Re: [PATCH v3 2/2] mm: mmap: add trace point of vm_unmapped_area

[Vlastimil Babka]

On 3/29/20 6:14 PM, Matthew Wilcox wrote:
> On Fri, Mar 20, 2020 at 02:58:23PM +0900, Jaewon Kim wrote:
>> +	TP_printk("addr=%lx err=%ld total_vm=0x%lx flags=0x%lx len=0x%lx lo=0x%lx hi=0x%lx mask=0x%lx ofs=0x%lx\n",
>> +		IS_ERR_VALUE(__entry->addr) ? 0 : __entry->addr,
>> +		IS_ERR_VALUE(__entry->addr) ? __entry->addr : 0,
> 
> I didn't see the IS_ERR_VALUE problem that Vlastimil mentioned get resolved?

Steven is fixing it in trace-cmd:
https://lore.kernel.org/r/20200324200956.821799393@xxxxxxxxxxx

> I might suggest ...
> 
> +++ b/include/linux/err.h
> @@ -19,7 +19,8 @@
>  
>  #ifndef __ASSEMBLY__
>  
> -#define IS_ERR_VALUE(x) unlikely((unsigned long)(void *)(x) >= (unsigned long)-MAX_ERRNO)
> +#define __IS_ERR_VALUE(x) ((unsigned long)(void *)(x) >= (unsigned long)-MAX_ERRNO)
> +#define IS_ERR_VALUE(x) unlikely(__IS_ERR_VALUE(x))

So this shouldn't be needed, as we are adding a new tracepoint, not "breaking"
an existing one?

>  static inline void * __must_check ERR_PTR(long error)
>  {
> 
> and then you can use __IS_ERR_VALUE() which removes the unlikely() problem.
>