Announcement

Collapse
No announcement yet.

Linux 5.19.1 Released With LTS Kernels Back To 4.19 For The "PBRSB" CPU Issue

Collapse
X
 
  • Filter
  • Time
  • Show
Clear All
new posts

  • tildearrow
    replied
    Originally posted by birdie View Post
    Oh, and 5.19 has an ext4 regression. Not critical, but quite serious for some. Never mind, some people here believe I hate open source. Must be a madman considering all the time I've invested to fix a metric ton of issues in Open Source over the past 25 years.
    Oh wow, thanks for the info!
    No wonder why am I still on 5.4...


    Now stop trying to start flamewars - that last paragraph was unnecessary.
    Here's how your post should have been like:

    Oh, and 5.19 has an ext4 regression. Not critical, but quite serious for some.
    Last edited by tildearrow; 12 August 2022, 04:31 PM.

    Leave a comment:


  • caligula
    replied
    Originally posted by reba View Post

    Huh? What are you talking about?
    Bluetooth even works when I start up in terminal only because it's part of PipeWire. I know it because my BT-headset indicates connection establishment.
    And of course pairing or managing Bluetooth devices works e.g. with KDE Plasma.
    Bluetooth INPUT devices don't work. You need to pair the keyboard BEFORE being able install the distro as most installers expect textual input (e.g. username, password). Unless you have 1) paired it beforehand, 2) have a keyboard with both 2,4GHz / BT, or 3) wired / BT modes, or 4) a spare keyboard which does not use BT. Or maybe you're using a distro that let's you pair a BT keyboard without having a keyboard plugged in. Not all of them do.

    FWIW, many PCs don't even let you use the menu in GRUB or other boot managers with BT keyboards. You can't access the UEFI BIOS setup with a BT keyboard. It's totally useless piece of shit on Linux/PC.

    Leave a comment:


  • reba
    replied
    Originally posted by JMB9 View Post

    Well, maybe this is correct after your experience - my view differs totally.
    I have never seen that BT works - it is not quite a real standard: in most cases it works in some aspects and others don't work at all.
    Additionally a BT keyboard can be read from a nice distance, so not even a key logger necessary.
    There are many possibilities for war driving - this is one of the easiest possibilities ...
    And any internet connection is considered a security risk - if security is really important.

    So in my world BT is not of acceptable quality - like all devices from IoT, smatphones or tablets - or car electronics.
    And I don't harden my GNU/Linux system but just keep them up-to-date - and I was responsible for
    hardening important servers.

    So my view on security may be a lot different to the current digital natives.
    A generation thing ...
    Ah, okay, your point was "Bluetooth as a protocol" not "Bluetooth as an implementation on Linux".
    Latter works fine for me (headset, mouse, etc.), former is out of my scope. Security is a progress: wait long (or short) enough and everything breaks.

    Leave a comment:


  • openminded
    replied
    Originally posted by Malsabku View Post
    I disabled all these mitigations on my productive system. Performance above all <3
    The difference is so minor that it doesn't worth fiddling with disabling of mitigations.

    Leave a comment:


  • JMB9
    replied
    Originally posted by reba View Post
    Huh? What are you talking about?
    Bluetooth even works ...
    Well, maybe this is correct after your experience - my view differs totally.
    I have never seen that BT works - it is not quite a real standard: in most cases it works in some aspects and others don't work at all.
    Additionally a BT keyboard can be read from a nice distance, so not even a key logger necessary.
    There are many possibilities for war driving - this is one of the easiest possibilities ...
    And any internet connection is considered a security risk - if security is really important.

    So in my world BT is not of acceptable quality - like all devices from IoT, smatphones or tablets - or car electronics.
    And I don't harden my GNU/Linux system but just keep them up-to-date - and I was responsible for
    hardening important servers.

    So my view on security may be a lot different to the current digital natives.
    A generation thing ...

    Leave a comment:


  • JMB9
    replied
    Originally posted by NotMine999 View Post

    The ext4 bug is offtopic, no? ...
    "No" is correct - as the headline is about the new kernel released with 5.19 now being promoted from mainline to stable in a sense that GKH is caring
    for it starting with 5.19.1.
    So if a regression is found in 5.19 in a relevant part like ext4, it would be woth mentioning and well in place here.
    But if one want to do it, some more info incl. link would be welcome.

    I don't want to judge about quality - but being on topic is not decisive for me - but here current kernel developmen is the topic.
    So blaming about off-topic when being really on topic is strange.

    That Intel - and really much more limited some 'friends' - is in the center is just a problem of people not caring about quality for tens of years.
    That's while Linus is quite clear here, like "Because Intel was pushing shit to consumers." (concerning 'ECC absolutely matters.');
    "Is Intel really planning on making this shit architectural?" (on macros to restrict/unrestrict Indirect Branch
    Speculation) ... this is an endless nightmare!

    And unfortunately, this is not off-topic, as one of the current problems is exactly this - and will stay for a long while.

    Technical people will currently not buy any CPU/GPU by Intel - and it is shocking that Intel does not work in the right direction.
    If you have an Intel CPU, you are lost - as you can NOT STOP performance drops as they are in largest part done by firmware updates.
    And this can not be revoked - so not covered by any benchmarks except you have original firmware parts - which
    is not the case when Phoronix shows benchmarks - if I am wrong, please correct me in that.

    So switching everything off concerning Linux mitigation will be smaller than a 10% effect of the real mitigation - I tested it with my
    Hasswell system (the last one a technical person could use when on Linux - Skylake was trash a long time - and then mitigation
    was clear for all).

    I still hope that Intel will make a new design avoiding ANY mitigation ... but it seems that this is not to be expected at all.
    And even AMD increase the amount of necessary mitigations due to benchmarks ... and unfortunately, this is more important
    than real quality today.

    Leave a comment:


  • reba
    replied
    Originally posted by caligula View Post

    You shouldn't use BT on Linux. For example when installing a distro, most desktop environments don't even have a way to initiate the BT connection, so, no way to even type the PIN codes. This is standard functionality on Mac because they ship with BT keyboards.
    Huh? What are you talking about?
    Bluetooth even works when I start up in terminal only because it's part of PipeWire. I know it because my BT-headset indicates connection establishment.
    And of course pairing or managing Bluetooth devices works e.g. with KDE Plasma.

    Leave a comment:


  • rene
    replied
    updated: https://www.youtube.com/watch?v=Ek9f50a_X18

    Leave a comment:


  • caligula
    replied
    Originally posted by birdie View Post
    Two weeks ago I asked all the three developers behind it to push it to stable. No one cares.

    What does it manifest in? After resume, your BT devices won't be able to connect automatically to your BT host/computer because the code is simply wrong.
    You shouldn't use BT on Linux. For example when installing a distro, most desktop environments don't even have a way to initiate the BT connection, so, no way to even type the PIN codes. This is standard functionality on Mac because they ship with BT keyboards.

    Leave a comment:


  • NotMine999
    replied
    Originally posted by birdie View Post
    Oh, and 5.19 has an ext4 regression. Not critical, but quite serious for some. Never mind, some people here believe I hate open source. Must be a madman considering all the time I've invested to fix a metric ton of issues in Open Source over the past 25 years.
    The ext4 bug is offtopic, no? After all, the Phoronix article is talking about CPU bugs, not FS bugs.

    And the sloppy attempt to hijack a forum thread for meaningless self-promotion that is highlighted in RED.

    Leave a comment:

Working...
X