No regressions noticed in either of my x8t_64 systms, aarch64 rpi 4b, i586 and
vb guests.

In the x86_64 laptop cpuinfo has ...
cpu family      : 23
model           : 96
model name      : AMD Ryzen 7 4800H with Radeon Graphics

Running the zenbleed poc shows
Spawning 8 Threads...
zenbleed: failed to start thread on specifed code 1: Success

before and after the kernel or prior microcode update.

CC: (none) => davidwhodgins

cpu family	: 23
model		: 104
model name	: AMD Ryzen 5 5500U with Radeon Graphics

Before installing these updates the zenbleed PoC generated reams of output.
After installing these updates it outputs

  % ./zenbleed 
  *** EMBARGOED SECURITY ISSUE --  DO NOT DISTRIBUTE! ***
  ZenBleed Testcase -- taviso@google.com

  NOTE: Try -h to see configuration options

  Spawning 12 Threads...
  Thread 0x7f4efd6ce640 running on CPU 0
  Thread 0x7f4efcecd640 running on CPU 1
  Thread 0x7f4efc6cc640 running on CPU 2
  Thread 0x7f4efaec9640 running on CPU 5
  Thread 0x7f4efb6ca640 running on CPU 4
  Thread 0x7f4ef9ec7640 running on CPU 7
  Thread 0x7f4efa6c8640 running on CPU 6
  Thread 0x7f4ef8ec5640 running on CPU 9
  Thread 0x7f4ef96c6640 running on CPU 8
  Thread 0x7f4ee3fff640 running on CPU 10
  Thread 0x7f4efbecb640 running on CPU 3
  Thread 0x7f4ee37fe640 running on CPU 11

and the journal contains

  kernel: Zenbleed: please update your microcode for the most optimal fix

No regressions noted so far.

CC: (none) => mageia

Mageia8, x86_64
AMD Ryzen 7 5700U

Installed all the files and rebooted.
Searched for the PoC
https://github.com/google/security-research/blob/master/pocs/cpus/zenbleed/README.md

Don't know how to use git so downloaded all the files individually and ran make.
$ make
Makefile:1: *** target pattern contains no '%'.  Stop.

What now?

CC: (none) => tarazed25

I downloaded the PoC from here:

https://web.archive.org/web/20230724160509/https://lock.cmpxchg8b.com/files/zenbleed-v5.tar.gz

then

  tar xf zenbleed-v5.tar.gz
  cd zenbleed
  make
  ./zenbleed

You will need gcc and nasm installed.

I used "wget https://seclists.org/oss-sec/2023/q3/att-59/zenbleed-v5_tar.gz"
to get it, then tar -xf zenbleed..., cd zenbleed, make, ./zenbleed to run it.

After removing nosmt=force from the kernel options on my laptop from comment 1,
zenbleed is using 100% (or slightly less for short periods) cpu on all 16 cores.

$ ./zenbleed 
*** EMBARGOED SECURITY ISSUE --  DO NOT DISTRIBUTE! ***
ZenBleed Testcase -- taviso@google.com

NOTE: Try -h to see configuration options

Spawning 16 Threads...
Thread 0x7fc62cbf5640 running on CPU 8
<snip>
Thread 0x7fc6293ee640 running on CPU 15

I killed it after 10 minutes with no further output.

That's with kernel and microcode updates.

Thanks for the pointers.
Note: No microcode update.  Using server kernel - linus later.

$ ./zenbleed
*** EMBARGOED SECURITY ISSUE --  DO NOT DISTRIBUTE! ***
[...]
Spawning 16 Threads...
Thread 0x7f5434f2e640 running on CPU 0
...
Thread 0x7f5410ff9640 running on CPU 15
Thread 0x7f54117fa640 running on CPU 14

# journalctl | grep Zenbleed
Jul 25 20:59:42 rutilicus kernel: Zenbleed: please update your microcode for the most optimal fix

gkrellm shows solid 99-100% CPU usage on all cores.  Killed it after a while.

Installed microcode and repeated test.  Same output as before and advice in the journal to update microcode.

My newest AMD processor is an A8-4555M, which as I misunderstand it is unenlightened where any kind of Zen is concerned, so should be unaffected by this issue. Testing the kernel anyway...

HP Pavilion 15, AMD A8-4555M APU, AMD HD 7600G graphics, MGA8-64 Plasma system. Updated desktop kernel and microcode in one operation. Did my usual kernel testing, no issues noted.

CC: (none) => andrewsfarm

5.15.122-desktop-1.mga8 x86_64
10-Core Intel Core i9-7900X
NVIDIA GP102 [GeForce GTX 1080 Ti]
Intel Ethernet I219-V driver: e1000e

Running Mate for several hours.  No issues.

An update for this issue has been pushed to the Mageia Updates repository.

https://advisories.mageia.org/MGASA-2023-0242.html

Status: NEW => RESOLVED
Resolution: (none) => FIXED