Re: [PATCH v6 09/10] dax/kmem: add sysfs interface for atomic whole-device hotplug
Hannes Reinecke <[email protected]>
| Newsgroups | dev.linux.lists.nvdimm,dev.linux.lists.driver-core,org.kernel.vger.linux-cxl,org.kernel.vger.linux-kernel,org.kernel.vger.linux-kselftest,org.kvack.linux-mm |
|---|---|
| Message-ID | <[email protected]> |
On 7/1/26 12:14 AM, Gregory Price wrote: > On Tue, Jun 30, 2026 at 05:18:41PM -0400, Gregory Price wrote: >> There is no atomic mechanism to offline and remove an entire >> multi-block DAX kmem device. This is presently done in two steps: > > ... snip snip snip ... > > Sashiko pointed out a false-positive, but adding a fixup patch > here that adds additional consistency. > > On total failure - release all resources. This makes the sysfs > interface consistent with the probe failure path. > Speaking of which ... With this patch we now have _two_ interfaces to do the same thing. And both will be generating uevents. Which is far from ideal (one could easily envision _conflicting_ udev rules, one set doing an 'online' on the old interface, and another set doing an 'offline' on the new interface...) Is there a way to not sending uevents for the old interface or to make it configurable? The old interface had the nasty side effect of generating _tons_ of uevents during booting, and on larger machines we even had seen udev acting as a fork-bomb during booting, taking down the entire machine (we had to restrict udev to 512 threads max to avoid that from happening). So if we could disable uevents for the old interface things would be _so_ much easier ... Cheers, Hannes -- Dr. Hannes Reinecke Kernel Storage Architect [email protected] +49 911 74053 688 SUSE Software Solutions GmbH, Frankenstr. 146, 90461 Nürnberg HRB 36809 (AG Nürnberg), GF: I. Totev, A. McDonald, W. Knoblich