Memtest86+ 7.0 Released With IMC Polling & Initial ECC Polling

[Eudyptula]

MemTest86+ is great except neither RedHat, nor Microsoft want to digitally sign it, thus it won't work with Secure Boot on which is a bummer, that's why I've long switched to MemTest86.

Secure Boot is indeed supported, and has been tested to work on many computers. However, memtest86+ binaries are not signed with Microsoft's key, or any usual distributor's key. For now, one needs to sign it with their own key, the validation key for which needs to be inserted into your BIOS' MOK, just like what happens e.g. with DKMS for Linux kernel modules

Why won't they digitally sign it?

[Mirox]

I can boot Memtest86+ with Secureboot enabled and UEFI. By copying the ISO in a Ventoy USB. Then booting Ventoy and boot Memtest86+ from the Ventoy menu.

[debrouxl]

Why won't they digitally sign it?

For Microsoft, ISTR from some of Sam's posts that it's fairly hard to achieve for a volunteer-driven project - something around complexity and/or time consumption. I might be mistaken.
For Linux distros, https://github.com/rhboot/shim-review/issues/314 , and I opened a corresponding https://github.com/memtest86plus/mem...iscussions/370 yesterday.

[bug77]

My mentor told me an awful long time ago that you'd have a hard time allocating a buffer to construct an error message for an out-of-memory error.

It depends on the context.
I saw this done on Java (thankfully only) once. Somebody thought it was useful to catch everything so they caught Throwable. At runtime, that meant it was also catching OutOfMemoryError. And here's the kicker, the GC would kick in, free a bit of memory, letting the code log a stack trace, before running out of memory again. Rinse and repeat. That thing would run out of disk space long before an admin would notice it was fubar.

[Old Grouch]

My mentor told me an awful long time ago that you'd have a hard time allocating a buffer to construct an error message for an out-of-memory error.

It's trivial if you allocate the buffer space for an out-of-memory error before the error occurs. Sometimes preparation is important. Then again, I'm a dinosaur from an age when all variables (including strings) had to be allocated before use. C's heap was regarded with suspicion, and stacks were used sparingly and carefully.

[Jabberwocky]

I recently stumbled on Ryzen DRAM Calculator v1.7.0 when I researched optimal timing for 128gb RAM on zen2 and zen3. Latency tests were an interesting bonus although I'm not sure how accurate it is. https://www.reddit.com/r/Amd/comment...lculator_v170/

In the past I would watch some arbitrary buildzoid / actual hardware overclocking video about ram timing then take very little of that information in. What was nice about the Ryzen DRAM Calculator app is that all the presets for the different modules were available. A spreadsheet would to the same job but it's nice to have everything consolidated in one place.

Would be nice to introduce some of those features into an open source project. I enjoy messing around with this even though I don't always know what I'm doing.

I hope everyone's RAM timings are lucky

[unwind-protect]

What is ECC polling, anyway?

[debrouxl]

What is ECC polling, anyway?

Some form of short for something along the lines of "Polling a suitably configured Error Detection And Correction (EDAC) controller for information about corrected and uncorrected memory errors in ECC memory".

[Particle]

I'm curious at this point if Memtest86 or Memtest86+ is a better tester of memory issues on modern systems. I know plus lacks a Rowhammer test for instance. Given that their testing algorithms aren't identical anymore, I'm just curious which generally is able to suss out subtle problems more often.

I have a recent high-end ASUS motherboard, and it performs some kind of automated memtest regularly on bootup (shown on the mb embedded screen for a few minutes).

You're probably referring to memory training rather than memory testing. Training is a process that adjusts and tests a bunch of parameters of the memory controller to try to discover the cleanest achievable signal. This allows data to fly over the wire reliably. It doesn't meaningfully test the storage capability of the memory array itself.