View previous topic :: View next topic |
Author |
Message |
chadzor n00b
Joined: 16 Aug 2005 Posts: 3
|
Posted: Tue Aug 16, 2005 8:51 pm Post subject: sandbox access violations for all ebuilds |
|
|
I'm currently trying to run a stage3/1 install with 2005.1, but I seem to run into a sandbox error during any emerge.
my make.conf
--------------------------------------
CHOST="i386-pc-linux-gnu"
CFLAGS="-O2 -mcpu=pentium -fomit-frame-pointer -pipe"
ACCEPT_KEYWORDS="x86"
CXXFLAGS="${CFLAGS}"
PORTDIR=/usr/portage
PORTAGE_TMPDIR=/var/tmp
PORTDIR_OVERLAY=/usr/local/portage
AUTOCLEAN="YES"
MAKEOPTS="-j2"
PORTAGE_NICENESS=3
FEATURES="ccache distlocks sandbox userpriv usersandbox"
CCACHE_SIZE="512M"
USE="nptl -oss -gnome -kde -gtk -arts"
error:
livecd / # emerge gcc-config
Calculating dependencies ...done!
>>> emerge (1 of 1) sys-devel/gcc-config-1.3.12-r2 to /
>>> md5 files gcc-config-1.3.10-r1.ebuild
>>> md5 files gcc-config-1.3.12-r2.ebuild
>>> md5 files gcc-config-1.3.12-r1.ebuild
>>> md5 files gcc-config-1.3.11-r4.ebuild
>>> md5 files gcc-config-1.3.12.ebuild
>>> md5 files gcc-config-1.3.11-r3.ebuild
>>> md5 files files/digest-gcc-config-1.3.10-r1
>>> md5 files files/digest-gcc-config-1.3.11-r3
>>> md5 files files/digest-gcc-config-1.3.11-r4
>>> md5 files files/digest-gcc-config-1.3.12-r1
>>> md5 files files/digest-gcc-config-1.3.12-r2
>>> md5 files files/wrapper-1.4.5.c
>>> md5 files files/wrapper-1.4.6.c
>>> md5 files files/wrapper-1.4.7.c
>>> md5 files files/gcc-config-1.3.10
>>> md5 files files/gcc-config-1.3.11
>>> md5 files files/gcc-config-1.3.12
>>> md5 files files/digest-gcc-config-1.3.12
ACCESS DENIED open_wr: /newroot/dev/pts/0
ACCESS DENIED open_wr: /newroot/dev/pts/0
>>> Unpacking source...
>>> Source unpacked.
--------------------------- ACCESS VIOLATION SUMMARY ---------------------------
LOG FILE = "/var/log/sandbox/sandbox-sys-devel_-_gcc-config-1.3.12-r2-12273.log"
open_wr: /newroot/dev/pts/0
open_wr: /newroot/dev/pts/0
--------------------------------------------------------------------------------
I've searched all around and the only information I could find is that it's an ebuild problem with sandbox, yet this is occuring with all ebuilds so i'm figuring it's my problem. Anyone have any idea? |
|
Back to top |
|
|
lnxz Guru
Joined: 03 Jul 2005 Posts: 472 Location: Earth
|
Posted: Tue Aug 16, 2005 9:12 pm Post subject: |
|
|
Don't know how to fix the issue with sandbox, but you could just disable it by adding -sandbox to FEATURES in make.conf. |
|
Back to top |
|
|
chadzor n00b
Joined: 16 Aug 2005 Posts: 3
|
Posted: Wed Aug 17, 2005 6:42 am Post subject: |
|
|
Fixed, forgot to edit fstab.. duh ^^ |
|
Back to top |
|
|
96140 Retired Dev
Joined: 23 Jan 2005 Posts: 1324
|
Posted: Wed Aug 17, 2005 7:28 am Post subject: |
|
|
Don't forget to put [SOLVED] in the thread's subject. |
|
Back to top |
|
|
chadzor n00b
Joined: 16 Aug 2005 Posts: 3
|
Posted: Wed Aug 17, 2005 9:17 am Post subject: |
|
|
Errr, a few new developments. I'm actually no longer sure if editing the fstab had anything to do with the workaround. This install around the fstab did not matter, the problem was mounting proc. After rebooting and not mounting proc the install is going smooth ^^. More updates to come when I figure more out. |
|
Back to top |
|
|
grx Apprentice
Joined: 19 Jan 2005 Posts: 173 Location: Maryland
|
Posted: Thu Aug 18, 2005 4:52 am Post subject: |
|
|
I'm doing the same type of install and I'm getting the same problem. umount /proc does seem to let it compile fine. Any ideas yet on the cause of the problem? |
|
Back to top |
|
|
sypher n00b
Joined: 14 Aug 2005 Posts: 26
|
Posted: Thu Aug 18, 2005 8:35 am Post subject: Re: sandbox access violations for all ebuilds |
|
|
chadzor wrote: | LOG FILE = "/var/log/sandbox/sandbox-sys-devel_-_gcc-config-1.3.12-r2-12273.log"
open_wr: /newroot/dev/pts/0
open_wr: /newroot/dev/pts/0
-------------------------------------------------------------------------------- |
I encountered this error when attempting to bootstrap. You can still leave sandbox in the FEATURES variable but you have to remove userpriv and usersandbox. It's annoying but it was the only way I could figure out how to get my sys to bootstrap. _________________ .Sypher |
|
Back to top |
|
|
SVN n00b
Joined: 15 Apr 2004 Posts: 45
|
Posted: Fri Sep 02, 2005 12:29 pm Post subject: |
|
|
I have also just encountered this error when doing a stage 1/3 install. Is there already a resolution or can someone explain me why I can not use userpriv and usersandbox with the stage 1/3 method?
edit: It has also something to do with ccache (see: http://gentoo-wiki.com/Ccache)
I'm going to try again without ccache in my make.conf. |
|
Back to top |
|
|
DevSolar Tux's lil' helper
Joined: 07 Sep 2004 Posts: 82
|
Posted: Wed Sep 14, 2005 7:13 am Post subject: |
|
|
/bump
Same problem for me. Before I disable sandbox (compromising security) or unmount /proc (deviating from handbook), does anyone have an explanation for what's happening? |
|
Back to top |
|
|
SVN n00b
Joined: 15 Apr 2004 Posts: 45
|
Posted: Wed Sep 14, 2005 8:14 am Post subject: |
|
|
I think there is something wrong with the 2005.1 install CD. I have got these problems with a stage 1/3 install, but also with a stage 3 install. I have done a successfull stage 1/3 install using a 2005.1 stage 3 tarball and the 2005.0 CD. |
|
Back to top |
|
|
dpetka2001 l33t
Joined: 04 Mar 2005 Posts: 804
|
Posted: Wed Sep 14, 2005 9:25 am Post subject: |
|
|
could someone explain what sandbox is?? thanks in advance... |
|
Back to top |
|
|
DevSolar Tux's lil' helper
Joined: 07 Sep 2004 Posts: 82
|
Posted: Wed Sep 14, 2005 2:13 pm Post subject: |
|
|
Sandboxing is a Gentoo feature where the package in question is build in a secure environment (i.e., the sandbox). Only after the package has finished building does Portage do the *real* installation. That means that Portage is aware of everything the installation changed in the system, instead of relying a non-Gentoo Makefile to not violate Gentoo rules. |
|
Back to top |
|
|
DevSolar Tux's lil' helper
Joined: 07 Sep 2004 Posts: 82
|
Posted: Wed Sep 14, 2005 2:16 pm Post subject: |
|
|
SVN wrote: | I have done a successfull stage 1/3 install using a 2005.1 stage 3 tarball and the 2005.0 CD. |
You would think they *TEST* those darn things before releasing them. That would be the *SECOND* Gentoo LiveCD that fails me on a stock stage1 / stage3 from-scratch installation. (Of all three I tried.)
I'll try the old Knoppix CD again this evening. And if that thing works while the Gentoo LiveCD doesn't, I'll be p***ed. |
|
Back to top |
|
|
hothead Apprentice
Joined: 02 Jul 2004 Posts: 277 Location: /dev/core
|
Posted: Wed Sep 14, 2005 3:35 pm Post subject: |
|
|
Hi,
I also ran into the same problem when doing a fresh stage1 install on my laptop today.
It seems that it is insufficient to just mount the /proc folder as described in the handbook.
Binding the /dev folder of the liveCD to the chroot /dev folder fixed the problem for me.
Binding /dev to /mount/gentoo/dev before chroot:
Code: | mount -o bind /dev /mnt/gentoo/dev |
I allready filed a documentation bug report, so hopefully this will be added to the documentation.
Regards
Ruben _________________ Got a question? - http://justfuckinggoogleit.com/ |
|
Back to top |
|
|
Ben2040 Guru
Joined: 07 May 2003 Posts: 445 Location: UK
|
Posted: Sun Sep 18, 2005 9:52 pm Post subject: |
|
|
hothead wrote: | Hi,
I also ran into the same problem when doing a fresh stage1 install on my laptop today.
It seems that it is insufficient to just mount the /proc folder as described in the handbook.
Binding the /dev folder of the liveCD to the chroot /dev folder fixed the problem for me.
Binding /dev to /mount/gentoo/dev before chroot:
Code: | mount -o bind /dev /mnt/gentoo/dev |
I allready filed a documentation bug report, so hopefully this will be added to the documentation.
Regards
Ruben |
Just confirming this works - Thanks!
Ben |
|
Back to top |
|
|
altrhombus Tux's lil' helper
Joined: 25 Sep 2004 Posts: 92
|
Posted: Wed Sep 21, 2005 2:59 am Post subject: |
|
|
Worked for me! |
|
Back to top |
|
|
torchZ06 Apprentice
Joined: 01 Nov 2003 Posts: 175 Location: the front range
|
Posted: Tue Sep 27, 2005 7:10 pm Post subject: |
|
|
worked for me too-- thanks! |
|
Back to top |
|
|
Bob P Advocate
Joined: 20 Oct 2004 Posts: 3355 Location: Jackass! Development Labs
|
|
Back to top |
|
|
Bob P Advocate
Joined: 20 Oct 2004 Posts: 3355 Location: Jackass! Development Labs
|
|
Back to top |
|
|
amne Bodhisattva
Joined: 17 Nov 2002 Posts: 6378 Location: Graz / EU
|
Posted: Sat Nov 12, 2005 3:48 pm Post subject: |
|
|
lnxz wrote: | Don't know how to fix the issue with sandbox, but you could just disable it by adding -sandbox to FEATURES in make.conf. |
Turning off the sandbox may cure the symptoms, but potentially harms your system and does not get the bug fixed. Please refrain from giving false advice to people. Sandbox issues should go to bugs.gentoo.org (please search if there is already bugreport about it).
Bob P wrote: | btw, this thread needed to have [SOLVED] added to the subject header months ago... |
Rather [BROKEN_WORKAROUND], plus there is no policy about adding solved-tags. _________________ Dinosaur week! (Ok, this thread is so last week) |
|
Back to top |
|
|
Bob P Advocate
Joined: 20 Oct 2004 Posts: 3355 Location: Jackass! Development Labs
|
Posted: Sat Nov 12, 2005 4:03 pm Post subject: |
|
|
BROKEN_WORKAROUND? i guess we're looking at different solutions.
@hothead posted the definitive solution to the problem:
hothead wrote: | Code: | mount -o bind /dev /mnt/gentoo/dev |
|
THAT will fix the problem. its the definitive cure, and it doesn't b0rk anything. i agree that those recommendations to turn of the sandbox were ill-advised. i make the mistake of trying that in my youth...
so i guess i'd say that [SOLVED] would have been a good idea in response to @hothead's fix.
now this surprises me -- is there really no official policy regarding putting [SOLVED] in headers when problems are solved? is it just a nice convention that alot of people follow for no good reason? _________________ .
Stage 1/3 | Jackass! | Rockhopper! | Thanks | Google Sucks |
|
Back to top |
|
|
amne Bodhisattva
Joined: 17 Nov 2002 Posts: 6378 Location: Graz / EU
|
Posted: Sat Nov 12, 2005 4:14 pm Post subject: |
|
|
Bob P wrote: |
@hothead posted the definitive solution to the problem:
|
Oh, overlooked that one with all that -sandbox stuff.
Bob P wrote: | now this surprises me -- is there really no official policy regarding putting [SOLVED] in headers when problems are solved? is it just a nice convention that alot of people follow for no good reason? |
Yupp. Some people are in favour of it, some not.
Also see Mark for solved threads (plus some other threads in GFF probably). _________________ Dinosaur week! (Ok, this thread is so last week) |
|
Back to top |
|
|
mvanitallie n00b
Joined: 13 Dec 2005 Posts: 7 Location: NY
|
Posted: Thu Dec 22, 2005 3:07 pm Post subject: |
|
|
amne wrote: |
Please refrain from giving false advice to people. |
Brushing your teeth gives you cancer
Quote: | hothead wrote:
mount -o bind /dev /mnt/gentoo/dev |
Works for me. |
|
Back to top |
|
|
zanzer7 n00b
Joined: 11 Oct 2004 Posts: 30 Location: Denmark
|
Posted: Thu Feb 16, 2006 12:28 pm Post subject: Re: sandbox access violations for all ebuilds |
|
|
sypher wrote: | You can still leave sandbox in the FEATURES variable but you have to remove userpriv and usersandbox. |
Yup, that did the job! _________________ Will divide by zero for food |
|
Back to top |
|
|
Exodus.vE n00b
Joined: 21 Jun 2006 Posts: 1
|
Posted: Wed Jun 21, 2006 6:24 pm Post subject: |
|
|
I found this problem while i searched on google for "ACCESS DENIED open_wr: /newroot/".
My hardware is a hp pavilion dv8210us laptop.
But the hardware doesn't matter in this error. Here's what i could make of the problem and how i fixed it; AND, how i think is the real and only fix.
I downloaded amd64 netinst, and since i've installed gentoo a couple of times now, and to a couple of people i just glanced at the handbook for new things in the 2006.0 install, and just a couple of things appeared. But due to the quickness in which i proceded, i had forgotten some things and had gotten this error.
Now if you get this error and you're system was already installed and functional, it's a diferent error than mine, and i guess a diferent or related issue must be addressed.
i proceded with install, downloaded the stage 3, downloaded the portage, I uncompressed the files in /mnt/gentoo as standard. But the whole problem revolves around chroot.
First i had a chroot error which i found that it was a incompatibility with the stage3 i had downloaded, i used a amd64 netinst, and was chroot'ing to a ia64 stage3. So i went to a correct mirror where i could download amd64 stage3, bothersome cause i have 512kbps conexion and took 30 - 40 mins.
Now having everything in order, and it being 4am, i ended up with that error after doing the whole chroot procedure. It was so late that i couldn't remember what i did heheh. So i got some sleep and tried again after 3 hrs sleep.
Rebooted the machine. And noticed that with patience, and paying attention, things get fixed.
I figured this is not a bug, it's just not follow the correct instructions on the handbook. If you don't mount the proc and dev in the new chroot, and the chroot, env-update, and source /etc/profile
that error comes up.
If those steps are taken care of correctly, the "problem" will not occur.
I hope this helps someone. Cheers, it's SOLVED to me.
Continuing, i had done this so quick that i couldn't remember if i did "source /etc/profile" or did some other stuff |
|
Back to top |
|
|
|