Re: [PATCH v5] erofs: use Z_EROFS_LCLUSTER_TYPE_MAX to simplify switches

[Gao Xiang <hsiangkao@linux.alibaba.com>]

Hi Chao,

On 2025/3/16 10:36, Chao Yu wrote:
> On 2025/2/10 11:29, Hongzhen Luo wrote:
>> There's no need to enumerate each type.  No logic changes.
>>
>> Signed-off-by: Hongzhen Luo <hongzhen@linux.alibaba.com>
> 
> Looks good to me, feel free to add:
> 
> Reviewed-by: Chao Yu <chao@kernel.org>
> 
> And one minor comment below.
> 
>> diff --git a/fs/erofs/zmap.c b/fs/erofs/zmap.c
>> index 689437e99a5a..d278ebd60281 100644
>> --- a/fs/erofs/zmap.c
>> +++ b/fs/erofs/zmap.c
>> @@ -265,26 +265,22 @@ static int z_erofs_extent_lookback(struct z_erofs_maprecorder *m,
>>           if (err)
>>               return err;
>> -        switch (m->type) {
>> -        case Z_EROFS_LCLUSTER_TYPE_NONHEAD:
>> +        if (m->type >= Z_EROFS_LCLUSTER_TYPE_MAX) {
>> +            erofs_err(sb, "unknown type %u @ lcn %lu of nid %llu",
>> +                  m->type, lcn, vi->nid);
>> +            DBG_BUGON(1);
>> +            return -EOPNOTSUPP;
>  > +        } else if (m->type == Z_EROFS_LCLUSTER_TYPE_NONHEAD) {>               lookback_distance = m->delta[0];
>>               if (!lookback_distance)
>> -                goto err_bogus;
>> +                break;
>>               continue;
>> -        case Z_EROFS_LCLUSTER_TYPE_PLAIN:
>> -        case Z_EROFS_LCLUSTER_TYPE_HEAD1:
>> -        case Z_EROFS_LCLUSTER_TYPE_HEAD2:
>> +        } else {
>>               m->headtype = m->type;
>>               m->map->m_la = (lcn << lclusterbits) | m->clusterofs;
>>               return 0;
>> -        default:
>> -            erofs_err(sb, "unknown type %u @ lcn %lu of nid %llu",
>> -                  m->type, lcn, vi->nid);
>> -            DBG_BUGON(1);
>> -            return -EOPNOTSUPP;
> 
> Should return EFSCORRUPTED here? is m->type >= Z_EROFS_LCLUSTER_TYPE_MAX a possible
> case?

It's impossible by the latest on-disk definition, see:
#define Z_EROFS_LI_LCLUSTER_TYPE_MASK	(Z_EROFS_LCLUSTER_TYPE_MAX - 1)

Previously, it was useful before Z_EROFS_LCLUSTER_TYPE_HEAD2 was
introduced, but the `default:` case is already deadcode now.

> 
> Btw, we'd better to do sanity check for m->type in z_erofs_load_full_lcluster(),
> then we can treat m->type as reliable variable later.
> 
>      advise = le16_to_cpu(di->di_advise);
>      m->type = advise & Z_EROFS_LI_LCLUSTER_TYPE_MASK;
>      if (m->type >= Z_EROFS_LCLUSTER_TYPE_MAX) {

It's always false here.

Thanks,
Gao Xiang