From mboxrd@z Thu Jan  1 00:00:00 1970
From: "Jan Beulich" <jbeulich@novell.com>
Subject: Re: [patch 1/2] Relocate VDSO ELF headers to match mapped
	location with COMPAT_VDSO
Date: Thu, 05 Apr 2007 08:10:40 +0100
Message-ID: <4614BD10.76E4.0078.0@novell.com>
References: <20070405045825.511024444@goop.org>
 <20070405045843.333498131@goop.org>
Mime-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Return-path: <linux-kernel-owner+glk-linux-kernel-3=40m.gmane.org-S1946015AbXDEHJz@vger.kernel.org>
In-Reply-To: <20070405045843.333498131@goop.org>
Content-Disposition: inline
Sender: linux-kernel-owner@vger.kernel.org
To: Jeremy Fitzhardinge <jeremy@goop.org>
Cc: Ingo Molnar <mingo@elte.hu>, Andrew Morton <akpm@linux-foundation.org>, virtualization@lists.osdl.org, Roland McGrath <roland@redhat.com>, Andi Kleen <ak@suse.de>, lkml <linux-kernel@vger.kernel.org>, Zachary Amsden <zach@vmware.com>, "Eric W. Biederman" <ebiederm@xmission.com>
List-Id: virtualization@lists.linuxfoundation.org

>+static __cpuinit void reloc_dyn(Elf32_Ehdr *ehdr, unsigned offset)
>+{
>+	Elf32_Dyn *dyn = (void *)ehdr + offset;
>+
>+	for(; dyn->d_tag != DT_NULL; dyn++)
>+		switch(dyn->d_tag) {
>+		case DT_PLTGOT:
>+		case DT_HASH:
>+		case DT_STRTAB:
>+		case DT_SYMTAB:
>+		case DT_RELA:
>+		case DT_INIT:
>+		case DT_FINI:
>+		case DT_REL:
>+		case DT_JMPREL:
>+		case DT_VERSYM:
>+		case DT_VERDEF:
>+		case DT_VERNEED:
>+			dyn->d_un.d_val += VDSO_HIGH_BASE;
>+		}
>+}

While there's a certain level of control on what DT_* may appear in the
vDSO, not even considering other than the above types seems fragile to
me. Since future additions to the set are supposedly following a fixed
scheme (distinguishing pointers and values via the low bit when below
OLD_DT_LOOS, and using sub-ranges when between DT_HIOS and
OLD_DT_HIOS), at least also handling those would seem like a good
idea, as would warning about unrecognized types.

Also, even though it shouldn't matter for the final result, if doing things
spec-conforming here you should use d_un.d_ptr.

In addition to Roland's remarks about missing symbol table relocation, I
would also assume section headers, if present, should be relocated.

Jan