Re: [PATCH 3/9] arm/gic: Get rid of READ/WRITE_SYSREG32

From: Julien Grall <julien@xen.org>
To: Michal Orzel <michal.orzel@arm.com>, xen-devel@lists.xenproject.org
Cc: Stefano Stabellini <sstabellini@kernel.org>,
	Volodymyr Babchuk <Volodymyr_Babchuk@epam.com>,
	bertrand.marquis@arm.com
Subject: Re: [PATCH 3/9] arm/gic: Get rid of READ/WRITE_SYSREG32
Date: Wed, 21 Apr 2021 11:21:17 +0100	[thread overview]
Message-ID: <7cac2363-6263-6b64-2f7e-c3f7cd588d15@xen.org> (raw)
In-Reply-To: <fd5924c5-685d-a9fc-1116-c1b8909c9aea@arm.com>

Hi Michal,

On 21/04/2021 08:48, Michal Orzel wrote:
> On 20.04.2021 15:28, Julien Grall wrote:
>> On 20/04/2021 08:08, Michal Orzel wrote:
>>> Signed-off-by: Michal Orzel <michal.orzel@arm.com>
>>>      static unsigned int gicv3_read_irq(void)
>>>    {
>>> -    unsigned int irq = READ_SYSREG32(ICC_IAR1_EL1);
>>> +    register_t irq = READ_SYSREG(ICC_IAR1_EL1);
>>>          dsb(sy);
>>>    -    return irq;
>>> +    /* Number of IRQs do not exceed 32bit. */
>>
>> If we want to be pedantic, the IRQs are encoded using 23-bit. So maybe we want to mask them below.
>>
>>> +    return (unsigned int)irq;
>>
>> NIT: We tend to avoid explicit cast unless they are strictly necessary because they can be more harmful than implicit cast (the compiler may not cast every conversion). So I would drop it and just keep the comment.
>>
> Ok. I will do:
> return (irq & 0xffffff);

Sounds good, so long we the value is not hardoced but provided through a 
define :).

Cheers,

-- 
Julien Grall