Re: [patch 16/21] Xen-paravirt: Add code into head.S to handle being booted by Xen

From: ebiederm@xmission.com (Eric W. Biederman)
To: Jeremy Fitzhardinge <jeremy@goop.org>
Cc: Andi Kleen <ak@muc.de>, Andrew Morton <akpm@osdl.org>,
	linux-kernel@vger.kernel.org, virtualization@lists.osdl.org,
	xen-devel@lists.xensource.com, Chris Wright <chrisw@sous-sol.org>,
	Zachary Amsden <zach@vmware.com>,
	Ian Campbell <Ian.Campbell@XenSource.com>
Subject: Re: [patch 16/21] Xen-paravirt: Add code into head.S to handle being booted by Xen
Date: Wed, 14 Feb 2007 13:10:42 -0700	[thread overview]
Message-ID: <m1zm7gpkjh.fsf@ebiederm.dsl.xmission.com> (raw)
In-Reply-To: <45D35ABE.1050504@goop.org> (Jeremy Fitzhardinge's message of "Wed, 14 Feb 2007 10:53:50 -0800")

Jeremy Fitzhardinge <jeremy@goop.org> writes:

> Andi Kleen wrote:
>>> +#ifdef CONFIG_XEN
>>> +#include "../xen/xen-head.S"
>>> +#endif
>>>     
>>
>> Can this really not be linked separately? 
>
> I did a patch to do this (attached).  In principle it should be pretty
> simple, but I think I'm running into toolchain issues.  If I link
> xen-head.S separately (with appropriate headers added), it compiles OK
> and contains the right notes, but they don't appear in the final vmlinux
> properly.  The note segment and section are there, but no tools will
> parse them as notes:
>
> : ezr:pts/1; readelf -n vmlinux 
> : ezr:pts/1; ../../unstable/tools/xcutils/readnotes vmlinux 
> : ezr:pts/1; readelf -l vmlinux 
>
> Elf file type is EXEC (Executable file)
> Entry point 0x2e1f70
> There are 3 program headers, starting at offset 52
>
> Program Headers:
>   Type           Offset   VirtAddr   PhysAddr   FileSiz MemSiz  Flg Align
>   LOAD           0x001000 0xc0100000 0x00100000 0x2a8920 0x2a8920 R E 0x1000
>   LOAD           0x2aa000 0xc03a9000 0x003a9000 0x5f085 0xaf000 RWE 0x1000
>   NOTE           0x26d2910 0x00239010 0x00239010 0x000e8 0x00000 R   0x4
>
>  Section to Segment mapping:
>   Segment Sections...
>    00 .text .text.head __ex_table .rodata .pci_fixup __ksymtab __ksymtab_gpl
> __kcrctab __kcrctab_gpl __ksymtab_strings __param __bug_table
>    01 .data .paravirtprobe .data_nosave .data.page_aligned
> .data.cacheline_aligned .data.read_mostly .data.init_task .init.text .init.data
> .init.setup .initcall.init .con_initcall.init .altinstructions
> .altinstr_replacement .parainstructions .exit.text .init.ramfs .bss
>    02     .notes 
> : ezr:pts/1; objdump -j .notes -s vmlinux 
>
> vmlinux:     file format elf32-i386
>
> Contents of section .notes:
>  239010 04000000 06000000 06000000 58656e00  ............Xen.
>  239020 6c696e75 78000000 04000000 04000000  linux...........
>  239030 07000000 58656e00 322e3600 04000000  ....Xen.2.6.....
>  239040 08000000 05000000 58656e00 78656e2d  ........Xen.xen-
>  239050 332e3000 04000000 04000000 03000000  3.0.............
>  239060 58656e00 000000c0 04000000 04000000  Xen.............
>  239070 01000000 58656e00 c80710c0 04000000  ....Xen.........
>  239080 04000000 02000000 58656e00 00b040c0  ........Xen...@.
>  239090 04000000 2a000000 0a000000 58656e00  ....*.......Xen.
>  2390a0 21777269 7461626c 655f7061 67655f74  !writable_page_t
>  2390b0 61626c65 737c7061 655f7067 6469725f  ables|pae_pgdir_
>  2390c0 61626f76 655f3467 62000000 04000000  above_4gb.......
>  2390d0 04000000 09000000 58656e00 79657300  ........Xen.yes.
>  2390e0 04000000 08000000 08000000 58656e00  ............Xen.
>  2390f0 67656e65 72696300                    generic.        
>
>
> I can't see what the problem is here, but it looks like a toolchain
> problem to me (I'm using binutils-2.17.50.0.6-2.fc6).

Well I did a little by hand parsing and the not I parsed looked ok.

How does the output differ from a what you get when xen-head.S is
included?

Eric