Re: [gentoo-user] Heads Up - glibc-2.27 breaks my system

[Neil Bothwick]

On 3 February 2018 17:34:11 GMT, Helmut Jarausch  wrote:
>On 02/03/2018 04:11:33 PM, Marc Joliet wrote:
>> Am Samstag, 3. Februar 2018, 10:50:53 CET schrieb Helmut Jarausch:
>> > On 02/03/2018 06:54:06 AM, Dale wrote:
>> > > While on this topic, I have a question about glibc.  I have it  
>> set in
>> > > make.conf to save the binary packages.  Generally I use it when I
> 
>> need
>> > > to go back shortly after a upgrade, usually Firefox or something.
>> > > However, this package is different since going back a version  
>> isn't a
>> > > good idea.  My question tho, what if one does go back a version  
>> using
>> > > those saved binary packages?  Has anyone ever did it and it work 
>
>> or
>> > > did
>> > > it and it fail miserably?
>> >
>> > I've tried to binary emerge my previous version. This didn't
>succeed
>> > since
>> > the ebuild disallows downgrading glibc.
>> >
>> > Luckily I had backuped my system just 20 hours ago.
>> 
>> Having up-to-date backups is always good :) .
>> 
>> > Does anybody know how to restore ONLY those files which are
>> > more recent on the target file system.
>> > (My whole back is 124 Gb large which is a lot to copy back)
>> 
>> If you can access the backups like a normal file system, then using  
>> rsync with
>> the --update option looks to me like what you want:
>> 
>> "-u, --updateskip files that are newer on the  
>> receiver".
>> 
>
>High Marc,
>I think I need the opposite :
>   only update files which are newer on the receiver
>
>Thanks,
>Helmut

Run the rsync in the opposite direction with - n as well as - u. That should 
give you a list of files that are newer on the live system. 
-- 
Sent from my Android device with K-9 Mail. Please excuse my brevity.

Re: [gentoo-user] Grub2 boot problem

[Tom H]

On Wed, Jan 31, 2018 at 2:58 PM, Magnus Johansson  wrote:
>
> I've got a fresh Gentoo installation that does not boot. I just end up in
> the Grub2 shell.
>
> However when there if I do 'set root=(md/0)' and 'configfile /grub/grub.cfg'
> I do get to the Grub2 menu where Gentoo boots just fine.
>
> /boot and / are both on mdadm devices.
>
> I've tried re-running grub-mkconfig and grub-install several times without
> luck.
>
> I've added domdadm to GRUB_CMDLINE_LINUX_DEFAULT in /etc/default/grub.
>
> Drives sda/sdb use GPT disklabels and have 20M BIOS boot partitions first.

To which device are you installing grub?

Check what "core.img" prefix and modules grub-install is using with

grub-install --debug $your_device 2>&1 | grep grub-mkimage

[I hit a similar problem with mdadm 6 or 7 years ago and had to create
a custom "core.img" to boot normally. You might have to do the same
but i would've thought that this problem's been solved. I haven't
encountered it since.]

Re: [gentoo-user] Peculiar problem: no su - to root

[Tom H]

On Sat, Feb 3, 2018 at 4:05 PM, Harry Putnam  wrote:
>
> I've just completed getting gentoo booted as guest in vbox vm.
>
> I'm having a peculiar problem. I cannot call `su -' or `su root' and
> login as root.
>
> I can still get to root by `ssh root@localhost' having set up
> /etc/sshd_config while still chrooted during install.
>
> Still no getting to root by way of `su -' or `su root'
>
> I just get the `Permission denied' message.
>
> I've never had this happen before over many installs of various linux
> distros.
>
> I did try ssh root@localhost and then resetting root passwd while
> logged in as root, but still no getting to root by `su -'.

Add the su-ing user to the "wheel" group (because of "auth required
pam_wheel.so" in "/etc/pam.d/su").

Re: [gentoo-user] Heads Up - glibc-2.27 breaks my system

[Helmut Jarausch]

On 02/04/2018 12:21:13 AM, Bill Kenworthy wrote:

On 04/02/18 01:34, Helmut Jarausch wrote:
> On 02/03/2018 04:11:33 PM, Marc Joliet wrote:
>> Am Samstag, 3. Februar 2018, 10:50:53 CET schrieb Helmut Jarausch:
>> > On 02/03/2018 06:54:06 AM, Dale wrote:
>> > > While on this topic, I have a question about glibc.  I have it  
set in
>> > > make.conf to save the binary packages.  Generally I use it  
when I

>> need
>> > > to go back shortly after a upgrade, usually Firefox or  
something.
>> > > However, this package is different since going back a version  
isn't a
>> > > good idea.  My question tho, what if one does go back a  
version using
>> > > those saved binary packages?  Has anyone ever did it and it  
work or

>> > > did

1. do another backup
2. take your last good binary package and unpack it in the root
directory - it is an "image" of that package as it sits in the file  
system.
3. rebuild that version of glibc by overiding emerge - comment out  
"die
"aborting to save your system" in  
/usr/portage/eclass/toolchain-glibc.eclass


I have done this a couple of times with gcc (when manual deletes have
gone rogue) but it should work with glibc as you have not recompiled  
any

new packages.



Many thanks Bil,
Helmut

[gentoo-user] eselect mystery

[Helmut Jarausch]

Hi,

could anybody please explain me how 'eselect' works?

My problem

I have installed dev-java/oracle-jdk-bin-9.04 and removed  
dev-java/oracle-jdk-bin-1.8.0.162


but eselect java-vm list shows

Available Java Virtual Machines:
grep: /usr/share/java-config-2/vm/oracle-jdk-bin-1.8: No such file or  
directory

  [1]   icedtea-8
  [2]   oracle-jdk-bin-1.8  system-vm

i.e. it still searches for oracle-jdk-bin-1.8 and doesn't know about  
oracle-jdk-bin-9


Is this a bug or is there any means to bring eselect up-to-date ?

Many thanks for a hint,
Helmut

Re: [gentoo-user] eselect mystery

[Floyd Anderson]

On Sun, 04 Feb 2018 12:57:40 +0100
Helmut Jarausch  wrote:


i.e. it still searches for oracle-jdk-bin-1.8 and doesn't know about 
oracle-jdk-bin-9


Is this a bug or is there any means to bring eselect up-to-date ?


Comment 2 on bug #641690 [1] say no. As mentioned there, the point 
seems to be dev-java/oracle-jdk-bin-9.0.4[-gentoo-vm].


You should also note the true-branch (when enabling gentoo-vm):

   ewarn "WARNING! You have enabled the gentoo-vm USE flag, making this JDK"
   ewarn "recognised by the system. This will almost certainly break things."

in the dev-java/oracle-jdk-bin-9.0.4 ebuild.


Link:
 - [1] 



--
Regards,
floyd

[gentoo-user] Failed builds of kbuild and cdrdao with "undefined reference to `__alloca'"

[tuxic]

Hi,

I still have the problem of failed builds due to an
'undefined reference to `__alloca''. I recompiled 
gcc/glibc and I am using linux-4.15.1 (from kernel.org)
with linux-headers 4.15. .

Affected are (at least) cdrdao and kbuild.

For me "alloca" (memory allocation_) seems to be quite fundamental
and if this would really be not availlable I think a lot more
builds would fail...

I am using gcc (Gentoo 7.3.0 p1.0) 7.3.0. I recompiled glibc.
I did https://wiki.gentoo.org/wiki/Upgrading_GCC

What can cause this problem and how can I fix it?

Cheers
Meino

[gentoo-user] Re: Heads Up - glibc-2.27 breaks my system

[Nikos Chantziaras]

On 03/02/18 16:08, Dale wrote:

Nikos Chantziaras wrote:

It is perfectly fine to downgrade glibc if you didn't emerge anything
that compiled binaries.

If you did, you can still downgrade, but then you need to rebuild the
packages that you emerged since the glibc upgrade. qlop is your friend
here; it lets you find out the dates on which you emerged packages.


That makes sense.  So, if worse comes to worse, downgrade, then emerge
-e world if unsure what all has been updated since.  If, using qlop or
friends, you can figure what was done since the upgrade, emerge those to
make sure the linking is correct.  At least that is a option that should
be doable.  That's better than thinking you can't downgrade for any
reason, period.


You might not be able to do that, if python (used by emerge) uses 
something that breaks when downgrading glibc. Or gcc. Or binutils. Or 
bash. Or anything else that's needed during an emerge.


So you need to check with qlop *before* downgrading, and if it looks 
like something critical was built against the new glibc, then all bets 
are off. Which is why the downgrade protection exists in the first place.


The only way out of this, is restoring from backup or fixing things by 
booting from a sysrescuecd or similar.


If only firefox or your media player and stuff like that got built 
against the new glibc, then it's fine to downgrade. Otherwise, you could 
end up bricking your system.

Re: [gentoo-user] Re: Heads Up - glibc-2.27 breaks my system

[Dale]

Nikos Chantziaras wrote:
> On 03/02/18 16:08, Dale wrote:
>> Nikos Chantziaras wrote:
>>> It is perfectly fine to downgrade glibc if you didn't emerge anything
>>> that compiled binaries.
>>>
>>> If you did, you can still downgrade, but then you need to rebuild the
>>> packages that you emerged since the glibc upgrade. qlop is your friend
>>> here; it lets you find out the dates on which you emerged packages.
>>
>> That makes sense.  So, if worse comes to worse, downgrade, then emerge
>> -e world if unsure what all has been updated since.  If, using qlop or
>> friends, you can figure what was done since the upgrade, emerge those to
>> make sure the linking is correct.  At least that is a option that should
>> be doable.  That's better than thinking you can't downgrade for any
>> reason, period.
>
> You might not be able to do that, if python (used by emerge) uses
> something that breaks when downgrading glibc. Or gcc. Or binutils. Or
> bash. Or anything else that's needed during an emerge.
>
> So you need to check with qlop *before* downgrading, and if it looks
> like something critical was built against the new glibc, then all bets
> are off. Which is why the downgrade protection exists in the first place.
>
> The only way out of this, is restoring from backup or fixing things by
> booting from a sysrescuecd or similar.
>
> If only firefox or your media player and stuff like that got built
> against the new glibc, then it's fine to downgrade. Otherwise, you
> could end up bricking your system.
>
>
>


I see.  That would cause problems.  Depending on how bad it is affected,
even emerge -k may not work same could be said for tar to I guess.  So,
while upgrading glibc is required, eventually, it is also risky unless
it is well, very well, tested. 

I searched the wiki, I don't see anything about this topic.  I don't
know how to do the wiki thing but it would be nice for someone who does
to create a wiki page for this.  It is likely a rare thing to happen but
the consequences of it are pretty serious and tricky to fix.  To keep
from hijacking this thread anymore, I'd be happy to start a new thread,
let people post what they know and should be on the wiki and then
whoever knows how to do a wiki page move whatever is agreed on to the
page. 

Any takers? 

Dale

:-)  :-) 

[gentoo-user] Re: Failed builds of kbuild and cdrdao with "undefined reference to `__alloca'"

[Ian Zimmerman]

On 2018-02-04 15:03, tu...@posteo.de wrote:

> I still have the problem of failed builds due to an
> 'undefined reference to `__alloca''. I recompiled 
> gcc/glibc and I am using linux-4.15.1 (from kernel.org)
> with linux-headers 4.15. .
> 
> Affected are (at least) cdrdao and kbuild.

[...]

> I am using gcc (Gentoo 7.3.0 p1.0) 7.3.0. I recompiled glibc.
> I did https://wiki.gentoo.org/wiki/Upgrading_GCC
> 
> What can cause this problem and how can I fix it?

To make progress against this, you'll have to look at the build logs, or
let others look at them.

Before doing that, I would try to build the affected packages outside of
portage, saving the output of "configure" and "make".  If you also see
the problem that way, try with the stable gcc and compare logs.

BTW alloca() is a very peculiar way of allocating memory and only a few
packages use it, fortunately.

-- 
Please don't Cc: me privately on mailing lists and Usenet,
if you also post the followup to the list or newsgroup.
To reply privately _only_ on Usenet, fetch the TXT record for the domain.

[gentoo-user] Re: Forced rebuild of a package...how?

[Nikos Chantziaras]

On 04/02/18 06:20, tu...@posteo.de wrote:

Hi,

after installing linux-4.15.1 (downloaded from kernel.org) I want to
reinstall (beside others) nvidia drivers.

Emerge told me:
|>emerge nvidia-drivers


As a side-note, I recommend using this instead:

  emerge @module-rebuild

which will rebuild all external modules, so you don't have to remember 
or know which packages install external kernel modules.

Re: [gentoo-user] Re: Forced rebuild of a package...how?

[tuxic]

On 02/05 01:39, Nikos Chantziaras wrote:
> On 04/02/18 06:20, tu...@posteo.de wrote:
> > Hi,
> > 
> > after installing linux-4.15.1 (downloaded from kernel.org) I want to
> > reinstall (beside others) nvidia drivers.
> > 
> > Emerge told me:
> > |>emerge nvidia-drivers
> 
> As a side-note, I recommend using this instead:
> 
>   emerge @module-rebuild
> 
> which will rebuild all external modules, so you don't have to remember or
> know which packages install external kernel modules.
> 
> 

Hi Nikos,

thank you! :)

Cheers!
Meino