On Sat, Jun 13, 2020 at 11:49:18PM +0800, Xiaolong Ye wrote:
> Set rte_errno as ENOMEM when allocation failure.
> 
> Fixes: 4958ca3a443a ("mbuf: support dynamic fields and flags")
> Cc: [email protected]
> 
> Signed-off-by: Xiaolong Ye <[email protected]>

Good catch, I was wrongly expecting that rte_malloc() functions sets
rte_errno on error.

I wonder if the malloc API shouldn't be modified, because I can see
several places where the same kind of fix would be needed: rte_efd,
rte_cuckoo_hash, rte_fbk_hash, rte_member, ...

Anatoly, what do you think?

Acked-by: Olivier Matz <[email protected]>

Reply via email to