On 03/12/2018 09:56, Jan Beulich wrote: >>>> On 30.11.18 at 19:01, <[email protected]> wrote: >> On Fri, Nov 30, 2018 at 05:09:42PM +0000, Ian Jackson wrote: >>> Wei Liu writes ("[PATCH v2 0/3] Remove tmem"): >>>> It is agreed that tmem can be removed from xen.git. See the thread >>>> starting >> >> >> >>>> from <[email protected]>. >>> Those are notes from some phone call amongst industry stakeholders. >>> None of the messages have a Subject line mentioning tmem. There is no >>> explanation of the basis for the decision; just a confirmation from >>> the current maintainers that they will ack the removal. >>> >>> I think this is not really an appropriate way to carry on! What if >>> there is someone else who wants to step up to maintain this ? What >>> about user communication ? Going straight from `Supported' to >>> `Deleted' seems rather vigorous. >> Step up to maintain> I would rather say step up to develop. >> >> The status in MAINTAINERS is wrong. According to SUPPORT.md, it is only >> experimental. Our definition of "experimental" is: >> >> Functional completeness: No >> Functional stability: Here be dragons >> Interface stability: Not stable >> Security supported: No > Exactly. Plus my proposal to remove it was posted to xen-devel > on Aug 30th. I don't think removal of an experimental feature > requires posting to xen-announce. Ian - please reconsider your > nack.
I concur with Wei and Jan. TMEM has been off by default due to being declared "full of security holes - don't use" since XSA-15. That was in 2012, and TMEM hasn't made its way back into security support in that time. In addition, it was never fixed to work with Migration v2. The save side doesn't query any TMEM state, and convert-legacy-stream raises TODO on encountering legacy TMEM data. I don't know about other distributions, but it has been compiled out of XenServer for all versions which have Kconfig. tl;dr It doesn't work, and at this point, it looks very unlikely to change. There is a non-zero cost for retaining obsolete functionality, and the hypervisor maintainers want it gone in 4.12, which we think is entirely reasonable given the circumstances. ~Andrew _______________________________________________ Xen-devel mailing list [email protected] https://lists.xenproject.org/mailman/listinfo/xen-devel
