Gentoo Websites Logo
Go to: Gentoo Home Documentation Forums Lists Bugs Planet Store Wiki Get Gentoo!

Bug 802360

Summary: >=dev-libs/libxml2-2.9.12-r3: fails build with python and ABI_X86="64 32"
Product: Gentoo Linux Reporter: David Haller <gentoo>
Component: Current packagesAssignee: Sam James <sam>
Status: RESOLVED NEEDINFO    
Severity: normal CC: base-system, gentoo
Priority: Normal Keywords: PATCH
Version: unspecified   
Hardware: All   
OS: Linux   
See Also: https://bugs.gentoo.org/show_bug.cgi?id=582130
Whiteboard:
Package list:
Runtime testing required: ---
Attachments: patch to multilib wrap xmlversion.h

Description David Haller 2021-07-15 22:58:17 UTC
Created attachment 724081 [details, diff]
patch to multilib wrap xmlversion.h

If both ABI_X86="32 64" and python are enabled (and probably more combinations),
the build fails due to header checksum mismatches between ABIs. The reason is

==== include/xmlversion.h.in ====
#if @WITH_MODULES@
#define LIBXML_MODULES_ENABLED
/**
 * LIBXML_MODULE_EXTENSION:
 *
 * the string suffix used by dynamic modules (usually shared libraries)
 */
#define LIBXML_MODULE_EXTENSION "@MODULE_EXTENSION@"
#endif
====

and the fix is to wrap the header, see attached patch against both '-r3' and '-r4'
Comment 1 Sam James archtester Gentoo Infrastructure gentoo-dev Security 2021-07-15 23:11:48 UTC
Thanks!
Comment 2 David Seifert gentoo-dev 2021-08-03 22:18:46 UTC
(In reply to David Haller from comment #0)
> Created attachment 724081 [details, diff] [details, diff]
> patch to multilib wrap xmlversion.h
> 
> If both ABI_X86="32 64" and python are enabled (and probably more
> combinations),
> the build fails due to header checksum mismatches between ABIs. The reason is
> 
> ==== include/xmlversion.h.in ====
> #if @WITH_MODULES@
> #define LIBXML_MODULES_ENABLED
> /**
>  * LIBXML_MODULE_EXTENSION:
>  *
>  * the string suffix used by dynamic modules (usually shared libraries)
>  */
> #define LIBXML_MODULE_EXTENSION "@MODULE_EXTENSION@"
> #endif
> ====
> 
> and the fix is to wrap the header, see attached patch against both '-r3' and
> '-r4'

Could you post the contents of the two differing files? Maybe even a diff or so? Given that we can't reproduce the bug, I'm wary of just wrapping something in MULTILIB_WRAPPED_HEADERS without understanding the root cause.