LinuxQuestions.org
Welcome to the most active Linux Forum on the web.
Go Back   LinuxQuestions.org > Forums > Linux Forums > Linux - Distributions > Slackware
User Name
Password
Slackware This Forum is for the discussion of Slackware Linux.

Notices


Reply
  Search this Thread
Old 08-31-2014, 03:56 AM   #61
moisespedro
Senior Member
 
Registered: Nov 2013
Location: Brazil
Distribution: Slackware
Posts: 1,223

Rep: Reputation: 195Reputation: 195

All this talk is completely beyond me, (udev, eudev, /dev/shm, etc). Do you guys have any good reading on this? Any good link?
 
Old 08-31-2014, 04:38 AM   #62
GazL
Senior Member
 
Registered: May 2008
Posts: 4,754
Blog Entries: 14

Rep: Reputation: Disabled
Slackware's rc.S skips udev startup if passed a "nohotplug" option on the kernel command line (also its possible that rc.udev might be 'chmod -x'), so it seems like rc.S would be the place to check for and create if necessary these two mountpoints before running 'mount -a'.
 
Old 08-31-2014, 04:45 AM   #63
Didier Spaier
LQ Addict
 
Registered: Nov 2008
Location: Paris, France
Distribution: Slint64-14.2.1 on Lenovo Thinkpad W520
Posts: 8,492

Original Poster
Rep: Reputation: 3242Reputation: 3242Reputation: 3242Reputation: 3242Reputation: 3242Reputation: 3242Reputation: 3242Reputation: 3242Reputation: 3242Reputation: 3242Reputation: 3242
Quote:
Originally Posted by moisespedro View Post
All this talk is completely beyond me, (udev, eudev, /dev/shm, etc). Do you guys have any good reading on this? Any good link?
About shm, type "man shm_overview" in a terminal. shm stands for "shared memory" and is a way for two processes to share data that are stored at RAM addresses known of both. The purpose of this line in your /etc/fstab:
Code:
tmpfs            /dev/shm         tmpfs       defaults         0   0
is explained in /usr/src/linux*/Documentation/filesystems/tmpfs.txt in your system, provided you installed the kernel-source package.

shm_open and shm_unlink mentioned in that document are specified by POSIX here and there.

udev is mostly a way to populate the /dev directory, creating devices nodes only for devices found in your system. This is a way to access these devices, e.g. keybord, printers, hard disks.

eudev is roughly a stand-alone (i.e. extracted from systemd) implementation of udev, see the project's home page.

To be able to actually use the shared memory /dev/shm should be mounted the way indicated in /etc/fstab, but first the device node /dev/shm should have been created during the boot sequence in some way, hence the discussion.

Last edited by Didier Spaier; 08-31-2014 at 05:04 AM.
 
1 members found this post helpful.
Old 08-31-2014, 06:28 AM   #64
ml4711
Member
 
Registered: Aug 2012
Location: Ryomgård, Danmark
Distribution: Slackware64
Posts: 146

Rep: Reputation: 101Reputation: 101
@Didier

Quote:
To be able to actually use the shared memory /dev/shm should be mounted the way indicated in /etc/fstab, but first the device node /dev/shm should have been created during the boot sequence in some way, hence the discussion.
This is not quite right.

From tmpfs.txt:

Quote:
Since tmpfs lives completely in the page cache and on swap,
all tmpfs pages currently in memory will show up as cached.
So tmpfs is a kernel thing, and /dev/shm is NOT a device node, but a mount point.
 
Old 08-31-2014, 06:41 AM   #65
Didier Spaier
LQ Addict
 
Registered: Nov 2008
Location: Paris, France
Distribution: Slint64-14.2.1 on Lenovo Thinkpad W520
Posts: 8,492

Original Poster
Rep: Reputation: 3242Reputation: 3242Reputation: 3242Reputation: 3242Reputation: 3242Reputation: 3242Reputation: 3242Reputation: 3242Reputation: 3242Reputation: 3242Reputation: 3242
Quote:
Originally Posted by ml4711 View Post
@Didier

This is not quite right. [...] tmpfs is a kernel thing, and /dev/shm is NOT a device node, but a mount point.
Indeed. From /usr/src/linux*/Documentation/devices.txt:
Code:
    Mount points

The following names are reserved for mounting special filesystems
under /dev.  These special filesystems provide kernel interfaces that
cannot be provided with standard device nodes.

/dev/pts    devpts       PTY slave filesystem
/dev/shm    tmpfs        POSIX shared memory maintenance access
I stand corrected.

This notwithstanding, the mount point should be created before use

Last edited by Didier Spaier; 08-31-2014 at 06:59 AM.
 
Old 08-31-2014, 07:17 AM   #66
ml4711
Member
 
Registered: Aug 2012
Location: Ryomgård, Danmark
Distribution: Slackware64
Posts: 146

Rep: Reputation: 101Reputation: 101
Quote:
This notwithstanding, the mount point should be created before use
Yes, that is true.

And have a wonderful Sunday :-)
 
Old 08-31-2014, 08:31 AM   #67
ReaperX7
LQ Guru
 
Registered: Jul 2011
Location: California
Distribution: Slackware64-Current
Posts: 6,438
Blog Entries: 15

Rep: Reputation: 2005Reputation: 2005Reputation: 2005Reputation: 2005Reputation: 2005Reputation: 2005Reputation: 2005Reputation: 2005Reputation: 2005Reputation: 2005Reputation: 2005
Quote:
Originally Posted by Didier Spaier View Post
Indeed. From /usr/src/linux*/Documentation/devices.txt:
Code:
    Mount points

The following names are reserved for mounting special filesystems
under /dev.  These special filesystems provide kernel interfaces that
cannot be provided with standard device nodes.

/dev/pts    devpts       PTY slave filesystem
/dev/shm    tmpfs        POSIX shared memory maintenance access
I stand corrected.

This notwithstanding, the mount point should be created before use
Yes. /dev/shm should point to /run/shm by default but /run has to be properly setup during the virtual file system mount at the start of the init process. tmpfs mounted at /run must be mounted first of all the kernel file systems and then directories created for lvm, user, lock, shm, and var, then /dev/shm if not existing, is created as a symlink to /run/shm. If it does exist it should be pointing to /run/shm already. If /etc/rc.d/rc.S isn't setting up the kernel file systems in the proper order, it needs adjustments to do so.
 
Old 08-31-2014, 08:44 AM   #68
Didier Spaier
LQ Addict
 
Registered: Nov 2008
Location: Paris, France
Distribution: Slint64-14.2.1 on Lenovo Thinkpad W520
Posts: 8,492

Original Poster
Rep: Reputation: 3242Reputation: 3242Reputation: 3242Reputation: 3242Reputation: 3242Reputation: 3242Reputation: 3242Reputation: 3242Reputation: 3242Reputation: 3242Reputation: 3242
Quote:
Originally Posted by ReaperX7 View Post
Yes. /dev/shm should point to /run/shm by default but /run has to be properly setup during the virtual file system mount at the start of the init process. tmpfs mounted at /run must be mounted first of all the kernel file systems and then directories created for lvm, user, lock, shm, and var, then /dev/shm if not existing, is created as a symlink to /run/shm. If it does exist it should be pointing to /run/shm already. If /etc/rc.d/rc.S isn't setting up the kernel file systems in the proper order, it needs adjustments to do so.
This kind of baseless[1] statement is off topic, again.

[1] If it's not baseless, please be kind enough to provide references or better, send these references instead to Mr. Volkerding if you think that despite having maintained Slackware since 21 years to its users' satisfaction he still needs to learn how an init system should work. Feel free too to explain him why Slackware should adopt runit or even better, just propose him a runit implementation adapted to Slackware as that's certainly easy for you but please, don't hijack this thread for that.

Last edited by Didier Spaier; 08-31-2014 at 01:19 PM.
 
Old 08-31-2014, 01:14 PM   #69
moisespedro
Senior Member
 
Registered: Nov 2013
Location: Brazil
Distribution: Slackware
Posts: 1,223

Rep: Reputation: 195Reputation: 195
Quote:
Originally Posted by Didier Spaier View Post
About shm, type "man shm_overview" in a terminal. shm stands for "shared memory" and is a way for two processes to share data that are stored at RAM addresses known of both. The purpose of this line in your /etc/fstab:
Code:
tmpfs            /dev/shm         tmpfs       defaults         0   0
is explained in /usr/src/linux*/Documentation/filesystems/tmpfs.txt in your system, provided you installed the kernel-source package.

shm_open and shm_unlink mentioned in that document are specified by POSIX here and there.

udev is mostly a way to populate the /dev directory, creating devices nodes only for devices found in your system. This is a way to access these devices, e.g. keybord, printers, hard disks.

eudev is roughly a stand-alone (i.e. extracted from systemd) implementation of udev, see the project's home page.

To be able to actually use the shared memory /dev/shm should be mounted the way indicated in /etc/fstab, but first the device node /dev/shm should have been created during the boot sequence in some way, hence the discussion.
Hmm ok, thanks for all the links
 
Old 08-31-2014, 04:28 PM   #70
unSpawn
Moderator
 
Registered: May 2001
Posts: 29,390
Blog Entries: 55

Rep: Reputation: 3563Reputation: 3563Reputation: 3563Reputation: 3563Reputation: 3563Reputation: 3563Reputation: 3563Reputation: 3563Reputation: 3563Reputation: 3563Reputation: 3563
Quote:
Originally Posted by Didier Spaier View Post
This kind of baseless[1] statement is off topic, again.
Up until this point you both managed to maintain a proper technical discussion. ReaperX7 played a seemingly helpful role and even if you would not agree with his last post its contents are purely technical not invective, no taunts or anything else that would warrant such an outburst.

Please do not do that again.
 
1 members found this post helpful.
Old 08-31-2014, 04:36 PM   #71
ReaperX7
LQ Guru
 
Registered: Jul 2011
Location: California
Distribution: Slackware64-Current
Posts: 6,438
Blog Entries: 15

Rep: Reputation: 2005Reputation: 2005Reputation: 2005Reputation: 2005Reputation: 2005Reputation: 2005Reputation: 2005Reputation: 2005Reputation: 2005Reputation: 2005Reputation: 2005
Times change and with them so do the needs of scripting Didier. Just because a script worked way back when 21 years ago doesn't mean a modern kernel is going to still work the same. In ways it may, but in other ways it may not, even a kernel from last year can have significant changes that require unlearning what you have learned, and learning anew.

To be honest, look at our project as a reference, you'd be shocked that Stage 1 script which is equal to rc.S in many ways went through five revisions at least just for the kernel virtual file systems to be properly mounted, and we gathered examples from Slackware, LFS, Arch-Ignite, VoidLinux, and others just to get it "working" before we could get everything else cleaned up.

If rc.S does need modernizations, it may need them to work with eudev.

Last edited by ReaperX7; 08-31-2014 at 04:43 PM.
 
Old 08-31-2014, 05:16 PM   #72
Didier Spaier
LQ Addict
 
Registered: Nov 2008
Location: Paris, France
Distribution: Slint64-14.2.1 on Lenovo Thinkpad W520
Posts: 8,492

Original Poster
Rep: Reputation: 3242Reputation: 3242Reputation: 3242Reputation: 3242Reputation: 3242Reputation: 3242Reputation: 3242Reputation: 3242Reputation: 3242Reputation: 3242Reputation: 3242
Quote:
Originally Posted by unSpawn View Post
Up until this point you both managed to maintain a proper technical discussion. ReaperX7 played a seemingly helpful role and even if you would not agree with his last post its contents are purely technical not invective, no taunts or anything else that would warrant such an outburst.

Please do not do that again.
I do feel that ReaperX7 way too often post content off topic on threads that I initiate, including in this case: expressing oneself in a technical way doesn't make a post relevant given the topic. This annoys me because that makes harder for other people to follow the discussion or more importantly can discourage those who have something relevant to bring to it to do so. I just expressed that frustration again, in case you didn't noticed.

Thus I don't think that your rebuke be deserved though I must listen to it as you are a moderator.

Quote:
Please do not do that again.
To make sure I don't I won't answer any post from ReaperX7 from now on.

If you think that I should be banned from this forum or from LQ, so be it.

Last edited by Didier Spaier; 08-31-2014 at 05:22 PM.
 
Old 08-31-2014, 05:49 PM   #73
unSpawn
Moderator
 
Registered: May 2001
Posts: 29,390
Blog Entries: 55

Rep: Reputation: 3563Reputation: 3563Reputation: 3563Reputation: 3563Reputation: 3563Reputation: 3563Reputation: 3563Reputation: 3563Reputation: 3563Reputation: 3563Reputation: 3563
Quote:
Originally Posted by Didier Spaier View Post
expressing oneself in a technical way doesn't make a post relevant given the topic.
That may be so but you could either point that out in a respectful way or else ignore it, as you choose to do now.


Quote:
Originally Posted by Didier Spaier View Post
This annoys me because that makes harder for other people to follow the discussion or more importantly can discourage those who have something relevant to bring to it to do so.
I understand that, yes.


Quote:
Originally Posted by Didier Spaier View Post
I just expressed that frustration again, in case you didn't noticed.
I did. Do note an apparent lack of moderation sometimes may be effective moderation as well. (Silence between the notes, if you dig what I mean.)


Quote:
Originally Posted by Didier Spaier View Post
If you think that I should be banned from this forum or from LQ, so be it.
I do not think so, no.
 
Old 08-31-2014, 06:45 PM   #74
ReaperX7
LQ Guru
 
Registered: Jul 2011
Location: California
Distribution: Slackware64-Current
Posts: 6,438
Blog Entries: 15

Rep: Reputation: 2005Reputation: 2005Reputation: 2005Reputation: 2005Reputation: 2005Reputation: 2005Reputation: 2005Reputation: 2005Reputation: 2005Reputation: 2005Reputation: 2005
Didier, I would not push Runit onto Patrick. I would offer, but that it. However, is Patrick a perfect human being that he feels people are beneath him and he doesn't need help from the peanut gallery? I doubt it.

Calm down man. I may not consider you "member of the year" but I don't feel like you don't or shouldn't have an opinion, and have nothing relevant to contribute. So you don't like me. Join the club. Sometimes I don't like me either.

Now with unSpawn's permission and possible help if he wants to, or can, let's get rc.S diagnosed and fixed up, and offer the proposal to Patrick. Hell you can slap your name on it all you want man. I'm not doing this for brownie points or recognition, I'm doing this because it's the right thing to do.

Last edited by ReaperX7; 08-31-2014 at 06:59 PM.
 
Old 09-05-2014, 02:44 PM   #75
hendrickxm
Member
 
Registered: Feb 2014
Posts: 215

Rep: Reputation: Disabled
Just wanted to say that rlworkman has updated the udev from systemd to 215.
 
  


Reply


Thread Tools Search this Thread
Search this Thread:

Advanced Search

Posting Rules
You may not post new threads
You may not post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is Off
HTML code is Off



Similar Threads
Thread Thread Starter Forum Replies Last Post
immutable ip address for testing purposes? stateless Linux - Networking 3 02-13-2014 02:34 PM
Slackware and eudev project jb.1234abcd Slackware 1 12-22-2012 07:14 PM
duplicating http-traffic for testing purposes pretos Linux - Server 1 01-20-2011 05:58 AM
[SOLVED] For learning purposes and tinkering: Slackware/Gentoo? (8 m.o newbie) Ubunoob001 Slackware 30 10-19-2010 10:38 PM
[Announce] LutelWall Firewall Configuration Tool - finally released for use/testing. GrapefruiTgirl Linux - Security 0 01-06-2010 06:17 PM

LinuxQuestions.org > Forums > Linux Forums > Linux - Distributions > Slackware

All times are GMT -5. The time now is 11:18 PM.

Main Menu
Advertisement
My LQ
Write for LQ
LinuxQuestions.org is looking for people interested in writing Editorials, Articles, Reviews, and more. If you'd like to contribute content, let us know.
Main Menu
Syndicate
RSS1  Latest Threads
RSS1  LQ News
Twitter: @linuxquestions
Facebook: linuxquestions Google+: linuxquestions
Open Source Consulting | Domain Registration