mount fail with option fault_injection < 20000 && >0

mount fail with option fault_injection < 20000 && >0

[cgxu519]

HI guys,

I found mount with option fault_injection < 20000 && >0 will fail with 
error -ENOMEM.
I understand fault_injection function is explicitly mocking some errors, 
but isn't it better
mocking errors after successful mount? Is it really allocating memory so 
many times
during mount process?

Thanks,
Chengguang

Re: mount fail with option fault_injection < 20000 && >0

[Chao Yu]

Hi Chengguang,

On 2018/9/18 19:02, cgxu519 wrote:
> HI guys,
> 
> I found mount with option fault_injection < 20000 && >0 will fail with error
> -ENOMEM.
> I understand fault_injection function is explicitly mocking some errors, but

If we skip injecting error into memory allocation functions during mount, we may
miss checking most error paths in fill_super(). So I think we really need this
during mount.

But if you want to inject error after mount, you can reenable error injection by
configuring fault_injection/fault_rate sysfs entry.

Thanks,

> isn't it better
> mocking errors after successful mount? Is it really allocating memory so many times
> during mount process?
> 
> Thanks,
> Chengguang
> 
> 
> 
> 
> 
> 
> 
> _______________________________________________
> Linux-f2fs-devel mailing list
> Linux-f2fs-devel@lists.sourceforge.net
> https://lists.sourceforge.net/lists/listinfo/linux-f2fs-devel