public inbox for developer@lists.illumos.org (since 2011-08)
 help / color / mirror / Atom feed
From: "gusev.vitaliy via illumos-developer" <developer@lists.illumos.org>
To: illumos-developer <developer@lists.illumos.org>,
	Tom Caputi <tcaputi@datto.com>
Subject: Re: [developer] Panic during pkg update
Date: Tue, 12 Nov 2024 09:18:25 +0300	[thread overview]
Message-ID: <3459173C-4C67-4F32-BDFF-27005E54922D@icloud.com> (raw)
In-Reply-To: <3959AC6D-6D8E-4139-A107-7809AB79B12B@icloud.com>

[-- Attachment #1: Type: text/plain, Size: 7851 bytes --]

> On 12 Nov 2024, at 00:11, gusev.vitaliy via illumos-developer <developer@lists.illumos.org> wrote:
> 
> Panic comes from this function:
> 
> usr/src/uts/common/fs/zfs/arc.c
> 
>  5369 /*                                                                                                                                               
>  5370  * XXX this should be changed to return an error, and callers                                                                                    
>  5371  * re-read from disk on failure (on nondebug bits).                                                                                              
>  5372  */                                                                                                                                              
>  5373 static void                                                                                                                                      
>  5374 arc_hdr_verify_checksum(spa_t *spa, arc_buf_hdr_t *hdr, const blkptr_t *bp)
> 
>   …
> 
>  5395                 err = zio_checksum_error_impl(spa, bp,                                                                                           
>  5396                     BP_GET_CHECKSUM(bp), abd, psize, 0, NULL);                                                                                   
>  5397                 if (err != 0) {                                                                                                                  
>  5398                         /*                                                                                                                       
>  5399                          * Use abd_copy_to_buf() rather than                                                                                     
>  5400                          * abd_borrow_buf_copy() so that we are sure to                                                                          
>  5401                          * include the buf in crash dumps.                                                                                       
>  5402                          */                                                                                                                      
>  5403                         void *buf = kmem_alloc(psize, KM_SLEEP);                                                                                 
>  5404                         abd_copy_to_buf(buf, abd, psize);                                                                                        
>  5405                         panic("checksum of cached data doesn't match BP "                                                                        
>  5406                             "err=%u hdr=%p bp=%p abd=%p buf=%p",                                                                                 
>  5407                             err, (void *)hdr, (void *)bp, (void *)abd, buf);          
> 
> OpenZFS though doesn’t have this piece of code and generally returns error as written in comment XXX.
> 
> Tom, do you think the panic() call should be replaced with returning an error?
> 

I mean Tom Caputi :) Since this code was added in "8727 Native data and metadata encryption for zfs”.

Tom, could you clarify,  in which cases zio_checksum_error_impl() can return error here: arc_read -> arc_hdr_verify_checksum -> zio_checksum_error_impl ? Is is so critical to panic?

—
Vitaliy Gusev


> —
> Vitaliy Gusev
> 
>> On 11 Nov 2024, at 21:36, Toomas Soome via illumos-developer <developer@lists.illumos.org> wrote:
>> 
>> 
>> 
>>> On 30. Oct 2024, at 20:57, Gary Mills <gary_mills@fastmail.fm> wrote:
>>> 
>>> I'm not sure if this is a bug or just ZFS being careful, but I got a
>>> panic and reboot while I was doing a "pkg update".  The system
>>> has an AMD 6-core CPU with B550 support hardware.  The next
>>> "pkg update" completed normally, without a panic.  Here's what
>>> I found in /var/adm/messages.  Does it look familiar?
>>> 
>>> Oct 30 09:14:31 b550 unix: [ID 836849 kern.notice] 
>>> Oct 30 09:14:31 b550 ^Mpanic[cpu4]/thread=fffffe2cc9e88780: 
>>> Oct 30 09:14:31 b550 genunix: [ID 129249 kern.notice] checksum of cached data doesn't match BP err=50 hdr=fffffe3d478f51c0 bp=fffffe0040433988 abd=fffffe3d478f7cc0 buf=fffffe3b5a6f9000
>>> Oct 30 09:14:31 b550 unix: [ID 100000 kern.notice] 
>>> Oct 30 09:14:31 b550 genunix: [ID 655072 kern.notice] fffffe0040433760 zfs:zfs_nfsshare_inited+378b87f0 ()
>>> Oct 30 09:14:31 b550 genunix: [ID 655072 kern.notice] fffffe0040433890 zfs:arc_read+de1 ()
>>> Oct 30 09:14:31 b550 genunix: [ID 655072 kern.notice] fffffe00404338e0 zfs:dbuf_issue_final_prefetch+77 ()
>>> Oct 30 09:14:31 b550 genunix: [ID 655072 kern.notice] fffffe0040433a70 zfs:dbuf_prefetch_impl+502 ()
>>> Oct 30 09:14:31 b550 genunix: [ID 655072 kern.notice] fffffe0040433b20 zfs:dmu_zfetch+2ed ()
>>> Oct 30 09:14:31 b550 genunix: [ID 655072 kern.notice] fffffe0040433bd0 zfs:dmu_buf_hold_array_by_dnode+321 ()
>>> Oct 30 09:14:31 b550 genunix: [ID 655072 kern.notice] fffffe0040433c70 zfs:dmu_read_uio_dnode+54 ()
>>> Oct 30 09:14:31 b550 genunix: [ID 655072 kern.notice] fffffe0040433cc0 zfs:dmu_read_uio_dbuf+51 ()
>>> Oct 30 09:14:31 b550 genunix: [ID 655072 kern.notice] fffffe0040433d60 zfs:zfs_read+19c ()
>>> Oct 30 09:14:31 b550 genunix: [ID 655072 kern.notice] fffffe0040433de0 genunix:fop_read+60 ()
>>> Oct 30 09:14:31 b550 genunix: [ID 655072 kern.notice] fffffe0040433f00 genunix:read+2b5 ()
>>> Oct 30 09:14:31 b550 genunix: [ID 655072 kern.notice] fffffe0040433f10 unix:brand_sys_syscall+1fe ()
>>> Oct 30 09:14:31 b550 unix: [ID 100000 kern.notice] 
>>> Oct 30 09:14:31 b550 genunix: [ID 111219 kern.notice] dumping to /dev/zvol/dsk/rpool/dump, offset 65536, content: kernel
>>> Oct 30 09:14:31 b550 ahci: [ID 405573 kern.info] NOTICE: ahci0: ahci_tran_reset_dport port 0 reset port
>>> Oct 30 09:14:32 b550 ahci: [ID 405573 kern.info] NOTICE: ahci0: ahci_tran_reset_dport port 1 reset port
>>> Oct 30 09:14:50 b550 genunix: [ID 100000 kern.notice] 
>>> Oct 30 09:14:50 b550 genunix: [ID 665016 kern.notice] ^M100% done: 859875 pages dumped, 
>>> Oct 30 09:14:50 b550 genunix: [ID 851671 kern.notice] dump succeeded
>>> Oct 30 09:15:34 b550 genunix: [ID 107833 kern.notice] ^MOpenIndiana Hipster 2022.10 Version illumos-806838751b 64-bit
>>> 
>>> 
>> 
>> Dan got blown up while running zfs-tests (rsend), and that resulted on me picking one series of updates from OpenZFS concerning dbuf and dmu. There are still few XXX notes for myself, but so far both debug and non-debug builds have been behaving nicely (debug build used to run zfs-tests). I have seen myself also panic from arc (ASSERT fired while running zfs-tests on debuilg build — that was before the work mentioned above). Most likely need to pick some arc bits as well.
>> 
>> The current wip branch is: https://github.com/tsoome/illumos-gate/tree/rsend if you like to test. The problem about those panics is that they seem to be random, or at least not easily repeatable.
>> 
>> rgds,
>> toomas
>> 
> 
> illumos <https://illumos.topicbox.com/latest> / illumos-developer / see discussions <https://illumos.topicbox.com/groups/developer> + participants <https://illumos.topicbox.com/groups/developer/members> + delivery options <https://illumos.topicbox.com/groups/developer/subscription>Permalink <https://illumos.topicbox.com/groups/developer/Te1153c7aaa3e05c7-M5484aceef4b12b181bfff6f0>

------------------------------------------
illumos: illumos-developer
Permalink: https://illumos.topicbox.com/groups/developer/Te1153c7aaa3e05c7-Mc08c98af25e077799cc2d612
Delivery options: https://illumos.topicbox.com/groups/developer/subscription

[-- Attachment #2: Type: text/html, Size: 34920 bytes --]

      reply	other threads:[~2024-11-12  6:21 UTC|newest]

Thread overview: 7+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2024-10-30 18:57 Gary Mills
2024-10-30 19:10 ` Peter Tribble
2024-10-30 20:57   ` Gary Mills
2024-11-11 17:19     ` Gergő Doma
2024-11-11 18:36 ` Toomas Soome via illumos-developer
2024-11-11 21:11   ` gusev.vitaliy via illumos-developer
2024-11-12  6:18     ` gusev.vitaliy via illumos-developer [this message]

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=3459173C-4C67-4F32-BDFF-27005E54922D@icloud.com \
    --to=developer@lists.illumos.org \
    --cc=tcaputi@datto.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).