This is the mail archive of the gcc-patches@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [v3] testsuite/ext cleanups


Andrew Pinski wrote:

> On Oct 24, 2004, at 1:58 PM, Aaron W. LaFramboise wrote:
> 
>>This breaks i686-pc-mingw32 bootstrap.
>>
>>testsuite/testsuite_abi.h includes <ext/hash_map>, and the command line
>>for testsuite/testsuite_abi.cc specifies "-I.", so GCC incorrectly
>>attempts to include testsuite/ext/hash_map, which is the new directory,
>>which obviously won't work.
> 
> Is this a cross compiler or a native one.  If it is a cross compiler and
> you are building on a reiser file system, this is PR 15772 but really 
> this
> is a bug in glibc/reiser.
> 
> If this is a native one there was bug about this, this is most likely a
> bug in mingw32 code, it is returning "Permission denied" when it should
> have returned a different error message.

This is a native i686-pc-mingw32 bootstrap on Windows XP SP2 with NTFS.

The confusing "Permission denied" error ultimately comes from the
behavior of CreateFile() on Windows, which is to set ERROR_ACCESS_DENIED
as the failure condition when no other error number particularly
applies.  Windows does not have an equivilent to the Unix EISDIR.

If this is a bug, it is a bug that has existed for almost two decades,
that is probably impossible to work around sanely in the C library.  GCC
could add special Windows-specific handling to provide a more
descriptive error message, but its probably not worth it.


Aaron W. LaFramboise


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]