linux-kernel - Re: [PATCH v1] irqchip: loongson-liointc: add hierarchy irq support

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]

Message-ID: <86ilinlbf2.wl-maz@kernel.org>
Date:   Wed, 07 Dec 2022 11:07:29 +0000
From:   Marc Zyngier <maz@...nel.org>
To:     Yinbo Zhu <zhuyinbo@...ngson.cn>
Cc:     Huacai Chen <chenhuacai@...nel.org>,
        Jiaxun Yang <jiaxun.yang@...goat.com>,
        Thomas Gleixner <tglx@...utronix.de>,
        linux-mips@...r.kernel.org, linux-kernel@...r.kernel.org
Subject: Re: [PATCH v1] irqchip: loongson-liointc: add hierarchy irq support

On Wed, 07 Dec 2022 10:50:56 +0000,
Yinbo Zhu <zhuyinbo@...ngson.cn> wrote:
> 
> 
> 在 2022/12/7 16:08, Marc Zyngier 写道:
> > On Wed, 07 Dec 2022 01:45:55 +0000,
> > Yinbo Zhu <zhuyinbo@...ngson.cn> wrote:
> >> When the irq of hierarchical interrupt chip was routed to liointc
> >> that asked liointc driver to support hierarchy irq and this patch
> >> was to add such support.
> >> 
> >> In addition, this patch only consider dts, and acpi hierarchy irq
> >> support will be added later as required.
> >> 
> >> Signed-off-by: Yinbo Zhu <zhuyinbo@...ngson.cn>
> >> ---
> >>   drivers/irqchip/irq-loongson-liointc.c | 31 ++++++++++++++++++++++++++
> >>   1 file changed, 31 insertions(+)
> >> 
> >> diff --git a/drivers/irqchip/irq-loongson-liointc.c b/drivers/irqchip/irq-loongson-liointc.c
> >> index 0da8716f8f24..58e43a2cd02e 100644
> >> --- a/drivers/irqchip/irq-loongson-liointc.c
> >> +++ b/drivers/irqchip/irq-loongson-liointc.c
> >> @@ -177,6 +177,32 @@ static const struct irq_domain_ops acpi_irq_gc_ops = {
> >>   	.xlate	= liointc_domain_xlate,
> >>   };
> >>   +#ifdef	CONFIG_IRQ_DOMAIN_HIERARCHY
> >> +static int liointc_domain_alloc(struct irq_domain *domain, unsigned int virq,
> >> +				unsigned int nr_irqs, void *arg)
> >> +{
> >> +	int i, ret;
> >> +	irq_hw_number_t hwirq;
> >> +	unsigned int type = IRQ_TYPE_NONE;
> >> +	struct irq_fwspec *fwspec = arg;
> >> +
> >> +	ret = irq_domain_translate_twocell(domain, fwspec, &hwirq, &type);
> >> +	if (ret)
> >> +		return ret;
> >> +
> >> +	for (i = 0; i < nr_irqs; i++)
> >> +		irq_map_generic_chip(domain, virq + i, hwirq + i);
> >> +
> >> +	return 0;
> >> +}
> >> +
> >> +static const struct irq_domain_ops of_irq_gc_ops = {
> >> +	.translate	= irq_domain_translate_twocell,
> >> +	.alloc		= liointc_domain_alloc,
> >> +	.free		= irq_domain_free_irqs_top,
> >> +};
> >> +#endif
> >> +
> >>   static int liointc_init(phys_addr_t addr, unsigned long size, int revision,
> >>   		struct fwnode_handle *domain_handle, struct device_node *node)
> >>   {
> >> @@ -218,8 +244,13 @@ static int liointc_init(phys_addr_t addr, unsigned long size, int revision,
> >>   		domain = irq_domain_create_linear(domain_handle, LIOINTC_CHIP_IRQ,
> >>   					&acpi_irq_gc_ops, priv);
> >>   	else
> >> +#ifdef	CONFIG_IRQ_DOMAIN_HIERARCHY
> >> +		domain = irq_domain_create_linear(domain_handle, LIOINTC_CHIP_IRQ,
> >> +					&of_irq_gc_ops, priv);
> >> +#else
> >>   		domain = irq_domain_create_linear(domain_handle, LIOINTC_CHIP_IRQ,
> >>   					&irq_generic_chip_ops, priv);
> >> +#endif
> > Two things:
> > 
> > - Why do we need three calls to create the same domains depending on
> >    what firmware is used and kernel configuration?
> yes, It depend on firmeware and kernel configuration.

Read again:

why do we need 3 different calls to irq_domain_create_linear when you
can *indirect* them with a pointer to the correct structure?

> > 
> > - who is going to decide whether to select the
> >    CONFIG_IRQ_DOMAIN_HIERARCHY option?
> The latest gpio driver will select  CONFIG_IRQ_DOMAIN_HIERARCHY

Then why do we need two different behaviours? The same kernel should
run everywhere.

> > 
> > I'd really like to see a statement from the Loongson folks about what
> > this whole DT stuff is all about. AFAICT, the core ACPICA stuff isn't
> > even fully merged (i.e. we still rely on arch-specific hacks).
> 
> The support of dts is mainly for Loongson embedded chips, such as
> LoongArch Loongson-2 series SoC.  and it use dts to descripte device
> and don't support acpi.

That doesn't answer my question. Please have a *consistent* approach
to your interrupt handling, and work with your ACPI colleagues.

	M.

-- 
Without deviation from the norm, progress is not possible.