On 2/21/2023 4:07 PM, Greg KH wrote:
On Tue, Feb 21, 2023 at 02:57:15PM +0530, Srinivasarao Pathipati wrote:
On 2/21/2023 12:24 AM, Greg KH wrote:
On Mon, Feb 20, 2023 at 11:43:59PM +0530, Srinivasarao Pathipati wrote:
Making show_rcu_tasks_rude_gp_kthread() function as 'inline' to
fix below compilation error.
This is applicable to only 5.10 kernels as code got modified
in latest kernels.
In file included from kernel/rcu/update.c:579:0:
kernel/rcu/tasks.h:710:13: error: ‘show_rcu_tasks_rude_gp_kthread’ defined but not used [-Werror=unused-function]
static void show_rcu_tasks_rude_gp_kthread(void) {}
Signed-off-by: Srinivasarao Pathipati <quic_c_spathi@xxxxxxxxxxx>
---
kernel/rcu/tasks.h | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/kernel/rcu/tasks.h b/kernel/rcu/tasks.h
index 8b51e6a..53ddb4e 100644
--- a/kernel/rcu/tasks.h
+++ b/kernel/rcu/tasks.h
@@ -707,7 +707,7 @@ static void show_rcu_tasks_rude_gp_kthread(void)
#endif /* #ifndef CONFIG_TINY_RCU */
#else /* #ifdef CONFIG_TASKS_RUDE_RCU */
-static void show_rcu_tasks_rude_gp_kthread(void) {}
+static inline void show_rcu_tasks_rude_gp_kthread(void) {}
#endif /* #else #ifdef CONFIG_TASKS_RUDE_RCU */
////////////////////////////////////////////////////////////////////////
--
2.7.4
What commit id caused this problem?
commit (8344496e8b49c4122c1808d6cd3f8dc71bccb595 rcu-tasks: Conditionally
compile show_rcu_tasks_gp_kthreads()) introduced this issue
So this has been around for a very very long time.
Why is this suddenly an issue now, almost 3 years later? What changed
to cause this to become an issue?
And please put this information in the changelog text.
We observed this issue on Android kernels with arch=UM after picking
this patch (3fe617ccafd6f5bb33c2391d6f4eeb41c1fd0151 Enable '-Werror' by
default for all kernel builds ) recently into Android kernels.
Earlier it was just warning and this issue depends on specific
configuration, so it might got missed.
This patch added conditional macros for definition of
show_rcu_tasks_rude_gp_kthread() but not for dummy definition.
And why isn't it an issue in newer kernels, what commit id fixed it and
why can't we just take that instead?
Later code got modified with patch (27c0f1448389 rcutorture: Make
grace-period kthread report match RCU flavor being tested) , with this
there won't be compilation issue.
This patch is part of below series, Not sure all these can be picked to this
5.10 stable branch so fixed issue by simply making function inline.
if you think it is better to pick this series, please merge to 5.10 branch.
https://lore.kernel.org/lkml/20201105233900.GA20676@paulmck-ThinkPad-P72/
[1/4] e1eb075ccf37 rcutorture: Make preemptible TRACE02 enable lockdep
[2/4] 77dc174103fd rcu-tasks: Convert rcu_tasks_wait_gp() for-loop to
while-loop
[3/4] 27c0f1448389 rcutorture: Make grace-period kthread report match RCU
flavor being tested
[4/4] 75dc2da5ecd6 rcu-tasks: Make the units of ->init_fract be jiffies
I would need you to test and verify that these actually work properly
and do not cause any regressions and sign off on them before I could
take them.
probably it would be better to pick this no risk patch instead of
picking above series as this is stable branch.
thanks,
greg k-h