Re: fs/zonefs/./trace.h:22:1: sparse: sparse: cast to restricted blk_opf_t

From: Damien Le Moal
Date: Wed Aug 10 2022 - 15:14:26 EST


On 2022/08/10 11:08, Bart Van Assche wrote:
> On 8/10/22 07:36, Damien Le Moal wrote:
>> On 2022/08/08 8:37, Bart Van Assche wrote:
>>> Thanks for having taken a look. Please help with verifying whether the
>>> following patch is sufficient to fix the reported warning: "[PATCH]
>>> tracing: Suppress sparse warnings triggered by is_signed_type()"
>>> (https://lore.kernel.org/all/20220717151047.19220-1-bvanassche@xxxxxxx/).
>>
>> With the current Linus tree, I do not see this sparse warning. However, applying
>> the above patch, "make M=fs/zonefs C=1" generates a lot of warnings:
>
> That doesn't make sense to me. My patch reduces the number of sparse
> warnings that are reported.
>
>> make -j64 M=fs/zonefs C=1
>> CC [M] fs/zonefs/super.o
>> CC [M] fs/zonefs/sysfs.o
>> CHECK fs/zonefs/sysfs.c
>> CHECK fs/zonefs/super.c
>> fs/zonefs/sysfs.c: note: in included file (through include/linux/bitops.h,
>> include/linux/kernel.h, arch/x86/include/asm/percpu.h,
>> arch/x86/include/asm/preempt.h, include/linux/preempt.h,
>> include/linux/spinlock.h, ...):
>> ./arch/x86/include/asm/bitops.h:66:1: warning: unreplaced symbol 'return'
>
> I think that you are hitting a bug in sparse. See also
> https://lore.kernel.org/all/e91d351c-3c16-e48d-7e9d-9f096c4acbc9@xxxxxxxxxx/T/.
> I also see the above warnings if I use the sparse binary from Debian
> testing. I do not see these sparse warnings if I download the sparse
> source code and compile that source code myself.

Good point. I was using Fedora 36 sparse package. Using sparse compiled from
source, I now see again the warnings without the patch and no warnings with the
patch applied. So the patch looks good. Are you going to send it as a fix for
6.0-rc1 ?

Cheers.


--
Damien Le Moal
Western Digital Research