From: Juergen Gross <jgross@suse.com>
To: Borislav Petkov <bp@alien8.de>
Cc: linux-kernel@vger.kernel.org, x86@kernel.org,
Thomas Gleixner <tglx@linutronix.de>,
Ingo Molnar <mingo@redhat.com>,
Dave Hansen <dave.hansen@linux.intel.com>,
"H. Peter Anvin" <hpa@zytor.com>
Subject: Re: [PATCH v5 03/15] x86/mtrr: replace some constants with defines
Date: Wed, 5 Apr 2023 09:55:59 +0200 [thread overview]
Message-ID: <1231926a-5d03-5631-a376-b0d738c7e25c@suse.com> (raw)
In-Reply-To: <20230403160323.GAZCr4y14GvOA4dZID@fat_crate.local>
[-- Attachment #1.1.1: Type: text/plain, Size: 3028 bytes --]
On 03.04.23 18:03, Borislav Petkov wrote:
> On Sat, Apr 01, 2023 at 08:36:40AM +0200, Juergen Gross wrote:
>> @@ -643,10 +646,12 @@ static bool set_mtrr_var_ranges(unsigned int index, struct mtrr_var_range *vr)
>> unsigned int lo, hi;
>> bool changed = false;
>>
>> +#define BASE_MASK (MTRR_BASE_TYPE_MASK | (size_and_mask << PAGE_SHIFT))
>> +#define MASK_MASK (MTRR_MASK_VALID | (size_and_mask << PAGE_SHIFT))
>
> No, "MASK_MASK" is too much. :-)
Any better suggestion for the name? :-)
> Anyway, so I started looking at this and here's what I'm seeing on my
> machine even with the old code:
>
> rdmsr(MTRRphysBase_MSR(index), lo, hi);
> if ((vr->base_lo & 0xfffff0ffUL) != (lo & 0xfffff0ffUL)
> || (vr->base_hi & (size_and_mask >> (32 - PAGE_SHIFT))) !=
> (hi & (size_and_mask >> (32 - PAGE_SHIFT)))) {
>
>
> size_and_mask: 0x0000000ffff00000
> the shifted version: 0x000000000000ffff
>
> which is bits [15:0]
>
> Now, looking at MTRRphysBase_MSR's definition, the high 32 bits are:
>
> [63 ... reserved ... ][ max_addr ... 32 ]
>
> and that second slice: from max_addr to the 32nd bit of the whole MSR is
> the second part of PhysBase.
>
> max_addr aka phys_addr comes from:
>
> phys_addr = cpuid_eax(0x80000008) & 0xff;
>
> on that machine, that value is 48.
>
> Which means, the physaddr slice should be [48 .. 32], i.e.,
>
> 0x0001ffff00000000
No. The "48" is the _number_ of physical address bits, so the 64 bit address
mask will be 0000ffff.ffffffff (48 bits set).
>
> and when you shift that by 32 so that it can be ANDed with the high
> portion of the MSR, it becomes
>
> 0x000000000001ffff
>
> i.e., bit 16 is set too.
>
> And that max address can be up to 51:
>
> "Range Physical Base-Address (PhysBase)—Bits 51:12. The memory-range base-address in
> physical-address space. PhysBase is aligned on a 4-Kbyte (or greater) address in the 52-bit
> physical-address space supported by the AMD64 architecture. PhysBase represents the most-
> significant 40-address bits of the physical address. Physical-address
> bits 11:0 are assumed to be 0."
>
> Long story short, size_and_mask needs to be done from
>
> size_and_mask = ~size_or_mask & 0xfffff00000ULL;
>
> to
>
> size_and_mask = ~size_or_mask & GENMASK_ULL(phys_addr, 20);
This must be phys_addr - 1, as we start to count with bit 0.
>
> size_or_mask bits already took into consideration the phys_addr:
>
> size_or_mask = SIZE_OR_MASK_BITS(phys_addr);
>
> and size_and_mask needs to do it too.
>
> Right?
>
> I'll test this on the boxes here everywhere. I guess this gets hit only
> on boxes where the phys_addr of the variable MTRRs goes over the 16
> bits.
>
> As to this patch: please make all the bit and mask definitions you're
> adding as close to the MTRR bit and mask definition names in the
> APM+SDM. I've started this already (ontop of yours):
Okay.
Juergen
[-- Attachment #1.1.2: OpenPGP public key --]
[-- Type: application/pgp-keys, Size: 3149 bytes --]
[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 495 bytes --]
next prev parent reply other threads:[~2023-04-05 7:56 UTC|newest]
Thread overview: 53+ messages / expand[flat|nested] mbox.gz Atom feed top
2023-04-01 6:36 [PATCH v5 00/15] x86/mtrr: fix handling with PAT but without MTRR Juergen Gross
2023-04-01 6:36 ` [PATCH v5 01/15] x86/mtrr: split off physical address size calculation Juergen Gross
2023-04-01 6:36 ` [PATCH v5 02/15] x86/mtrr: optimize mtrr_calc_physbits() Juergen Gross
2023-04-01 6:36 ` [PATCH v5 03/15] x86/mtrr: replace some constants with defines Juergen Gross
2023-04-03 16:03 ` Borislav Petkov
2023-04-05 7:55 ` Juergen Gross [this message]
2023-04-05 20:26 ` Borislav Petkov
2023-04-11 13:32 ` Juergen Gross
2023-04-01 6:36 ` [PATCH v5 04/15] x86/mtrr: support setting MTRR state for software defined MTRRs Juergen Gross
2023-04-03 2:25 ` Huang, Kai
2023-04-03 7:10 ` Juergen Gross
2023-04-03 9:27 ` Huang, Kai
2023-04-03 9:35 ` Juergen Gross
2023-04-03 9:44 ` Huang, Kai
2023-04-03 9:43 ` Huang, Kai
2023-04-11 13:20 ` Borislav Petkov
2023-04-11 13:31 ` Juergen Gross
2023-04-11 17:14 ` Borislav Petkov
2023-04-11 13:59 ` Michael Kelley (LINUX)
2023-04-11 14:04 ` Juergen Gross
2023-04-11 14:26 ` Borislav Petkov
2023-04-11 15:57 ` Juergen Gross
2023-04-11 17:15 ` Borislav Petkov
2023-04-12 8:30 ` Juergen Gross
2023-04-11 14:25 ` Borislav Petkov
2023-04-01 6:36 ` [PATCH v5 05/15] x86/hyperv: set MTRR state when running as SEV-SNP Hyper-V guest Juergen Gross
2023-04-02 2:36 ` Michael Kelley (LINUX)
2023-04-01 6:36 ` [PATCH v5 06/15] x86/xen: set MTRR state when running as Xen PV initial domain Juergen Gross
2023-04-01 6:36 ` [PATCH v5 07/15] x86/mtrr: replace vendor tests in MTRR code Juergen Gross
2023-04-12 8:45 ` Borislav Petkov
2023-04-12 10:25 ` Juergen Gross
2023-04-01 6:36 ` [PATCH v5 08/15] x86/mtrr: have only one set_mtrr() variant Juergen Gross
2023-04-12 12:30 ` Borislav Petkov
2023-04-12 12:56 ` Juergen Gross
2023-04-12 20:09 ` Borislav Petkov
2023-04-01 6:36 ` [PATCH v5 09/15] x86/mtrr: allocate mtrr_value array dynamically Juergen Gross
2023-04-12 21:11 ` Borislav Petkov
2023-04-13 10:07 ` Juergen Gross
2023-04-01 6:36 ` [PATCH v5 10/15] x86/mtrr: add get_effective_type() service function Juergen Gross
2023-04-01 6:36 ` [PATCH v5 11/15] x86/mtrr: construct a memory map with cache modes Juergen Gross
2023-04-20 12:15 ` Borislav Petkov
2023-04-20 12:30 ` Juergen Gross
2023-04-20 12:30 ` Juergen Gross
2023-04-20 13:01 ` Borislav Petkov
2023-04-20 13:57 ` Juergen Gross
2023-04-20 14:54 ` Borislav Petkov
2023-04-20 15:10 ` Juergen Gross
2023-04-21 11:23 ` Borislav Petkov
2023-04-21 14:35 ` Juergen Gross
2023-04-01 6:36 ` [PATCH v5 12/15] x86/mtrr: use new cache_map in mtrr_type_lookup() Juergen Gross
2023-04-01 6:36 ` [PATCH v5 13/15] x86/mtrr: don't let mtrr_type_lookup() return MTRR_TYPE_INVALID Juergen Gross
2023-04-01 6:36 ` [PATCH v5 14/15] x86/mm: only check uniform after calling mtrr_type_lookup() Juergen Gross
2023-04-01 6:36 ` [PATCH v5 15/15] x86/mtrr: remove unused code Juergen Gross
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=1231926a-5d03-5631-a376-b0d738c7e25c@suse.com \
--to=jgross@suse.com \
--cc=bp@alien8.de \
--cc=dave.hansen@linux.intel.com \
--cc=hpa@zytor.com \
--cc=linux-kernel@vger.kernel.org \
--cc=mingo@redhat.com \
--cc=tglx@linutronix.de \
--cc=x86@kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox
all inboxes | Powered by JetHome®