AMD Threadripper 3900 Series MCE Fix Queued In RAS/Core But Not Yet Mainlined

[agd5f]

Fixes usually have to land in master before they get applied to stable. You can add an explicit cc to stable if the patch applies to the stable kernels you want to apply it to, otherwise, you need to backport the change to the kernels and send them directly to stable. On top of that, Sasha Levin's AI stuff will generally pick up all patches with a "fixes" tag and attempt to apply it to any relevant stable kernels based on when the commit in the fixes tags landed.

[Michael]

I'm a bit fuzzy on the details (vacation brain) but AFAIK the "Fixes" tag in the patch marks it for going into stable as well as mainline, and that has worked well on previous fixes:

Usually there is "Cc: <[email protected]>" for copying to stable, granted, patches without that are still routinely picked up for back-porting.

While these patches are in ras/core, there is ras/urgent which is where the "fixes" PR tend to be sent in from RAS area for the current cycle. ras/urgent saw a new patch today but not the patch in question (https://git.kernel.org/pub/scm/linux.../?h=ras/urgent) so still looking like it might not come until 5.6.

[nuetzel]

Thanks !



I'm a bit fuzzy on the details (vacation brain) but AFAIK the "Fixes" tag in the patch marks it for going into stable as well as mainline, and that has worked well on previous fixes:

Jup. As soon as it lands in Linus's git (with stable tag), w'll see it in stable, too.
Maybe faster in (open)SUSE kernels...

[stormcrow]

Given that AMD's core market is the enthusiast & developer crowd with the mid to higher end of their CPUs, I'm continually amazed they're consistently dropping the ball on getting support to Linux, if not the BSDs, before launch to make sure those people have smooth experiences. This is especially puzzling with Threadripper releases, since these are definitely not for the gaming enthusiast crowd nor the typical desktop. These are workstation and server class CPUs, most of which will go to Linux stations and servers with some BSD users sprinkled about.

[bridgman]

Clarified the sentence to point it out more as yes he is one of the maintainers, just wanted to bring it up that Boris formerly worked for AMD and encouraging more AMD Linux efforts for 2020

Thanks !

Since this is a bug fix, why isn't it merged to a point release of the latest stable kernel?

I'm a bit fuzzy on the details (vacation brain) but AFAIK the "Fixes" tag in the patch marks it for going into stable as well as mainline, and that has worked well on previous fixes:

Fixes: 68627a697c19 ("x86/mce/AMD, EDAC/mce_amd: Enumerate Reserved SMCA bank type")

[Michael]

AFAIK the patches are being routed through Borislav because he and Tony Luck are the RAS maintainers rather than anything to do with SUSE/AMD.

Clarified the sentence to point it out more as yes he is one of the maintainers, just wanted to bring it up that Boris formerly worked for AMD and encouraging more AMD Linux efforts for 2020

[bridgman]

Last week though the patch was picked up into the ras/core tree. Though as of writing isn't in mainline. And it's a bit ironic still the AMD MCE patches continue to be routed through Borislav Petkov thanks to his currently employment by SUSE

AFAIK the patches are being routed through Borislav because he and Tony Luck are the RAS maintainers rather than anything to do with SUSE/AMD.

I'm not sure about workflow details for patches that are headed for both mainline and stable, but my impression was that going through ras/core was part of the flow.

I believe the intent is for the patch to go into mainline and stable as well but it's being managed on the CPU side of the house. I checked with them and they seemed to be aware and on top of it.

[Venemo]

Since this is a bug fix, why isn't it merged to a point release of the latest stable kernel?

[phoronix]

Phoronix: AMD Threadripper 3900 Series MCE Fix Queued In RAS/Core But Not Yet Mainlined

The AMD Ryzen Threadripper 3960X/3970X are incredibly fast and trounce the competition, but as noted on launch-day most (all?) Linux distributions have a boot issue with them over a machine check exception. There is an easy workaround to let these core-happy CPUs boot and run Linux while the proper fix was queued last week in ras/core in what looks like it will wait until Linux 5.6 for merging...

http://www.phoronix.com/scan.php?page=news_item&px=Threadripper-3-MCE-In-RAS-Core