[gentoo-user] Any consequences to package.mask'ing newer kernels?

[gentoo-user] Any consequences to package.mask'ing newer kernels?

[Walter Dnes]

  Partial "df" output before unmerging a bunch of kernels 
Filesystem           1K-blocks      Used Available Use% Mounted on
/dev/sda1             11726996   7325372   4401624  63% /


  Partial "df" output after  unmerging a bunch of kernels 
Filesystem           1K-blocks      Used Available Use% Mounted on
/dev/sda1             11726996   4183616   7543380  36% /

  Yes folks, 3.14 gigs.  Having gotten rather tired of doing this
manually... again... I went into /etc/portage/package.mask and added

>sys-kernel/gentoo-sources-2.6.19-r5

  It won't hurt me now, but is there anything that might depend on newer
kernels?  It's assumed I'll upgrade when required by a security alert.
Other than that, how long can I get away between kernel upgrades?

-- 
Walter Dnes <waltdnes@waltdnes.org> In linux /sbin/init is Job #1
Q. Mr. Ghandi, what do you think of Microsoft security?
A. I think it would be a good idea.
-- 
gentoo-user@gentoo.org mailing list

Re: [gentoo-user] Any consequences to package.mask'ing newer kernels?

[Neil Bothwick]

[-- Attachment #1: Type: text/plain, Size: 1711 bytes --]

On Sat, 31 Mar 2007 01:55:10 -0400, Walter Dnes wrote:

>   Partial "df" output before unmerging a bunch of kernels 
> Filesystem           1K-blocks      Used Available Use% Mounted on
> /dev/sda1             11726996   7325372   4401624  63% /
> 
> 
>   Partial "df" output after  unmerging a bunch of kernels 
> Filesystem           1K-blocks      Used Available Use% Mounted on
> /dev/sda1             11726996   4183616   7543380  36% /
> 
>   Yes folks, 3.14 gigs.  Having gotten rather tired of doing this
> manually... again... 

At around 300MB per kernel, that's ten excess kernels, so you can't be
doing it that often. Once you're happy with the current kernel, you only
need "emerge -P gentoo-sources" to remove the rest. I use a script that
removes all but the last two, and also cleans out /lib/modules and /boot.

> I went into /etc/portage/package.mask and added
> 
> >sys-kernel/gentoo-sources-2.6.19-r5  
> 
>   It won't hurt me now, but is there anything that might depend on newer
> kernels?  It's assumed I'll upgrade when required by a security alert.
> Other than that, how long can I get away between kernel upgrades?

As long as you like, some people are still running 2.4 kernels! As long
as you don't add hardware supported only by a newer kernel, your system
will work exactly the same in six months as it does now, although you
should bear in mind that -r updates are generally problem fixes.

I usually read the Changelog when a new kernel is released and then
decide whether it's worth installing. There's no point in forcing a
reboot when the old kernel works for me.


-- 
Neil Bothwick

Reality is for people who can't handle Star Trek

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-user] Any consequences to package.mask'ing newer kernels?

[Walter Dnes]

On Sat, Mar 31, 2007 at 09:21:22AM +0100, Neil Bothwick wrote

> At around 300MB per kernel, that's ten excess kernels, so you can't be
> doing it that often.

  I ran df and ll between each individual unmerge.  The individual
kernels take approx 250 megs, freshly emerged.  Compiling generates
another 200 megs worth of object code, etc.  Here's partial output of
"ll" before the cleanup.  Note that 2.6.16-r7, 2.6.17-r7, and 2.6.18-r3
were compiled, as well as the current 2.6.19-r5.

drwxr-xr-x 19 root root  744 Sep  6  2006 linux-2.6.16-gentoo-r13
drwxr-xr-x 19 root root  744 May  4  2006 linux-2.6.16-gentoo-r6
drwxr-xr-x 20 root root 1488 Oct 14 02:14 linux-2.6.16-gentoo-r7
drwxr-xr-x 19 root root  744 Jun 13  2006 linux-2.6.16-gentoo-r9
drwxr-xr-x 19 root root  712 Jul 29  2006 linux-2.6.17-gentoo-r4
drwxr-xr-x 20 root root 1448 Sep  6  2006 linux-2.6.17-gentoo-r7
drwxr-xr-x 19 root root  744 Sep 16  2006 linux-2.6.17-gentoo-r8
drwxr-xr-x 19 root root  712 Nov 12 09:01 linux-2.6.18-gentoo-r2
drwxr-xr-x 20 root root 1328 Feb 17 18:51 linux-2.6.18-gentoo-r3
drwxr-xr-x 19 root root  712 Dec 24 22:14 linux-2.6.18-gentoo-r5
drwxr-xr-x 19 root root  712 Jan 14 20:15 linux-2.6.18-gentoo-r6
drwxr-xr-x 20 root root 1328 Mar  8 19:32 linux-2.6.19-gentoo-r5

  I don't mind the 30 or 40 megs for the source tarball+patches in my
distfiles directory.  But the quarter gig for each minor "r" bump, most
of which I never build, is a bit much.

-- 
Walter Dnes <waltdnes@waltdnes.org> In linux /sbin/init is Job #1
Q. Mr. Ghandi, what do you think of Microsoft security?
A. I think it would be a good idea.
-- 
gentoo-user@gentoo.org mailing list

Re: [gentoo-user] Any consequences to package.mask'ing newer kernels?

[Neil Bothwick]

[-- Attachment #1: Type: text/plain, Size: 434 bytes --]

On Mon, 2 Apr 2007 01:02:32 -0400, Walter Dnes wrote:

>   I don't mind the 30 or 40 megs for the source tarball+patches in my
> distfiles directory.  But the quarter gig for each minor "r" bump, most
> of which I never build, is a bit much.

Why install it if you're not going to build it?

r bumps are necessarily minor, as they often contain security fixes.


-- 
Neil Bothwick

(A)bort (R)etry (T)ake an axe to it?

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-user] Any consequences to package.mask'ing newer kernels?

[Boyd Stephen Smith Jr.]

[-- Attachment #1: Type: text/plain, Size: 879 bytes --]

On Saturday 31 March 2007 00:55:10 Walter Dnes wrote:
> Having gotten rather tired of doing this
> manually... again... I went into /etc/portage/package.mask and added
>
> >sys-kernel/gentoo-sources-2.6.19-r5
>
>   It won't hurt me now, but is there anything that might depend on newer
> kernels?  It's assumed I'll upgrade when required by a security alert.
> Other than that, how long can I get away between kernel upgrades?

Just watch your GLSAs, and you should be fine.  If a package depends on a 
certain kernel version, it will list that in it's depend line, and emerge 
will complain that it can't satisfy that dependency.

-- 
Boyd Stephen Smith Jr.                     ,= ,-_-. =. 
bss03@volumehost.net                      ((_/)o o(\_))
ICQ: 514984 YM/AIM: DaTwinkDaddy           `-'(. .)`-' 
http://iguanasuicide.org/                      \_/     

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-user] Any consequences to package.mask'ing newer kernels?

[Hemmann, Volker Armin]

On Samstag, 31. März 2007, Boyd Stephen Smith Jr. wrote:
> On Saturday 31 March 2007 00:55:10 Walter Dnes wrote:
> > Having gotten rather tired of doing this
> > manually... again... I went into /etc/portage/package.mask and added
> >
> > >sys-kernel/gentoo-sources-2.6.19-r5
> >
> >   It won't hurt me now, but is there anything that might depend on newer
> > kernels?  It's assumed I'll upgrade when required by a security alert.
> > Other than that, how long can I get away between kernel upgrades?
>
> Just watch your GLSAs, and you should be fine.  If a package depends on a
> certain kernel version, it will list that in it's depend line, and emerge
> will complain that it can't satisfy that dependency.

Well, there are many security related kernel bugs (just read the Changelogs of 
the .18, .19 and .20 'stable' releases), but hardly any kernel GLSAs. It 
looks like kernels are not covered by GLSAs.
--
gentoo-user@gentoo.org mailing list

Re: [gentoo-user] Any consequences to package.mask'ing newer kernels?

[Hemmann, Volker Armin]

On Samstag, 31. März 2007, Walter Dnes wrote:
>   Partial "df" output before unmerging a bunch of kernels
> Filesystem           1K-blocks      Used Available Use% Mounted on
> /dev/sda1             11726996   7325372   4401624  63% /
>
>
>   Partial "df" output after  unmerging a bunch of kernels
> Filesystem           1K-blocks      Used Available Use% Mounted on
> /dev/sda1             11726996   4183616   7543380  36% /
>
>   Yes folks, 3.14 gigs.  Having gotten rather tired of doing this
> manually... again... I went into /etc/portage/package.mask and added
>
> >sys-kernel/gentoo-sources-2.6.19-r5
>
>   It won't hurt me now, but is there anything that might depend on newer
> kernels?  It's assumed I'll upgrade when required by a security alert.
> Other than that, how long can I get away between kernel upgrades?
>

besides critical bug fixes, security fixes and driver updates?

IMHO masking never kernels is a really bad idea.
--
gentoo-user@gentoo.org mailing list

Re: [gentoo-user] Any consequences to package.mask'ing newer kernels?

[b.n.]

Hemmann, Volker Armin ha scritto:
  > besides critical bug fixes, security fixes and driver updates?
> 
> IMHO masking never kernels is a really bad idea.

Why? I upgrade my kernel once in a blue moon -that is, when I need to 
because of new features I need, because of incompatibility with current 
system (i.e. upgrade to udev) or just because I need to recompile to 
enable/disable something and, since I'm recompiling anyway, I also 
upgrade.(note that I don't use genkernel)

Otherwise, I personally don't bother. If my kernel currently serves me 
well and there is nothing new I really need, why having to recompile, 
with the risk to make a mess?

m.
-- 
gentoo-user@gentoo.org mailing list

Re: [gentoo-user] Any consequences to package.mask'ing newer kernels?

[Hemmann, Volker Armin]

On Samstag, 31. März 2007, b.n. wrote:
> Hemmann, Volker Armin ha scritto:
>   > besides critical bug fixes, security fixes and driver updates?
> >
> > IMHO masking never kernels is a really bad idea.
>
> Why? 

because of:

- filesystem bugs (2.6.17 and XFS for example)

- security problems (local and remote holes). 

In almost every kernel release a security problem is found, that is fixed in a 
stable release.

> I upgrade my kernel once in a blue moon -that is, when I need to 
> because of new features I need, because of incompatibility with current
> system (i.e. upgrade to udev) or just because I need to recompile to
> enable/disable something and, since I'm recompiling anyway, I also
> upgrade.(note that I don't use genkernel)

and between that blue moons, your box is wide open to attacks.

>
> Otherwise, I personally don't bother. If my kernel currently serves me
> well and there is nothing new I really need, why having to recompile,
> with the risk to make a mess?

Which risk? Which mess? There is not a risk, if you use oldconfig. But there 
is a big risk in security holes.
--
gentoo-user@gentoo.org mailing list

Re: [gentoo-user] Any consequences to package.mask'ing newer kernels?

[b.n.]

Hemmann, Volker Armin ha scritto:

> In almost every kernel release a security problem is found, that is fixed in a 
> stable release.

Stable release? AFAIK, *all* 2.6.x releases are stable releases. The 
days of double trees (2.4.x and 2.5.x) are gone.
Probably I don't get what you mean. I use x86 kernels, not ~x86: that's 
what you mean as stable? I don't understand.

> 
> and between that blue moons, your box is wide open to attacks.

Well, if in *every* kernel there is *always* a security problem, my box 
is always open to attacks... :)

(I understand your point, however. I didn't realize the linux kernel was 
so full of security holes. I thought it was one of the most secure 
components. Why aren't there GLSAs for the kernel?)

> Which risk? Which mess? There is not a risk, if you use oldconfig. 

oldconfig doesn't always work well between major releases (2.6.x vs 
2.6.x+1).

>But there 
> is a big risk in security holes.

True, but can you explain me the points above?

m.
-- 
gentoo-user@gentoo.org mailing list

Re: [gentoo-user] Any consequences to package.mask'ing newer kernels?

[Hemmann, Volker Armin]

On Sonntag, 1. April 2007, b.n. wrote:
> Hemmann, Volker Armin ha scritto:
> > In almost every kernel release a security problem is found, that is fixed
> > in a stable release.
>
> Stable release? AFAIK, *all* 2.6.x releases are stable releases.

No, they aren't. There are the 'normal' releases (for example 2.6.20) and 
the 'stable' releases which fix important bugs and security holes (like, for 
example 2.6.20.2).

> The 
> days of double trees (2.4.x and 2.5.x) are gone.

Today we have at least 4 trees.
Linus.
Morton.
The 'stable releases' (2.6.XY.Z)
Bunk's 2.6.16.XY


> > Which risk? Which mess? There is not a risk, if you use oldconfig.
>
> oldconfig doesn't always work well between major releases (2.6.x vs
> 2.6.x+1).
>

I works like a charm for me....
-- 
gentoo-user@gentoo.org mailing list

Re: [gentoo-user] Any consequences to package.mask'ing newer kernels?

[b.n.]

Hemmann, Volker Armin ha scritto:
> On Sonntag, 1. April 2007, b.n. wrote:
>> Hemmann, Volker Armin ha scritto:
>>> In almost every kernel release a security problem is found, that is fixed
>>> in a stable release.
>> Stable release? AFAIK, *all* 2.6.x releases are stable releases.
> 
> No, they aren't. There are the 'normal' releases (for example 2.6.20) and 
> the 'stable' releases which fix important bugs and security holes (like, for 
> example 2.6.20.2).

Yes, I know that. I didn't call them "unstable" and "stable", that's why 
I was confused, however I know.
Now my questions are:
1)I only see gentoo-sources-2.6.X-rY, I never see 
gentoo-sources-2.6.X.a.b-rY .What am I installing when I install 
gentoo-sources-2.6.x-rY?

2)How do the binary distribution people cope with this?

>> The 
>> days of double trees (2.4.x and 2.5.x) are gone.
> 
> Today we have at least 4 trees.
> Linus.
> Morton.
> The 'stable releases' (2.6.XY.Z)
> Bunk's 2.6.16.XY

Well, there have ALWAYS been a lot of different trees, but Morton, for 
example, AFAIK is not an "official" tree (although it is maintained 
closely to the official).

However that's just nitpicking. :)

>>> Which risk? Which mess? There is not a risk, if you use oldconfig.
>> oldconfig doesn't always work well between major releases (2.6.x vs
>> 2.6.x+1).
>>
> I works like a charm for me....

Not for me. And I've sometimes read of newer kernels breaking things on 
the gentoo mailing list. Upgrading a kernel is never straightforward, 
imho (maybe it's me being unexperienced, however it's my years-old only 
desktop box and I hate to b0rk it).

m.
-- 
gentoo-user@gentoo.org mailing list

Re: [gentoo-user] Any consequences to package.mask'ing newer kernels?

[Hemmann, Volker Armin]

On Montag, 2. April 2007, b.n. wrote:
> Hemmann, Volker Armin ha scritto:
> > On Sonntag, 1. April 2007, b.n. wrote:
> >> Hemmann, Volker Armin ha scritto:
> >>> In almost every kernel release a security problem is found, that is
> >>> fixed in a stable release.
> >>
> >> Stable release? AFAIK, *all* 2.6.x releases are stable releases.
> >
> > No, they aren't. There are the 'normal' releases (for example 2.6.20) and
> > the 'stable' releases which fix important bugs and security holes (like,
> > for example 2.6.20.2).
>
> Yes, I know that. I didn't call them "unstable" and "stable", that's why
> I was confused, however I know.
> Now my questions are:
> 1)I only see gentoo-sources-2.6.X-rY, I never see
> gentoo-sources-2.6.X.a.b-rY .What am I installing when I install
> gentoo-sources-2.6.x-rY?

look into the changelogs ;) 
I don't use gentoo-sources, but AFAIK, the -rX releases are related to the 
vanilla .X releases.

>
> 2)How do the binary distribution people cope with this?

backporting patches. That is why you get kernels named '2.6.17-201' and stuff 
like that.

>
> >> The
> >> days of double trees (2.4.x and 2.5.x) are gone.
> >
> > Today we have at least 4 trees.
> > Linus.
> > Morton.
> > The 'stable releases' (2.6.XY.Z)
> > Bunk's 2.6.16.XY
>
> Well, there have ALWAYS been a lot of different trees, but Morton, for
> example, AFAIK is not an "official" tree (although it is maintained
> closely to the official).

It is the official testing tree. Every new feature and lots of patches and 
drivers have to 'mature' in Morton's tree - and he decides, together with the 
maintainers, which stuff goes to Linus.
-- 
gentoo-user@gentoo.org mailing list

Re: [gentoo-user] Any consequences to package.mask'ing newer kernels?

[Neil Bothwick]

[-- Attachment #1: Type: text/plain, Size: 846 bytes --]

On Sun, 1 Apr 2007 23:35:25 +0200, Hemmann, Volker Armin wrote:

> > 1)I only see gentoo-sources-2.6.X-rY, I never see
> > gentoo-sources-2.6.X.a.b-rY .What am I installing when I install
> > gentoo-sources-2.6.x-rY?  
> 
> look into the changelogs ;) 
> I don't use gentoo-sources, but AFAIK, the -rX releases are related to
> the vanilla .X releases.

Not necessarily, which is why you need to read the changelogs. For
example, 2.6.21-r1 may be released to fix something with 2.6.20, so when
2.6.21.1 is released, it will be in 2.6.21-r2. Bit it is reasonable to
assume that the latest -r release is based on the latest revision of the
kernel.

Note that stable has different meanings depending on whether you apply
it to the kernel or the ebuild.


-- 
Neil Bothwick

If this were an actual tagline, it would be funny.

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-user] Any consequences to package.mask'ing newer kernels?

[Walter Dnes]

On Sat, Mar 31, 2007 at 04:11:42PM +0200, Hemmann, Volker Armin wrote

> Which risk? Which mess? There is not a risk, if you use oldconfig.

  With oldconfig, 99% of the updates seem to consist of added support
for exotic raid controllers or network cards.  Since my system has been
running OK for the past couple of years without the new features, I
obviously don't need them.  I end up hitting "N" all the time.

  I got bitten in the latest stable kernel (2.6.19-r5).  It moved SATA
support out of SCSI, and into a separate section altogether.  I plowed
through "make oldconfig", hitting "N" for every option.  Because I have
a SATA drive, the result was kernel panic when I rebooted into the new
kernel.

-- 
Walter Dnes <waltdnes@waltdnes.org> In linux /sbin/init is Job #1
Q. Mr. Ghandi, what do you think of Microsoft security?
A. I think it would be a good idea.
-- 
gentoo-user@gentoo.org mailing list

Re: [gentoo-user] Any consequences to package.mask'ing newer kernels?

[Neil Bothwick]

[-- Attachment #1: Type: text/plain, Size: 742 bytes --]

On Mon, 2 Apr 2007 01:35:42 -0400, Walter Dnes wrote:

>   I got bitten in the latest stable kernel (2.6.19-r5).  It moved SATA
> support out of SCSI, and into a separate section altogether.  I plowed
> through "make oldconfig", hitting "N" for every option.  Because I have
> a SATA drive, the result was kernel panic when I rebooted into the new
> kernel.

Same here, and it was the only time I've had a problem with make
oldconfig since switching to 2.6 however long ago. That really was a one
off situation that should have been addressed with a suitable ewarn, but
hindsight makes that easy.


-- 
Neil Bothwick

deja vous - the act of forgetting someone's name /again/ despite being
introduced to them several times.

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-user] Any consequences to package.mask'ing newer kernels?

[Bo Ørsted Andresen]

[-- Attachment #1: Type: text/plain, Size: 872 bytes --]

On Friday 13 April 2007 01:59:47 Neil Bothwick wrote:
> > Provided you have gentoolkit something as simple as this works:
> >
> > # emerge -Cva $(equery -q list gentoo-sources | head -n -2)
>
> That only cleans out /usr/src, it's slightly different to what I use
> (which rm's the directories first to speed things up) but does basically
> the same. You also need to clear out /lib/modules and /boot with

True.

> Here's the script I use, which is guaranteed to work when it doesn't
> fail. When it does break, you can keep the pieces.
[SNIP]
> equery --quiet --nocolor list --duplicates gentoo-sources | awk '{print $1}' | head -n -2 | xargs --no-run-if-empty emerge --unmerge &>/dev/null

Out of interest:

1) Why --duplicates (i.e. am I missing something ;).
2) Why the awk? Is there ever more than one column without --no-pipe ?
-- 
Bo Andresen

[-- Attachment #2: Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-user] Any consequences to package.mask'ing newer kernels?

[Neil Bothwick]

[-- Attachment #1: Type: text/plain, Size: 981 bytes --]

On Fri, 13 Apr 2007 02:11:58 +0200, Bo Ørsted Andresen wrote:

> > equery --quiet --nocolor list --duplicates gentoo-sources | awk
> > '{print $1}' | head -n -2 | xargs --no-run-if-empty emerge --unmerge
> > &>/dev/null  
> 
> Out of interest:
> 
> 1) Why --duplicates (i.e. am I missing something ;).

No idea, I wrote the script a couple of years ago, if I'd wanted to
remember how it worked I'd have commented it. :) It was never intended
for per review, just a quick hack top free up some disk space. 
I originally used qpkg, but you're right that equery doesn't need
--duplicates.

> 2) Why the awk? Is there ever more than one column without --no-pipe ?

I don't think so, but it doesn't hurt to leave it in. I used awk
because the terminal output adds extra fields and I possibly didn't even
bother to check whether they were still there when using a pipe.


-- 
Neil Bothwick

Q:  Why is top-posting evil?
A: backwards read don't humans because

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-user] Any consequences to package.mask'ing newer kernels?

[Neil Bothwick]

[-- Attachment #1: Type: text/plain, Size: 642 bytes --]

On Fri, 13 Apr 2007 08:40:10 +0100, Neil Bothwick wrote:

> > > equery --quiet --nocolor list --duplicates gentoo-sources | awk
> > > '{print $1}' | head -n -2 | xargs --no-run-if-empty emerge
> > > --unmerge &>/dev/null    
> > 
> > Out of interest:
> > 
> > 1) Why --duplicates (i.e. am I missing something ;).  
> 
> No idea, I wrote the script a couple of years ago, if I'd wanted to
> remember how it worked I'd have commented it. :)

I remember now. If you have only one kernel installed, --duplicates
prevents it being uninstalled - quite a useful feature ;-)


-- 
Neil Bothwick

Oxymoron: Clearly Misunderstood.

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-user] Any consequences to package.mask'ing newer kernels?

[Bo Ørsted Andresen]

[-- Attachment #1: Type: text/plain, Size: 807 bytes --]

On Monday 16 April 2007 15:00:30 Neil Bothwick wrote:
> On Fri, 13 Apr 2007 08:40:10 +0100, Neil Bothwick wrote:
> > > > equery --quiet --nocolor list --duplicates gentoo-sources | awk
> > > > '{print $1}' | head -n -2 | xargs --no-run-if-empty emerge
> > > > --unmerge &>/dev/null
> > >
> > > Out of interest:
> > >
> > > 1) Why --duplicates (i.e. am I missing something ;).
> >
> > No idea, I wrote the script a couple of years ago, if I'd wanted to
> > remember how it worked I'd have commented it. :)
>
> I remember now. If you have only one kernel installed, --duplicates
> prevents it being uninstalled - quite a useful feature ;-)

head -n -2 would prevent that anyway. As well as preventing the deletion 
from /boot and /lib/modules (where it really matters)..

-- 
Bo Andresen

[-- Attachment #2: This is a digitally signed message part. --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-user] Any consequences to package.mask'ing newer kernels?

[Neil Bothwick]

[-- Attachment #1: Type: text/plain, Size: 593 bytes --]

On Mon, 16 Apr 2007 15:06:54 +0200, Bo Ørsted Andresen wrote:

> > I remember now. If you have only one kernel installed, --duplicates
> > prevents it being uninstalled - quite a useful feature ;-)  
> 
> head -n -2 would prevent that anyway. As well as preventing the
> deletion from /boot and /lib/modules (where it really matters)..

Of course it will, I'm not thinking straight today. I suppose I could
remove the --duplicates and if you don't hear from me again, you'll know
there was a good reason for it ;-)


-- 
Neil Bothwick

A Microsoft joke (is that a tautology?)

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 189 bytes --]

Re: [gentoo-user] Any consequences to package.mask'ing newer kernels?

[Anthony E. Caudel]

Neil Bothwick wrote:
> On Fri, 13 Apr 2007 01:33:43 +0200, Bo Ørsted Andresen wrote:
> 
>>>> At around 300MB per kernel, that's ten excess kernels, so you can't
>>>> be doing it that often. Once you're happy with the current kernel,
>>>> you only need "emerge -P gentoo-sources" to remove the rest. I use
>>>> a script that removes all but the last two, and also cleans
>>>> out /lib/modules and /boot.  
>>> Neil, any chance we could get that script?  
>> Provided you have gentoolkit something as simple as this works:
>>
>> # emerge -Cva $(equery -q list gentoo-sources | head -n -2)
> 
> That only cleans out /usr/src, it's slightly different to what I use
> (which rm's the directories first to speed things up) but does basically
> the same. You also need to clear out /lib/modules and /boot with
> 
> Here's the script I use, which is guaranteed to work when it doesn't
> fail. When it does break, you can keep the pieces.
> 
> #!/bin/bash
> 
> # clean /lib/modules
> cd /lib/modules
> ls -1rt | head -n -2 | xargs --no-run-if-empty rm -fr
> 
> # clean /boot
> grep --quiet /boot /etc/fstab && mount /boot -o remount,rw
> 
> cd /boot
> ls -1rt config-* | head -n -2 | while read f; do
> 	bzip2 -9 $f
> 	mv $f.bz2 oldconfigs/
> 	done
> 
> ls -1rt System.map-* | head -n -2 | xargs --no-run-if-empty rm -f
> if [ -f vmlinux ]; then
> 	ls -1rt vmlinux-* | head -n -2 | xargs --no-run-if-empty rm -f
> else
> 	ls -1rt vmlinuz-* | head -n -2 | xargs --no-run-if-empty rm -f 
> 	fi
> 
> # clean /usr/src
> cd /usr/src
> ls -1drt linux-* | head -n -2 | xargs --no-run-if-empty rm -fr
> equery --quiet --nocolor list --duplicates gentoo-sources | awk '{print $1}' | head -n -2 | xargs --no-run-if-empty emerge --unmerge &>/dev/null
> 
> grep --quiet /boot /etc/fstab && mount /boot -o remount,ro
> # END
> 
> The vmlinuz/vmlinux stuff is because I have a PPC system too, which calls the kernel vmlinux.
> 
> 

Thanks Neil.

Tony

-- 
Those who would give up essential Liberty, to purchase a little
temporary Safety, deserve neither Liberty nor Safety.
   -- Benjamin Franklin
-- 
gentoo-user@gentoo.org mailing list