On Thu, 26 Apr 2018, Benoit Lecocq wrote: > > > On 04/25/18 22:43, Peter Hessler wrote: > > On 2018 Apr 25 (Wed) at 22:39:13 +0200 (+0200), Jeremie Courreges-Anglas > > wrote: > > :On Wed, Apr 25 2018, Peter Hessler <[email protected]> wrote: > > :> fixes the build on armv7 (and very likely on aarch64 as well), OK > > : > > :Markus' diff makes sense and ftpcopy/ls work both on amd64 and arm; > > :ok jca@ > > : > > :One note though, __regparm__ looks mostly like premature optimization to > > :me. I had a wip diff that just did > > : > > : #define attribute_regparm(x) /* nothing */ > > : > > :and worked equally fine. > > : > > :-- > > :jca | PGP : 0x1524E7EE / 5135 92C1 AD36 5293 2BDF DDCC 0DFA 74AE 1524 E7EE > > > > I'm also happy with that. Either way works for me. > > > > > > ok benoit@
Reducing the define to nothing like Jeremie suggested is the better way I think. Here is the updated diff: Index: patches/patch-src_attributes_h =================================================================== RCS file: patches/patch-src_attributes_h diff -N patches/patch-src_attributes_h --- /dev/null 1 Jan 1970 00:00:00 -0000 +++ patches/patch-src_attributes_h 26 May 2018 06:09:14 -0000 @@ -0,0 +1,14 @@ +$OpenBSD$ + +Index: src/attributes.h +--- src/attributes.h.orig ++++ src/attributes.h +@@ -68,7 +68,7 @@ + #define attribute_inline __inline__ + + #if GNUC_MINIMUM(2,7) /* doesn't work reliable before, IIRC */ +-# define attribute_regparm(x) __attribute__((__regparm__((x)))) ++# define attribute_regparm(x) /* nothing */ + #else + # define attribute_regparm(x) + #endif
