NFS core dump blocks

This is not an issue in 2.5, but in 2.4:

When running a large app over NFS, it crashes and dumps core. This takes some time, during which ps, top, and other customers of /proc hang while they wait for the mmap_sem. Is there an argument against changing the

        down_write(&current->mm->mmap_sem);

in fs/binfmt_elf.c:elf_core_dump to

        down_read(&current->mm->mmap_sem);

Thanks,
Chad