Re: [DNG] Request file system reviews and recomendations.

2017-12-30 Thread J. Fahrner

Am 2017-12-29 21:35, schrieb Adam Borowski:

resize2fs -b $DEV
tune2fs -O metadata_csum $DEV
fsck.ext4 -D $DEV


According to this wiki:
https://ext4.wiki.kernel.org/index.php/Ext4_Metadata_Checksums

should we also use option journal_checksum on mount?

Jochen
___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng


Re: [DNG] ext4 checksums was: Request file system reviews and recomendations.

2017-12-30 Thread J. Fahrner
Am 2017-12-29 21:35, schrieb Adam Borowski: 

> resize2fs -b $DEV
> tune2fs -O metadata_csum $DEV
> fsck.ext4 -D $DEV

One thing is unclear to me. Reading this wiki:
https://ext4.wiki.kernel.org/index.php/Ext4_Metadata_Checksums

They say: 

* Install Linux 3.6+ and e2fsprogs 1.43-WIP.
* modprobe crc32c-intel
* mkfs.ext4 -O metadata_csum,64bit /dev/path/to/disk
* mount /dev/path/to/disk /mountpoint -o journal_checksum

and: 

Some recent CPUs (Intel and SPARC) provide a hardware accelerated CRC32c
implementation. For best performance you should ensure that these
modules load before any ext4 filesystems with checksums. 

doing a lsmod on my system, I see: 

$ lsmod|fgrep crc
crct10dif_pclmul 16384 0
crc32_pclmul 16384 0
crc16 16384 2 bluetooth,ext4
crc32c_generic 16384 0
crc32c_intel 24576 2

crc16 is used by ext4. Does that mean I am not using hardware
accelerated crc32c? 

If not, what can I do to use hardware accelerated crc32c? 

Is it possible to turn an existing ext4fs into one with crc32c? 

How can I load module crc32c_intel before filesystems get mounted? 

Jochen___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng


Re: [DNG] ext4 checksums was: Request file system reviews and recomendations.

2017-12-30 Thread Adam Borowski
On Sat, Dec 30, 2017 at 12:56:58PM +0100, J. Fahrner wrote:
> Some recent CPUs (Intel and SPARC) provide a hardware accelerated CRC32c
> implementation. For best performance you should ensure that these
> modules load before any ext4 filesystems with checksums. 
> 
> doing a lsmod on my system, I see: 
> 
> $ lsmod|fgrep crc
> crct10dif_pclmul 16384 0
> crc32_pclmul 16384 0
> crc16 16384 2 bluetooth,ext4
> crc32c_generic 16384 0
> crc32c_intel 24576 2
> 
> crc16 is used by ext4. Does that mean I am not using hardware
> accelerated crc32c? 

Loading module dependencies on demand is error-prone, and you really don't
want your / filesystem to fail at runtime for such a reason.  Thus, these
modules get loaded unconditionally.

Also, if I understand correctly, ext4 can't switch crc implementations after
it's initialized, which is a yet another reason to load them all.

> If not, what can I do to use hardware accelerated crc32c? 

It's loaded automatically, and used if your hardware supports acceleration.
I don't see ext4 being chatty wrt what it uses, you can load the btrfs
module, which I believe uses the same code, and which says:
on Braswell:
Btrfs loaded, crc32c=crc32c-intel
on Phenom II:
Btrfs loaded, crc32c=crc32c-generic
on Allwinner A64:
Btrfs loaded, crc32c=crc32c-generic
on Exynos 4412:
Btrfs loaded, crc32c=crc32c-generic

> Is it possible to turn an existing ext4fs into one with crc32c? 

You need "resize2fs -b" to switch to the so-called "64 bit" format, as the
old format did not have enough unused space.  Then you do the switch, then
"fsck.ext4 -D" to rewrite all directories to the new format.

> How can I load module crc32c_intel before filesystems get mounted? 

I believe you would need extra steps to _not_ load it when appropriate.


Meow!
-- 
// If you believe in so-called "intellectual property", please immediately
// cease using counterfeit alphabets.  Instead, contact the nearest temple
// of Amon, whose priests will provide you with scribal services for all
// your writing needs, for Reasonable And Non-Discriminatory prices.
___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng


[DNG] SOLVED: Trouble getting copied system to boot

2017-12-30 Thread Hendrik Boom
On Sat, Dec 16, 2017 at 05:55:46PM -0500, Hendrik Boom wrote:
> I m copying my entire Devuan system to new  partitions on the same computer. 
> but I can't get the copy system to boot.
> 
> I plan to upgrade the copy to ascii, keeping the old system in  dual-boot 
> scenario just in case  I'm not expecting problems, but I've had them on 
> Debian upgrades years ago, and they sometimes arise from causes that have 
> nothing to do with the quality of the old and new systems.
> 
> The messge I get is:
> 
> vmlinuz-3.16.0-4-686-pae not found
> you need to load the kernel first
> 
> But vmlinuz-3.16.0-4-686-pae *is* the kernel.  I don't know how to load the 
> ernel before it can find the kernel.

I've solved the problem.  I've been embarrassingly stupid.

I used the exact, correct UUID, checked hex digit by hex digit 
multiple times, of the *wrong partition*.  Of course it didn't find 
the kernel.  There was no kernel on the wrong partition.

-- hendrik
___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng


[DNG] listing partitions conveniently

2017-12-30 Thread Hendrik Boom
Disk partitions seem to have many names.  THis is convenient, but also 
confusing. 

Names might be
the current mount point
the mount point mentioned in /etc/fstab (usually, but not always, the 
same)
the /dev/disk/by-uuid name
the /dev/disk/by-label name
the /dev/diskby-id nane
the uuid
the /dev/dm-* name

etc., etc.

Is there some simple, readable way to list them all, partition by 
partition, giving all the names for each one?

-- hendrik

___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng


Re: [DNG] listing partitions conveniently

2017-12-30 Thread Michael Siegel
Am 31.12.2017 um 00:32 schrieb Hendrik Boom:
> Disk partitions seem to have many names.  THis is convenient, but also 
> confusing. 
> 
> Names might be
>   the current mount point
>   the mount point mentioned in /etc/fstab (usually, but not always, the 
> same)
>   the /dev/disk/by-uuid name
>   the /dev/disk/by-label name
>   the /dev/diskby-id nane
>   the uuid
>   the /dev/dm-* name
> 
> etc., etc.
> 
> Is there some simple, readable way to list them all, partition by 
> partition, giving all the names for each one?

Isn't 'lsblk' able to do this?


msi

___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng


Re: [DNG] listing partitions conveniently

2017-12-30 Thread Hendrik Boom
On Sun, Dec 31, 2017 at 01:57:00AM +0100, Michael Siegel wrote:
> Am 31.12.2017 um 00:32 schrieb Hendrik Boom:
> > Disk partitions seem to have many names.  THis is convenient, but also 
> > confusing. 
> > 
> > Names might be
> > the current mount point
> > the mount point mentioned in /etc/fstab (usually, but not always, the 
> > same)
> > the /dev/disk/by-uuid name
> > the /dev/disk/by-label name
> > the /dev/diskby-id nane
> > the uuid
> > the /dev/dm-* name
> > 
> > etc., etc.
> > 
> > Is there some simple, readable way to list them all, partition by 
> > partition, giving all the names for each one?
> 
> Isn't 'lsblk' able to do this?


Thanks.  It's a lovely listing.

> 
> 
> msi
> 
> ___
> Dng mailing list
> Dng@lists.dyne.org
> https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng
___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng


Re: [DNG] Summary of investigations on Lid open/close in ASCII. A simple solution

2017-12-30 Thread Steve Litt
On Fri, 29 Dec 2017 15:35:24 +0100
Didier Kryn  wrote:

> Le 29/12/2017 à 12:35, J. Fahrner a écrit :
> > Am 2017-12-29 12:27, schrieb Didier Kryn:  
> >>     Actualy xscreensaver also prevents the suspend. I didn't notice
> >> the first time because it seems to take a logout-login before the
> >> presence of the daemon to be noticed.
> >>
> >>     There is a hackish workaround: comment the following 3 lines in
> >> /etc/acpi/lid.sh:
> >> if { CheckPolicy || HasLogindAndSystemd1Manager; }; then
> >> exit
> >> fi  
> >
> > You see all this rubbish crap!
> > This is why I love minimalism. No DE, no bloated packages. The only 
> > thing you need is acpid and some self written 3-line shell scripts
> > to act on the events.  
> 
>      Sure, but it comes just naked with essentially the right-click
> to offer you to open a terminal emulator. That's not a very
> productive way of working, compared to xfce4's panel. I've seen on
> documentation that you can add a number of feature

Didier, you're right if the only alternative to xfce is unmanaged X, or
bloatsters KDE and Gnome. However, lightweights such as LXDE without
its addons, Openbox, ctwm, jwm, i3, IceWM, WindowMaker, fvwm don't take
it upon themselves to mess with your system functions such as mounting
and sleep.

To make all of these instantly productive, all you need to do is
install and configure dmenu as your way of running programs. Besides
that, most of them have menus of greater or lesser value.
 
SteveT

Steve Litt 
December 2017 featured book: Thriving in Tough Times
http://www.troubleshooters.com/thrive
___
Dng mailing list
Dng@lists.dyne.org
https://mailinglists.dyne.org/cgi-bin/mailman/listinfo/dng