[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[Bug binutils/15951] New: binutils testsuite builds status wrapper uncon

From: akolesov at synopsys dot com
Subject: [Bug binutils/15951] New: binutils testsuite builds status wrapper unconditionally
Date: Fri, 13 Sep 2013 11:44:01 +0000


            Bug ID: 15951
           Summary: binutils testsuite builds status wrapper
           Product: binutils
           Version: 2.24 (HEAD)
            Status: NEW
          Severity: normal
          Priority: P2
         Component: binutils
          Assignee: unassigned at sourceware dot org
          Reporter: akolesov at synopsys dot com

Created attachment 7195
  --> https://sourceware.org/bugzilla/attachment.cgi?id=7195&action=edit
Suggested patch

binutils/testsuite/binutils-all/objcopy.exp always builds Dejagnu status
wrapper and links test prog with it. This causes problems on architectures that
doesn't implement function _exit() and testing is done using gdb-comm generic
configuration. Testglue wraps around function _exit regardless of its existence
and if this functions doesn't exists and symbol _exit will always point to
_wrap__exit. On the other hand Dejagnu gdb-comm.exp when performing execution
tests sets breakpoints at functions abort() and _exit(), and of the latter one
doesn't exist then at exit(). However because of status wrapper function _exit
always exists even if never called. As a result gdb-comm "waits" for an
application exit at an invalid location until Dejagnu will reach timeout.

Attached patch makes objcopy.exp aware of target_info field
needs_status_wrapper. This field is already used by GCC test suite. So if
status wrapper is disabled for GCC it will automatically be disabled for
objcopy.exp. Default behavior is unchanged, if this field is not set then
status wrapper will be used.

This patch resolves problem in my setup, when function _exit doesn't exists and
status wrapper is not required. It will not help with situation when _exit is
not defined but status wrapper is required.

You are receiving this mail because:
You are on the CC list for the bug.

reply via email to

[Prev in Thread] Current Thread [Next in Thread]