Re: [RFC PATCH v0.2] PCI: Add support for tango PCIe host bridge

From: Mason <slash.tmp@free.fr>
To: Thomas Gleixner <tglx@linutronix.de>
Cc: Marc Zyngier <marc.zyngier@arm.com>,
	Bjorn Helgaas <helgaas@kernel.org>,
	Robin Murphy <robin.murphy@arm.com>,
	Lorenzo Pieralisi <lorenzo.pieralisi@arm.com>,
	Liviu Dudau <liviu.dudau@arm.com>,
	David Laight <david.laight@aculab.com>,
	linux-pci <linux-pci@vger.kernel.org>,
	Linux ARM <linux-arm-kernel@lists.infradead.org>,
	Thibaud Cornic <thibaud_cornic@sigmadesigns.com>,
	Phuong Nguyen <phuong_nguyen@sigmadesigns.com>,
	LKML <linux-kernel@vger.kernel.org>
Subject: Re: [RFC PATCH v0.2] PCI: Add support for tango PCIe host bridge
Date: Mon, 27 Mar 2017 17:18:23 +0200	[thread overview]
Message-ID: <4f426de0-b24a-c8dc-76a0-d4ec607a86be@free.fr> (raw)
In-Reply-To: <alpine.DEB.2.20.1703271644210.3616@nanos>

On 27/03/2017 16:46, Thomas Gleixner wrote:

> On Mon, 27 Mar 2017, Mason wrote:
>
>> On 24/03/2017 19:22, Marc Zyngier wrote:
>>
>>> You cannot directly use a pointer to a u32 in any of the bitmap
>>> operations. You need to copy the value to an unsigned long, and
>>> apply the bitmap op on that.
>>
>> On my platform, find_first_zero_bit() resolves to
>>
>>   int _find_first_zero_bit_le(const void * p, unsigned size);
>>
>> If the underlying implementation actually expects an unsigned long
>> pointer, should the function prototype be changed?
> 
> Errm? Why are you worrying about the underlying implementations?
> 
> find_first_zero_bit() is what you are supposed to use in your code. And
> that explicitely takes a unsigned long pointer.

I don't think so.

If the prototype for find_first_zero_bit() specified the first
argument as an unsigned long pointer, then the compiler would
have rejected my code like this:

  CC      drivers/pci/host/pcie-tango.o
In file included from ./include/linux/bitops.h:36:0,
                 from ./include/linux/kernel.h:10,
                 from ./include/linux/list.h:8,
                 from ./include/linux/smp.h:11,
                 from ./include/linux/irq.h:12,
                 from ./include/linux/irqchip/chained_irq.h:21,
                 from drivers/pci/host/pcie-tango.c:1:
drivers/pci/host/pcie-tango.c: In function 'tango_irq_domain_alloc':
drivers/pci/host/pcie-tango.c:122:28: error: passing argument 1 of '_find_first_zero_bit_le' from incompatible pointer type [-Werror=incompatible-pointer-types]
  pos = find_first_zero_bit(&mask, 32);
                            ^
./arch/arm/include/asm/bitops.h:199:59: note: in definition of macro 'find_first_zero_bit'
 #define find_first_zero_bit(p,sz) _find_first_zero_bit_le(p,sz)
                                                           ^
./arch/arm/include/asm/bitops.h:162:12: note: expected 'const long unsigned int *' but argument is of type 'u32 * {aka unsigned int *}'
 extern int _find_first_zero_bit_le(const unsigned long * p, unsigned size);
            ^
cc1: some warnings being treated as errors
make[1]: *** [drivers/pci/host/pcie-tango.o] Error 1
make: *** [drivers/pci/host/pcie-tango.o] Error 2

But, in fact, the compiler remained silent, specifically because
the situation on my platform is:

  #define find_first_zero_bit(p,sz)	_find_first_zero_bit_le(p,sz)
  int _find_first_zero_bit_le(const void * p, unsigned size);

So I asked if the prototype could/should be changed, to have the
compiler catch the error as early as possible.

Regards.