other/10109: documentation of target macro LIBCALL_VALUE

[a.cilio@et.tudelft.nl]

>Number:         10109
>Category:       other
>Synopsis:       documentation of target macro LIBCALL_VALUE
>Confidential:   no
>Severity:       non-critical
>Priority:       low
>Responsible:    unassigned
>State:          open
>Class:          doc-bug
>Submitter-Id:   net
>Arrival-Date:   Sun Mar 16 14:56:01 UTC 2003
>Closed-Date:
>Last-Modified:
>Originator:     Andrea Cilio
>Release:        any from 2.7 to 3.2
>Organization:
>Environment:
not relevant
>Description:
The description of target macro LIBCALL_VALUE ( ) in the internals manual states:
"[...] The definition of LIBRARY_VALUE[sic] need not be concerned [with] aggregate data types, because none of the library functions returns such types."
I assume that LIBCALL_VALUE is intended where LIBRARY_VALUE is mentioned (the latter does not exist).
Another problem is that it is not clear whether this macro must take into account that some library support functions, although not returning aggregate types, may require to return their result in memory like aggregates and BLKmode values. This is the case of functions that return "double" and emulate double precision floating point on targets with only 32-bit integer registers.
Or should "need not be concerned" be interpreted as the fact that gcc cannot handle support functions that return their result in memory?
>How-To-Repeat:

>Fix:

>Release-Note:
>Audit-Trail:
>Unformatted: