This is the mail archive of the libc-hacker@sources.redhat.com mailing list for the glibc project.

Note that libc-hacker is a closed list. You may look at the archives of this list, but subscription and posting are not open.


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

Re: Recent glibc locale changes on ia64.


   From: Ulrich Drepper <drepper@redhat.com>
   Date: 07 Sep 2000 17:59:20 -0700

   "H . J . Lu" <hjl@valinux.com> writes:

   > My last successful "make check" in glibc 2.2 on ia64 was made on Sep. 1.
   > Since then, I got,

   It fails on PPC as well for whatever reason.  It doesn't on x86 and
   this is my measuring point.  Somebody will have to fix it and it's
   defeinitely not me since I still have no working IA-64 system with
   glibc 2.2.

From the descriptions I've seen of this problem, it sounds a lot like
yet another unaligned memory access problem in the locale code.

Isn't there a way to make the x86 signal an cpu exception when data is
accessed in an unaligned fashion?  Ulrich, you could make use of that
if it were true.  (I am rather positive that 486 and later provide
such a switch, perhaps in the eflags register)

This is important beyond correctness for risc-list cpus, it can turn
into a quiet performance problem even on x86 which doesn't trap but
executes those memory accesses much more slowly.

Later,
David S. Miller
davem@redhat.com

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