On 19/12/2023 4:28 pm, Joe Tretter wrote: > On 12/19/23 10:05, Andrew Cooper wrote: >> Is it always the same test which fails, or is it random? > Which test fails seems to be random (see attached screenshot). >> Looking at https://github.com/Tarsnap/scrypt it's only a trivial piece >> of userspace crypto. >> >> The fact that running multiple instances makes it fail more easily >> points towards some kind of register handling issue, but the fact that >> it repros only under Xen, and even with eager-fpu (which isn't the >> default on AMD, sadly), is weird. >> >> Looking at the scrypt source, it has alternative routines for the AESNI >> and SHANI instruction groups. However, because it's a Zen1, we don't >> have a useful way of filtering visible for PV dom0 userspace. >> >> First of all, can you get the exact CPU model and microcode version. >> `head /proc/cpuinfo` will be enough. But while you're at it, can you >> include `xl dmesg` too just in case there's something obvious showing up >> there too. >> > I have attachted text files with the (full) cpuinfo and the dmesg.
microcode : 0x8001129 That's 0x08001129 when not rendered brokenly, and the up-to-date version is 0x08001138 (which itself dates from 2019). If you can, get a firmware update. Given that it's a Dell, there's a good chance it's already on LFVS/fwupd. This is definitely the preferred option. If not, and you're willing to experiment in definitely unsupported territory, then move /lib/firmware/amd-ucode/microcode_amd_fam17h.bin sideways in dom0, and replace it with the attached SummitRidge-08001138 file (it's important to still be named microcode_amd_fam17h.bin in the end), then rebuild the initrd and reboot. You already have ucode=scan on Xen's command line, so after the reboot you should see some messages about updating microcode too. Irritatingly, AMD don't put client microcode into linux-firmware, but there are various collections of blobs found in the wild online. I've picked the one which I think is right for your CPU, and packaged it it appropriately for Xen. Anyway, I'm not sure if this will fix anything, but life is too short to be debugging stuff like this on out-of-date firmware/ucode. ~Andrew
SummitRidge-08001138
Description: Binary data
