Putting a (temporarly) build dependency on devel/openssl to make grpc happy.

Could you please upload a diff with extra context? In particular, see https://wiki.freebsd.org/Phabricator#Create_a_Revision_via_Web_Interface . Ditto for D30864.

@mmel sure, thanks for the heads up. Please provide the pointer once available.

I'm working on this - but situation about return value becomes more clear now. Please see comment in ofw_if.m https://cgit.freebsd.org/src/tree/sys/dev/ofw/ofw_bus_if.m#n148 and andrew just pointed me to this commit https://cgit.freebsd.org/src/commit/?h=0d8d9edaaaca1

Thanks for spotting this!

In D30867#694432, @ceri wrote:

In D30867#694382, @dbaio wrote:

just to be clear, DDG still supports multiple sites.

This doens't work anymore:

site:docs.FreeBSD.org/mail,lists.FreeBSD.org/pipermail

This works:

(site:www.FreeBSD.org OR site:docs.FreeBSD.org OR site:lists.FreeBSD.org OR site:wiki.FreeBSD.org OR site:forums.FreeBSD.org)

go to https://www.freebsd.org or https://www.freebsd.org/search/ (first form) and search for pgp, this is working there.

You can do a search that way, but the results are not as good. For instance, if you then change the date range to "past week", the multiple site version returns zero results while the single site version shows a commit made to net/keycloak. Additionally, clicking "All Results" on the multiple site version gives you a search for "pgp OR OR OR OR" rather than "pgp".

VS_LOCK/VS_UNLOCK in vi_interrupt causes deadlock in case MSI-X is disabled.

In D30867#694382, @dbaio wrote:

just to be clear, DDG still supports multiple sites.

This doens't work anymore:

site:docs.FreeBSD.org/mail,lists.FreeBSD.org/pipermail

This works:

(site:www.FreeBSD.org OR site:docs.FreeBSD.org OR site:lists.FreeBSD.org OR site:wiki.FreeBSD.org OR site:forums.FreeBSD.org)

go to https://www.freebsd.org or https://www.freebsd.org/search/ (first form) and search for pgp, this is working there.

Hi @mmel The patch is IMO correct and plan to commit it, so that to remove breakage for Armada platforms. In case any further improvements are required, feel free to continue the discussion.

mail/dovecot-pigeonhole: regen distinfo

In D30715#693653, @wosch wrote:

In D30715#693461, @fernape wrote:

Hi Wolfram,

Thanks for all the input. I think I'm close but I am not familiar with all the makefiles and macros we use in the different *.mk files. For the missing man pages that you mentioned, they are now installed but only in the general directory and not in the arch-specific one.
So for nvram.8, it is installed under /usr/share/man/man8/ but not under /usr/share/man/man8/powerpc

For the manual pages in man4 it was simple to do because we visit the architecture specific makefiles and they just need to be interpreted and executed. But in this case, I don't want to build and install the programs, just the man pages. MLINKS does not create directories, so how can I create the /i386 and /powerpc from the top level directory?

I have no idea about the MLINKS stuff and the arch sub-directories. Albeit not being perfect this patch is much better than what we currently have and good enough. Many thanks!.

Note: this patch should be merged 4 weeks later into stable/13 and stable/12, so it will affect the next stable releases.

Upstream MR https://gitlab.freedesktop.org/mesa/glu/-/merge_requests/5

In D30870#694401, @kib wrote:

You are locking (supposedly devfs) vnode for very long interval.

You are locking (supposedly devfs) vnode for very long interval. For instance, ffs_flushfiles() now runs under this lock. I do not like it. If we ever lock devvp inside SU code, this would recurse.

Embedding vnode into inode is indeed a lot of churn, together with some secondary troubles. For instance, unmount of the last filesystem for given vfs type no longer means that fs-specific zone can be destroyed. So although the idea of embedding have its merit, its scope is much larger (not only due to required code churn) than struct vnode shrinking by not allocating bufobj part, which is my goal in the patch.

Matt's suggestion is what @jbeich suggested in PR241568, I'll get a patch to freedesktop and amend this PR if it passes his review:

jhb's recommendations

If going for multiple zones, I think the way to go is to get rid of the current notion of vnodes as separate allocations to begin with and rework dooming/forced unmount semantics.

The main issue with devfs is that metadata io does not lock devvp. In other words, we cannot guarantee the liveness of devfs vnode for the duration of io, which results in problems with liveness of bufobj. I.e. bufobj could go away while owned buffer is under io. It is somewhat similar to mntfs reason, but have its own nuances. My conclusion was that the lifetime of bufobj must be equal to lifetime of vnode and not inode.

just to be clear, DDG still supports multiple sites.

This diff based on changes from https://reviews.freebsd.org/D30840 for file: sbin/devd/devd.conf.5

Add var old_name to keep if_xname.

Move code down, update date.

Ping.

The current error is grpc++ pkgconfig requiring grpc requiring openssl. Since you're not installing ports openssl and since base doesn't provide an openssl pkgconfig (/usr/libdata/pkgconfig/) openssl can't be found by pkgconfig.

Any other comments? Anyone able to approve?

In D30175#694137, @rozhuk.im-gmail.com wrote:

In D30175#693715, @donner wrote:

May you please add a paragraph to the rc man page, too?

I am not native speaker, it is a bit hard to me to write for humans.

Thank you for adding context to the diff.