linux-kernel - Re: [PATCH] x86/sev: Apply RMP table fixups for kexec.

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: <b2da0a15-c0f5-49a7-b34c-eb408e9aa293@amd.com>
Date: Tue, 2 Apr 2024 17:31:09 -0500
From: "Kalra, Ashish" <ashish.kalra@....com>
To: Tom Lendacky <thomas.lendacky@....com>, linux-tip-commits@...r.kernel.org
Cc: bp@...en8.de, michael.roth@....com, x86@...nel.org,
 linux-kernel@...r.kernel.org
Subject: Re: [PATCH] x86/sev: Apply RMP table fixups for kexec.


On 4/2/2024 5:09 PM, Tom Lendacky wrote:
> On 3/12/24 13:47, Ashish Kalra wrote:
>> From: Ashish Kalra <ashish.kalra@....com>
>>
>> RMP table start and end physical range may not be aligned to 2MB in
>> the e820 tables causing fatal RMP page faults during kexec boot when
>> new page allocations are done in the same 2MB page as the RMP table.
>> Check if RMP table start and end physical range in e820_table is not
>> aligned to 2MB and in that case use e820__range_update() to map this
>> range to reserved.
>>
>> Override e820__memory_setup_default() to check and apply these RMP table
>> fixups in e820_table before e820_table is used to setup
>> e280_table_firmware and e820_table_kexec.
>>
>> Fixes: c3b86e61b756 ("x86/cpufeatures: Enable/unmask SEV-SNP CPU 
>> feature")
>> Signed-off-by: Ashish Kalra <ashish.kalra@....com>
>> ---
>>   arch/x86/virt/svm/sev.c | 52 +++++++++++++++++++++++++++++++++++++++++
>>   1 file changed, 52 insertions(+)
>>
>> diff --git a/arch/x86/virt/svm/sev.c b/arch/x86/virt/svm/sev.c
>> index cffe1157a90a..e0d7584df28f 100644
>> --- a/arch/x86/virt/svm/sev.c
>> +++ b/arch/x86/virt/svm/sev.c
>> @@ -65,6 +65,8 @@ static u64 probed_rmp_base, probed_rmp_size;
>>   static struct rmpentry *rmptable __ro_after_init;
>>   static u64 rmptable_max_pfn __ro_after_init;
>>   +static char *__init snp_rmptable_e820_fixup(void);
>> +
>>   static LIST_HEAD(snp_leaked_pages_list);
>>   static DEFINE_SPINLOCK(snp_leaked_pages_list_lock);
>>   @@ -160,9 +162,59 @@ bool snp_probe_rmptable_info(void)
>>       pr_info("RMP table physical range [0x%016llx - 0x%016llx]\n",
>>           probed_rmp_base, probed_rmp_base + probed_rmp_size - 1);
>>   +    /*
>> +     * Override e820__memory_setup_default() to do any RMP table fixups
>> +     * for kexec if required.
>> +     */
>> +    x86_init.resources.memory_setup = snp_rmptable_e820_fixup;
>
> This produces a build warning:
>
> WARNING: modpost: vmlinux: section mismatch in reference: 
> snp_probe_rmptable_info+0x95 (section: .text) -> x86_init (section: 
> .init.data)
> WARNING: modpost: vmlinux: section mismatch in reference: 
> snp_probe_rmptable_info+0x99 (section: .text) -> 
> snp_rmptable_e820_fixup (section: .init.text)
>
Oh, so this requires snp_probe_rmptable_info() to be fixed to use the 
__init macro.

I believe that snp_probe_rmptable_info() should be anyway using the 
__init macro and this fix for snp_probe_rmptable_info() needs to be sent 
as a separate patch and regardless of this patch getting merged or not.

Thanks, Ashish

> Thanks,
> Tom
>
>> +
>>       return true;
>>   }