Re: git: 5670b8cc3672 - main - libthr: Preresolve selected EABI symbols on arm.
Date: Sat, 27 Jul 2024 04:01:35 UTC
On Jul 26, 2024, at 20:28, Philip Paeps <philip@freebsd.org> wrote: > On 2024-07-27 07:57:38 (+0800), Mark Millard wrote: >> Michal Meloun <mmel_at_FreeBSD.org> wrote on >> Date: Thu, 25 Jul 2024 16:25:09 UTC : >> >>> The branch main has been updated by mmel: >>> >>> URL: https://cgit.FreeBSD.org/src/commit/?id=5670b8cc3672d5a6bc2c41eb48d7d01343c43ad0 >>> >>> commit 5670b8cc3672d5a6bc2c41eb48d7d01343c43ad0 >>> Author: Michal Meloun <mmel@FreeBSD.org> >>> AuthorDate: 2024-07-24 15:11:27 +0000 >>> Commit: Michal Meloun <mmel@FreeBSD.org> >>> CommitDate: 2024-07-25 16:24:22 +0000 >>> >>> libthr: Preresolve selected EABI symbols on arm. >>> >>> Add the ability to pre-resolve architecture-specific EABI symbols and >>> use it on arm for selected EABI functions. These functions can be called >>> with rtld bind lock write-locked, so they should be resolved in forward. >>> >>> Reported by: Mark Millard <marklmi@yahoo.com>, John F Carr <jfc@mit.edu> >>> Reviewed by: kib, imp >>> MFC after: 1 week >>> Differential Revision: https://reviews.freebsd.org/D46104 >> >> Philip Paeps is likely going to want to know if any releng/13.* or releng/14.* >> that would be in some jail(s) on ampere1 or ampere3 will be updated to have >> the change. He likely would do a round of updating the jail content vintages >> in use for any such updated releng/13.* or releng/14.* . >> >> So: Any plans to have any already-supported release builds updated? >> >> Similarly: What of the upcoming 13.4-RELEASE that has made it to -PRERELEASE >> so far? >> >> (I'm not aware of active port building runs based on stable/13 or >> stable/14 . But, if there are some, they would likely track the MFC updates >> sometime after the MFC update was in place.) > > We don't build packages on stable branches. Not normally. But there the following existed at one time: 14stable-i386-default on beefy11.nyi.freebsd.org <http://beefy11.nyi.freebsd.org/> 14stable-amd64-default on beefy12.nyi.freebsd.org <http://beefy12.nyi.freebsd.org/> They may have been very temporary and only for special test runs for all I know. While the actual build history is gone now, https://portsfallout.com/server still shows the names under "build environment". > Packages are produced on the earliest supported releng/x branch. Yep. My questions for releng/1[34].* do get into what will be the definition of "supported versions" for at least armv7. The armv7 jails need the fix if they are to avoid being brittle, which gets into if a EN would be done to change the definition of "support version" in order to avoid the armv7 status. The rule about "earliest supported" would then need to track the update if that gets such an EN update. > I don't actually know how/when the builder jails get upgraded. For main to be fixed, the jails have to have been updated. The host version of main does not (at least relative to the code at hand). For any existing, supported releng/1[34].* to avoid being brittle like main was, it would need an EN update and both the base installation and the jails that would then been intended to have avoided the armv7 status would need to be updated. > That's a question for Antoine (Cc:ed) who manages the actual building. clusteradm only manages the base installations. It may be that only main will be fixed unless an example of the brittle status is observed in some supported version (or new releng/*.* branch not yet released). === Mark Millard marklmi at yahoo.com