Willem de Bruijn <willemdebruijn.kernel@xxxxxxxxx> wrote: > Is the MSG_CONFIRM needed to trigger this? It's not actually necessary. The syz test was doing it. > Is this indeed trivially sidestepped if downgrading from splicing to > regular copying with fragmentation? That works. The logging looks like: ==>splice_to_socket() 5535 udp_sendmsg(8,8) __ip_append_data(copy=-1,len=8, mtu=8192 skblen=8189 maxfl=8188) copy 8 = 9 - 0 - 1 - 0 length 8 -= 8 + 0 <==splice_to_socket() = 8 It looks like pagedlen being non-zero might be the issue. David
