linux-kernel - Re: [PATCH] x86/head/64: Switch to KERNEL

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: <8a250f71-bd21-a0a7-f1f8-5e6a3bffb18b@amd.com>
Date:   Fri, 26 May 2023 08:07:44 -0500
From:   Tom Lendacky <thomas.lendacky@....com>
To:     Dave Hansen <dave.hansen@...el.com>, linux-kernel@...r.kernel.org,
        x86@...nel.org
Cc:     Thomas Gleixner <tglx@...utronix.de>,
        Ingo Molnar <mingo@...hat.com>, Borislav Petkov <bp@...en8.de>,
        Dave Hansen <dave.hansen@...ux.intel.com>,
        "H. Peter Anvin" <hpa@...or.com>,
        Michael Roth <michael.roth@....com>,
        Ashish Kalra <ashish.kalra@....com>,
        Joerg Roedel <jroedel@...e.de>
Subject: Re: [PATCH] x86/head/64: Switch to KERNEL_CS as soon as new GDT is
 installed

On 5/25/23 18:17, Dave Hansen wrote:
> On 5/17/23 09:26, Tom Lendacky wrote:
>> However, a recent patchset that looked to avoid using the legacy
>> decompressor during an EFI boot exposed this bug. At entry to startup_64,
>> the CS value is that of EFI and is not mapped in the new kernel GDT. So
>> when a #VC exception occurs, the CS value used by IRETQ is not valid and
>> the guest boot crashes.
> 
> This confused me a bit.  Nobody merged that patchset yet, right?  You
> just happened across this issue when debugging a crash in that *other* set?

Correct, it was Ard's EFI decompressor patchset series that he submitted, 
but has not yet been accepted, that I was testing:

[PATCH 0/6] efi/x86: Avoid legacy decompressor during EFI boot
   https://lore.kernel.org/lkml/20230424165726.2245548-1-ardb@kernel.org/

I reported the problem to Ard and submitted this patch before I realized 
that he also included a patch in his next version of the series:

https://lore.kernel.org/lkml/20230508070330.582131-16-ardb@kernel.org/

> 
>> Fix this issue by moving the block that switches to the KERNEL_CS value to
>> be done immediately after returning from startup_64_setup_env().
>>
>> Fixes: bcce82908333 ("x86/sev: Detect/setup SEV/SME features earlier in boot")
>> Signed-off-by: Tom Lendacky <thomas.lendacky@....com>
> 
> Any thoughts on whether we want this in stable@?

It probably doesn't need to go to stable since the current decompressor 
code switches to a new GDT and uses the same kernel CS value that is being 
setup in arch/x86/kernel/head_64.S (which is why it hasn't been a problem).

> 
> I also wonder whether we need a comment in that little chunk of code
> something along the lines of:
> 
> 	/*
> 	 * Do not add anything which might take a fault or exception.
> 	 * IRET does not work here.
> 	 */
> 
> Michael, do you think you would have spotted something like this had it
> been in the code when you were patching it?

Let me know if you would like a v2 with that comment. I'm also fine if you 
  just want to add it as part of applying the patch, too.

Thanks,
Tom