[PATCH v3] x86/vmware: use unsigned integer for shifting

[PATCH v3] x86/vmware: use unsigned integer for shifting

[Shreenidhi Shedi]

From: Shreenidhi Shedi <sshedi@vmware.com>

Shifting signed 32-bit value by 31 bits is implementation-defined
behaviour. Using unsigned is better option for this.

Fixes: 4cca6ea04d31 ("x86/apic: Allow x2apic without IR on VMware platform")
Cc: stable@vger.kernel.org
Signed-off-by: Shreenidhi Shedi <sshedi@vmware.com>
---
 arch/x86/kernel/cpu/vmware.c | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/arch/x86/kernel/cpu/vmware.c b/arch/x86/kernel/cpu/vmware.c
index c04b933f4..02039ec35 100644
--- a/arch/x86/kernel/cpu/vmware.c
+++ b/arch/x86/kernel/cpu/vmware.c
@@ -476,8 +476,8 @@ static bool __init vmware_legacy_x2apic_available(void)
 {
 	uint32_t eax, ebx, ecx, edx;
 	VMWARE_CMD(GETVCPU_INFO, eax, ebx, ecx, edx);
-	return (eax & (1 << VMWARE_CMD_VCPU_RESERVED)) == 0 &&
-	       (eax & (1 << VMWARE_CMD_LEGACY_X2APIC)) != 0;
+	return !(eax & BIT(VMWARE_CMD_VCPU_RESERVED)) &&
+		(eax & BIT(VMWARE_CMD_LEGACY_X2APIC));
 }

 #ifdef CONFIG_AMD_MEM_ENCRYPT
--
2.36.1

Re: [PATCH v3] x86/vmware: use unsigned integer for shifting

[Peter Zijlstra]

On Fri, May 27, 2022 at 11:27:37PM +0530, Shreenidhi Shedi wrote:
> From: Shreenidhi Shedi <sshedi@vmware.com>
> 
> Shifting signed 32-bit value by 31 bits is implementation-defined
> behaviour. Using unsigned is better option for this.

The kernel builds with -fno-strict-overflow and as such this behaviour
is well defined.

> Fixes: 4cca6ea04d31 ("x86/apic: Allow x2apic without IR on VMware platform")

Nothing broken, therefore nothing fixed.

Re: [PATCH v3] x86/vmware: use unsigned integer for shifting

[Srivatsa S. Bhat]

On 5/28/22 6:52 AM, Peter Zijlstra wrote:
> On Fri, May 27, 2022 at 11:27:37PM +0530, Shreenidhi Shedi wrote:
>> From: Shreenidhi Shedi <sshedi@vmware.com>
>>
>> Shifting signed 32-bit value by 31 bits is implementation-defined
>> behaviour. Using unsigned is better option for this.
> 
> The kernel builds with -fno-strict-overflow and as such this behaviour
> is well defined.
>

Ah, I see. Thank you, Peter!
 
>> Fixes: 4cca6ea04d31 ("x86/apic: Allow x2apic without IR on VMware platform")
> 
> Nothing broken, therefore nothing fixed.
> 

Agreed.

I think using the BIT() macro still provides a nice readability
improvement. So, Shreenidhi, could you spin a new version of the patch
with the same code changes but with a different commit message about
using the BIT() macro to simplify the code, and also include a
clarification as to why the existing code is correct (which Peter
pointed out), please?

Thank you!

Regards,
Srivatsa
VMware Photon OS