Re: [lfs-dev] L1TF : more pain for intel virtual machines

[Ken Moffat]

On Tue, Aug 14, 2018 at 08:00:16PM +0100, Ken Moffat wrote:
> 
> People may wish to read the docs patch at
> https://lore.kernel.org/patchwork/patch/974280/
> 
In a later patch, Yonah processors were removed from 'not
vulnerable' because that has not been confirmed.  But I hope nobody
is using those for VMs now.

Also, when I quoted that non-VMs are protected, that obviously only
applies to the current -rc stable kernels and (presumably) linus's
current tree.  I saw a reference to tglx's pull request on phoronix,
but so far I haven't noticed it in my mailbox.

ĸen
-- 
   Entropy not found, thump keyboard to continue

-- 
http://lists.linuxfromscratch.org/listinfo/lfs-dev
FAQ: http://www.linuxfromscratch.org/faq/
Unsubscribe: See the above information page

[lfs-dev] L1TF : more pain for intel virtual machines

[Ken Moffat]

L1TF is the latest variation in the meltdown/spectre saga.  Again,
it only applies to intel CPUs.

As part of the 4.18.1-rc1 patchset, which is available at
https://www.kernel.org/pub/linux/kernel/v4.x/stable-review/patch-4.18.1-rc1.xz
there are a whole set of patches, including documentation.
All being well, 4.18.1 will be reeleased on Thursday late afternoon
European time.  There are also -rc tarballs for 4.17.15 and 4.14.63.

People may wish to read the docs patch at
https://lore.kernel.org/patchwork/patch/974280/

quoting for physical machines -

+   The Linux kernel contains a mitigation for this attack vector, PTE
+   inversion, which is permanently enabled and has no performance
+   impact. The kernel ensures that the address bits of PTEs, which are not
+   marked present, never point to cacheable physical memory space.

ĸen
-- 
   Entropy not found, thump keyboard to continue

-- 
http://lists.linuxfromscratch.org/listinfo/lfs-dev
FAQ: http://www.linuxfromscratch.org/faq/
Unsubscribe: See the above information page

Re: [lfs-dev] Multilib patch

[Anthony Jagers]

>
> 
> >Why did you use CLFS? I thought
> the patch was intended for LFS.


I didn't find out about your patch until after I completed it. I've grown
accustom
to the following the book anyway.
-- 
http://lists.linuxfromscratch.org/listinfo/lfs-dev
FAQ: http://www.linuxfromscratch.org/faq/
Unsubscribe: See the above information page

Re: [lfs-dev] Multilib patch

[Alain Toussaint]

Le mardi 14 août 2018 à 13:29 -0400, Anthony Jagers a écrit :
> I just ran through the CLFS book with updated tarballs from SVN.
> I posted my notes here:
> 
> https://www.linuxquestions.org/questions/linux-from-scratch-13/clfs-s
> till-works-4175636269/

Why did you use CLFS? I thought the patch was intended for LFS.

Alain
-- 
http://lists.linuxfromscratch.org/listinfo/lfs-dev
FAQ: http://www.linuxfromscratch.org/faq/
Unsubscribe: See the above information page

Re: [lfs-dev] Multilib patch

[Anthony Jagers]

I just ran through the CLFS book with updated tarballs from SVN.
I posted my notes here:

https://www.linuxquestions.org/questions/linux-from-scratch-13/clfs-still-works-4175636269/
-- 
http://lists.linuxfromscratch.org/listinfo/lfs-dev
FAQ: http://www.linuxfromscratch.org/faq/
Unsubscribe: See the above information page

[lfs-dev] Multilib patch

[Thomas Trepl]

Hi,

for those who are interested in doing {,B}LFS with multilib support,
here is a patch on the LFS book adding some instructions to build the
LFS core system with multilib support.

Apply the patch to the LFS sources and run "make ARCH=multilib". If
leaving out the ARCH= parameter or set it to ARCH=default will produce
the book without any ML stuff in it.

Running thru the book created with ARCH=multilib will provide you a
system with m64, mx32 and m32 ABI.

The patch is based on the latest development LFS book (20180808), not
yet tested on the 8.3-rc1.

I had to split up building glibc in chapter 6 into two sections, the
"usual" build of 64-bit glibc and than the 32-bit glibc with having the
adjustment section in between. Otherwise i ran into the "glibc builds
endlessly"-issue.
Root directory gets a bit polluted by two additional dirs(symlinks)
named /lib32 and /libx32. I think there could be a way to avoid this
when diving deeper into the gcc-specs-stuff and such.

The patch also adds ISL as a new package. I'm not sure whether there is
a real dependency on that, I just included it as DJ did that in his ML-
book, too.

Whats missing:
* The kernel chapter should be enhanced with instructions to enable the
required emulation support.
* My building system has multilib support allready builtin. Need to
redo testing from a single-64bit-ABI-system to ensure that there are no
silent cross-references. 

And finally:
I'm pretty sure that there are some bugs/flaws/whatever in the patch
(and the instructions it provides), consider the patch as an alpha
version. Please be invited to report bugs, provide tweaks, comments and
enhancements!
Yes, ML goes a bit beyond the basic educational approach of LFS but
there is still software out there (in my case a printer driver from
DELL) which is only provided as 32bit-binary and this renders a pure-
64bit LFS useless to some extend.

And big thanks to DJ, Nathan and William F. for their recent work on
multilib support!

--
Thomas
diff -Naur -x .svn -x 'lfs-bootscripts-*.tar.bz2' -x appendices BOOK-orig/chapter01/askforhelp.xml BOOK-multilib/chapter01/askforhelp.xml
--- BOOK-orig/chapter01/askforhelp.xml	2018-08-10 07:50:45.0 +0200
+++ BOOK-multilib/chapter01/askforhelp.xml	2018-08-12 21:15:20.298266973 +0200
@@ -39,7 +39,8 @@
 
   
 The version of the book being used (in this case 
-  &version;
+  &version;
+  &version;-multilib
   &versiond;)
   
   
diff -Naur -x .svn -x 'lfs-bootscripts-*.tar.bz2' -x appendices BOOK-orig/chapter01/changelog.xml BOOK-multilib/chapter01/changelog.xml
--- BOOK-orig/chapter01/changelog.xml	2018-08-10 07:50:45.0 +0200
+++ BOOK-multilib/chapter01/changelog.xml	2018-08-12 21:13:51.669498909 +0200
@@ -11,7 +11,8 @@
   Changelog
 
   This is version 
-&version;
+&version;
+&version;-multilib
 &versiond;
   of the Linux From Scratch book, dated
   &releasedate;. If this book is more than six months old, a newer and better
diff -Naur -x .svn -x 'lfs-bootscripts-*.tar.bz2' -x appendices BOOK-orig/chapter03/packages.xml BOOK-multilib/chapter03/packages.xml
--- BOOK-orig/chapter03/packages.xml	2018-08-10 07:50:45.0 +0200
+++ BOOK-multilib/chapter03/packages.xml	2018-08-10 09:43:39.531077226 +0200
@@ -356,6 +356,15 @@
 
 
 
+  ISL (&isl-version;) - &isl-size;:
+  
+Home page: 
+Download: 
+MD5 sum: &isl-md5;
+  
+
+
+
   Kbd (&kbd-version;) - &kbd-size;:
   
 Home page: 
diff -Naur -x .svn -x 'lfs-bootscripts-*.tar.bz2' -x appendices BOOK-orig/chapter04/settingenviron.xml BOOK-multilib/chapter04/settingenviron.xml
--- BOOK-orig/chapter04/settingenviron.xml	2018-08-10 07:50:46.0 +0200
+++ BOOK-multilib/chapter04/settingenviron.xml	2018-08-12 13:38:55.044890706 +0200
@@ -37,7 +37,7 @@
   .bashrc file instead. Create the
   .bashrc file now:
 
-cat > ~/.bashrc << "EOF"
+cat > ~/.bashrc << "EOF"
 set +h
 umask 022
 LFS=/mnt/lfs
@@ -46,6 +46,17 @@
 PATH=/tools/bin:/bin:/usr/bin
 export LFS LC_ALL LFS_TGT PATH
 EOF
+cat > ~/.bashrc << "EOF"
+set +h
+umask 022
+LFS=/mnt/lfs
+LC_ALL=POSIX
+LFS_TGT=x86_64-lfs-linux-gnu
+LFS_TGT32=i686-lfs-linux-gnu
+LFS_TGTX32=x86_64-lfs-linux-gnux32
+PATH=/tools/bin:/bin:/usr/bin
+export LFS LC_ALL LFS_TGT LFS_TGT32 LFS_TGTX32 PATH
+EOF
 
   The set +h command turns off
   bash's hash function. Hashing is ordinarily a useful
diff -Naur -x .svn -x 'lfs-bootscripts-*.tar.bz2' -x appendices BOOK-orig/chapter05/binutils-pass1.xml BOOK-multilib/chapter05/binutils-pass1.xml
--- BOOK-orig/chapter05/binutils-pass1.xml	2018-08-10 07:50:46.0 +0200
+++ BOOK-multilib/chapter05/binutils-pass1.xml	2018-08-12 17:07:56.147972097 +0200
@@ -72,12 +72,20 @@
 
 Now prepare Binutils for compilation:
 
-../configure --prefix=/tools\
+../configure --prefix=/tools \
  --with-sysroot=$LFS\
  -