On 2026-02-26 20:12, Nathan Chancellor wrote:

FWIW, the ClangBuiltLinux CI sees a boot failure with sparc64_defconfig,
which does not appear to be clang specific. I can reproduce it here
with:

I found the issue with the info provided by Heiko. Thanks for the
reproducer!

This series is not bisectable to see which specific patch causes this,
as I get

   In file included from mm/init-mm.c:2:
   include/linux/mm_types.h:1419:57: error: 
'PERCPU_COUNTER_TREE_ITEMS_STATIC_SIZE' undeclared here (not in a function)
    1419 |         [0 ... sizeof(cpumask_t) + MM_CID_STATIC_SIZE + 
PERCPU_COUNTER_TREE_ITEMS_STATIC_SIZE. - 1] = 0  \
         |                                                         
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
   mm/init-mm.c:50:27: note: in expansion of macro 
'MM_STRUCT_FLEXIBLE_ARRAY_INIT'
      50 |         .flexible_array = MM_STRUCT_FLEXIBLE_ARRAY_INIT,
         |                           ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~
   include/linux/mm_types.h:1419:10: error: array index in initializer not of 
integer type
    1419 |         [0 ... sizeof(cpumask_t) + MM_CID_STATIC_SIZE + 
PERCPU_COUNTER_TREE_ITEMS_STATIC_SIZE - 1] = 0  \
         |          ^
   mm/init-mm.c:50:27: note: in expansion of macro 
'MM_STRUCT_FLEXIBLE_ARRAY_INIT'
      50 |         .flexible_array = MM_STRUCT_FLEXIBLE_ARRAY_INIT,
         |                           ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~
   include/linux/mm_types.h:1419:10: note: (near initialization for 
'init_mm.flexible_array')
    1419 |         [0 ... sizeof(cpumask_t) + MM_CID_STATIC_SIZE + 
PERCPU_COUNTER_TREE_ITEMS_STATIC_SIZE - 1] = 0  \
         |          ^
   mm/init-mm.c:50:27: note: in expansion of macro 
'MM_STRUCT_FLEXIBLE_ARRAY_INIT'
      50 |         .flexible_array = MM_STRUCT_FLEXIBLE_ARRAY_INIT,
         |                           ^~~~~~~~~~~~~~~~~~~~~~~~~~~~~

prior to this change that removes PERCPU_COUNTER_TREE_ITEMS_STATIC_SIZE.

Good catch. I will fix this as well for v18.

Thanks!

Mathieu

--
Mathieu Desnoyers
EfficiOS Inc.
https://www.efficios.com

Reply via email to