Requests for -current (14.2-->15.0)

GazL · 01-30-2019, 04:11 AM

initrd is needed under certain circumstances (LUKS/LVM/RAID rootfs being some of the primary ones¹).

While an initrd may not be required on an individual host level, at the distro level it seems quite reasonable to say: "If some of the users will need one, then lets just use one for everyone, whether it's strictly needed or not, and keep things consistent".

¹ early firmware loading also being another one, which is more important than it used to be post spectre/meltdown.

timsoft · 01-30-2019, 01:18 PM

i must confess I always use huge, and if i compile my own kernels I include the fs support (ext2/3/4) etc. to avoid having to create an initrd, so I'm greatful for those "huge" kernels. it does follow the KISS principle, and if you have very limited ram and don't want to do your own kernel, you can still use what is available.

ZhaoLin1457 · 01-30-2019, 01:33 PM

I don't think that is something in common between the initrd usage and not following whatever KISS principles.

And regarding the memory consumption, the default "huge" kernel is rather memory wasteful, because the allocated memory for initrd is freed after switching to real root, if I understand right from what other distributions says.

Nobby6 · 01-30-2019, 02:39 PM

Quote:

Originally Posted by GazL

initrd is needed under certain circumstances (LUKS/LVM/RAID rootfs being some of the primary ones¹).

While an initrd may not be required on an individual host level, at the distro level it seems quite reasonable to say: "If some of the users will need one, then lets just use one for everyone, whether it's strictly needed or not, and keep things consistent".

¹ early firmware loading also being another one, which is more important than it used to be post spectre/meltdown.

Fair enough, but if I was betting person, most people would be ignorant of teh fact and not need it.
I've always used hardware RAID, even at home since 2000ish, so I admit not looking into it, but, if it goes in an initrd image, then surely it can be used in boot kernel, I cant imagine linux in 2019 needing it for software RAID, but.. as vI said I admit ignorance in software RAID.

atelszewski · 01-30-2019, 03:12 PM

Hi,

Quote:

Originally Posted by GazL

early firmware loading also being another one, which is more important than it used to be post spectre/meltdown.

Although I'm initrd fan, sometime ago somebody emailed me that it is possible to load microcode with the huge kernel.
There will come the day I'll investigate it ;-)

--
Best regards,
Andrzej Telszewski

ponce · 01-30-2019, 03:27 PM

Quote:

Originally Posted by atelszewski

Hi,

Although I'm initrd fan, sometime ago somebody emailed me that it is possible to load microcode with the huge kernel.
There will come the day I'll investigate it ;-)

you can just load the microcode in place of the initrd, like (excerpt from lilo.conf)

Code:

image = /boot/vmlinuz-huge
  initrd = /boot/intel-ucode.cpio
  root = /dev/sda1
  label = Linux
  read-only

ZhaoLin1457 · 01-30-2019, 03:48 PM

But, but, but /boot/intel-ucode.cpio is still an initrd, which by its own existence and usage, defeats the claim that you can load the microcode without an initrd.

That the kernel is capable to load multiple (concatenated) initrds, that's entirely another story.

ponce · 01-30-2019, 03:53 PM

to be precise, I was replying to the above "it is possible to load microcode with the huge kernel".

mlangdn · 01-30-2019, 05:39 PM

No more interest in no more initrd.

I built a custom kernel this morning. That is something I hadn't done in maybe 4 years. I just really got tired of building an initrd on each kernel upgrade. I was really tired of kernel upgrades in current. I blacklisted those things and all its parts.

I took out loads of stuff from the generic kernel, especially for embedded and stuff I was sure about. I also used the latest stable release - 4.20.5. That way, I won't be tempted to upgrade so often.

The only boo-boo was not building the nvidia driver for the second kernel. The three finger salute sent me back to correct that self-inflicted bludgeoning.

So far, everything works, and it seems to have shaved about 20 seconds. I'm too lazy to go back and time that again.

SCerovec · 01-31-2019, 01:55 AM

Quote:

Originally Posted by mlangdn

No more interest in no more initrd.

I built a custom kernel this morning. That is something I hadn't done in maybe 4 years. I just really got tired of building an initrd on each kernel upgrade. I was really tired of kernel upgrades in current. I blacklisted those things and all its parts.

I took out loads of stuff from the generic kernel, especially for embedded and stuff I was sure about. I also used the latest stable release - 4.20.5. That way, I won't be tempted to upgrade so often.

The only boo-boo was not building the nvidia driver for the second kernel. The three finger salute sent me back to correct that self-inflicted bludgeoning.

So far, everything works, and it seems to have shaved about 20 seconds. I'm too lazy to go back and time that again.

I was just about to ask what happened to roll your own people

qunying · 01-31-2019, 01:55 PM

Please update the libimobiledevice from the git HEAD in https://github.com/libimobiledevice/libimobiledevice.

The current version was released 4 years ago and does not work on newer iOS and iTunes:
https://github.com/libimobiledevice/...ice/issues/380

USUARIONUEVO · 01-31-2019, 03:13 PM

glibc-2.29

release notes --> https://sourceware.org/ml/libc-annou.../msg00000.html
sources --> http://ftp.gnu.org/gnu/glibc/glibc-2.29.tar.xz

cwizardone · 01-31-2019, 04:49 PM

Mesa 18.3.3

The release notes,

https://www.mesa3d.org/relnotes/18.3.3.html

SCerovec · 02-01-2019, 02:59 AM

kernel got 5.0 ?

_gin · 02-01-2019, 06:49 AM

Qt 5.12.1

http://blog.qt.io/blog/2019/02/01/qt-5-12-1-released/
http://download.qt.io/official_relea...-5.12.1.tar.xz