Re: [PATCH] lib/test_vmalloc.c: demote vmalloc_test_init to late_initcall

[Harry Yoo <harry.yoo@oracle.com>]

On Fri, Jun 20, 2025 at 10:24:48PM +0800, David Wang wrote:
> Commit 2d76e79315e4 ("lib/test_vmalloc.c: allow built-in execution")
> enable test_vmalloc module to be built into kernel directly, but
> vmalloc_test_init depends on alloc_tag module via alloc_tag_top_users().
>
> When a kernel build with following config:
> 
> CONFIG_TEST_VMALLOC=y
> CONFIG_MEM_ALLOC_PROFILING=y
> CONFIG_MEM_ALLOC_PROFILING_ENABLED_BY_DEFAULT=y
> CONFIG_MEM_ALLOC_PROFILING_DEBUG=y
> 
> If vmalloc_test_init() run before alloc_tag_init(), memory
> failure tests would invoke alloc_tag_top_users() which is not
> ready to use and cause kernel BUG:
> 
>  [  135.116045] BUG: kernel NULL pointer dereference, address: 0000000000000030
>  [  135.116063] #PF: supervisor read access in kernel mode
>  [  135.116074] #PF: error_code(0x0000) - not-present page
>  [  135.116085] PGD 0 P4D 0
>  [  135.116094] Oops: Oops: 0000 [#1] SMP NOPTI
>  [  135.116123] Tainted: [E]=UNSIGNED_MODULE
>  [  135.116132] Hardware name: QEMU Standard PC (Q35 + ICH9, 2009), BIOS 1.16.2-debian-1.16.2-1 04/01/2014
>  [  135.116148] RIP: 0010:down_read_trylock+0x1d/0x80
>  [  135.116188] RSP: 0000:ffffb5e481a9b8f8 EFLAGS: 00010246
>  [  135.116200] RAX: ffff93dc8a5ac700 RBX: 0000000000000030 RCX: 8000000000000007
>  [  135.116214] RDX: 0000000000000001 RSI: 000000000000000a RDI: ffffffff93d2e733
>  [  135.116228] RBP: ffffb5e481a9b9a0 R08: 0000000000000000 R09: 0000000000000003
>  [  135.116241] R10: ffffb5e481a9b860 R11: ffffffff94ec6328 R12: ffffb5e481a9b9b0
>  [  135.116255] R13: 0000000000000003 R14: 0000000000000001 R15: ffffffff94e0c580
>  [  135.116271] FS:  00007fd41947e540(0000) GS:ffff93dd6654a000(0000) knlGS:0000000000000000
>  [  135.116286] CS:  0010 DS: 0000 ES: 0000 CR0: 0000000080050033
>  [  135.116298] CR2: 0000000000000030 CR3: 00000001099f8000 CR4: 0000000000350ef0
>  [  135.116314] Call Trace:
>  [  135.116321]  <TASK>
>  [  135.116328]  codetag_trylock_module_list+0x9/0x20
>  [  135.116342]  alloc_tag_top_users+0x153/0x1b0
>  [  135.116354]  ? srso_return_thunk+0x5/0x5f
>  [  135.116365]  ? _printk+0x57/0x80
>  [  135.116378]  __show_mem+0xeb/0x210
>  [  135.116394]  ? dump_header+0x2ce/0x3e0
>  [  135.116405]  dump_header+0x2ce/0x3e0
> 
> Demote vmalloc_test_init to late_initcall can make sure alloc_tag
> module got initialized before test_vmalloc module.

I'm not sure this is the right place to fix it.

The bug can be triggered by any early memory allocation failure,
before alloc_tag_init() is called (yeah, that's not that likely).

There is nothing specific to vmalloc that triggers the bug.

-- 
Cheers,
Harry / Hyeonggon

> Link: https://urldefense.com/v3/__https://lore.kernel.org/lkml/20250620100258.595495-1-00107082@163.com/__;!!ACWV5N9M2RV99hQ!NXhzLP0lE5O2YKK9PfCt3LDTk4qWGsy1ebNXBQETNNJrL2JS3R01iunwBVXbDA4_kKjrbyQWfzNa7iN5RQ$ 
> Reported-by: kernel test robot <oliver.sang@intel.com>
> Closes: https://urldefense.com/v3/__https://lore.kernel.org/oe-lkp/202506181351.bba867dd-lkp@intel.com__;!!ACWV5N9M2RV99hQ!NXhzLP0lE5O2YKK9PfCt3LDTk4qWGsy1ebNXBQETNNJrL2JS3R01iunwBVXbDA4_kKjrbyQWfzP95R5wqA$ 
> Fixes: 2d76e79315e4 ("lib/test_vmalloc.c: allow built-in execution")
> Signed-off-by: David Wang <00107082@163.com>
> ---
>  lib/test_vmalloc.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/lib/test_vmalloc.c b/lib/test_vmalloc.c
> index 1b0b59549aaf..5af009df56ad 100644
> --- a/lib/test_vmalloc.c
> +++ b/lib/test_vmalloc.c
> @@ -598,7 +598,7 @@ static int __init vmalloc_test_init(void)
>  	return IS_BUILTIN(CONFIG_TEST_VMALLOC) ? 0:-EAGAIN;
>  }
>  
> -module_init(vmalloc_test_init)
> +late_initcall(vmalloc_test_init)
>  
>  MODULE_LICENSE("GPL");
>  MODULE_AUTHOR("Uladzislau Rezki");
> -- 
> 2.39.2
> 
>