linux-ext4 - [Bug 217965] ext4(?) regression since 6.5.0 on sata hdd

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date: Wed, 15 Nov 2023 22:39:39 +0000
From: bugzilla-daemon@...nel.org
To: linux-ext4@...r.kernel.org
Subject: [Bug 217965] ext4(?) regression since 6.5.0 on sata hdd

https://bugzilla.kernel.org/show_bug.cgi?id=217965

--- Comment #26 from Eyal Lebedinsky (bugzilla@...l.emu.id.au) ---
Ojaswin,

For my understanding: the (dirty) buffers hold data blocks that need to be
written. Clearing dirty buffers does not include fs activity,
which was already performed during the copy. Is this correct? If so then why
are we talking about ext4?
    I expect that my understanding is incorrect.
The copy itself is very fast, completing in a few seconds.

I assume that the requested mount option is safe. Is it safe to use this way
   $ sudo mount -o remount,nodelalloc /data1
or should I also add the other current options, 'noatime' (from fstab) and
'stripe=640' (automatic)?
or should I reboot with this option in fstab?

Below is what I did now, is this what you need?

$ sudo perf record -p 1366946 sleep 60
[ perf record: Woken up 37 times to write data ]
[ perf record: Captured and wrote 9.079 MB perf.data (237339 samples) ]

$ sudo perf report --no-children --stdio -i perf.data
# To display the perf.data header info, please use --header/--header-only
options.
#
#
# Total Lost Samples: 0
#
# Samples: 237K of event 'cycles:P'
# Event count (approx.): 258097341907
#
# Overhead  Command          Shared Object      Symbol                          
# ........  ...............  ................. 
..........................................
#
    53.90%  kworker/u16:4+f  [kernel.kallsyms]  [k] ext4_mb_good_group
    13.71%  kworker/u16:4+f  [kernel.kallsyms]  [k] ext4_get_group_info
     7.39%  kworker/u16:4+f  [kernel.kallsyms]  [k]
ext4_mb_find_good_group_avg_frag_lists
     6.96%  kworker/u16:4+f  [kernel.kallsyms]  [k] __rcu_read_unlock
     5.01%  kworker/u16:4+f  [kernel.kallsyms]  [k] ext4_mb_scan_aligned
     4.51%  kworker/u16:4+f  [kernel.kallsyms]  [k] __rcu_read_lock
     1.70%  kworker/u16:4+f  [kernel.kallsyms]  [k] mb_find_order_for_block
     0.89%  kworker/u16:4+f  [kernel.kallsyms]  [k] xas_descend
     0.85%  kworker/u16:4+f  [kernel.kallsyms]  [k] filemap_get_entry
     0.81%  kworker/u16:4+f  [kernel.kallsyms]  [k] ext4_mb_regular_allocator
     0.77%  kworker/u16:4+f  [kernel.kallsyms]  [k] ext4_mb_load_buddy_gfp
     0.54%  kworker/u16:4+f  [kernel.kallsyms]  [k] xas_load
     0.50%  kworker/u16:4+f  [kernel.kallsyms]  [k] ext4_mb_unload_buddy
     0.42%  kworker/u16:4+f  [kernel.kallsyms]  [k] _raw_read_unlock
     0.40%  kworker/u16:4+f  [kernel.kallsyms]  [k] mb_find_extent
     0.40%  kworker/u16:4+f  [kernel.kallsyms]  [k] _raw_spin_trylock
     0.25%  kworker/u16:4+f  [kernel.kallsyms]  [k] __filemap_get_folio
     0.24%  kworker/u16:4+f  [kernel.kallsyms]  [k] _raw_read_lock
     0.13%  kworker/u16:4+f  [kernel.kallsyms]  [k] xas_start
     0.12%  kworker/u16:4+f  [kernel.kallsyms]  [k] mb_find_buddy
     0.10%  kworker/u16:4+f  [kernel.kallsyms]  [k] pagecache_get_page
     0.08%  kworker/u16:4+f  [kernel.kallsyms]  [k] folio_mark_accessed
     0.05%  kworker/u16:4+f  [kernel.kallsyms]  [k] folio_test_hugetlb
     0.04%  kworker/u16:4+f  [kernel.kallsyms]  [k] _raw_spin_unlock
     0.02%  kworker/u16:4+f  [kernel.kallsyms]  [k] __cond_resched
     0.02%  kworker/u16:4+f  [kernel.kallsyms]  [k]
perf_adjust_freq_unthr_context
     0.01%  kworker/u16:4+f  [kernel.kallsyms]  [k] gen8_irq_handler
     0.01%  kworker/u16:4+f  [kernel.kallsyms]  [k] _raw_spin_lock
     0.01%  kworker/u16:4+f  [kernel.kallsyms]  [k] native_irq_return_iret
     0.01%  kworker/u16:4+f  [kernel.kallsyms]  [k] native_write_msr
     0.01%  kworker/u16:4+f  [kernel.kallsyms]  [k] timekeeping_advance
     0.01%  kworker/u16:4+f  [kernel.kallsyms]  [k] _raw_spin_lock_irqsave
     0.01%  kworker/u16:4+f  [kernel.kallsyms]  [k] native_read_msr
<then many zero% items>

-- 
You may reply to this email to add a comment.

You are receiving this mail because:
You are watching the assignee of the bug.