On Thu, 13 Feb 2020 12:25:36 +0530 rohit maheshwari wrote: > On 13/02/20 9:39 AM, Jakub Kicinski wrote: > >> diff --git a/net/tls/tls_device.c b/net/tls/tls_device.c > >> index cd91ad812291..2898517298bf 100644 > >> --- a/net/tls/tls_device.c > >> +++ b/net/tls/tls_device.c > >> @@ -602,7 +602,8 @@ struct tls_record_info *tls_get_record(struct > >> tls_offload_context_tx *context, */ > >> info = > >> list_first_entry_or_null(&context->records_list, struct > >> tls_record_info, list); > >> - if (!info) > >> + /* return NULL if seq number even before the 1st > >> entry. */ > >> + if (!info || before(seq, info->end_seq - > >> info->len)) > > Is it not more appropriate to use between() in the actual comparison > > below? I feel like with this patch we can get false negatives. > > If we use between(), though record doesn't exist, we still go and > compare each record, > > which I think, should actually be avoided. You can between() first and last element on the list at the very start of the search. > >> return NULL; > >> record_sn = context->unacked_record_sn; > >> } > > If you post a v2 please add a Fixes tag and CC maintainers of this > > code.
