Messages in this thread | | | Subject | Re: [PATCH 1/1] net: bcmgenet: fix incorrect return value checks | From | Florian Fainelli <> | Date | Mon, 24 Apr 2017 12:52:09 -0700 |
| |
On 04/24/2017 12:48 PM, David Miller wrote: > From: Pan Bian <bianpan201602@163.com> > Date: Sun, 23 Apr 2017 18:01:05 +0800 > >> From: Pan Bian <bianpan2016@163.com> >> >> Function platform_get_irq() will return a negative value on errors. >> However, in function bcmgenet_probe(), 0 is considered as a flag of >> error. This patch fixes the bug by checking whether the return value of >> platform_get_irq() is less than 0. >> >> Signed-off-by: Pan Bian <bianpan2016@163.com> > > I'm definitely not confident enough to apply this. > > On some platform zero IRQs are invalid. > > There are also lots of pieces of code that check the return value "> > 0" as success. >
I don't think we are fixing any real issue by applying this patch, but I will do a check on ARM, ARM64 and MIPS where this driver is used to see if it is even remotely possible to have a 0 IRQ. -- Florian
| |