RE: Why is (2 < 2) true? Is it a gcc bug?

From: Dorau, Lukasz
Date: Fri Jan 17 2014 - 12:06:27 EST


On Friday, January 17, 2014 5:40 PM Sebastian Riemer <sebastian.riemer@xxxxxxxxxxxxxxxx> wrote:
> On 17.01.2014 14:55, Dorau, Lukasz wrote:
> >
> > Some additional information:
> >
> > The loop 'for' in macro ' for_each_isci_host ' defined as
> (drivers/scsi/isci/host.h:313):
> >
> > #define for_each_isci_host(id, ihost, pdev) \
> > for (id = 0, ihost = to_pci_info(pdev)->hosts[id]; \
> > id < ARRAY_SIZE(to_pci_info(pdev)->hosts) && ihost; \
> > ihost = to_pci_info(pdev)->hosts[++id])
> >
> > should be executed only for i = 0 and 1, because:
> > ARRAY_SIZE(to_pci_info(pdev)->hosts) = SCI_MAX_CONTROLLERS = 2
> >
> > but it was executed also for i=2 regardless the above loop's end condition.
>
> to_pci_info() can return NULL in dev_get_drvdata(). The use of
> ARRAY_SIZE() is inappropriate.
>
> #define ARRAY_SIZE(arr) (sizeof(arr) / sizeof((arr)[0]) +
> __must_be_array(arr))
>
> #define __must_be_array(a) BUILD_BUG_ON_ZERO(__same_type((a), &(a)[0]))
>
> #define BUILD_BUG_ON_ZERO(e) (sizeof(struct { int:-!!(e); }))
>
> I would say that this was supposed to trigger a build error but it
> didn't and added 1 to the loop end condition.
>
> Can you test putting SCI_MAX_CONTROLLERS to the loop end condition, please?
>

Replacing 'ARRAY_SIZE(to_pci_info(pdev)->hosts)' with 'SCI_MAX_CONTROLLERS' in the definition of the ' for_each_isci_host ' macro does not help. I have checked it.
The following patch helps:
http://marc.info/?l=linux-scsi&m=138987823011697&w=2

Lukasz


--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at http://vger.kernel.org/majordomo-info.html
Please read the FAQ at http://www.tux.org/lkml/