Re: Missing "standard" headers in FC37 arm (aarch64) C/C++ cross-compiler RPMs ??

[Jeffrey Walton]

On Thu, Jun 1, 2023 at 4:30 PM ron flory via users
 wrote:
>
> Hi- am hopefully missing something dumb/obvious here---
>
>  Am trying to cross-compile for raspberry-pi, but some basic headers appear 
> to be missing, or not redirecting to 'generics'.
>
> -
> Install the cross-compiler(s):
> dnf install gcc-aarch64-linux-gnu  gcc-c++-aarch64-linux-gnu
>
> -
> Compile hello_world sample:
> aarch64-linux-gnu-gcc  hello_world.cpp -o hello_world
>
> Results:
> hello_world.c:1:10: fatal error: stdio.h: No such file or directory
> 1 | #include 
>   |  ^
> compilation terminated.
>
> -
> Hello_world sample consists of:
>
> #include 
> #include 
>
> int main(int argc, char *argv[])
> {
> printf("Hello World\n");
> return(0);
> }

When cross-compiling, you typically need to setup the tools (for the
cross-compile) and the include and library path (for the
cross-compile). Assuming aarch64-linux-gnu-gcc is the only tool you
need, you are still missing include and library paths.

You can add include and library paths via --isysroot and --sysroot. So
something like:

AARCH64_INCL="..."
AARCH64_LIB="..."
aarch64-linux-gnu-gcc -o hello_world \
--isysroot="${AARCH64_INCL}" hello_world.cpp \
--sysroot="${AARCH64_LIB}"

When you are building for the host, like x86_64, stuff is found in
/usr by default, and you don't need to specify --isysroot and
--sysroot.

When building for aarch64, your cross-compile headers will be
somewhere like /usr/aarch64/include, and libraries will be in
/usr/aarch64/lib. So you would use AARCH64_INCL=/usr/aarch64 and
AARCH64_LIB=/usr/aarch64. --isysroot should add the trailing include/
for you, and --sysroot should add the trailing lib/ for you.

In reality, it is usually a little more complex, like when building a
library for Android or iOS. In this case you usually want to set your
tools, too. (Tools like CC, CXX, AR and LD). For an example of scripts
that setup those environments, see
https://github.com/weidai11/cryptopp/blob/master/TestScripts/setenv-android.sh
and https://github.com/weidai11/cryptopp/blob/master/TestScripts/setenv-ios.sh
.

Jeff
___
users mailing list -- users@lists.fedoraproject.org
To unsubscribe send an email to users-le...@lists.fedoraproject.org
Fedora Code of Conduct: 
https://docs.fedoraproject.org/en-US/project/code-of-conduct/
List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines
List Archives: 
https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org
Do not reply to spam, report it: 
https://pagure.io/fedora-infrastructure/new_issue

Re: screen not waking from sleep - Fedora 38

[Sbob]

Yes I installed the rpmfusion repos and did a dnf -y update

(and rebooted)


On 6/4/23 18:42, George N. White III wrote:

On Sun, Jun 4, 2023 at 8:26 PM Sbob  wrote:

Hi all;


I just installed Fedora 38 and when the laptop puts the screen to
sleep
it never wakes up.


Have you also installed the recent updates?


Running Fedora 38 on a Dell XPS 17


There were reports like this when F38 first appeared on 
, but
that seems to have died down -- either people notice previous reports 
and don't generate new

ones or updates have fixed the issues.


Thoughts?
___
users mailing list -- users@lists.fedoraproject.org
To unsubscribe send an email to users-le...@lists.fedoraproject.org
Fedora Code of Conduct:
https://docs.fedoraproject.org/en-US/project/code-of-conduct/
List Guidelines:
https://fedoraproject.org/wiki/Mailing_list_guidelines
List Archives:
https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org
Do not reply to spam, report it:
https://pagure.io/fedora-infrastructure/new_issue



--
George N. White III


___
users mailing list --users@lists.fedoraproject.org
To unsubscribe send an email tousers-le...@lists.fedoraproject.org
Fedora Code of 
Conduct:https://docs.fedoraproject.org/en-US/project/code-of-conduct/
List Guidelines:https://fedoraproject.org/wiki/Mailing_list_guidelines
List 
Archives:https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org
Do not reply to spam, report 
it:https://pagure.io/fedora-infrastructure/new_issue___
users mailing list -- users@lists.fedoraproject.org
To unsubscribe send an email to users-le...@lists.fedoraproject.org
Fedora Code of Conduct: 
https://docs.fedoraproject.org/en-US/project/code-of-conduct/
List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines
List Archives: 
https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org
Do not reply to spam, report it: 
https://pagure.io/fedora-infrastructure/new_issue

Re: screen not waking from sleep - Fedora 38

[George N. White III]

On Sun, Jun 4, 2023 at 8:26 PM Sbob  wrote:

> Hi all;
>
>
> I just installed Fedora 38 and when the laptop puts the screen to sleep
> it never wakes up.
>

Have you also installed the recent updates?

>
> Running Fedora 38 on a Dell XPS 17
>

There were reports like this when F38 first appeared on <
https://discussion.fedoraproject.org>, but
that seems to have died down -- either people notice previous reports and
don't generate new
ones or updates have fixed the issues.


>
> Thoughts?
> ___
> users mailing list -- users@lists.fedoraproject.org
> To unsubscribe send an email to users-le...@lists.fedoraproject.org
> Fedora Code of Conduct:
> https://docs.fedoraproject.org/en-US/project/code-of-conduct/
> List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines
> List Archives:
> https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org
> Do not reply to spam, report it:
> https://pagure.io/fedora-infrastructure/new_issue
>


-- 
George N. White III
___
users mailing list -- users@lists.fedoraproject.org
To unsubscribe send an email to users-le...@lists.fedoraproject.org
Fedora Code of Conduct: 
https://docs.fedoraproject.org/en-US/project/code-of-conduct/
List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines
List Archives: 
https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org
Do not reply to spam, report it: 
https://pagure.io/fedora-infrastructure/new_issue

Re: How to set up dhcpd.conf to serve different UEFI files per OS

[Thomas Cameron]

On 6/4/23 17:30, Chris Adams wrote:

Again, the DHCP request that gets a response "use this file" comes from
the firmware, not the OS.

It goes something like:

- BIOS/UEFI configured for network boot sends DHCP request
- DHCP server says "use this file (aka shim)"
- BIOS/UEFI loads that file and runs it
- shim loads grub2 or pxelinux, they get their configs/menus
- you choose an OS to load
- grub2/pxelinux fetches kernel and checks the signature and fails it it
   doesn't match

The step that has to change between Fedora and RHEL is step #2, when the
DHCP says "use this file", and at that point, the only thing involved is
the BIOS/UEFI firmware.  The only available info at that point is some
very minimal hardware info like the MAC address.  After that, it's too
late to change.

Fair enough.

--
Thomas
___
users mailing list -- users@lists.fedoraproject.org
To unsubscribe send an email to users-le...@lists.fedoraproject.org
Fedora Code of Conduct: 
https://docs.fedoraproject.org/en-US/project/code-of-conduct/
List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines
List Archives: 
https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org
Do not reply to spam, report it: 
https://pagure.io/fedora-infrastructure/new_issue

Re: How to set up dhcpd.conf to serve different UEFI files per OS

[Thomas Cameron]

On 6/4/23 17:12, Samuel Sieb wrote:
The part you're missing is that it isn't the OS that's sending the 
DHCP request.  It's the BIOS.  There's no OS loaded yet, that's what 
you're trying to boot. 
The hardware definitely sends a DHCP request when it tries to PXE boot. 
But when the OS actually loads, it sends a separate DHCP request. I was 
hoping to find that and use it for an if/then/else sort of menu.


But that's probably too late, if I understand the boot and installation 
processes.


Thomas
___
users mailing list -- users@lists.fedoraproject.org
To unsubscribe send an email to users-le...@lists.fedoraproject.org
Fedora Code of Conduct: 
https://docs.fedoraproject.org/en-US/project/code-of-conduct/
List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines
List Archives: 
https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org
Do not reply to spam, report it: 
https://pagure.io/fedora-infrastructure/new_issue

screen not waking from sleep - Fedora 38

[Sbob]

Hi all;


I just installed Fedora 38 and when the laptop puts the screen to sleep 
it never wakes up.


Running Fedora 38 on a Dell XPS 17

Thoughts?
___
users mailing list -- users@lists.fedoraproject.org
To unsubscribe send an email to users-le...@lists.fedoraproject.org
Fedora Code of Conduct: 
https://docs.fedoraproject.org/en-US/project/code-of-conduct/
List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines
List Archives: 
https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org
Do not reply to spam, report it: 
https://pagure.io/fedora-infrastructure/new_issue

Re: grub2-mkconfig

[Stephen Morris]

On 4/6/23 04:33, Patrick Dupre wrote:

  vgs
   VG#PV #LV #SN Attr   VSize   VFree
   VolSys_1b   1   3   0 wz--n- <40.00g  <4.00g
   VolSys_21   3   0 wz--n-  84.50g  0
   VolSys_2b   1   2   0 wz--n- <40.00g <14.00g
# lvs
   LVVGAttr   LSize   Pool Origin Data%  Meta%  Move Log 
Cpy%Sync Convert
   iso-image VolSys_1b -wi-ao  10.00g
   tmp   VolSys_1b -wi-a-   1.00g
   usr-src   VolSys_1b -wi-ao  25.00g
   root  VolSys_2  -wi-a-  74.53g
   tmp   VolSys_2  -wi-a- 992.00m
   usr_local VolSys_2  -wi-a-   9.00g
   tmp   VolSys_2b -wi-a-   1.00g
   usr-src   VolSys_2b -wi-a-  25.00g
# pvs
   PV VGFmt  Attr PSize   PFree
   /dev/sda7  VolSys_2  lvm2 a--   84.50g  0
   /dev/sdb5  VolSys_1b lvm2 a--  <40.00g  <4.00g
   /dev/sdb6  VolSys_2b lvm2 a--  <40.00g <14.00g


===
  Patrick DUPRÉ | | email: pdu...@gmx.com
===



Sent: Saturday, June 03, 2023 at 8:11 PM
From: "Roger Heflin" 
To: "Community support for Fedora users" 
Cc: "stan" 
Subject: Re: grub2-mkconfig

try typing vgs/lvs/pvs and see if that also gets the warning.
Some prior centos version did also have this error for a while, I
never noticed any real issue with the programs reporting this.

For the most part this is just some attempt at reporting a "leaked"
file descriptor.I am not sure how one leaks a file descriptor as
all file descriptors owned by a process disappear when the process
goes away.  They may simply be reporting file descriptors that are
still "open" when the program exits (not properly closed), which so
long as it is not a file you are writing to(and have an unsynced
buffer in program ram still), then it is not an actual problem (it is
likely somewhat poor coding practices).   It likely means someone
forgot to turn off some debugging options and also forgot to close
some file that not closing it has no real impact on the program
function.

On Sat, Jun 3, 2023 at 9:09 AM stan via users
 wrote:

On Fri, 2 Jun 2023 13:18:47 +0200
Patrick Dupre  wrote:


Hello,

On a machine with a triple boot (an 3 HD, all fedora), I get a
bunches of

File descriptor 3 (pipe:[240388]) leaked on vgs invocation. Parent
PID 75317: grub2-probe

when I run grub2-mkconfig
I have /boot/efi (UEFI) on /dev/sda3 (in fat16)
and
grub2 core.img on /dev/sda1

Can I fix this?

In addition,
What are your recommendations for running (which systems)
grub2-mkconfig
and
grub2-install /dev/sda
Hi Patrick, just on this front, I use grub2-mkconfig on my UEFI F38 
system as grubby, which is used by default in Fedora, sometimes does not 
configure grub to load the initramfs at boot time. If your boot system 
is UEFI, then you don't need grub2-install /dev/sda, as that is only 
required for legacy bios boots, but even in the legacy boot mode you can 
still use grub2-mkconfig to build your grub.cfg file.


regards,
Steve


I never see that error when I run grub2-mkconfig, in both uefi and
legacy.  So, I suspect there is something about the way your system is
set up causing the issue.

A search turned up this link that describes the cause and a workaround.
https://superuser.com/questions/1437911/file-descriptor-3-leaked-on-vgcfgbackup-invocation-parent-pid-bash
Other links explained that this is happening in the lvm utilities, and
is harmless, basically a warning.  You must have lvm enabled, and the
grub2-mkconfig is probing them, and triggering lvm to issue this error.
___
users mailing list -- users@lists.fedoraproject.org
To unsubscribe send an email to users-le...@lists.fedoraproject.org
Fedora Code of Conduct: 
https://docs.fedoraproject.org/en-US/project/code-of-conduct/
List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines
List Archives: 
https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org
Do not reply to spam, report it: 
https://pagure.io/fedora-infrastructure/new_issue

___
users mailing list -- users@lists.fedoraproject.org
To unsubscribe send an email to users-le...@lists.fedoraproject.org
Fedora Code of Conduct: 
https://docs.fedoraproject.org/en-US/project/code-of-conduct/
List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines
List Archives: 
https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org
Do not reply to spam, report it: 
https://pagure.io/fedora-infrastructure/new_issue


___
users mailing list -- users@lists.fedoraproject.org
To unsubscribe send an email to users-le...@lists.fedoraproject.org
Fedora Code of Conduct: 
https://docs.fedoraproject.org/en-US/project/code-of-conduct/
List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines
List Archives: 

Re: How to set up dhcpd.conf to serve different UEFI files per OS

[Chris Adams]

Once upon a time, Thomas Cameron  said:
> I really wish that there was something in the OS that would identify
> itself when it sends a DHCP broadcast.

Again, the DHCP request that gets a response "use this file" comes from
the firmware, not the OS.

It goes something like:

- BIOS/UEFI configured for network boot sends DHCP request
- DHCP server says "use this file (aka shim)"
- BIOS/UEFI loads that file and runs it
- shim loads grub2 or pxelinux, they get their configs/menus
- you choose an OS to load
- grub2/pxelinux fetches kernel and checks the signature and fails it it
  doesn't match

The step that has to change between Fedora and RHEL is step #2, when the
DHCP says "use this file", and at that point, the only thing involved is
the BIOS/UEFI firmware.  The only available info at that point is some
very minimal hardware info like the MAC address.  After that, it's too
late to change.
-- 
Chris Adams 
___
users mailing list -- users@lists.fedoraproject.org
To unsubscribe send an email to users-le...@lists.fedoraproject.org
Fedora Code of Conduct: 
https://docs.fedoraproject.org/en-US/project/code-of-conduct/
List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines
List Archives: 
https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org
Do not reply to spam, report it: 
https://pagure.io/fedora-infrastructure/new_issue

Re: How to set up dhcpd.conf to serve different UEFI files per OS

[Samuel Sieb]

On 6/4/23 15:00, Thomas Cameron wrote:
I really wish that there was something in the OS that would identify 
itself when it sends a DHCP broadcast. I've read up 


The part you're missing is that it isn't the OS that's sending the DHCP 
request.  It's the BIOS.  There's no OS loaded yet, that's what you're 
trying to boot.

___
users mailing list -- users@lists.fedoraproject.org
To unsubscribe send an email to users-le...@lists.fedoraproject.org
Fedora Code of Conduct: 
https://docs.fedoraproject.org/en-US/project/code-of-conduct/
List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines
List Archives: 
https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org
Do not reply to spam, report it: 
https://pagure.io/fedora-infrastructure/new_issue

Re: How to set up dhcpd.conf to serve different UEFI files per OS

[Thomas Cameron]

On 6/4/23 16:25, Barry wrote:

I have always seen this done by having tooling that read a database of hardware 
mac addresses mapped to config.

With that setup you “just” edit the database to switch the os you want and 
rebuild
your dhcpd/tftpd config.
Unfortunately, the vast majority of my systems are virtual machines, so 
the MAC addresses are dynamically generated. And even with the physical 
servers, I go back and forth between RHEL and Fedora on a pretty regular 
basis. So that MAC address mapping doesn't necessarily stay the same.


I think the easiest thing in my little environment is to write a little 
shell script that copies the RHEL or Fedora shim.efi to 
/var/lib/tftpboot before I kickstart a machine. I suspect that it will 
take a little bit before I do it out of habit, but I install machines so 
frequently for different projects that it will become a habit pretty 
quickly.


I really wish that there was something in the OS that would identify 
itself when it sends a DHCP broadcast. I've read up 
https://kb.isc.org/docs/isc-dhcp-44-manual-pages-dhcp-options and it 
suggested that I add:


set vendor-string = option vendor-class-identifier;

to my dhcpd.conf, and I tried it. But all that I see in the logs when I 
get a dhcp request is:


lease 172.31.100.165 {
  starts 0 2023/06/04 21:55:20;
  ends 0 2023/06/04 22:55:20;
  cltt 0 2023/06/04 21:55:20;
  binding state active;
  next binding state free;
  rewind binding state free;
  hardware ethernet 30:e1:71:51:24:18;
  uid "\0010\341qQ$\030";
  set vendor-string = "anaconda-Linux";
  set vendor-class-identifier = "anaconda-Linux";
}

I feel like there's more info in the dhcp requests than I know how to 
find. Still digging.

--
Thomas
___
users mailing list -- users@lists.fedoraproject.org
To unsubscribe send an email to users-le...@lists.fedoraproject.org
Fedora Code of Conduct: 
https://docs.fedoraproject.org/en-US/project/code-of-conduct/
List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines
List Archives: 
https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org
Do not reply to spam, report it: 
https://pagure.io/fedora-infrastructure/new_issue

Re: Missing "standard" headers in FC37 arm (aarch64) C/C++ cross-compiler RPMs ??

[Barry]

> On 4 Jun 2023, at 21:48, Ron Flory via users  
> wrote:
> 
>  If another noarch or arm header-only RPM happens to contain the missing 
> headers (which I have not stumbled upon yet), it would be helpful for this to 
> be marked a co-dependency so dnf can install it too.

I use dnf search to look for the stdio.h and there are no rpms that contain 
that header.

Try raising a bug against the cross compiler package. Or ask on fedora devel 
list.
This needs the maintainer to comment.

Barry
___
users mailing list -- users@lists.fedoraproject.org
To unsubscribe send an email to users-le...@lists.fedoraproject.org
Fedora Code of Conduct: 
https://docs.fedoraproject.org/en-US/project/code-of-conduct/
List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines
List Archives: 
https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org
Do not reply to spam, report it: 
https://pagure.io/fedora-infrastructure/new_issue

Re: How to set up dhcpd.conf to serve different UEFI files per OS

[Barry]

> On 4 Jun 2023, at 19:43, Thomas Cameron  
> wrote:
> 
> Or am I going about this the wrong way?

I have always seen this done by having tooling that read a database of hardware 
mac addresses mapped to config.

With that setup you “just” edit the database to switch the os you want and 
rebuild
your dhcpd/tftpd config.

Barry

___
users mailing list -- users@lists.fedoraproject.org
To unsubscribe send an email to users-le...@lists.fedoraproject.org
Fedora Code of Conduct: 
https://docs.fedoraproject.org/en-US/project/code-of-conduct/
List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines
List Archives: 
https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org
Do not reply to spam, report it: 
https://pagure.io/fedora-infrastructure/new_issue

Re: Missing "standard" headers in FC37 arm (aarch64) C/C++ cross-compiler RPMs ??

[Ron Flory via users]

On 6/3/23 08:55, stan via users wrote:
On Thu, 01 Jun 2023 20:29:54 - ron flory via users 
 wrote:
Results: hello_world.c:1:10: fatal error: stdio.h: No such file or 
directory 1 | #include  | ^ compilation terminated. 
- Hello_world sample consists of: #include  #include 
 int main(int argc, char *argv[]) { printf("Hello 
World\n"); return(0); } 
I find at this link 
https://stackoverflow.com/questions/19580758/gcc-fatal-error-stdio-h-no-such-file-or-directorythis 
comment, You can see where gcc is looking for header files by doing 
echo "#include " | gcc -v -x c - and examining the search 
paths. – Christian Ternus Oct 25, 2013 at 3:53 It is old, and when I 
tried it, it didn't work. But it might give you an idea for things to 
try. This also seems relevant, "it seems implausible that there would 
be a release which generated errors upon importing stdio" Is the file 
present in /usr/include? If it is, you could use the -I option to 
point there, -I /usr/include, and see if it works. If it does, that 
indicates that it is a problem in the configuration of gcc; it isn't 
looking in the system include directory. 


 it was sloppy of me to assume everybody understood the host is x86-64.

 There are only 3 relevant compiler/binutils RPMs for aarch64 in FC37: 
(excluding other langs/libs)


dnf list all | grep aarch | grep -v "^rust\|^qemu\|^edk2\|^ipxe"
  gcc-aarch64-linux-gnu
  gcc-c++-aarch64-linux-gnu
  binutils-aarch64-linux-gnu

 ...the binutils is automatically installed as a dependency (similar 
deal for the "gcc-arm-linux" sibling).



The (very few) related headers are installed in 
/usr/lib/gcc/aarch64-linux-gnu/12/include/ as confirmed by 1)
  dnf repoquery -l gcc-aarch64-linux-gnu gcc-c++-aarch64-linux-gnu | 
grep /st | grep "\.h$" | sort


and 2) which are not present when the cross-compiler is removed, but are 
present when the cross-compiler is re-installed.


  Having built many cross-compilers over the years, a look at this 
header include directory is very, very sparse, missing the majority of 
our usual .h friends.


  It --may-- be the RPMs were just built 'for successful compilation' 
without running a minimal test suite to confirm the output artifacts are 
functional-  ...maybe.   ( at least a 'hello-world' should be 
cross-compiled to confirm output-artifacts are produced, and for the 
correct target arch )


 If another noarch or arm header-only RPM happens to contain the 
missing headers (which I have not stumbled upon yet), it would be 
helpful for this to be marked a co-dependency so dnf can install it too.


- - -

 As a side note, for reference only, please no flames intended; the 
corresponding two aarch64 gcc/c++ packages for the popular debian-based 
distro beginning with "U" (which shall not be spoken here) does compile 
and link C/C++ programs correctly, which is why I suspect it could be an 
RPM packaging issue.


ron___
users mailing list -- users@lists.fedoraproject.org
To unsubscribe send an email to users-le...@lists.fedoraproject.org
Fedora Code of Conduct: 
https://docs.fedoraproject.org/en-US/project/code-of-conduct/
List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines
List Archives: 
https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org
Do not reply to spam, report it: 
https://pagure.io/fedora-infrastructure/new_issue

Re: How to set up dhcpd.conf to serve different UEFI files per OS

[Chris Adams]

Once upon a time, Joe Zeff  said:
> On 06/04/2023 01:40 PM, Chris Adams wrote:
> >It'd be nice if there was a way to chainload one shim from another
> 
> If memory serves, you could have GRUB boot Windows by giving it the
> command chainload +X, where X represented the number of sectors to
> load.  I've no idea if GRUB2 still does this, but if so, it might be
> what's needed.

I poked around at that at one point, and couldn't find a way to get it
to chainload another shim from the network.
-- 
Chris Adams 
___
users mailing list -- users@lists.fedoraproject.org
To unsubscribe send an email to users-le...@lists.fedoraproject.org
Fedora Code of Conduct: 
https://docs.fedoraproject.org/en-US/project/code-of-conduct/
List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines
List Archives: 
https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org
Do not reply to spam, report it: 
https://pagure.io/fedora-infrastructure/new_issue

Re: How to set up dhcpd.conf to serve different UEFI files per OS

[Joe Zeff]

On 06/04/2023 01:40 PM, Chris Adams wrote:


It'd be nice if there was a way to chainload one shim from another


If memory serves, you could have GRUB boot Windows by giving it the 
command chainload +X, where X represented the number of sectors to load. 
 I've no idea if GRUB2 still does this, but if so, it might be what's 
needed.

___
users mailing list -- users@lists.fedoraproject.org
To unsubscribe send an email to users-le...@lists.fedoraproject.org
Fedora Code of Conduct: 
https://docs.fedoraproject.org/en-US/project/code-of-conduct/
List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines
List Archives: 
https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org
Do not reply to spam, report it: 
https://pagure.io/fedora-infrastructure/new_issue

Re: How to set up dhcpd.conf to serve different UEFI files per OS

[Chris Adams]

Once upon a time, Thomas Cameron  said:
> Yeah, that's why I was hoping there was maybe some magic in the
> vendor-class-identifier response that I could use. It would make
> life a LOT easier.

All the DHCP communication happens before shim is loaded (and then it's
too late to change), so all you can see is the base hardware info.  If
you know your hardware, you could configure MAC addresses in your DHCP
config to control the response, pointing to a different shim/config
depending on which MAC is requesting.

Otherwise, I suppose you might be able to do something odd like serve up
one shim/config via TFTP and a different one via HTTP, so you could
choose UEFI PXE for one OS and UEFI HTTP for the other.  It would
probably be confusing after the fact though.

-- 
Chris Adams 
___
users mailing list -- users@lists.fedoraproject.org
To unsubscribe send an email to users-le...@lists.fedoraproject.org
Fedora Code of Conduct: 
https://docs.fedoraproject.org/en-US/project/code-of-conduct/
List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines
List Archives: 
https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org
Do not reply to spam, report it: 
https://pagure.io/fedora-infrastructure/new_issue

Re: How to set up dhcpd.conf to serve different UEFI files per OS

[Thomas Cameron]

On 6/4/23 14:40, Chris Adams wrote:

As far as I can tell, you cannot configure network boot for different
OSes in a UEFI Secure Boot environment.  The shim is loaded first,
before you get to the point of choosing which kernel to boot, and a
given distribution's shim will only load other Linux things signed by
that distribution's key.

It'd be nice if there was a way to chainload one shim from another
(they're all signed by the MS firmware-trusted key, so it seems like
this should be possible and still meet the security requirements), so
you could have a menu option "Switch to RHEL" that would load the RHEL
shim+bootloader, but I don't think that's possible today.  I'm using
grub2 for network book rather than syslinux, but I couldn't figure out a
way to make that work.

The only way to handle it would be to distinguish the clients at the
DHCP server (use separate VLANs, pre-configure MAC addresses, etc.).
Once the DHCP server sends an answer, it's too late to change.
Yeah, that's why I was hoping there was maybe some magic in the 
vendor-class-identifier response that I could use. It would make life a 
LOT easier.


I've started using cobbler again 
(https://cobbler.readthedocs.io/en/latest/quickstart-guide.html). 
Cobbler was HORRIBLY broken in Fedora 37, but it seems to be fixed in 
38. I do love that, when I set up a distro, it automatically creates the 
grub.cfg file in /var/lib/tftpboot/grub and creates the menu files for 
me. I actually have my kickstarts set up such that I can install Fedora 
38 and RHEL 9, except for this weird shim issue. I guess when I'm going 
to install RHEL machines, I need to copy the RHEL shim.efi to 
/var/lib/tftpboot, and when I'm going to install Fedora machines, copy 
the Fedora shim.efi. It's kind of a pain in the backside. But I can 
script it pretty easily. I just need to remember to run the script when 
I decide to change what I'm installing.


Thanks so much for the answer, I appreciate it, Chris.

--
Thomas
___
users mailing list -- users@lists.fedoraproject.org
To unsubscribe send an email to users-le...@lists.fedoraproject.org
Fedora Code of Conduct: 
https://docs.fedoraproject.org/en-US/project/code-of-conduct/
List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines
List Archives: 
https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org
Do not reply to spam, report it: 
https://pagure.io/fedora-infrastructure/new_issue

Re: How to set up dhcpd.conf to serve different UEFI files per OS

[Chris Adams]

Once upon a time, Thomas Cameron  said:
> Is it that the shim.efi file is signed for UEFI environments, and
> the RHEL kernel is expecting the signature for the RHEL shim.efi
> file? If so, how do I specify which shim.efi file I want to use
> based on the kernel? I would assume I'd need to add the correct
> shim.efi file in /var/lib/tftpboot/images/[kickstart_os] the same as
> I add the vmlinuz and initrd.img. But how do I tell the machine
> being kickstarted where to get the correct shim.efi? Is there a
> vendor-class-identifier I can check to see what the OS is, and then
> point the machine being kickstarted to that file?

As far as I can tell, you cannot configure network boot for different
OSes in a UEFI Secure Boot environment.  The shim is loaded first,
before you get to the point of choosing which kernel to boot, and a
given distribution's shim will only load other Linux things signed by
that distribution's key.

It'd be nice if there was a way to chainload one shim from another
(they're all signed by the MS firmware-trusted key, so it seems like
this should be possible and still meet the security requirements), so
you could have a menu option "Switch to RHEL" that would load the RHEL
shim+bootloader, but I don't think that's possible today.  I'm using
grub2 for network book rather than syslinux, but I couldn't figure out a
way to make that work.

The only way to handle it would be to distinguish the clients at the
DHCP server (use separate VLANs, pre-configure MAC addresses, etc.).
Once the DHCP server sends an answer, it's too late to change.
-- 
Chris Adams 
___
users mailing list -- users@lists.fedoraproject.org
To unsubscribe send an email to users-le...@lists.fedoraproject.org
Fedora Code of Conduct: 
https://docs.fedoraproject.org/en-US/project/code-of-conduct/
List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines
List Archives: 
https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org
Do not reply to spam, report it: 
https://pagure.io/fedora-infrastructure/new_issue

How to set up dhcpd.conf to serve different UEFI files per OS

[Thomas Cameron]

I am trying to kickstart multiple versions of Linux. Some of my systems 
are BIOS based, and some are UEFI based.


I have a stanza in my dhcpd.conf file that looks like this:

    class "pxeclients" {
    match if substring (option vendor-class-identifier, 0, 9) = 
"PXEClient";

    next-server 172.31.100.1;
    if option architecture-type = 00:07 {
    filename "shim.efi";
    } else {
    filename "pxelinux.0";
    }
  }

I got the shim.efi I am serving from the tftpboot directory from the 
Fedora 38 shim-x64 package.


The problem is, when I try to kickstart a RHEL machine, it errors out 
saying the shim is invalid.


Is it that the shim.efi file is signed for UEFI environments, and the 
RHEL kernel is expecting the signature for the RHEL shim.efi file? If 
so, how do I specify which shim.efi file I want to use based on the 
kernel? I would assume I'd need to add the correct shim.efi file in 
/var/lib/tftpboot/images/[kickstart_os] the same as I add the vmlinuz 
and initrd.img. But how do I tell the machine being kickstarted where to 
get the correct shim.efi? Is there a vendor-class-identifier I can check 
to see what the OS is, and then point the machine being kickstarted to 
that file?


Or am I going about this the wrong way?

Thomas
___
users mailing list -- users@lists.fedoraproject.org
To unsubscribe send an email to users-le...@lists.fedoraproject.org
Fedora Code of Conduct: 
https://docs.fedoraproject.org/en-US/project/code-of-conduct/
List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines
List Archives: 
https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org
Do not reply to spam, report it: 
https://pagure.io/fedora-infrastructure/new_issue

Re: Chrome DRM supported content fails

[Patrick O'Callaghan]

On Sun, 2023-06-04 at 09:41 -0400, Alex wrote:
> Hi,
> 
> I'm using fedora37 on my desktop, and it appears a recent update
> caused google-chrome-stable-114.0.5735.90-1.x86_64 to no longer play
> DRM
> protected content like Netflix. I've installed all the latest
> updates.
> Anyone know what changed?
> 
> I've confirmed that protected content is enabled
> at chrome://settings/content/protectedContent
> 
> I've also learned this could be related to ffmpeg and/or widevine
> support.
> 
> Widevine Content Decryption Module - Version: 4.10.2652.1
> ffmpeg-libs-5.1.3-3.fc37.x86_64
> 
> Any ideas greatly appreciated. Very frustrating.

Works for me (on F38), but I have ffmpeg-libs-6.0-11.fc38.x86_64

poc
___
users mailing list -- users@lists.fedoraproject.org
To unsubscribe send an email to users-le...@lists.fedoraproject.org
Fedora Code of Conduct: 
https://docs.fedoraproject.org/en-US/project/code-of-conduct/
List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines
List Archives: 
https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org
Do not reply to spam, report it: 
https://pagure.io/fedora-infrastructure/new_issue

Chrome DRM supported content fails

[Alex]

Hi,

I'm using fedora37 on my desktop, and it appears a recent update
caused google-chrome-stable-114.0.5735.90-1.x86_64 to no longer play DRM
protected content like Netflix. I've installed all the latest updates.
Anyone know what changed?

I've confirmed that protected content is enabled
at chrome://settings/content/protectedContent

I've also learned this could be related to ffmpeg and/or widevine support.

Widevine Content Decryption Module - Version: 4.10.2652.1
ffmpeg-libs-5.1.3-3.fc37.x86_64

Any ideas greatly appreciated. Very frustrating.

Thanks,
Alex
___
users mailing list -- users@lists.fedoraproject.org
To unsubscribe send an email to users-le...@lists.fedoraproject.org
Fedora Code of Conduct: 
https://docs.fedoraproject.org/en-US/project/code-of-conduct/
List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines
List Archives: 
https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org
Do not reply to spam, report it: 
https://pagure.io/fedora-infrastructure/new_issue

Re: cannot remove disk if it has logical volumes [SOLVED]

[Eyal Lebedinsky]

Thanks Mike,

On 04/06/2023 14.02, Mike Wright wrote:

On 6/3/23 19:42, Eyal Lebedinsky wrote:

I am on f38, but I had this problem before with f36.

I attach an external USB disk that has CentOS.
I see that beyond the basic device and partitions, three other devices 
/dev/dm-{0,1,2} show up.
Without doing anything else I want to remove the disk safely. I could not do 
it, see below.

The question at the end is: how do I cleanly remove the external USB disk. It 
is an SSD in a USB3.0 caddy.






This stays until a reboot. I have some cron jobs that log the state of the 
machine and they raise errors in the log
from looking into these devices.


  Hi Eyal,

Partition /dev/sdi2 is now a pv so you can't work with it as a physical; you 
must work with it under the VG/LV names.


If it was not clear, nothing was mounted so no swapoff/umount is required in my 
case.


Remove the swap: swapoff /dev/cl/swap

Unmount root: umount /dev/cl/root

Unmount home: umount /dev/cl/home

Deactivate the LVs: vgchange --activate n cl (or vgchange -a n cl).


This is what I needed and it solved my problem. I learn something every day.


The deactivation step may not be necessary but it won't hurt to make sure.


It was necessary here.


All LVM stuff is cleared out.  You must now refer to it by its physical name.

Note: the VG/LVs still exist and can be remounted by referring to them with 
their physical names: /dev/cl/{root,home} or added to swap with swapon 
/dev/cl/swap.

eject /dev/sdi -- you may now safely remove your device.

Hope that explains everything clearly.


It does.

cheers,
Eyal


Mike Wright


--
Eyal Lebedinsky (fed...@eyal.emu.id.au)
___
users mailing list -- users@lists.fedoraproject.org
To unsubscribe send an email to users-le...@lists.fedoraproject.org
Fedora Code of Conduct: 
https://docs.fedoraproject.org/en-US/project/code-of-conduct/
List Guidelines: https://fedoraproject.org/wiki/Mailing_list_guidelines
List Archives: 
https://lists.fedoraproject.org/archives/list/users@lists.fedoraproject.org
Do not reply to spam, report it: 
https://pagure.io/fedora-infrastructure/new_issue