Re: OCaml / aarch64 / binutils / coq mess

2017-09-14 Thread Jerry James
On Thu, Sep 14, 2017 at 6:49 AM, Richard W.M. Jones wrote: > More news. > > https://github.com/ocaml/ocaml/pull/1330 > > Upstream came up with a different fix. It's slower since every > reference (even within compilation units) now goes through the GOT, > but it's likely to be safer and it's th

Re: OCaml / aarch64 / binutils / coq mess

2017-09-14 Thread Richard W.M. Jones
More news. https://github.com/ocaml/ocaml/pull/1330 Upstream came up with a different fix. It's slower since every reference (even within compilation units) now goes through the GOT, but it's likely to be safer and it's the one they went with for the future 4.06 release. Therefore I have add

Re: OCaml / aarch64 / binutils / coq mess

2017-09-07 Thread Richard W.M. Jones
On Wed, Sep 06, 2017 at 08:38:57PM -0600, Jerry James wrote: > On Wed, Sep 6, 2017 at 2:20 PM, Richard W.M. Jones wrote: > > Nearly done it. Unfortunately I updated ‘why’ to the latest upstream > > version (the previous one didn't know about OCaml 4.05), however that > > will require updating ‘fr

Re: OCaml / aarch64 / binutils / coq mess

2017-09-06 Thread Jerry James
On Wed, Sep 6, 2017 at 2:20 PM, Richard W.M. Jones wrote: > Nearly done it. Unfortunately I updated ‘why’ to the latest upstream > version (the previous one didn't know about OCaml 4.05), however that > will require updating ‘frama-c’: > > configure: WARNING: bad Frama-c version "Silicon-201611

Re: OCaml / aarch64 / binutils / coq mess

2017-09-06 Thread Richard W.M. Jones
Nearly done it. Unfortunately I updated ‘why’ to the latest upstream version (the previous one didn't know about OCaml 4.05), however that will require updating ‘frama-c’: configure: WARNING: bad Frama-c version "Silicon-20161101", you need version Phosphorus At that point I stopped. Would y

Re: OCaml / aarch64 / binutils / coq mess

2017-09-06 Thread Richard W.M. Jones
I believe I have come up with a patch which works well enough that we can push it to Fedora 27+ aarch64: https://pagure.io/fedora-ocaml/c/e7fdaf008e047c445fd7a6acf9362d8b5940bf4b?branch=fedora-27-4.05.0 Note this patch is not upstream but I've brought it to the attention of the upstream mainta

Re: OCaml / aarch64 / binutils / coq mess

2017-08-31 Thread Richard W.M. Jones
On Wed, Aug 30, 2017 at 07:23:04PM -0600, Jerry James wrote: > BTW, most of the packages that sit on top of coq have new versions > available, so when this is fixed, I will want to update the other > packages anyway. If you could let me know when a fix for this issue > is available, I will take ca

Re: OCaml / aarch64 / binutils / coq mess

2017-08-30 Thread Jerry James
On Wed, Aug 30, 2017 at 2:57 PM, Dan Horák wrote: > On Tue, 29 Aug 2017 14:51:49 +0100 > "Richard W.M. Jones" wrote: > >> (1) Continue having broken dependencies for the affected packages on >> aarch64 for a bit and see if upstream come up with anything. > > I would lean to that "solution" Me, t

Re: OCaml / aarch64 / binutils / coq mess

2017-08-30 Thread Dan Horák
On Tue, 29 Aug 2017 14:51:49 +0100 "Richard W.M. Jones" wrote: > OCaml 4.05 was added to Fedora 27+ recently. Unfortunately, on > aarch64 only, it interacts badly with a change made in binutils 2.29 > which tightens up the rules on relocations for PC-relative addresses. > More details: > > ht

OCaml / aarch64 / binutils / coq mess

2017-08-29 Thread Richard W.M. Jones
OCaml 4.05 was added to Fedora 27+ recently. Unfortunately, on aarch64 only, it interacts badly with a change made in binutils 2.29 which tightens up the rules on relocations for PC-relative addresses. More details: https://caml.inria.fr/mantis/view.php?id=7585detailed discussion