From:   Thomas Gleixner <tglx@linutronix.de>
To:     Usama Arif <usama.arif@bytedance.com>, dwmw2@infradead.org,
        kim.phillips@amd.com, brgerst@gmail.com
Cc:     piotrgorski@cachyos.org, oleksandr@natalenko.name,
        arjan@linux.intel.com, mingo@redhat.com, bp@alien8.de,
        dave.hansen@linux.intel.com, hpa@zytor.com, x86@kernel.org,
        pbonzini@redhat.com, paulmck@kernel.org,
        linux-kernel@vger.kernel.org, kvm@vger.kernel.org,
        rcu@vger.kernel.org, mimoja@mimoja.de, hewenliang4@huawei.com,
        thomas.lendacky@amd.com, seanjc@google.com, pmenzel@molgen.mpg.de,
        fam.zheng@bytedance.com, punit.agrawal@bytedance.com,
        simon.evans@bytedance.com, liangma@liangbit.com,
        David Woodhouse <dwmw@amazon.co.uk>,
        Usama Arif <usama.arif@bytedance.com>
Subject: Re: [PATCH v12 07/11] x86/smpboot: Remove early_gdt_descr on 64-bit
In-Reply-To: <20230226110802.103134-8-usama.arif@bytedance.com>
References: <20230226110802.103134-1-usama.arif@bytedance.com>
 <20230226110802.103134-8-usama.arif@bytedance.com>
Date:   Tue, 28 Feb 2023 22:01:05 +0100
Message-ID: <878rghmrn2.ffs@tglx>
MIME-Version: 1.0
Content-Type: text/plain
Precedence: bulk

On Sun, Feb 26 2023 at 11:07, Usama Arif wrote:
> @@ -265,7 +265,12 @@ SYM_INNER_LABEL(secondary_startup_64_no_verify, SYM_L_GLOBAL)
>  	 * addresses where we're currently running on. We have to do that here
>  	 * because in 32bit we couldn't load a 64bit linear address.
>  	 */
> -	lgdt	early_gdt_descr(%rip)
> +	subq	$16, %rsp
> +	movw	$(GDT_SIZE-1), (%rsp)
> +	leaq	gdt_page(%rdx), %rax

Even on !SMP gdt_page is in the 0...__per_cpu_end range. Which means
that on !SMP this results in:

      leaq    0xb000(%rdx),%rax

and RDX is 0. That's not really a valid GDT pointer, right?

> +	movq	%rax, 2(%rsp)
> +	lgdt	(%rsp)

and obviously that's equally broken for the task stack part:

>	movq	pcpu_hot + X86_current_task(%rdx), %rax

This needs:

--- a/arch/x86/kernel/head_64.S
+++ b/arch/x86/kernel/head_64.S
@@ -239,7 +239,7 @@ SYM_INNER_LABEL(secondary_startup_64_no_
 	/* Get the per cpu offset for the given CPU# which is in ECX */
 	movq	__per_cpu_offset(,%rcx,8), %rdx
 #else
-	xorl	%edx, %edx
+	leaq	INIT_PER_CPU_VAR(fixed_percpu_data)(%rip), %rdx
 #endif /* CONFIG_SMP */
 
 	/*

in the initial_stack patch, which then allows to remove this hunk in the
initial_gs patch:

@@ -286,9 +286,6 @@ SYM_INNER_LABEL(secondary_startup_64_no_
 	 * the per cpu areas are set up.
 	 */
 	movl	$MSR_GS_BASE,%ecx
-#ifndef CONFIG_SMP
-	leaq	INIT_PER_CPU_VAR(fixed_percpu_data)(%rip), %rdx
-#endif
 	movl	%edx, %eax
 	shrq	$32, %rdx
 	wrmsr

Maybe we should enforce CONFIG_SMP=y first :)

Thanks,

        tglx