Bug 262053

Summary:	Fix sys-process/procps for arm and gcc < 4.4
Product:	Gentoo Linux	Reporter:	Sven 'sleipnir' Rebhan <OdinsHorse>
Component:	[OLD] Core system	Assignee:	Gentoo's Team for Core System packages <base-system>
Status:	RESOLVED WORKSFORME
Severity:	normal	CC:	embedded, maxposedon
Priority:	High
Version:	unspecified
Hardware:	All
OS:	Linux
Whiteboard:
Package list:		Runtime testing required:	---
Bug Depends on:	261111
Bug Blocks:
Attachments:	Fix for arm and gcc < 4.4 + cross-compile include fix

Description Sven 'sleipnir' Rebhan 2009-03-10 21:44:20 UTC

Compilation of procps fails for gcc < 4.4 because -frename-registers does not work on that version (see http://gcc.gnu.org/bugzilla/show_bug.cgi?id=36209 for details).

Furthermore, a wrong include path for cross-compilation is fixed.

Reproducible: Always

Comment 1 Sven 'sleipnir' Rebhan 2009-03-10 21:46:46 UTC

Created attachment 184627 [details]
Fix for arm and gcc < 4.4 + cross-compile include fix

The patch fixes a bug for arm and gcc < 4.4. Furthermore, it contains a cross-compile include fix.

Comment 2 SpanKY gentoo-dev

2009-03-10 22:30:39 UTC

Comment on attachment 184627 [details]
Fix for arm and gcc < 4.4 + cross-compile include fix

these changes are no good

if gcc is generating bad code, file a bug for that

as for ncurses, using ROOT in src_* ebuilds is wrong.  simply delete the -I paths and see if it works.

Comment 3 Sven 'sleipnir' Rebhan 2009-03-11 20:33:04 UTC

The gcc fix was submitted to bug #261111. The package cross-compiles even with the wrong ncurses include path. If a patch is appreciated, I will provide the according sed line.

Comment 4 SpanKY gentoo-dev

2009-03-11 22:24:54 UTC

yes, please post a sed to remove the -I to /usr/include/....

Comment 5 Sven 'sleipnir' Rebhan 2009-03-12 11:29:02 UTC

I took a look at the sources of procps and see that there are quite some includes refereing to ncurses.h. I think, removing the -I/usr/include is wrong. So how do we proceed here?

Comment 6 SpanKY gentoo-dev

2009-03-14 12:03:27 UTC

if you read the ebuild you'll see we already clobber CPPFLAGS.  so whatever the package itself is doing is irrelevant.

the gcc issue has its own bug which means there's nothing left here