| Message ID | 20180801214526.24599-1-chris.packham@alliedtelesis.co.nz (mailing list archive) |
|---|---|
| State | New, archived |
| Headers | show |
| Series | arm_pmu: fix compiler warning in arm_pmu_device_probe | expand |
Hi Chris, On 01/08/18 22:45, Chris Packham wrote: > GCC warns > > arm_pmu_platform.c:234:5: error: 'err' may be used uninitialized in this function [-Werror=maybe-uninitialized] > > This is because we rely on the for_each_cpu loop in armpmu_request_irqs > to initialise err. The warning is a little bogus because we know if > there were 0 CPUs this code would not be running. > > Initialise err to 0 to avoid the warning. Maybe initialising to something like -EINVAL would be more appropriate, just in case we did ever manage to get here with armpmu->supported_cpus unset? Robin. > Signed-off-by: Chris Packham <chris.packham@alliedtelesis.co.nz> > --- > This has been reported before in https://lkml.org/lkml/2018/3/5/508 I'm not > sure if it was dismmissed as "meh, gcc is wrong" or if it was just wainting for > someone with some round tuits. > > > drivers/perf/arm_pmu_platform.c | 2 +- > 1 file changed, 1 insertion(+), 1 deletion(-) > > diff --git a/drivers/perf/arm_pmu_platform.c b/drivers/perf/arm_pmu_platform.c > index 971ff336494a..96075cecb0ae 100644 > --- a/drivers/perf/arm_pmu_platform.c > +++ b/drivers/perf/arm_pmu_platform.c > @@ -160,7 +160,7 @@ static int pmu_parse_irqs(struct arm_pmu *pmu) > static int armpmu_request_irqs(struct arm_pmu *armpmu) > { > struct pmu_hw_events __percpu *hw_events = armpmu->hw_events; > - int cpu, err; > + int cpu, err = 0; > > for_each_cpu(cpu, &armpmu->supported_cpus) { > int irq = per_cpu(hw_events->irq, cpu); >
On Thu, Aug 02, 2018 at 09:45:26AM +1200, Chris Packham wrote: > GCC warns > > arm_pmu_platform.c:234:5: error: 'err' may be used uninitialized in this function [-Werror=maybe-uninitialized] > > This is because we rely on the for_each_cpu loop in armpmu_request_irqs > to initialise err. The warning is a little bogus because we know if > there were 0 CPUs this code would not be running. > > Initialise err to 0 to avoid the warning. > > Signed-off-by: Chris Packham <chris.packham@alliedtelesis.co.nz> > --- > This has been reported before in https://lkml.org/lkml/2018/3/5/508 I'm not > sure if it was dismmissed as "meh, gcc is wrong" or if it was just wainting for > someone with some round tuits. Which version of GCC are you using? I don't see this warning locally. Will
On 02/08/18 23:05, Will Deacon wrote: > On Thu, Aug 02, 2018 at 09:45:26AM +1200, Chris Packham wrote: >> GCC warns >> >> arm_pmu_platform.c:234:5: error: 'err' may be used uninitialized in this function [-Werror=maybe-uninitialized] >> >> This is because we rely on the for_each_cpu loop in armpmu_request_irqs >> to initialise err. The warning is a little bogus because we know if >> there were 0 CPUs this code would not be running. >> >> Initialise err to 0 to avoid the warning. >> >> Signed-off-by: Chris Packham <chris.packham@alliedtelesis.co.nz> >> --- >> This has been reported before in https://lkml.org/lkml/2018/3/5/508 I'm not >> sure if it was dismmissed as "meh, gcc is wrong" or if it was just wainting for >> someone with some round tuits. > > Which version of GCC are you using? I don't see this warning locally. > My local cross-compiler is $ arm-softfloat-linux-gnueabi-gcc --version arm-softfloat-linux-gnueabi-gcc (crosstool-NG crosstool-ng-1.22.0) 4.9.3 Old-ish but as far as I was aware still supported for building the kernel. I've just tried 7.3.0 and 8.1.0 neither of which complain. A quick look at the report linked to above seems they probably were using 4.6.3 and it still appears in current build reports http://kisskb.ellerman.id.au/kisskb/buildresult/13455291/
diff --git a/drivers/perf/arm_pmu_platform.c b/drivers/perf/arm_pmu_platform.c index 971ff336494a..96075cecb0ae 100644 --- a/drivers/perf/arm_pmu_platform.c +++ b/drivers/perf/arm_pmu_platform.c @@ -160,7 +160,7 @@ static int pmu_parse_irqs(struct arm_pmu *pmu) static int armpmu_request_irqs(struct arm_pmu *armpmu) { struct pmu_hw_events __percpu *hw_events = armpmu->hw_events; - int cpu, err; + int cpu, err = 0; for_each_cpu(cpu, &armpmu->supported_cpus) { int irq = per_cpu(hw_events->irq, cpu);
GCC warns arm_pmu_platform.c:234:5: error: 'err' may be used uninitialized in this function [-Werror=maybe-uninitialized] This is because we rely on the for_each_cpu loop in armpmu_request_irqs to initialise err. The warning is a little bogus because we know if there were 0 CPUs this code would not be running. Initialise err to 0 to avoid the warning. Signed-off-by: Chris Packham <chris.packham@alliedtelesis.co.nz> --- This has been reported before in https://lkml.org/lkml/2018/3/5/508 I'm not sure if it was dismmissed as "meh, gcc is wrong" or if it was just wainting for someone with some round tuits. drivers/perf/arm_pmu_platform.c | 2 +- 1 file changed, 1 insertion(+), 1 deletion(-)