linux-kernel - Re: [PATCH 00/02] kexec: Move segment code to assembly files

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [day] [month] [year] [list]

Message-ID: <aec7e5c30612050708w3fe2e0e4ye4d435dfb9667c41@mail.gmail.com>
Date:	Wed, 6 Dec 2006 00:08:25 +0900
From:	"Magnus Damm" <magnus.damm@...il.com>
To:	vgoyal@...ibm.com
Cc:	"Magnus Damm" <magnus@...inux.co.jp>,
	"Linux Kernel" <linux-kernel@...r.kernel.org>,
	fastboot@...ts.osdl.org, ebiederm@...ssion.com,
	"Andrew Morton" <akpm@...l.org>, "Rik van Riel" <riel@...hat.com>
Subject: Re: [PATCH 00/02] kexec: Move segment code to assembly files

On 12/5/06, Vivek Goyal <vgoyal@...ibm.com> wrote:
> On Tue, Dec 05, 2006 at 10:37:57PM +0900, Magnus Damm wrote:
> > kexec: Move segment code to assembly files
> >
> > The following patches rearrange the lowlevel kexec code to perform idt,
> > gdt and segment setup code in assembly on the code page instead of doing
> > it in inline assembly in the C files.
> >
>
> I don't think we should be doing this. I would rather prefer code to
> keep in C for easier debugging, readability and maintenance.

I prefer to write code in C too, but I don't see how wrapping assembly
instructions in inline C makes the code any easier compared to raw
assembly. Either you understand the assembly or you don't.

> > Our dom0 Xen port of kexec and kdump executes the code page from the
> > hypervisor when kexec:ing into a new kernel. Putting as much code as
> > possible on the code page allows us to keep the amount of duplicated
> > code low.
> >
>
> Is Xen going upstream now? I heard now lhype+KVM seems to be the way.
> Even if it is required, we should do it once Xen goes in.

I am not sure about status of the Xen merging effort. domU seemed to
be the top priority last time I heard something, but this change only
affects dom0 so it is probably even further away.

> You have already moved page table setup code to assembly and we should
> be getting rid of that code too.

This was recommended to me by Eric if I'm not mistaken, but if we can
move out parts of the assembly code to C then that would be great.

> I would rather live with duplicated code than moving more code in assembly
> which can be written in C. Understanding and debugging assembly code
> is such a big pain.

Again, I think that is true for C code - not for inline assembly in C
files. But I guess you are talking about the already merged page table
a patches. My first version implemented the code in C, have a look at
the function create_mapping() which I think is very clear:

http://lists.osdl.org/pipermail/fastboot/2006-May/002838.html

The important question IMO is if this should be merged ahead of the
rest of the Xen stuff, and maybe it shouldn't.

Thanks,

/ magnus
-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/