This is the mail archive of the
gcc-bugs@gcc.gnu.org
mailing list for the GCC project.
Bug in 3.0.4 (prerelease) in decl.c:start_method()
- From: "Kaveh R. Ghazi" <ghazi at caip dot rutgers dot edu>
- To: jason at redhat dot com, mark at codesourcery dot com
- Cc: gcc-bugs at gcc dot gnu dot org, gcc-patches at gcc dot gnu dot org
- Date: Sun, 17 Feb 2002 12:29:51 -0500 (EST)
- Subject: Bug in 3.0.4 (prerelease) in decl.c:start_method()
I noticed this new warning, since 3.0.3, in the 3.0.4 prerelease:
> cp/decl.c:14147: warning: passing arg 1 of `cplus_decl_attributes'
> from incompatible pointer type
I think it may indicate a latent bug because in fact on the 3.0 branch
cplus_decl_attributes is expecting a tree and gets passed a tree*.
I suspect the reason for this is that the patch which introduced this
was imported from the trunk where cplus_decl_attributes does in fact
expect a tree*.
I haven't had time to test this fix, other than to make sure it makes
the warning go away. But I wanted you to be aware of this since 3.0.4
is due. I'll do a full regtest in the mean time and let you know how
it goes.
Ok to install assuming no regressions?
--Kaveh
PS: I wonder why the code in PR907 (which was the genesis for this)
doesn't generate a tree checking failure somewhere. Someone who
understands this bit better should examine the resulting code to make
sure the issues in PR907 are being addressed and/or create a suitable
testcase.
2002-02-17 Kaveh R. Ghazi <ghazi@caip.rutgers.edu>
* decl.c (start_method): Fix error in last change.
--- egcc-3.0-CVS20020216/gcc/cp/decl.c~ Wed Feb 13 16:38:31 2002
+++ egcc-3.0-CVS20020216/gcc/cp/decl.c Sun Feb 17 11:28:48 2002
@@ -14144,7 +14144,7 @@ start_method (declspecs, declarator, att
return NULL_TREE;
if (attrlist)
- cplus_decl_attributes (&fndecl, attrlist, 0);
+ cplus_decl_attributes (fndecl, attrlist, 0);
/* Pass friends other than inline friend functions back. */
if (fndecl == void_type_node)