linux-kernel - Re: Early boot regression from f0551af0213 ("x86/topology: Ignore non-present APIC IDs in a present package")

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: <87wmntkhak.ffs@tglx>
Date: Thu, 16 May 2024 15:38:59 +0200
From: Thomas Gleixner <tglx@...utronix.de>
To: Lyude Paul <lyude@...hat.com>, "Linux regression tracking
 (Thorsten Leemhuis)" <regressions@...mhuis.info>
Cc: x86@...nel.org, linux-kernel@...r.kernel.org, Mario Limonciello
 <mario.limonciello@....com>, Borislav Petkov <bp@...en8.de>, Linux kernel
 regressions list <regressions@...ts.linux.dev>
Subject: Re: Early boot regression from f0551af0213 ("x86/topology: Ignore
 non-present APIC IDs in a present package")

Lyude!

On Wed, May 15 2024 at 19:15, Lyude Paul wrote:
> On Tue, 2024-05-14 at 10:25 +0200, Thomas Gleixner wrote:
>> 
>> Which one of the debug patches did you use?
>
> The one you sent on 4/18, when you also asked me for the output of
> /sys/kernel/debug/x86/topo/
>
> (I thought I remembered that patch not booting on previous kernels but maybe
> I'm misremembering)

It booted with the -rc kernel too. :)

I found an interesting and probably related difference in the boot logs
though. Up to 8 possible CPUs the kernel uses logical destination mode
for the APIC. With more than 8 it uses physical destination mode.

Can you please test the patch below on top of 6.9 and validate that it
boots w/o any magic command line parameter?

Thanks,

        tglx
---
diff --git a/arch/x86/kernel/apic/apic.c b/arch/x86/kernel/apic/apic.c
index 66fd4b2a37a3..db5e93a7c194 100644
--- a/arch/x86/kernel/apic/apic.c
+++ b/arch/x86/kernel/apic/apic.c
@@ -1883,6 +1883,8 @@ static inline void try_to_enable_x2apic(int remap_mode) { }
 static inline void __x2apic_enable(void) { }
 #endif /* !CONFIG_X86_X2APIC */
 
+int irq_remap_mode = -1;
+
 void __init enable_IR_x2apic(void)
 {
 	unsigned long flags;
@@ -1915,6 +1917,8 @@ void __init enable_IR_x2apic(void)
 
 	if (ir_stat < 0)
 		restore_ioapic_entries();
+	else
+		irq_remap_mode = ir_stat;
 	legacy_pic->restore_mask();
 	local_irq_restore(flags);
 }
diff --git a/arch/x86/kernel/apic/apic_flat_64.c b/arch/x86/kernel/apic/apic_flat_64.c
index f37ad3392fec..5cc386db3557 100644
--- a/arch/x86/kernel/apic/apic_flat_64.c
+++ b/arch/x86/kernel/apic/apic_flat_64.c
@@ -12,6 +12,7 @@
 #include <linux/export.h>
 #include <linux/acpi.h>
 
+#include <asm/irq_remapping.h>
 #include <asm/jailhouse_para.h>
 #include <asm/apic.h>
 
@@ -130,7 +131,8 @@ static int physflat_acpi_madt_oem_check(char *oem_id, char *oem_table_id)
 
 static int physflat_probe(void)
 {
-	return apic == &apic_physflat || num_possible_cpus() > 8 || jailhouse_paravirt();
+	return apic == &apic_physflat || irq_remap_mode == IRQ_REMAP_XAPIC_MODE ||
+		num_possible_cpus() > 8 || jailhouse_paravirt();
 }
 
 static struct apic apic_physflat __ro_after_init = {
diff --git a/arch/x86/kernel/apic/local.h b/arch/x86/kernel/apic/local.h
index 842fe28496be..f633ab6dfa9f 100644
--- a/arch/x86/kernel/apic/local.h
+++ b/arch/x86/kernel/apic/local.h
@@ -22,6 +22,8 @@ void x2apic_send_IPI_allbutself(int vector);
 void x2apic_send_IPI_self(int vector);
 extern u32 x2apic_max_apicid;
 
+extern int irq_remap_mode;
+
 /* IPI */
 
 DECLARE_STATIC_KEY_FALSE(apic_use_ipi_shorthand);