Date: Mon, 12 Dec 2022 20:08:16 +0000 From: bugzilla-noreply@freebsd.org To: net@FreeBSD.org Subject: [Bug 264094] VNET and DPCPU defined in modules panic on use on arm64 after LLVM 14 import due to new linker relaxation Message-ID: <bug-264094-7501-VV4bXwkpYd@https.bugs.freebsd.org/bugzilla/> In-Reply-To: <bug-264094-7501@https.bugs.freebsd.org/bugzilla/> References: <bug-264094-7501@https.bugs.freebsd.org/bugzilla/>
next in thread | previous in thread | raw e-mail | index | archive | help
https://bugs.freebsd.org/bugzilla/show_bug.cgi?id=3D264094 --- Comment #21 from Mark Johnston <markj@FreeBSD.org> --- (In reply to Jessica Clarke from comment #20) Oh, right. So kmod_syms.awk is defeating the hack in VNET_DEFINE_STATIC. = That can be bandaged easily enough. I tried implementing your suggestion of adding an explicit offset with each VNET definition. It's a bit painful due to consumers doing things like: VNET_DEFINE(LIST_HEAD(, foo), n) =3D LIST_HEAD_INITIALIZER(); I don't think it's possible to implement VNET_DEFINE this way without either outlawing static initializers (or making them an argument to the macro itse= lf instead of using assignment), or outlawing anonymous types, i.e., the above would have to be: LIST_HEAD(foo_list, foo); VNET_DEFINE(struct foo_list, n) =3D LIST_HEAD_INITIALIZER(); I would go with the second option, but it does require a bunch of consumers= to be patched. I haven't yet compared the generated code with what we had before. --=20 You are receiving this mail because: You are on the CC list for the bug.=
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?bug-264094-7501-VV4bXwkpYd>