All of lore.kernel.org
 help / color / mirror / Atom feed
From: Harald Hoyer <harald-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>
To: Dave Young <dyoung-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org>,
	Felix Miata <mrmazda-ihVZJaRskl1bRRN4PJnoQQ@public.gmane.org>
Cc: initramfs-u79uwXL29TY76Z2rM5mHXA@public.gmane.org
Subject: Re: kernel/initrd loading delays
Date: Fri, 13 Nov 2015 13:49:38 +0100	[thread overview]
Message-ID: <5645DC62.3090903@redhat.com> (raw)
In-Reply-To: <20151014090447.GB4795-sa4SJRhfYT4qE/hkJvsb+hcY2uh10dtjAL8bYrjMMd8@public.gmane.org>

On 14.10.2015 11:04, Dave Young wrote:
> On 10/12/15 at 12:29pm, Felix Miata wrote:
>> On more and more installations since most distros have replaced mkinitrd with
>> dracut, on selecting a bootloader menu choice the "root
>> (hd0,...filesystemtype..[Linux-bzImage, setup, ... initrd /boot/initrd"
>> message stays on screen 40 seconds or more before the time-stamped boot
>> messages begin scrolling the screen. IIRC, these delays first began appearing
>> last January or February, and appear on multiple machines. All kernel and
>> initrd images are on smallish EXT3 or EXT4 (mostly the latter) filesystems
>> using 1024 blocksize on BIOS logical partitions on rotating rust (IIRC, all
>> with native 512b sector size, manufactured in 2011 or earlier). All
>> installations are configured with static IP. All have Plymouth either not
>> installed, or are booted with noplymouth included on kernel cmdline.
>>
>> e.g, on my fastest system, host msi85, a 3.0 GHz dual core Haswell, total
>> time from boot stanza select on BIOS host msi85 to Rawhide 4.3.0-0.rc3.git4.1
>> (dracut initramfs.img 11,326,223) multi-user.target shell prompt ready is
>> about 80 seconds following that 40 second delay, with last time stamp showing
>> 36.844484.
>>
>> Same msi85 system booting LMDE 2 3.16.0-4 (non-dracut? initrd.img 27,362,964;
>> initramfs-tools 0.120) delays start of boot messages nearly 80 seconds,
>> reaching DM greeter ready nearly 140 seconds after stanza selection.
>>
>> Same system booting openSUSE Leap 42.1 4.1.8 (dracut initrd 8,082,360)
>> exhibits no perceptible delay, reaching multi-user.target shell prompt in
>> under 47 seconds from boot stanza selection.
>>
>> Same system booting openSUSE 13.2 3.16.7 (dracut initrd 5,351,996) delays
>> start of boot messages about 40 seconds, reaching multi-user.target bash
>> prompt about 90 seconds after stanza selection.
>>
>> Same system booting Mageia 5 4.1.8 (dracut initrd.img 9,391,664) exhibits no
>> delay, reaching multi-user.target bash prompt in about 45 seconds.
>>
>> Same system booting Wheezy 3.2.0 (non-dracut initrd.img 10,537,321) exhibits
>> no delay, reaching DM greeter in less than about 45 seconds.
>>
>> Same system booting openSUSE 13.1 3.12.44 (non-dracut initrd 7,679,390)
>> exhibits no delay, reaching multi-user.target bash prompt in less than 45
>> seconds.
>>
>> Same system booting Fedora 23 4.2.2 (dracut initramfs.img 11,255,214) reaches
>> multi-user.target shell prompt in about 75 seconds after unknown
>> kernel/initrd delay (puts display into sleep mode until boot messages appear).
>>
>> Same system booting openSUSE Tumbleweed 4.2.1 (dracut initrd 8,970,760)
>> exhibits no delay, reaching multi-user.target bash prompt in less than 45
>> seconds.
>>
>> Same system booting (second, on sda28, vs. other on sda23) openSUSE
>> Tumbleweed 4.2.1 (dracut initrd 8,965,124) exhibits ~40 second delay,
>> reaching multi-user.target bash prompt in about 100 seconds.
>>
>> Same system booting Kubuntu 14.10 3.16.0 (non-dracut initrd 20,550,608)
>> exhibits no initial (linux/initrd) delay, reaching KDM greeter in about 90
>> seconds.
>>
>> Same system booting Fedora 22 4.1.8 (dracut initrd 10,788,392) exhibits no
>> delay, reaching multi-user.target bash prompt in less than 50 seconds.
> 
> Wow, you are using so many distributions.
> 
> From your previous tests, I feel it should not be a dracut problem.
> But you need first find where is the delay from, bootloader, kernel, or somewhere else.
> I think you can select one distribution delaying happend, try older kernel
> if the older kernel does not have the delay then it should be kernel problem.
> 
> As for boot loader, they may have some debug options but I do not know that
> much.
> 
> Thanks
> Dave

Also rawhide kernels have debugging enabled, which slows down significantly.

Adding "rd.debug" might help.

https://www.kernel.org/pub/linux/utils/boot/dracut/dracut.html#_troubleshooting

Also "debug" for systemd/udevd debug messages.

  parent reply	other threads:[~2015-11-13 12:49 UTC|newest]

Thread overview: 4+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2015-10-12 16:29 kernel/initrd loading delays Felix Miata
     [not found] ` <561BDFDF.1030303-ihVZJaRskl1bRRN4PJnoQQ@public.gmane.org>
2015-10-14  9:04   ` Dave Young
     [not found]     ` <20151014090447.GB4795-sa4SJRhfYT4qE/hkJvsb+hcY2uh10dtjAL8bYrjMMd8@public.gmane.org>
2015-11-13 12:49       ` Harald Hoyer [this message]
2016-03-22  8:49   ` Felix Miata

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=5645DC62.3090903@redhat.com \
    --to=harald-h+wxahxf7alqt0dzr+alfa@public.gmane.org \
    --cc=dyoung-H+wXaHxf7aLQT0dZR+AlfA@public.gmane.org \
    --cc=initramfs-u79uwXL29TY76Z2rM5mHXA@public.gmane.org \
    --cc=mrmazda-ihVZJaRskl1bRRN4PJnoQQ@public.gmane.org \
    /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 an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.