linux-kernel - Re: [PATCH v2] irqchip/qcom-pdc: add support for v3.2 HW

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 for Android: free password hash cracker in your pocket

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Wed, 23 Aug 2023 09:46:31 +0200
From:   Neil Armstrong <neil.armstrong@...aro.org>
To:     "Maulik Shah (mkshah)" <quic_mkshah@...cinc.com>,
        Andy Gross <agross@...nel.org>,
        Bjorn Andersson <andersson@...nel.org>,
        Konrad Dybcio <konrad.dybcio@...aro.org>,
        Thomas Gleixner <tglx@...utronix.de>,
        Marc Zyngier <maz@...nel.org>
Cc:     linux-arm-msm@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH v2] irqchip/qcom-pdc: add support for v3.2 HW

Hi,

On 23/08/2023 07:35, Maulik Shah (mkshah) wrote:
> Hi Neil,
> 
> @@ -142,8 +163,17 @@ static int qcom_pdc_gic_set_type(struct irq_data *d, unsigned int type)
>>           return -EINVAL;
>>       }
>> -    old_pdc_type = pdc_reg_read(IRQ_i_CFG, d->hwirq);
>> -    pdc_reg_write(IRQ_i_CFG, d->hwirq, pdc_type);
>> +    if (pdc_version < PDC_VERSION_3_2) {
>> +        old_pdc_type = pdc_reg_read(IRQ_i_CFG, d->hwirq);
>> +        pdc_reg_write(IRQ_i_CFG, d->hwirq, pdc_type);
>> +    } else {
>> +        u32 val;
>> +
>> +        val = pdc_reg_read(IRQ_i_CFG, d->hwirq);
>> +        old_pdc_type = val & IRQ_i_CFG_TYPE_MASK;
>> +        pdc_reg_write(IRQ_i_CFG, d->hwirq,
>> +                  pdc_type | (val & IRQ_i_CFG_IRQ_ENABLE));
>> +    }
> While above is correct, i don't think we need version check in qcom_pdc_gic_set_type() as bits 0-2 are always for the type in old/new version as mentioned in v1.
> 
> Adding one line after reading old_pdc_type should be good enough.

Yes I understood, but while looking at the IRQ_i_CFG bits, I wanted to keep the original
driver behavior intact by setting remaining bits to 0.

Adding this single line changes that behavior and keeps bits 3-31
to the default register value, which may have some consequences.

If you consider it's an ok change, then I'll reduce it to this single line.

> 
>    old_pdc_type = pdc_reg_read(IRQ_i_CFG, d->hwirq);
> + pdc_type |= (old_pdc_type & ~IRQ_i_CFG_TYPE_MASK);
>> +            if (pdc_version < PDC_VERSION_3_2) {
>> +                u32 irq_index, reg_index;
>> +
>> +                reg_index = (i + pdc_region[n].pin_base) >> 5;
>> +                irq_index = (i + pdc_region[n].pin_base) & 0x1f;
>> +                val = pdc_reg_read(IRQ_ENABLE_BANK, reg_index);
>> +                val &= ~BIT(irq_index);
> can use  __assign_bit(irq_index, &val, 0); here.
>> +                pdc_reg_write(IRQ_ENABLE_BANK, reg_index, val);
>> +            } else {
>> +                u32 irq;
>> +
>> +                irq = i + pdc_region[n].pin_base;
>> +                val = pdc_reg_read(IRQ_i_CFG, irq);
>> +                val &= ~IRQ_i_CFG_IRQ_ENABLE;
> 
> __assign_bit(IRQ_i_CFG_IRQ_ENABLE, &val,  0); here.

Ack I'll update those.

> 
> other than this..
> Reviewed-by: Maulik Shah <quic_mkshah@...cinc.com>
> 
> Thanks,
> Maulik
> 

Thanks,
Neil