Red Hat Bugzilla – Full Text Bug Listing
|Summary:||long double redirects unavailable in <math.h> and <complex.h> on ARM|
|Product:||[Fedora] Fedora||Reporter:||Daniel Drake <dsd>|
|Component:||glibc||Assignee:||Jeff Law <law>|
|Status:||CLOSED RAWHIDE||QA Contact:||Fedora Extras Quality Assurance <extras-qa>|
|Version:||15||CC:||fweimer, jakub, schwab|
|Fixed In Version:||Doc Type:||Bug Fix|
|Doc Text:||Story Points:||---|
|Last Closed:||2012-02-09 12:09:15 EST||Type:||---|
|oVirt Team:||---||RHEL 7.3 requirements from Atomic Host:|
Description Daniel Drake 2011-10-09 12:23:48 EDT
Functions such as creall() and cimagl() are not available on ARM (reproduced with glibc-2.14-7), due to failing to meet these preprocessor conditions in complex.h: #if (__STDC__ - 0 || __GNUC__ - 0) \ && (!defined __NO_LONG_DOUBLE_MATH || defined __LDBL_COMPAT) Packages which use these functions and use -Werror during build (such as libmpc) are failing to compile as a result. The use of -Werror is indeed questionable, but this does seem to be a glibc bug. Probably this one: http://sourceware.org/bugzilla/show_bug.cgi?id=10103 I confirmed that Jakub's patch still hasn't been applied in glibc HEAD.
Comment 1 Daniel Drake 2011-10-26 17:42:00 EDT
Created attachment 530394 [details] alternative fix This alternative fix was submitted by a Fedora-ARM contributor. Not sure if it is better or worse than Jakub's patch from 2009. Please let us know how we can help this issue get fixed.
Comment 2 Jakub Jelinek 2011-10-26 18:01:19 EDT
That is a bad idea, because then you penalize glibc quite a bit.
Comment 3 Daniel Drake 2011-10-26 18:06:30 EDT
OK, thanks for the quick response. Is your 2009 patch the right way to go then? Would it help you if we rediff/retest that patch against current glibc?
Comment 4 Fedora Admin XMLRPC Client 2011-11-14 14:45:16 EST
This package has changed ownership in the Fedora Package Database. Reassigning to the new owner of this component.
Comment 5 Jeff Law 2012-02-09 12:09:15 EST
Looks like this was fixed in the upstream sources on 11/4 and thus would be included in rawhide. If this is still an issue, please reopen the report with further details as to why the upstream fix was insufficient. Thanks, Jeff