linux-kernel - Re: [PATCH v8 01/13] x86/apic: Construct a selector for the interrupt delivery mode

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:   Thu, 7 Sep 2017 12:19:31 +0800
From:   Dou Liyang <douly.fnst@...fujitsu.com>
To:     Baoquan He <bhe@...hat.com>
CC:     <x86@...nel.org>, <linux-kernel@...r.kernel.org>,
        <tglx@...utronix.de>, <mingo@...nel.org>, <hpa@...or.com>,
        <rjw@...ysocki.net>, <bp@...en8.de>, <indou.takao@...fujitsu.com>,
        <izumi.taku@...fujitsu.com>
Subject: Re: [PATCH v8 01/13] x86/apic: Construct a selector for the interrupt
 delivery mode

Hi Baoquan

I am wordy one ah:
our target is checking if BIOS supports APIC, no matter what
type(separated/integrated) it is. if not, go to PIC mode.

Let‘s discuss the original logic and the smp_found_config,
then take about your code.

The existing logic is:

	if (!boot_cpu_has(X86_FEATURE_APIC) && !smp_found_config) ...(1)
		return -1;

	if (!boot_cpu_has(X86_FEATURE_APIC) &&
	                APIC_INTEGRATED(boot_cpu_apic_version)) { ...(2)
		pr_err(....);

why smp_found_config has to be checked in (1)?

Because, In case of discrete (pretty old) apics we may not set
X86_FEATURE_APIC bit in cpuid, with 82489DX we can't rely on apic
feature bit retrieved via cpuid(boot_cpu_has(X86_FEATURE_APIC)).[1]
So we assume that if SMP configuration is found from MP table
(smp_found_config = 1) in above case, there maybe a separated
chip in our pc.

After passing the check of (1), we in (2), check whether local APIC
is detected or not, If we have a BIOS bug.

[1] Commit 8312136fa8b0("x86, apic: Fix missed handling of discrete apics")

At 09/06/2017 06:17 PM, Baoquan He wrote:
> Hi Dou,
>
> On 08/28/17 at 11:20am, Dou Liyang wrote:
>> +static int __init apic_intr_mode_select(void)
>> +{
>> +	/* Check kernel option */
>> +	if (disable_apic) {
>> +		pr_info("APIC disabled via kernel command line\n");
>> +		return APIC_PIC;
>> +	}
>> +
>
> I am not very familiar with cpu registers, not sure if we can adjust
> below code flow as:
>
> 	/* If APIC is integrated, check local APIC only */
> 	if (lapic_is_integrated() && !boot_cpu_has(X86_FEATURE_APIC)) {
> 		disable_apic = 1;
> 		pr_info("APIC disabled by BIOS\n");
> 		return APIC_PIC;
> 	}
>
> 	/* If APIC is on a separate chip, check if smp_found_config is found*/
> 	if (!lapic_is_integrated() && !smp_found_config) {
> 		disable_apic = 1;
> 		return APIC_PIC;
> 	}

Yes, Awesome, we first consider it from APIC register space, then
the BOIS and software configration. let me do more investigation.

I rewrite it based on you, any comments will welcome.

	/* If APIC is not integrated, check if SMP configuration is
	 * found from MP table. If not too, no 82489DX. switch to
	 * PIC mode
	 *
	 * Else APIC is integrated, check if the BIOS allows local APIC
	 *
	 */
	if (!lapic_is_integrated()) {
		if (!smp_found_config) {
			disable_apic = 1;
			return APIC_PIC;
		}
	} else if(!boot_cpu_has(X86_FEATURE_APIC)) {
			disable_apic = 1;
			pr_info("APIC disabled by BIOS\n");
			return APIC_PIC;
		}
	}

BTW, As the macro APIC_INTEGRATED(x) has already wrapped by
CONFIG_X86_32, I will cleanup the lapic_is_integrated() for readablity
like that:

diff --git a/arch/x86/kernel/apic/apic.c b/arch/x86/kernel/apic/apic.c
index 7834f73..63b3ae9 100644
--- a/arch/x86/kernel/apic/apic.c
+++ b/arch/x86/kernel/apic/apic.c
@@ -211,11 +211,7 @@ static inline int lapic_get_version(void)
   */
  static inline int lapic_is_integrated(void)
  {
-#ifdef CONFIG_X86_64
-       return 1;
-#else
         return APIC_INTEGRATED(lapic_get_version());
-#endif
  }


Do you think so. ;-)


Thanks,
	dou.


> 	~~~~ Now, I haven't think of why smp_found_config has to be
>              checked here.
>
> In this way, we don't need the CONFIG_X86_64 checking since it's
> contained in lapic_is_integrated() already. And the checking is obvious
> for understanding. Just not very sure if the checking is adequate.
>
> Just my personal opinion.
>
>> +	/* Check BIOS */
>> +#ifdef CONFIG_X86_64
>> +	/* On 64-bit, the APIC must be integrated, Check local APIC only */
>> +	if (!boot_cpu_has(X86_FEATURE_APIC)) {
>> +		disable_apic = 1;
>> +		pr_info("APIC disabled by BIOS\n");
>> +		return APIC_PIC;
>> +	}
>> +#else
>> +	/*
>> +	 * On 32-bit, check whether there is a separate chip or integrated
>> +	 * APIC
>> +	 */
>> +
>> +	/* Has a separate chip ? */
>> +	if (!boot_cpu_has(X86_FEATURE_APIC) && !smp_found_config) {
>> +		disable_apic = 1;
>> +
>> +		return APIC_PIC;
>> +	}
>> +
>> +	/* Has a local APIC ? */
>> +	if (!boot_cpu_has(X86_FEATURE_APIC) &&
>> +		APIC_INTEGRATED(boot_cpu_apic_version)) {
>> +		disable_apic = 1;
>> +		pr_err(FW_BUG "Local APIC %d not detected, force emulation\n",
>> +				       boot_cpu_physical_apicid);
>> +
>> +		return APIC_PIC;
>> +	}
>> +#endif
>> +
>> +	/* Check MP table or ACPI MADT configuration */
>> +	if (!smp_found_config) {
>> +		disable_ioapic_support();
>> +
>> +		if (!acpi_lapic)
>> +			pr_info("APIC: ACPI MADT or MP tables are not detected\n");
>> +
>> +		return APIC_VIRTUAL_WIRE;
>> +	}
>> +
>> +	return APIC_SYMMETRIC_IO;
>> +}
>> +
>>  /*
>>   * An initial setup of the virtual wire mode.
>>   */
>> --
>> 2.5.5
>>
>>
>>
>
>
>