Re: [PATCH] arm64/cache: fix -Woverride-init compiler warnings

[Will Deacon]

On Mon, Aug 05, 2019 at 07:47:37AM -0400, Qian Cai wrote:
> 
> 
> > On Aug 5, 2019, at 5:52 AM, Will Deacon <will@xxxxxxxxxx> wrote:
> > 
> > On Fri, Aug 02, 2019 at 11:32:24AM -0400, Qian Cai wrote:
> >> The commit 155433cb365e ("arm64: cache: Remove support for ASID-tagged
> >> VIVT I-caches") introduced some compiation warnings from GCC,
> >> 
> >> arch/arm64/kernel/cpuinfo.c:38:26: warning: initialized field
> >> overwritten [-Woverride-init]
> >>  [ICACHE_POLICY_VIPT]  = "VIPT",
> >>                          ^~~~~~
> >> arch/arm64/kernel/cpuinfo.c:38:26: note: (near initialization for
> >> 'icache_policy_str[2]')
> >> arch/arm64/kernel/cpuinfo.c:39:26: warning: initialized field
> >> overwritten [-Woverride-init]
> >>  [ICACHE_POLICY_PIPT]  = "PIPT",
> >>                          ^~~~~~
> >> arch/arm64/kernel/cpuinfo.c:39:26: note: (near initialization for
> >> 'icache_policy_str[3]')
> >> arch/arm64/kernel/cpuinfo.c:40:27: warning: initialized field
> >> overwritten [-Woverride-init]
> >>  [ICACHE_POLICY_VPIPT]  = "VPIPT",
> >>                           ^~~~~~~
> >> arch/arm64/kernel/cpuinfo.c:40:27: note: (near initialization for
> >> 'icache_policy_str[0]')
> >> 
> >> because it initializes icache_policy_str[0 ... 3] twice.
> >> 
> >> Fixes: 155433cb365e ("arm64: cache: Remove support for ASID-tagged VIVT I-caches")
> >> Signed-off-by: Qian Cai <cai@xxxxxx>
> >> ---
> >> arch/arm64/kernel/cpuinfo.c | 4 ++--
> >> 1 file changed, 2 insertions(+), 2 deletions(-)
> >> 
> >> diff --git a/arch/arm64/kernel/cpuinfo.c b/arch/arm64/kernel/cpuinfo.c
> >> index 876055e37352..193b38da8d96 100644
> >> --- a/arch/arm64/kernel/cpuinfo.c
> >> +++ b/arch/arm64/kernel/cpuinfo.c
> >> @@ -34,10 +34,10 @@
> >> static struct cpuinfo_arm64 boot_cpu_data;
> >> 
> >> static char *icache_policy_str[] = {
> >> -	[0 ... ICACHE_POLICY_PIPT]	= "RESERVED/UNKNOWN",
> >> +	[ICACHE_POLICY_VPIPT]		= "VPIPT",
> >> +	[ICACHE_POLICY_VPIPT + 1]	= "RESERVED/UNKNOWN",
> >> 	[ICACHE_POLICY_VIPT]		= "VIPT",
> >> 	[ICACHE_POLICY_PIPT]		= "PIPT",
> >> -	[ICACHE_POLICY_VPIPT]		= "VPIPT",
> > 
> > I really don't like this patch. Using "[0 ... MAXIDX] = <default>" is a
> > useful idiom and I think the code is more error-prone the way you have
> > restructured it.
> > 
> > Why are you passing -Woverride-init to the compiler anyway? There's only
> > one Makefile that references that option, and it's specific to a pinctrl
> > driver.
> 
> Those extra warnings can be enabled by âmake W=1â. â-Woverride-init â seems to be useful
> to catch potential developer mistakes with unintented double-initializations. It is normal to
> start to fix the most of false-positives first before globally enabling the flag by default just like
> â-Wimplicit-fallthroughâ mentioned in,
> 
> https://lwn.net/Articles/794944/

I think this case is completely different to the implicit fallthrough stuff.
The solution there was simply to add a comment without restructuring the
surrounding code. What your patch does here is actively make the code harder
to understand.

Initialising a static array with a non-zero pattern is a useful idiom and I
don't think we should throw that away just to appease a silly compiler
warning that appears only with non-default build options. Have a look at
the way we use PERF_MAP_ALL_UNSUPPORTED in the Arm PMU code, for example.

Will