This is the mail archive of the
libc-alpha@sources.redhat.com
mailing list for the glibc project.
Re: Problem with p_align and AT_PAGESZ check
- From: "H. J. Lu" <hjl at lucon dot org>
- To: Ulrich Drepper <drepper at redhat dot com>
- Cc: GNU C Library <libc-alpha at sources dot redhat dot com>, binutils at sources dot redhat dot com
- Date: Sat, 14 Jun 2003 09:41:45 -0700
- Subject: Re: Problem with p_align and AT_PAGESZ check
- References: <20030614151821.GA1741@lucon.org> <3EEB4B7A.8000209@redhat.com>
On Sat, Jun 14, 2003 at 09:21:14AM -0700, Ulrich Drepper wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> H. J. Lu wrote:
> > On i386, it should be
> >
> > if (p_align != 0x1000
> > || ((GL(dl_pagesize) & (p_align - 1)) != 0)
> > error
> >
> > or
> > if (p_align != 0x1000)
> > error
>
> No. The file can specify higher alignment requirements, there is no
> problem with that as long as the p_align is a multiple of the pagesize.
>
> Whether we should fail if the actual alignment is not a multiple of the
> pagesize can be discussed. I probably would agree that it should be
> fatal since the extra work necessary and the subtle changes in the
> semantics make supporting this undesirable.
>
We want to try 16KB page size on existing i386 binaries if hardware
supports it. But ld.so won't load them. The main problem I can think of
is mmap and mprotect. I can it can be dealt with.
Maybe it is the time to increase MAXPAGESIZE for ELF/i386. Any
objections to change it to 16KB or 64KB? We still have COMMONPAGESIZE
set to 4KB. We shouldn't waste much memory.
H.J.