linux-kernel - Re: [patch 57/58] x86/apic: Provide static call infrastructure for APIC callbacks

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: <20230718091950.GA3472843@hirez.programming.kicks-ass.net>
Date:   Tue, 18 Jul 2023 11:19:50 +0200
From:   Peter Zijlstra <peterz@...radead.org>
To:     Thomas Gleixner <tglx@...utronix.de>
Cc:     LKML <linux-kernel@...r.kernel.org>, x86@...nel.org,
        Linus Torvalds <torvalds@...ux-foundation.org>,
        Andrew Cooper <andrew.cooper3@...rix.com>,
        Tom Lendacky <thomas.lendacky@....com>,
        Paolo Bonzini <pbonzini@...hat.com>,
        Wei Liu <wei.liu@...nel.org>,
        Arjan van de Ven <arjan@...ux.intel.com>,
        Juergen Gross <jgross@...e.com>
Subject: Re: [patch 57/58] x86/apic: Provide static call infrastructure for
 APIC callbacks

On Tue, Jul 18, 2023 at 01:16:03AM +0200, Thomas Gleixner wrote:
> Declare and define the static calls for the hotpath APIC callbacks. Note
> this deliberately uses STATIC_CALL_NULL() because otherwise it would be
> required to have the definitions in the 32bit and the 64bit default APIC
> implementations and it's hard to keep the calls in sync. The other option
> would be to have stub functions for each callback type. Not pretty either
> 
> So the NULL capable calls are used and filled in during early boot after
> the static key infrastructure has been initialized. The calls will be
> static_call() except for the wait_irc_idle() callback which is valid to be
> NULL for X2APIC systems.
> 
> Update the calls when a new APIC driver is installed and when a callback
> override is invoked.
> 
> Export the trampolines for the two calls which are used in KVM and MCE
> error inject modules.
> 
> Test the setup and let the next step convert the inline wrappers to make it
> effective.

Perhaps preserve some of that in a comment?


> --- a/arch/x86/kernel/apic/init.c
> +++ b/arch/x86/kernel/apic/init.c
> @@ -5,6 +5,28 @@
>  
>  #include "local.h"
>  

/*
 * Uses DEFINE_STATIC_CALL_NULL() to avoid having to provide a (stub)
 * function at this time.
 *
 * All except wait_icr_idle *MUST* be set before usage.
 */

> +#define DEFINE_APIC_CALL(__cb)						\
> +	DEFINE_STATIC_CALL_NULL(apic_call_##__cb, *apic->__cb)
> +
> +DEFINE_APIC_CALL(eoi);
> +DEFINE_APIC_CALL(native_eoi);
> +DEFINE_APIC_CALL(icr_read);
> +DEFINE_APIC_CALL(icr_write);
> +DEFINE_APIC_CALL(read);
> +DEFINE_APIC_CALL(send_IPI);
> +DEFINE_APIC_CALL(send_IPI_mask);
> +DEFINE_APIC_CALL(send_IPI_mask_allbutself);
> +DEFINE_APIC_CALL(send_IPI_allbutself);
> +DEFINE_APIC_CALL(send_IPI_all);
> +DEFINE_APIC_CALL(send_IPI_self);
> +DEFINE_APIC_CALL(wait_icr_idle);
> +DEFINE_APIC_CALL(wakeup_secondary_cpu);
> +DEFINE_APIC_CALL(wakeup_secondary_cpu_64);
> +DEFINE_APIC_CALL(write);
> +
> +EXPORT_STATIC_CALL_TRAMP_GPL(apic_call_send_IPI_mask);
> +EXPORT_STATIC_CALL_TRAMP_GPL(apic_call_send_IPI_self);