linux-kernel - Re: [PATCH v1 0/3] x86/boot: Supply boot_param in rdi instead of rsi from startup

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

Message-ID: <CAMzpN2jBOW3PVJQLvua=knQNu_1mZ6RZYi8JvNenFd5t30UC1Q@mail.gmail.com>
Date: Mon, 23 Jun 2025 08:33:05 -0400
From: Brian Gerst <brgerst@...il.com>
To: Khalid Ali <khaliidcaliy@...il.com>
Cc: tglx@...utronix.de, mingo@...hat.com, bp@...en8.de, 
	dave.hansen@...ux.intel.com, ubizjak@...il.com, x86@...nel.org, hpa@...or.com, 
	linux-efi@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH v1 0/3] x86/boot: Supply boot_param in rdi instead of rsi
 from startup_64()

On Mon, Jun 23, 2025 at 4:16 AM Khalid Ali <khaliidcaliy@...il.com> wrote:
>
> From: Khalid Ali <khaliidcaliy@...il.com>
>
> The current kernel entry point takes one argument which is boot_param
> from RSI. The only argument entry point recieves is pointer to
> boot_param.
>
> In order to comply with the ABI calling convension the entry point must
> recieve the boot_param from RDI instead of RSI. There were no specific
> use case used for RDI, so the kernel can safely recieve argument from
> that register to better comply with ABI.
>
> This patch makes the kernel to recieve boot_param which is the only
> argument it recieves, from RDI instead of RSI. All changes needed for
> stability and clarity have being changed.
>
> Changelog:
>  * Kernel uncompressed entry point expects boot_param from RDI instead
>    of RSI.
>  * The decompressor has being adjusted to supply argument from RDI
>    instead RSI.
>  * libstub has being adjusted to supply argument from RDI instead of RSI.
>
> After throughly tested there were no regression and UDs has being
> observed. Looking forward for feedback.
>
>  arch/x86/boot/compressed/head_64.S      | 2 +-
>  arch/x86/kernel/head_64.S               | 4 ++--
>  drivers/firmware/efi/libstub/x86-stub.c | 4 ++--
>  3 files changed, 5 insertions(+), 5 deletions(-)

This was never intended to conform to the C ABI, why is it necessary
to change it?

Also, you cannot break this up into three patches.  Every patch must
be fully functional so that git bisect will work.


Brian Gerst