[PATCH RFC] x86/apic: skip cpu slot reservation logic on boot in case of NR_CPUS == 1

[PATCH RFC] x86/apic: skip cpu slot reservation logic on boot in case of NR_CPUS == 1

[Yury Norov]

generic_processor_info() ensures that there's always a room for boot-cpu
in phys_cpu_present_map. This logic is unneeded when there's a single CPU
in the system, and it makes smatch warning when aliasing nr_cpu_ids to
NR_CPUS in this patch:

https://www.spinics.net/lists/kernel/msg4493525.html

smatch warnings:
arch/x86/kernel/apic/apic.c:2437 generic_processor_info() warn: always true condition '(num_processors >= (1) - 1) =>
+(0-u32max >= 0)'

Reported-by: kernel test robot <lkp@intel.com>
Signed-off-by: Yury Norov <yury.norov@gmail.com>
---
This is RFC because I'm not sure that this is not a false positive from
smatch. If NR_CPUS is 1, and the condition is always true, it's simply
optimized out by a compiler, and nothing wrong happens, except maybe a
couple of checks on boot.

 arch/x86/kernel/apic/apic.c | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/arch/x86/kernel/apic/apic.c b/arch/x86/kernel/apic/apic.c
index 6d303d1d276c..6a43faaea5ac 100644
--- a/arch/x86/kernel/apic/apic.c
+++ b/arch/x86/kernel/apic/apic.c
@@ -2430,6 +2430,7 @@ int generic_processor_info(int apicid, int version)
 		return -ENODEV;
 	}
 
+#if NR_CPUS > 1
 	/*
 	 * If boot cpu has not been detected yet, then only allow upto
 	 * nr_cpu_ids - 1 processors and keep one slot free for boot cpu
@@ -2445,6 +2446,7 @@ int generic_processor_info(int apicid, int version)
 		disabled_cpus++;
 		return -ENODEV;
 	}
+#endif
 
 	if (num_processors >= nr_cpu_ids) {
 		int thiscpu = max + disabled_cpus;
-- 
2.34.1

Re: [PATCH RFC] x86/apic: skip cpu slot reservation logic on boot in case of NR_CPUS == 1

[Dave Hansen]

On 9/2/22 09:42, Yury Norov wrote:
> This is RFC because I'm not sure that this is not a false positive from
> smatch. If NR_CPUS is 1, and the condition is always true, it's simply
> optimized out by a compiler, and nothing wrong happens, except maybe a
> couple of checks on boot.

I think it's a reasonable warning, but it's also not something we need
to hack around.  We can surely only land in here with
boot_cpu_detected==true if NR_CPUS==1, so the rest of the expression is
moot.

I don't think it's worth adding the #ifdef.

Re: [PATCH RFC] x86/apic: skip cpu slot reservation logic on boot in case of NR_CPUS == 1

[Yury Norov]

On Fri, Sep 02, 2022 at 10:14:51AM -0700, Dave Hansen wrote:
> On 9/2/22 09:42, Yury Norov wrote:
> > This is RFC because I'm not sure that this is not a false positive from
> > smatch. If NR_CPUS is 1, and the condition is always true, it's simply
> > optimized out by a compiler, and nothing wrong happens, except maybe a
> > couple of checks on boot.
> 
> I think it's a reasonable warning, but it's also not something we need
> to hack around.  We can surely only land in here with
> boot_cpu_detected==true if NR_CPUS==1, so the rest of the expression is
> moot.
> 
> I don't think it's worth adding the #ifdef.

Thanks Dave. Then I'll add a comment and move my series as-is.

Thanks,
Yury