Re: [PATCH v2 1/6] x86/mm/pat: Change PAT to support non-default PAT MSR

From: Borislav Petkov
Date: Tue Mar 22 2016 - 12:57:43 EST


$Subject is misleading - there's no non-default PAT MSR - the setting is
non-default.

On Wed, Mar 16, 2016 at 06:44:57PM -0600, Toshi Kani wrote:
> In preparation to fix a regression caused by 'commit 9cd25aac1f44
> ("x86/mm/pat: Emulate PAT when it is disabled")', PAT needs to
> support a case that PAT MSR is initialized with a non-default
> value.
>
> When pat_init() is called in PAT disable state, it initializes

is called and PAT is disabled

> PAT table with the BIOS default value. Xen, however, sets PAT MSR
> with a non-default value to enable WC. This causes inconsistency
> between PAT table and PAT MSR when PAT is set to disable on Xen.
>
> Change pat_init() to handle the PAT disable cases properly. Add
> pat_keep_handoff_state() to handle two cases when PAT is set to
> disable.
> 1. CPU supports PAT: Set PAT table to be consistent with PAT MSR.
> 2. CPU does not support PAT: Set PAT table to be consistent with
> PWT and PCD bits in a PTE.
>
> Signed-off-by: Toshi Kani <toshi.kani@xxxxxxx>
> Cc: Borislav Petkov <bp@xxxxxxx>
> Cc: Luis R. Rodriguez <mcgrof@xxxxxxxx>
> Cc: Juergen Gross <jgross@xxxxxxxx>
> Cc: Ingo Molnar <mingo@xxxxxxxxxx>
> Cc: H. Peter Anvin <hpa@xxxxxxxxx>
> Cc: Thomas Gleixner <tglx@xxxxxxxxxxxxx>
> ---
> arch/x86/mm/pat.c | 80 +++++++++++++++++++++++++++++++++++++++++------------
> 1 file changed, 62 insertions(+), 18 deletions(-)
>
> diff --git a/arch/x86/mm/pat.c b/arch/x86/mm/pat.c
> index 04e2e71..e0a34b0 100644
> --- a/arch/x86/mm/pat.c
> +++ b/arch/x86/mm/pat.c
> @@ -207,9 +207,6 @@ static void pat_bsp_init(u64 pat)
> return;
> }
>
> - if (!pat_enabled())
> - goto done;
> -
> rdmsrl(MSR_IA32_CR_PAT, tmp_pat);
> if (!tmp_pat) {
> pat_disable("PAT MSR is 0, disabled.");
> @@ -218,15 +215,11 @@ static void pat_bsp_init(u64 pat)
>
> wrmsrl(MSR_IA32_CR_PAT, pat);
>
> -done:
> pat_init_cache_modes(pat);
> }
>
> static void pat_ap_init(u64 pat)
> {
> - if (!pat_enabled())
> - return;
> -
> if (!cpu_has_pat) {
> /*
> * If this happens we are on a secondary CPU, but switched to
> @@ -238,18 +231,43 @@ static void pat_ap_init(u64 pat)
> wrmsrl(MSR_IA32_CR_PAT, pat);
> }
>
> -void pat_init(void)
> +/**
> + * pat_keep_handoff_state - Set PAT table to the handoff state
> + *
> + * This function keeps PAT in the BIOS handoff state. When CPU supports
> + * PAT, it sets PAT table to be consistent with PAT MSR. When CPU does not
> + * support PAT, it emulates PAT by setting PAT table consistent with PWT
> + * and PCD bits in a PTE.
> + *
> + * The PAT table is global to all CPUs, which is initialized once at
> + * boot-time. Any subsequent calls to this function have no effect.
> + */
> +static void pat_keep_handoff_state(void)

Static function, no need for "pat_" prefix. Also, no need for the
kernel-doc comment.

Also, no need for all that handoff nomenclature etc, just call it
setup_pat(). Because it does exactly that - it sets up the PAT bits
unconditionally, regardless of enabled or not.

> {
> - u64 pat;
> - struct cpuinfo_x86 *c = &boot_cpu_data;
> + u64 pat = 0;
> + static int set_handoff_done;

s/set_handoff_done/pat_setup_done/

--
Regards/Gruss,
Boris.

SUSE Linux GmbH, GF: Felix ImendÃrffer, Jane Smithard, Graham Norton, HRB 21284 (AG NÃrnberg)
--