linux-kernel - Re: [PATCH 2/9] ARM: PXA: Kill use of irq_create_strict

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]

Date:   Tue, 27 Apr 2021 05:56:21 -0700
From:   Guenter Roeck <linux@...ck-us.net>
To:     Marc Zyngier <maz@...nel.org>
Cc:     linux-arm-kernel@...ts.infradead.org, linux-kernel@...r.kernel.org,
        linux-mips@...r.kernel.org, linuxppc-dev@...ts.ozlabs.org,
        linux-sh@...r.kernel.org, Daniel Mack <daniel@...que.org>,
        Robert Jarzmik <robert.jarzmik@...e.fr>,
        Haojian Zhuang <haojian.zhuang@...il.com>,
        Yoshinori Sato <ysato@...rs.sourceforge.jp>,
        Rich Felker <dalias@...c.org>,
        Thomas Bogendoerfer <tsbogend@...ha.franken.de>,
        Michael Ellerman <mpe@...erman.id.au>,
        Thomas Gleixner <tglx@...utronix.de>
Subject: Re: [PATCH 2/9] ARM: PXA: Kill use of irq_create_strict_mappings()

On 4/27/21 1:30 AM, Marc Zyngier wrote:
> Hi Guenter,
> 
> Thanks for the heads up.
> 
> On Mon, 26 Apr 2021 23:39:42 +0100,
> Guenter Roeck <linux@...ck-us.net> wrote:
>>
>> On Tue, Apr 06, 2021 at 10:35:50AM +0100, Marc Zyngier wrote:
>>> irq_create_strict_mappings() is a poor way to allow the use of
>>> a linear IRQ domain as a legacy one. Let's be upfront about
>>> it and use a legacy domain when appropriate.
>>>
>>> Signed-off-by: Marc Zyngier <maz@...nel.org>
>>> ---
>>
>> When running the "mainstone" qemu emulation, this patch results
>> in many (32, actually) runtime warnings such as the following.
>>
>> [    0.528272] ------------[ cut here ]------------
>> [    0.528285] WARNING: CPU: 0 PID: 1 at kernel/irq/irqdomain.c:550 irq_domain_associate+0x194/0x1f0
>> [    0.528315] error: virq335 is not allocated
> 
> [...]
> 
> This looks like a case of CONFIG_SPARSE_IRQ, combined with a lack of
> brain engagement. I've come up with the following patch, which lets
> the kernel boot in QEMU without screaming (other than the lack of a
> rootfs...).
> 
> Please let me know if this helps.
> 

It does.

Tested-by: Guenter Roeck <linux@...ck-us.net>

Thanks,
Guenter

> Thanks,
> 
> 	M.
> 
> From 4d7f6ddbbfdff1c9f029bafca79020d3294dc32c Mon Sep 17 00:00:00 2001
> From: Marc Zyngier <maz@...nel.org>
> Date: Tue, 27 Apr 2021 09:00:28 +0100
> Subject: [PATCH] ARM: PXA: Fix cplds irqdesc allocation when using legacy mode
> 
> The Mainstone PXA platform uses CONFIG_SPARSE_IRQ, and thus we
> cannot rely on the irq descriptors to be readilly allocated
> before creating the irqdomain in legacy mode. The kernel then
> complains loudly about not being able to associate the interrupt
> in the domain -- can't blame it.
> 
> Fix it by allocating the irqdescs upfront in the legacy case.
> 
> Fixes: b68761da0111 ("ARM: PXA: Kill use of irq_create_strict_mappings()")
> Reported-by: Guenter Roeck <linux@...ck-us.net>
> Signed-off-by: Marc Zyngier <maz@...nel.org>
> ---
>  arch/arm/mach-pxa/pxa_cplds_irqs.c | 7 ++++++-
>  1 file changed, 6 insertions(+), 1 deletion(-)
> 
> diff --git a/arch/arm/mach-pxa/pxa_cplds_irqs.c b/arch/arm/mach-pxa/pxa_cplds_irqs.c
> index ec0d9b094744..bddfc7cd5d40 100644
> --- a/arch/arm/mach-pxa/pxa_cplds_irqs.c
> +++ b/arch/arm/mach-pxa/pxa_cplds_irqs.c
> @@ -121,8 +121,13 @@ static int cplds_probe(struct platform_device *pdev)
>  		return fpga->irq;
>  
>  	base_irq = platform_get_irq(pdev, 1);
> -	if (base_irq < 0)
> +	if (base_irq < 0) {
>  		base_irq = 0;
> +	} else {
> +		ret = devm_irq_alloc_descs(&pdev->dev, base_irq, base_irq, CPLDS_NB_IRQ, 0);
> +		if (ret < 0)
> +			return ret;
> +	}
>  
>  	res = platform_get_resource(pdev, IORESOURCE_MEM, 0);
>  	fpga->base = devm_ioremap_resource(&pdev->dev, res);
>