>      kunmap_local(sb);
> 
> Call trace is as follows:
> 
> [   21.427462] Oops: general protection fault, probably for 
> non-canonical address 0x8730d3f80000028: 0000 [#1] PREEMPT SMP NOPTI
> [   21.440104] CPU: 56 UID: 0 PID: 1531 Comm: mdadm Not tainted 
> 6.13.0-master.20250121.ol8.x86_64 #1
> [   21.450019] Hardware name: Oracle Corporation ORACLE SERVER 
> X9-2L/ASM,MTHRBD,2U, BIOS 62110100 07/15/2024
> [   21.460710] RIP: 0010:bitmap_get_stats+0x2b/0xa0
> [   21.465872] Code: 0f 1e fa 0f 1f 44 00 00 48 89 f2 48 85 ff 74 7d 48 
> 8b 4f 50 48 2b 0d dc 9f e5 00 48 8b 35 e5 9f e5 00 48 c1 f9 06 48 c1 e1 
> 0c <48> 8b 4c 31 28 48 89 4a 20 48 8b 4f 18 48 89 4a 10 48 8b 4f 10 48
> [   21.486849] RSP: 0018:ff3e5f658fc3fb18 EFLAGS: 00010206
> [   21.492690] RAX: ffffffff8d17d660 RBX: ff27d0600af69690 RCX: 
> 094b3d0000000000
> [   21.500663] RDX: ff3e5f658fc3fb28 RSI: ff27d03f80000000 RDI: 
> ff27d06008cd9c00
> [   21.507233] mlx5_core 0000:b1:00.0: Rate limit: 127 rates are 
> supported, range: 0Mbps to 97656Mbps
> [   21.508629] RBP: ff27d0604a737418 R08: 0000000000000000 R09: 
> 0000000000000000
> [   21.508631] R10: 0000000000000000 R11: 0000000000000000 R12: 
> 00000000012c2000
> [   21.508631] R13: ff27d0604a737018 R14: ff27d0604a737000 R15: 
> ff27d0604a737018
> [   21.508632] FS:  00007f61a01c98c0(0000) GS:ff27d07f7f600000(0000) 
> knlGS:0000000000000000
> [   21.508634] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
> [   21.508635] CR2: 000056503c28f458 CR3: 00000020c000c004 CR4: 
> 0000000000771ef0
> [   21.518772] mlx5_core 0000:b1:00.0: E-Switch: Total vports 27, per 
> vport: max uc(128) max mc(2048)
> [   21.526600] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 
> 0000000000000000
> [   21.526601] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 
> 0000000000000400
> [   21.526602] PKRU: 55555554
> [   21.526603] Call Trace:
> [   21.526604]  <TASK>
> [   21.535111] mlx5_core 0000:b1:00.0: Flow counters bulk query buffer 
> size increased, bulk_query_len(8)
> [   21.542533]  ? show_trace_log_lvl+0x1b0/0x300
> [   21.542537]  ? show_trace_log_lvl+0x1b0/0x300
> [   21.556126] mlx5_core 0000:b1:00.0: mlx5_pcie_event:301:(pid 529): 
> PCIe slot advertised sufficient power (27W).
> [   21.557983]  ? md_seq_show+0x2d2/0x5b0
> [   21.557988]  ? __die_body.cold+0x8/0x12
> [   21.641128]  ? die_addr+0x3c/0x60
> [   21.645080]  ? exc_general_protection+0x17d/0x400
> [   21.650574]  ? asm_exc_general_protection+0x26/0x30
> [   21.656267]  ? __pfx_bitmap_get_stats+0x10/0x10
> [   21.661568]  ? bitmap_get_stats+0x2b/0xa0
> [   21.666277]  md_seq_show+0x2d2/0x5b0
> [   21.670507]  seq_read_iter+0x2b9/0x470
> [   21.674924]  seq_read+0x12f/0x180
> [   21.678853]  proc_reg_read+0x57/0xb0
> [   21.683074]  vfs_read+0xf6/0x380
> [   21.686902]  ? __seccomp_filter+0x30b/0x520
> [   21.691786]  ksys_read+0x6c/0xf0
> [   21.695607]  do_syscall_64+0x82/0x170
> [   21.699909]  ? arch_exit_to_user_mode_prepare.isra.0+0x1e/0xd0
> [   21.706637]  ? syscall_exit_to_user_mode+0x37/0x1a0
> [   21.712295]  ? __memcg_slab_free_hook+0xf7/0x160
> [   21.717660]  ? __x64_sys_close+0x3c/0x80
> [   21.722248]  ? kmem_cache_free+0x400/0x460
> [   21.727028]  ? syscall_exit_to_user_mode_prepare+0x174/0x1b0
> [   21.733553]  ? arch_exit_to_user_mode_prepare.isra.0+0x1e/0xd0
> [   21.740270]  ? syscall_exit_to_user_mode+0x37/0x1a0
> [   21.745913]  ? do_syscall_64+0x8e/0x170
> [   21.750388]  ? do_syscall_64+0x8e/0x170
> [   21.754857]  ? clear_bhb_loop+0x45/0xa0
> [   21.759318]  ? clear_bhb_loop+0x45/0xa0
> [   21.763772]  ? clear_bhb_loop+0x45/0xa0
> [   21.768218]  entry_SYSCALL_64_after_hwframe+0x76/0x7e
> [   21.774014] RIP: 0033:0x7f619f862585
> [   21.778170] Code: fe ff ff 50 48 8d 3d 52 a8 06 00 e8 e5 08 02 00 0f 
> 1f 44 00 00 f3 0f 1e fa 48 8d 05 d5 71 2a 00 8b 00 85 c0 75 0f 31 c0 0f 
> 05 <48> 3d 00 f0 ff ff 77 53 c3 66 90 41 54 49 89 d4 55 48 89 f5 53 89
> [   21.799471] RSP: 002b:00007ffe50c2d3c8 EFLAGS: 00000246 ORIG_RAX: 
> 0000000000000000
> [   21.808099] RAX: ffffffffffffffda RBX: 000056503c2802a0 RCX: 
> 00007f619f862585
> [   21.816240] RDX: 0000000000000400 RSI: 000056503c28d000 RDI: 
> 0000000000000004
> [   21.824382] RBP: 0000000000000d68 R08: 0000000000000008 R09: 
> 0000000000000001
> [   21.832518] R10: 0000000000000000 R11: 0000000000000246 R12: 
> 00007f619fb00860
> [   21.840654] R13: 00007f619fb013a0 R14: 000056503c280a50 R15: 
> 000056503c281480
> [   21.848789]  </TASK>
> [   21.851389] Modules linked in: raid1 mgag200 drm_client_lib 
> drm_shmem_helper drm_kms_helper sd_mod sg raid0 mlx5_core(+) ahci 
> libahci drm crct10dif_pclmul ghash_clmulni_intel mlxfw sha512_ssse3 igb 
> nvme sha256_ssse3 libata tls sha1_ssse3 megaraid_sas nvme_core 
> pci_hyperv_intf psample dca nvme_auth i2c_algo_bit nfit(+) libnvdimm 
> aesni_intel gf128mul crypto_simd cryptd
> [   21.888253] ---[ end trace 0000000000000000 ]---
> [   22.452319] RIP: 0010:bitmap_get_stats+0x2b/0xa0
> [   22.457699] Code: 0f 1e fa 0f 1f 44 00 00 48 89 f2 48 85 ff 74 7d 48 
> 8b 4f 50 48 2b 0d dc 9f e5 00 48 8b 35 e5 9f e5 00 48 c1 f9 06 48 c1 e1 
> 0c <48> 8b 4c 31 28 48 89 4a 20 48 8b 4f 18 48 89 4a 10 48 8b 4f 10 48
> [   22.479037] RSP: 0018:ff3e5f658fc3fb18 EFLAGS: 00010206
> [   22.485067] RAX: ffffffff8d17d660 RBX: ff27d0600af69690 RCX: 
> 094b3d0000000000
> [   22.493217] RDX: ff3e5f658fc3fb28 RSI: ff27d03f80000000 RDI: 
> ff27d06008cd9c00
> [   22.501372] RBP: ff27d0604a737418 R08: 0000000000000000 R09: 
> 0000000000000000
> [   22.509527] R10: 0000000000000000 R11: 0000000000000000 R12: 
> 00000000012c2000
> [   22.517686] R13: ff27d0604a737018 R14: ff27d0604a737000 R15: 
> ff27d0604a737018
> [   22.525845] FS:  00007f61a01c98c0(0000) GS:ff27d07f7f600000(0000) 
> knlGS:0000000000000000
> [   22.535089] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
> [   22.541701] CR2: 000056503c28f458 CR3: 00000020c000c004 CR4: 
> 0000000000771ef0
> [   22.549866] DR0: 0000000000000000 DR1: 0000000000000000 DR2: 
> 0000000000000000
> [   22.558040] DR3: 0000000000000000 DR6: 00000000fffe0ff0 DR7: 
> 0000000000000400
> [   22.566202] PKRU: 55555554
> [   22.569425] Kernel panic - not syncing: Fatal exception
> [   22.576477] Kernel Offset: 0xb600000 from 0xffffffff81000000 
> (relocation range: 0xffffffff80000000-0xffffffffbfffffff)
> [   22.654941] Rebooting in 60 seconds..
> 
> 
> I would be happy to try any patches.

Can you try the following patch on latest kernel?

Thanks for the report!
Kuai

diff --git a/drivers/md/md.c b/drivers/md/md.c
index 94166b2e9512..b07e9c595a7c 100644
--- a/drivers/md/md.c
+++ b/drivers/md/md.c
@@ -8429,12 +8429,14 @@ static void md_bitmap_status(struct seq_file 
*seq, struct mddev *mddev)
         unsigned long chunk_kb;
         int err;

+       /* prevent bitmap to be freed after checking */
+       mutex_lock(&mddev->bitmap_info.mutex);
         if (!md_bitmap_enabled(mddev))
-               return;
+               goto out;

         err = mddev->bitmap_ops->get_stats(mddev->bitmap, &stats);
         if (err)
-               return;
+               goto out;

         chunk_kb = mddev->bitmap_info.chunksize >> 10;
         used_pages = stats.pages - stats.missing_pages;
@@ -8450,6 +8452,9 @@ static void md_bitmap_status(struct seq_file *seq, 
struct mddev *mddev)
         }

         seq_putc(seq, '\n');
+
+out:
+       mutex_unlock(&mddev->bitmap_info.mutex);
  }

  static int md_seq_show(struct seq_file *seq, void *v)

> 
> Thanks,
> Harshit
> 

Powered by blists - more mailing lists