diff mbox series

clocksource/arm_arch_timer: Fix bogus -Wsometimes-uninitialized warning

Message ID 20211107090144.3172241-1-oupton@google.com (mailing list archive)
State New, archived
Headers show
Series clocksource/arm_arch_timer: Fix bogus -Wsometimes-uninitialized warning | expand

Commit Message

Oliver Upton Nov. 7, 2021, 9:01 a.m. UTC
Since commit 4775bc63f880 ("clocksource/arm_arch_timer: Add build-time
guards for unhandled register accesses"), clang builds emit the
following warning:

>> drivers/clocksource/arm_arch_timer.c:156:3: warning: variable 'val' is used uninitialized whenever switch default is taken [-Wsometimes-uninitialized]
                   default:
                   ^~~~~~~
   drivers/clocksource/arm_arch_timer.c:163:9: note: uninitialized use occurs here
           return val;
                  ^~~

which is of course meaningless, as we break the build if the default
case is ever taken in the switch statement. Clang does static analysis
before deciding if the branch is ever taken, leading to the warning.

Fix the bogus warning by initializing val on the default branch.

Reported-by: kernel test robot <lkp@intel.com>
Fixes: 4775bc63f880 ("clocksource/arm_arch_timer: Add build-time guards for unhandled register accesses")
Signed-off-by: Oliver Upton <oupton@google.com>
---
Heh, I had caught this earlier but didn't hit send before starting the
weekend. Saw the bot email, so sending out.

 drivers/clocksource/arm_arch_timer.c | 2 ++
 1 file changed, 2 insertions(+)

Comments

Oliver Upton Nov. 7, 2021, 9:11 a.m. UTC | #1
On Sun, Nov 7, 2021 at 1:02 AM Oliver Upton <oupton@google.com> wrote:
>
> Since commit 4775bc63f880 ("clocksource/arm_arch_timer: Add build-time
> guards for unhandled register accesses"), clang builds emit the
> following warning:
>
> >> drivers/clocksource/arm_arch_timer.c:156:3: warning: variable 'val' is used uninitialized whenever switch default is taken [-Wsometimes-uninitialized]
>                    default:
>                    ^~~~~~~
>    drivers/clocksource/arm_arch_timer.c:163:9: note: uninitialized use occurs here
>            return val;
>                   ^~~
>
> which is of course meaningless, as we break the build if the default
> case is ever taken in the switch statement. Clang does static analysis
> before deciding if the branch is ever taken, leading to the warning.
>
> Fix the bogus warning by initializing val on the default branch.
>
> Reported-by: kernel test robot <lkp@intel.com>
> Fixes: 4775bc63f880 ("clocksource/arm_arch_timer: Add build-time guards for unhandled register accesses")
> Signed-off-by: Oliver Upton <oupton@google.com>

My inbox searching was poor, there is another thread on this issue
where Miguel proposed tweaking the assertion such that clang no longer
emits the warning. No longer having issues with the patch applied [1]

[1] http://lore.kernel.org/r/20211014132331.GA4811@kernel.org/

--
Oliver
Miguel Ojeda Nov. 8, 2021, 11:10 a.m. UTC | #2
On Sun, Nov 7, 2021 at 10:11 AM 'Oliver Upton' via Clang Built Linux
<clang-built-linux@googlegroups.com> wrote:
>
> My inbox searching was poor, there is another thread on this issue
> where Miguel proposed tweaking the assertion such that clang no longer
> emits the warning. No longer having issues with the patch applied [1]

Glad it helped! The fix is in mainline now.

Cheers,
Miguel
diff mbox series

Patch

diff --git a/drivers/clocksource/arm_arch_timer.c b/drivers/clocksource/arm_arch_timer.c
index 9a04eacc4412..8e2814fcea11 100644
--- a/drivers/clocksource/arm_arch_timer.c
+++ b/drivers/clocksource/arm_arch_timer.c
@@ -172,6 +172,7 @@  u32 arch_timer_reg_read(int access, enum arch_timer_reg reg,
 			val = readl_relaxed(timer->base + CNTP_CTL);
 			break;
 		default:
+			val = 0;
 			BUILD_BUG();
 		}
 	} else if (access == ARCH_TIMER_MEM_VIRT_ACCESS) {
@@ -181,6 +182,7 @@  u32 arch_timer_reg_read(int access, enum arch_timer_reg reg,
 			val = readl_relaxed(timer->base + CNTV_CTL);
 			break;
 		default:
+			val = 0;
 			BUILD_BUG();
 		}
 	} else {