linux-kernel - Re: [PATCH v7 14/22] x86/decompressor: Merge trampoline cleanup with switching code

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: <CAMj1kXF+=W_oe7MS87rWdJ_z555iz2A8iR42fJ1mwZ+bgCgjmw@mail.gmail.com>
Date:   Tue, 1 Aug 2023 14:11:21 +0200
From:   Ard Biesheuvel <ardb@...nel.org>
To:     Borislav Petkov <bp@...en8.de>
Cc:     linux-efi@...r.kernel.org, linux-kernel@...r.kernel.org,
        Evgeniy Baskov <baskov@...ras.ru>,
        Andy Lutomirski <luto@...nel.org>,
        Dave Hansen <dave.hansen@...ux.intel.com>,
        Ingo Molnar <mingo@...hat.com>,
        Peter Zijlstra <peterz@...radead.org>,
        Thomas Gleixner <tglx@...utronix.de>,
        Alexey Khoroshilov <khoroshilov@...ras.ru>,
        Peter Jones <pjones@...hat.com>,
        Gerd Hoffmann <kraxel@...hat.com>,
        Dave Young <dyoung@...hat.com>,
        Mario Limonciello <mario.limonciello@....com>,
        Kees Cook <keescook@...omium.org>,
        Tom Lendacky <thomas.lendacky@....com>,
        "Kirill A . Shutemov" <kirill.shutemov@...ux.intel.com>,
        Linus Torvalds <torvalds@...ux-foundation.org>,
        Joerg Roedel <jroedel@...e.de>
Subject: Re: [PATCH v7 14/22] x86/decompressor: Merge trampoline cleanup with
 switching code

On Tue, 1 Aug 2023 at 14:09, Borislav Petkov <bp@...en8.de> wrote:
>
> On Fri, Jul 28, 2023 at 11:09:08AM +0200, Ard Biesheuvel wrote:
> > diff --git a/arch/x86/boot/compressed/pgtable_64.c b/arch/x86/boot/compressed/pgtable_64.c
> > index 4016444e6788304f..4f50af23a0854f18 100644
> > --- a/arch/x86/boot/compressed/pgtable_64.c
> > +++ b/arch/x86/boot/compressed/pgtable_64.c
> > @@ -101,9 +101,10 @@ static unsigned long find_trampoline_placement(void)
> >       return bios_start - TRAMPOLINE_32BIT_SIZE;
> >  }
> >
> > -asmlinkage void set_paging_levels(void *rmode)
> > +asmlinkage void set_paging_levels(void *rmode, void *pgtable)
>
> Please get rid of this silly rmode arg which gets passed in here as
> boot_params and then gets assigned to an extern pointer to boot_params.
> This is just silly. Just do what the other functions get from %r15 now
> - a struct boot_params *bp arg.
>
> But perhaps in a separate patch.
>

OK

> >  {
> >       void (*toggle_la57)(void *trampoline, bool enable_5lvl);
> > +     void *trampoline_pgtable;
> >       bool l5_required = false;
> >
> >       /* Initialize boot_params. Required for cmdline_find_option_bool(). */
> > @@ -133,7 +134,7 @@ asmlinkage void set_paging_levels(void *rmode)
> >        * the desired one.
> >        */
> >       if (l5_required == !!(native_read_cr4() & X86_CR4_LA57))
> > -             return;
> > +             goto out;
> >
> >       trampoline_32bit = (unsigned long *)find_trampoline_placement();
> >
> > @@ -163,6 +164,8 @@ asmlinkage void set_paging_levels(void *rmode)
> >        * The new page table will be used by trampoline code for switching
> >        * from 4- to 5-level paging or vice versa.
> >        */
> > +     trampoline_pgtable = trampoline_32bit +
> > +                          TRAMPOLINE_32BIT_PGTABLE_OFFSET / sizeof(unsigned long);
> >
> >       if (l5_required) {
> >               /*
> > @@ -182,31 +185,21 @@ asmlinkage void set_paging_levels(void *rmode)
> >                * may be above 4G.
> >                */
> >               src = *(unsigned long *)__native_read_cr3() & PAGE_MASK;
> > -             memcpy(trampoline_32bit + TRAMPOLINE_32BIT_PGTABLE_OFFSET / sizeof(unsigned long),
> > -                    (void *)src, PAGE_SIZE);
> > +             memcpy(trampoline_pgtable, (void *)src, PAGE_SIZE);
> >       }
> >
> >       toggle_la57(trampoline_32bit, l5_required);
> > -}
> > -
> > -void cleanup_trampoline(void *pgtable)
> > -{
> > -     void *trampoline_pgtable;
> > -
> > -     trampoline_pgtable = trampoline_32bit + TRAMPOLINE_32BIT_PGTABLE_OFFSET / sizeof(unsigned long);
> >
> >       /*
> > -      * Move the top level page table out of trampoline memory,
> > -      * if it's there.
> > +      * Move the top level page table out of trampoline memory.
> >        */
> > -     if ((void *)__native_read_cr3() == trampoline_pgtable) {
> > -             memcpy(pgtable, trampoline_pgtable, PAGE_SIZE);
> > -             native_write_cr3((unsigned long)pgtable);
> > -     }
> > +     memcpy(pgtable, trampoline_pgtable, PAGE_SIZE);
> > +     native_write_cr3((unsigned long)pgtable);
> >
> >       /* Restore trampoline memory */
> >       memcpy(trampoline_32bit, trampoline_save, TRAMPOLINE_32BIT_SIZE);
> >
> > +out:
> >       /* Initialize variables for 5-level paging */
> >  #ifdef CONFIG_X86_5LEVEL
> >       if (__read_cr4() & X86_CR4_LA57) {
> > @@ -215,4 +208,5 @@ void cleanup_trampoline(void *pgtable)
> >               ptrs_per_p4d = 512;
> >       }
> >  #endif
> > +     return;
>
> No need for that one. It'll return without it. :-)
>

Removing it breaks the build for !CONFIG_X86_5LEVEL

However, I can move these assignments into the conditional at the top
of the function (the one that sets l5_required) so we don't need the
label at all. Will fix for v8