linux-kernel - Re: [PATCH] Fix resume on x86-32 machines

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]

Date:   Sun, 10 Dec 2017 13:58:23 -0800
From:   Andy Lutomirski <luto@...capital.net>
To:     Pavel Machek <pavel@....cz>
Cc:     Linus Torvalds <torvalds@...ux-foundation.org>,
        Zhang Rui <rui.zhang@...el.com>,
        Andrew Lutomirski <luto@...nel.org>,
        Thomas Gleixner <tglx@...utronix.de>,
        Jarkko Nikula <jarkko.nikula@...ux.intel.com>,
        "Rafael J. Wysocki" <rjw@...ysocki.net>,
        Linux Kernel Mailing List <linux-kernel@...r.kernel.org>,
        the arch/x86 maintainers <x86@...nel.org>
Subject: Re: [PATCH] Fix resume on x86-32 machines



> On Dec 10, 2017, at 1:38 PM, Pavel Machek <pavel@....cz> wrote:
> 
> 
> After 4.15-rc2, suspend stopped working on Thinkpad X60. 5b06bbc
> (unintentionally?) reordered stuff with respect to
> fix_processor_context() on 32-bit and 64-bit. We undo that change on
> 32-bit.
> 

Can you explain what was wrong with the reordering?  Your patch certainly *looks* incorrect.

I'm guessing that the real issue is that 32-bit needs %fs restored early for TLS.

> While we are at it, fix a comment.
> 
> Signed-off-by: Pavel Machek <pavel@....cz>
> Fixes: 5b06bbcfc2c621da3009da8decb7511500c293ed
> 
> diff --git a/arch/x86/power/cpu.c b/arch/x86/power/cpu.c
> index 5191de1..af7b613 100644
> --- a/arch/x86/power/cpu.c
> +++ b/arch/x86/power/cpu.c
> @@ -216,8 +216,8 @@ static void notrace __restore_processor_state(struct saved_context *ctxt)
>    write_cr0(ctxt->cr0);
> 
>    /*
> -     * now restore the descriptor tables to their proper values
> -     * ltr is done i fix_processor_context().
> +     * Now restore the descriptor tables to their proper values
> +     * ltr is done in fix_processor_context().
>     */
> #ifdef CONFIG_X86_32
>    load_idt(&ctxt->idt);
> @@ -235,13 +235,11 @@ static void notrace __restore_processor_state(struct saved_context *ctxt)
>    wrmsrl(MSR_GS_BASE, ctxt->gs_base);
> #endif
> 
> -    fix_processor_context();
> -
> +#ifdef CONFIG_X86_32
>    /*
> -     * Restore segment registers.  This happens after restoring the GDT
> -     * and LDT, which happen in fix_processor_context().
> +     * Restore segment registers.
>     */
> -#ifdef CONFIG_X86_32
> +
>    loadsegment(es, ctxt->es);
>    loadsegment(fs, ctxt->fs);
>    loadsegment(gs, ctxt->gs);
> @@ -252,8 +250,17 @@ static void notrace __restore_processor_state(struct saved_context *ctxt)
>     */
>    if (boot_cpu_has(X86_FEATURE_SEP))
>        enable_sep_cpu();
> +
> +    fix_processor_context();
> #else
> /* CONFIG_X86_64 */
> +    /*
> +     * Restore segment registers.  This happens after restoring the GDT
> +     * and LDT, which happen in fix_processor_context().
> +     */
> +    
> +    fix_processor_context();
> +
>    asm volatile ("movw %0, %%ds" :: "r" (ctxt->ds));
>    asm volatile ("movw %0, %%es" :: "r" (ctxt->es));
>    asm volatile ("movw %0, %%fs" :: "r" (ctxt->fs));
> 
> 
> -- 
> (english) http://www.livejournal.com/~pavelmachek
> (cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html